M IR P U B L IS H E R S M O S C O W

H. M SEJMEB

COnPOTVIBJIEHME MATEPMAJIOB

MSAAtUbCroO >HAyKA» MOCKBA

 N. M. BELYAEV

Strength of Materials

translated from the Russian

byN. R . Mehia

MIR PUBLISHERS MOSCOW

First published 1979
Revised from the 1976 Russian edition

Ha aHZAu&CKOM asuKe

© HaflareatcTBO cHaysa*, 1976

© English translation, Mir Publishers, 1979
Printed in the Union of Soviet Socialist Republics
 Nikolai Mikhailovich Belyaev
(1890—1944)

Nikolai Mikhailovich Belyaev occupied a leading position among

eminent Soviet scientists who worked on the technical application oT
theory of elasticity and strength of materials and structures.
After graduating from the St. Petersburg Institute of Railway Engi­
neering in 1916, Nikolai Mikhailovich Belyaev was invited to stay at
the Strength of Materials Department, where he worked under S. P. Ti­
moshenko.
Nikolai Mikhailovich Belyaev was associated with this institute
(now the Leningrad Institute of Railway Engineering) throughout
his life. At the institute he taught subjects like engineering structures,
bridges, theoretical mechanics, strength of materials and theory of
elasticity, and from 1924 to the end of his life was Head of the Strength
of Materials Department.
All his life Nikolai Mikhailovich Belyaev was a leading engineer
and research worker. He was the first to formulate and solve the prob­
lem of stability of prismatic bars under variable axial loading—a
problem interesting from the theoretical aspect and important from
the point of view of applications. Simultaneously, Nikolai Mikhailo­
vich Belyaev worked on the problem of local stresses in bodies in conlact
under compression. Here he considerably developed the works of Hertz.
The work first published by Nikolai Mikhailovich Belyaev in 1924 has
completely retained its value to this day.
In the Soviet Union Belyaev was one of the first to undertake the
study of the theory of plastic deformation, and he contributed a lot
towards the development of this field.
Nikolai Mikhailovich Belyaev spent the last years of his life in
fruitful research on problems of creep and relaxation of metals under
high temperatures.
Nikolai Mikhailovich Belyaev was a rare talent who successfully
combined theory with experimental research. In 1924 he took over as
Head of the mechanical engineering laboratory of the Leningrad In­
stitute of Railway Engineering, and in the course of 16 years of admi­
nistration changed the laboratory into a leading scientific research
centre.
New technical specifications ensuring long and reliable performance
of rails were compiled as a result of the research conducted at the
laboratory under the guidance and with direct participation of Niko­
lai Mikhailovich Belyaev. These specifications with minor additions
are in force to this day.
Research done by Nikolai Mikhailovich Belyaev in the field of
technology of concrete won wide acclaim all over the Soviet Union.
6 Nikolai Mikhailovich Belyaev

The pedagogical activity of Nikolai Mikhailovich Belyaev was not

restricted to the Leningrad Institute of Railway Engineering. He
worked at the Leningrad Technological Institute (1919-1926), Lening­
rad Institute of Civil Aviation (193M934), and from 1934 onwards
was Head of the Strength of Materials Department at the Leningrad
Polytechnical Institute—the biggest institute in the country.
In 1939 Nikolai Mikhailovich Belyaev was elected Corresponding
Member of the USSR Academy of Sciences, and from 1942 occupied
the post of Deputy Director of the Institute of Mechanics of the Aca­
demy of Sciences of the USSR.
His book Strength of Materials has won wide recognition in the
 Preface
to the Fifteenth Russian Edition

The new edition of Strength of Materials by N. M. Belyaev has

been published after II years. In 33 years that lapsed between the
publication by N. M. Belyaev of the first edition in 1932 and the last
fourteenth edition in 1965 a total of 675 000 copies of the book were
sold, testifying to its wide popularity. During this period the book was
periodically enlarged and revised by N. M. Belyaev and, after his
death in 1944, by a group of four of his co-workers. This group, which
prepared from the filth to the fourteenth editions for publication, did
not consider it proper to make substantia] changes in the original
work of N. M. Belyaev. Additions were done at one time or another
only when they became absolutely necessary due to changes in stan­
dards and technical specifications and in the light of recent research.
In the present edition, prepared by the same group, a number of
topics have been dropped either owing to their irrelevance to strength
of materials or because they are rarely taught in the main course. The
topics that have been dropped include Contact Stresses, Riveted Beams,
Reinforced Concrete Beams, Approximate Methods for Calculating
Deflection of Beams, Beams on Elastic Foundation, Design of Thin-
wailed Bars, all graphical methods, and a part of Complicated Prob­
lems of Stability Analysis, the other part of the last topic being pre-
sented in an abbreviated version. The reader may refer to the earlier
editions of this book or special monographs in case information is
required on these topics.
Considering the availability of a large number of problem books
(see, for instance, Problems on Strength of Materials edited by V. K. Ka-
churin) on the market, most of the examples have been dropped from
the present edition. Only examples that are essential for the explana­
tion of theoretical part have been retained.
For greater compactness the problem of design for safe loads has
now been included in Chapter 26 For the first time the chapter inclu­
des the principles of design for limiting states, which though beyond
the limits of the basic course of strength of materials are important
enough to require an exposition of the basic concepts even at this stage
of teaching.
The problems of strength, which in the previous editions occupied
two chapters, have been grouped into one. The part dealing with actual
stresses has been transferred to Chapter 2, where it has been presented
in a sufficiently detailed manner.
The tables containing data on materials have been dropped from the
appendices. A part of the data on materials has been transferred to
8 Preface to the Fifteenth Russian Edition

corresponding sections. The obsolete steel proliles grading has been

replaced by new ones.
As in the previous editions it was our endeavour to preserve Belyaev’s
style and method of presentation of material. Therefore the author's
text has in general been preserved. If Nikolai Mikhailovich Belyaev
were alive today he would possibly write many things in a different
way. However, since the book won wide popularity as written by
N. M. Belyaev, we tried to preserve the original text as far as possible.
The work involved In preparing the fifteenth edition for publica­
tion was distributed among the group as follows: Chapter 13, § 80
of Chapter 14, Chapters 15-19, 24-25—L. A. Belyavskii; Chapters 6,
8*12, 27-28-*-Ya. I. Kipnis; Chapters 1-5, 26 and appendices—N. Yu.
Kushelev; Chapter 7, § 79 of Chapter 14, Chapters 20-23, 29-32—
A- K. Sinitskii.
A . /(. Sinitskii
March 1976
 Contents

Nikolai Mikhailovich Belyaev 5

Preface to the Fifteenth Russian Edition 7

PART 1. Introduction. Tension and Compression

Chapter 1. Introduction 17
§ 1. The science of strength of materials 17
! 2. Classification of forces acting on elements of structures 18

3. Deformations and stresses 21

4. Scheme of a solution of the fundamental problem of strength of
materials 23
§ 5. Types of deformations 27
Chapter 2. Stress and Strain in Tension and Compression Within the Elastic Limit
Selection of Cross-sectional Area 27
§ 6. Determining the stresses in planes perpendicular io tile axis
of the bar 27
§7. Permissible stresses. Selecting the cross-sectional area 30
1 8. Deformations under tension and compression. Hooke's law 32
§9 . Lateral deformation coefficient. Poisson's ratio 36
Chapter 3. Experimental Study of Tension and Compression In Various Materials and
the Basis of Selecting the Permissible Stresses 40
§ 10. Tension test diagram. Mechanical properties of materials 40
1 11. Stress-strain diagram 47
§ 12. True stress-strain diagram 48
§ 13. Stress-strain diagram for ductile and brittle materials 62
§ 14. Rupture in compression of brittle and ductile materials. Compres­
sion test diagram 64
§ 16. Comparative study of the mechanical properties of ductile and
brittle materials 57
§ 16. Considerations in selection of safety factoi 59
§ 17. Permissible stresses under tension and compression for various
materials 64
PART II. Complicated Cases of Tension and
Compression
Chapter 4. Design of Statically Indeterminate Systems for Permissible Stresses 66
$ 18. Statically indeterminate systems 66
1 19. The effect of manufacturing inaccuracies on the forces acting in the
elements of statically indeterminate structures 73
§20. Tension and compression in bars made of heterogeneous mate­
rials 77
§21. Stresses due to temperature change 79
§22. Simultaneous account for various factors 82
§23. More complicated cases of statically indeterminate structures 85
10 Contents

Chapter 5. Account for Dead Weight In Tension and Compression* Design of Flexible
Strings 88
§ 24. Selecting the cross-sectional area with the account for the dead
weight (in tension and compression) 86
§25. Deformations due to dead weight 81
§26. Flexible cables 92

Chapter 6. Compound Stressed State. Stress and Strain 99

§ 27. Stresses along Inclined sections under axial tension or compression
(uniaxial stress) 99
§ 28. Concept of principal stresses. Types of stresses of materials 101
§ 29. Examples oil biaxial and triaxiaf stresses. Design of a cylindrical
reservoir 103
§30. Stresses In a biaxial stressed state 107
§31. Graphic determination of stresses (Mohr’s circle) 110
§32. Determination of the. principal stresses with the help of the stress
circle 114
33. Stresses in triaxia! stressed state 117
S34. Deformations In the compound stress 121

35. Potential energy of elastic deformation in compound stress 124

36. Pure shear. Stresses and strains. Hooke’s law. Potential
energy 127
Chapter 7. Strength of Materials in Compound Stress 132
§ 37. Resistance to failure. Rupture and shear 132
§38. Strength theories 136
§39. Theories of brittle failure (theories of rupture) 138
§40. Theories of ductile failure (theories of shear) 140
§41. Reduced stresses according to different strength theories 147
§42. Permissible stresses in pure shear 149

PART III. Shear and Torsion

Chapter 8. Practical Methods of Design on Shear 151
§43. Design of riveted and bolted joints 151
§44. Design of welded joints 158

Chapter 9. Torsion. Strength and Rigidity of Twisted Bars 164

45. Torque 164
! 46. Calculation of torques transmitted to the shaft 167

47. Determining stresses in a round shaft under torsion 168

48. Determination of polar moments of inertia and section moduli
of a shaft section 174
§49. Strength condition in torsion 176
§50. Deformations in torsion. Rigidity condition 176
§ 51. Stresses under torsion in a section inclined to the shaft axis 178
§52. Potential energy of torsion 180
§53. Stress and strain In dose-coiled helical springs 181
1 54. Torsion in rods of non-circular section 187
 Contents II
PART IV. Bending. Strength of Beams
Chapter 10. Internal Forces In Bending. Shearing-force and Bendfng-momenl
Diagrams 198
$ 55. Fundamental concepts of deformation in bending. Construction of
beam supports 195
§56. Nature of stresses in a beam. Bending moment and shealine
force 200 *
§57. Differential relation between the intensity of a continuous load,
shearing force and bending moment 205
§58. Plotting bending-moment and shearing-force diagrams 207
§59. Plotting bending-moment and shearing-force diagrams for more
complicated loads 214
1 60. The check of proper plotting of Qr and M-diagrams 221
§ 61. Application of the principle of superposition of forces In plotting
shearing-force and bending-moment diagrams 223
Chapter 1|. Determination of Normal Stresses in Bending and Strength of
Beams 225
§62. Experimental investigation of the working of materials in pure
bending 225
§ 63. Determination of norma) stresses In bending. Hooke’s law and po­
tential energy of bending 228
§ 64. Application of the results derived above in checking the strength of
beams 235
Chapter 12. Determination of Moments of Inertia of Plane Figures 239
§65. Determination of moments of inertia and section moduli for simple
sections 239
§66. General method of calculating the moments of inertia of complex
sections 244
§ 67. Relation between moments of inertia about two parallel axes one
of which is the central axis 246
§68. Relation between the moments of inertia under rotation of
axes 247
§69. Principal axes of inertia and principal moments of inertia 250
§70. The maximum and minimum values oi the central moments of
inertia 254
§ 71. Application of the formula for determining normal stresses to
beams of non-symmetricai sections 254
§72. Radii of inertia. Concept of the momenta! ellipse 256
§ 73. Strength check, choice of section and determination of permissible
load in bending 258
Chapter 13. Shearing and Principal Stresses In Beams 263
§ 74. Shearing stresses in a beam of rectangular section 263
1 75. Shearing stresses in I-beams 270
§ 76, Shearing stresses in beams of circular and ring sections 272
§77. Strength check for principal stresses 275
§ 78. Directions the principal stresses 280
Chapter 14. Shear Centre. Composite Beams 283
§ 79. Shearing stresses parallel to the neutral axis.
Concept of shear centre 233
§80, Riveted and welded beams 289
12 Contents

PART V. Deformation of Beams due to Bending

Chapter 15. Analytical Method of Determining Deformations 292
§ 81. Deflection and rotation oi beam sections 292
$ 82. Differential equation of the deflected axis 294
§ 83. Integration oi the differentia) equation of the deflected axis of a
beam fixed at one end 296
84. Integrating the differential equation of the deflected axis of a simply
supported beam 299
§ 85. Method of equating the constants of integration oi differential
equations when the beam has a number of differently loaded
portions 301
§ 86. Method of initial parameters for determining displacements in
beams 304
87, Simply supported beam unsymmettically loaded by a force 305
! 88. Integrating the differential equation for a hinged beam 307

89. Superposition of forces 310

90. Differential relations in bending 312
Chapter 16. Graph-analytic Method of Calculating Displacement in Bending 3)3
§ 91. Graph-analytic method 313
§ 92. Examples of determining deformations by the graph-analytic
method 317
§ 93. The graph-analytic method applied to curvilinear bending-moment
diagrams 320
Chapter 17. Non-uniform Beams 324
§ 94. Selecting the section in beams of uniform strength 324
| 95. Practical examples of beams of uniform strength 325
§ 96. Displacements in non-uniform beams 326
PART VI. Potential Energy. Statically Indeterminate
Beams
Chapter 18. Application of the Concept of Potential Energy in Determining Displace­
ments 331
§ 97. Statement of the problem 331
| 98. Potential energy in the simplest cases of loading 333
§ 99. Potential energy ior the case of several forces 334
§ 100. Calculating bending energy using internal forces 336
§ 101. Castigliano’s theorem 337
$ 102. Examples of application of Castigiiano’s theorem 341
$ 103. Method of introducing an external force 344
§ 104. Theorem of reciprocity of works 346
§ 105. The theorem of Maxwell and Mohr 347
§ 106. Vereshchagin’s method 349
1 107. Displacements In frames 351
1 108. Deflection of beams due to shearing force 353
Chapter 19. Statically indeterminate Beams 356
§ 109. Fundamental concepts 356
§ 110. Removing static indeterminacy via the differential equation of the
deflected beam axis 357
 Contents 13

§111. Concepts of redundant unknown and base beam 359

1 112. Method of comparison of displacements 360
§113. Application of the theorems of Castigliano and Mohr and Vereshcha­
gin's method 362
§ 114. solution of a simple statically Indeterminate frame 364
§ 1(5. Analysis of continuous beams 366
1 116. The theorem of three moments 366
1 117. An example on application erf the theorem of three moments 372
§118. Continuous beams with cantilevers. Beams with rigidly fixed
ends 375

PART VII. Resistance Under Compound Loading

Chapter 20. Unsymmetrlc Bending 378
§ 119. Fundamental concepts 378
£ 120. Unsymmetrlc bending.Determination ofstresses 379
§ 121. Determining displacements inunsymmetrlcbending 365

Chapter 21. Combined Bending and Tension or Compression 389

§ 122. Deflection of a beam subjected to axial and lateral forces 389
1 123. Eccentric tension or compression 392
§ 124. Core of section 396

Chapter 22. Combined bending and torsion 401

§ 125. Determination erf twisting and bending moments 401
§ 126. Determination of stresses and strength check In combined bending
and torsion 404

Chapter 23. General Compound Loading 408

§ 127. Stresses in a bar section subjected to general compound
loading 408
§ 128. Determination of normal stresses 410
129. Determination of shearing stresses 413
1 130. Determination of displacements 414
131. Design of a simple crank rod 417
Chapter 24. Curved Bars 423
§ 132. General concepts 423
§ 133. Determination of bending moments and normal and shearing
forces 424
§ 134. Determination of stresses due to normal and shearing forces 420
§ 135. Determination of stresses due to bending moment 427
§ 136. Computation of the radius of curvature of the neutral layer in a
rectangular section 433
§ 137. Determination of the radius of curvature oi the neutral layer for
circle and trapezoid 434
§ 138. Determining the location of neutral layer from tables 436
§ 139. Analysis of the formula for normal stresses In a curved bat 436
§ 140. Additional remarks on the formula for normal stresses 439
§ 141. An example on determining stresses in a curved bar 441
1 142. Determination of displacements in curved bars 442
§ 143. Analysis of a circular ring 445
14 Contents

Chapter 25. Thick-walled and Thin-walled Vessels 446

$ 144. Analysis of thick-walled cylinders 446
1 145. Stresses in thick spherical vessels 453
§ 146. Analysis of thin-walled vessels 454
Chapter 26. Design for Permissible toads. Design for Limiting States 467
§ 147. Design for permissible loads. Application to stattcally deter­
minate systems 457
§ 146. Design or statically indeterminate systems under tension or
compression by the method of permissible loads 458
§ 149. Determination of limiting lifting capacity of a twisted rod 462
§ 150. Selecting beam section Tor permissible loads 465
1 151. Design of statically indeterminate beams for permissible loads.
The nindamentals. Analysis of a two-span beam 468
6 152. Analysis of a three-span beam 472
§ 153. Fundamentals of design by the method of limiting states 474

PART VIII. Stability of Clements of Structures

Chapter 27. Stability ot Ban Under Compression 477
§ 154. Introduction. Fundamentals of stability of shape of compressed
bars 477
6 155. Euler's formula for critical force 480
§ 156. Effect of constraining the bar ends 484
§ 157. Limits of applicability of Euler’s formula. Plotting of the diagram
of total critical stresses 488
§ 158. 'The stability check of compressed bars 494
§ 159. Selection of the type of section and material 498
§ 160. Practical importance of stability check 502
Chapter 28. More Complicated Questions of Stability in Elements of Structures 604
§ 161. Stability of plane surface in bending of beams 504
§ 162. Design of compressed-bent b an 512
§ 163. Effect of eccentric compressive force and initial curvature
of bar 517

PART IX. Dynamic Action of Forces

Chapter 29. Effect of Forces of Inertia. Stresses due to Vibrations 521
§ 164. Introduction 521
§ 165. Determining stresses in uniformly accelerated motion of
bodies 523
§ 166. Stresses In a rotating ring (flywheel rim) 524
§ 167. Stresses in connecting rods 525
$ 168. Rotating disc of uniform thickness 529
1 169. Disc o f uniform strength 533
§ 170. Effect of resonance on the magnitude of stresses 535
§ 171. Determination of stresses in elements subjected to vibration 536
§ 172. The effect of mass of the elastic system on vibrations 541
Chapter 30. Stresses Linder Impact Loading 548
§ 173. Fundamental concepts 548
§ 174. General method of determining stresses under impact loading 549
 Contents 15

§ 175. Concrete cases of determining stresses and conducting strength

checks under impact 5S4
$ 176. Impact stresses in a non-uniform bat 559
177. Practical conclusions from tbe derived results 660
1 178. The effect of mass of the elastic system on impact 562

179. Impact testing for failure 565

180. Effect of various factors on the results of impact testing 568

Chapter 31. Strength Check of Materials Under Variable Loading 571

§ 181. Basic ideas concerning the effect of variable stresses on the -strength
of materials 571
§ 182. Cyclic stresses 573
§ 183. Strength condition under variable stresses 575
§ 184. Determination of endurance limit in a symmetrical cycle 576
1 185. Endurance limit in an unsymmetrlcsl cycle 579
1 186. Local stresses 682
§ 187. Effect of size of part and other factors onendurance limit 589
§ 188. Practical examples of failure undo: variableloading.Causes of
emergence and development of fatigue cracks 593
§ 189. Selection of permissible stresses 597
§ 190. Strength check under variable stresses and compound stressed
state 600
§ 191. Practical measures for preventing fatigue failure 602
Chapter 32. Fundamentals of Creep Analysis 605
§ 192. Effect of high temperatures on mechanical properties oi
metals 605
§ 193. Creep and after-effect 607
§ 194. Creep and after-effect curves 609
§ 195. Fundamentals of creep design 615
§ 196. Examples on creep design 620
Appendix 630
Name index 639
Subject index 641
 PART I
Introduction.
Tension and Compression

CHAPTER 1
Introduction

§ 1. The Science of Strength of Materials

In designing structures and machines, an engineer has to select the
material and the cross-sectional area of each element of the structure
or machine so that it enables the element to have strength to resist
external forces transmitted to it by adjacent elements of the structure
without failure of strength or distortion of shape, i. e. the element
should function properly. Strength of materials provides the engineer
with fundamentals for a proper solution of this problem.
Strength of materials deals with the behaviour of various materials
under the action of external forces and points out how to select the
appropriate material and the cross-sectional area of each element of
the structure so as to prpyide fully reliable functioning and the most
economic design.
Sometimes, strength of materials has to deal with the problem in a
modified form—to check the dimensions of a designed or existing
structure.
The conditions for maximum economy in design and reliability of
functioning are contradictory. The former demand minimum consump­
tion of materials whereas the latter lead to increase in consumption.
This contradiction forms the basis of the technique, which has facili­
tated the development of strength of materials.
Often the existing methods of checking the strength and the availab­
le materials are unable to meet the practical requirements for providing
answers to new problems (for example, attaining high speeds in engi­
neering in general and in aerostatics in particular, long-span structures',
dynamic stability, etc.). This initiates a search for new materials
and study of their properties, and inspires research for improving the
existing methods of designing and devising the new ones. Strength of
materials must keep pace with the general development of engineering
and technology.
2-3310
16 introduction. Tension and Compression [Part I

Sometimes, besides the chief requirements of maximum reliability

and economy, an engineer has to ensure fulfilment of other conditions
too, such as quick building (when restoring broken structures), mini­
mum weight (in aircraft design), etc. These conditions influence the
dimensions, the shape and the material of the various elements compris­
ing the structure.
The emergence of strength of materials as a separate science dates
back to 162® and is intimately connected with the works of Galileo
Galilei, the great Italian scientist. Galileo was a professor of mathe­
matics at Padua. Me lived in a period which saw the disintegration of
the feudal system, the development of trade capital and international
maritime transport,and thebirth of mining and metallurgical indus­
tries.
The rapid economic developments of those times called for speedy
solutions of new technological problems. Increase 'in international
maritime trade perpetuated the need for bigger ships which in turn
entailed changes in their design; at the same time it became necessary
to reconstruct the existing and to build new internal waterways, in­
cluding canals and sluices. These new technical problems could not be
solved by simply copying the existing designs of ships; it became
necessary to judge the strength of elements keeping in mind their size
and the* forces acting upon them.
Galileo devoted a considerable part of his work to the study of the
dependence between the dimensions of beams and bars and the loads
they could withstand. He pointed out that the results of his experiments
may prove very useful in building big ships, especially in strengthening
the deck and covering because low weight is very important in struc­
tures of this type. Galileo’s works have been published in his book
Discorsi e Dimoslrazioni Maiematiche . . . (“Dialogue on Two New
Sciences . . . ”) (1638, Leiden, Holland).
Further development of strength of materials went on in step with
the progress of mechanical and civil engineering, and materialized
owing to the research work done by a large number of eminent scien­
tists, mathematicians, physicists and engineers. Russian and Soviet
scientists occupy an important place amongst them. Brief informative
sketches about the role played by individual scientists in the develop­
ment of some problems of strength of materials are given in correspond­
ing chapters of the book.

§ 2. Classification of Forces Acting on Elements

of Structures
When in operation, the elements of structures and machines are
subjected to external loads, which they transmit to one another. A dam
bears its own weight and the pressure of water that it holds and trans­
mits these forces to the foundation. The steel trusses of bridges take
Ch. 11 Introduction, 19

the weight of the train through the wheels and rails and transmit it to
the stone supports, and the latter, in turn, communicate this load to
the foundation. The steam pressure in the cylinder of a steam engine is
transmitted to a piston rod. The pulling force of the locomotive is
transmitted to the train through a coupler which connects the tender
with the wagons. Hence, the elements of structures are subject to either
volume forces acting on each element of the structure (dead weight) or
forces of interaction* between the element under consideration anti
adjoining elements or between the element and the surrounding medi­
um (water, steam or air). In future, when we say that an external force
is being applied to an element of the structure, this will imply the
transmission of force of pressure (motion) to the element under consid­
eration from adjoining elements of the structure or the surrounding
medium.
The forces may be classified according to a number of criteria.
We distinguish between the concentrated and distributed forces.
A concentrated force is defined as the force of pressure transmitted to
the element of structure through an area which is very small as compa­
red to the size of the element, for example, the pressure of the wheels
of a moving train on the Tails.
In practice the concentrated force is considered to be acting at a point
owing fo the small area through which the pressure is transmitted.
We must keep in mind that this is an approximation which has been
introduced to simplify the calculations; actually, no pressure can be
transmitted through a point. However, the error due to this approxi­
mation is so small that it may be generally ignored.
A distributed force is defined as the force applied continually over a
certain length or area of the structure. A layer of sand of uniform thick­
ness spread over the sidewalk of a bridge represents a force which is
uniformly distributed over a certain area; if the thickness of the sand
layer is not uniform we shall obtain a non-unifornily distributed load.
The dead weight of a beam in the ceiling represents a load distributed
over its length.
The concentrated loads are measured in units of force (tons, kilog­
rams, newtons **); the loads distributed over an area are measured in
terms of force per unit area <tf/m*. kgf/cm*, N/m*, etc.); the loads
distributed along the length of an element are expressed as force per
unit length (kgf/m, N/m, etc.).
The loads may further be classified as permanent and temporary.
The permanent loads act throughout the whole life of Ihe structure,
e.g. dead weight. The temporary loads act on the structure only for a

* To be precise, the weight of a body is the force of interaction between the

body and the earth.
** In the SI system, which is now preferred and recommended, the force is mea­
sured In newtons (1 N«0.102 kgl).
2*
20 Introduction. Tension and Compression [Part /

certain period of time—the weight of the train moving along the bridge
may be cited as an example.
According to the nature of action, the loads may be classified as
static and dynamic.
Sialic loads act on the structure gradually, after being applied to the
structure they either do not change at all or change insignificantly;
the majority of loads acting in civil and hydraulic structures are of
this nature. Under the influence of static loading all elements of the
construction remain in equilibrium; accelerations in the elements of
the structure are either totally absent or so small that they may be
neglected.
If, however, the acceleration is considerable and the change in
velocity of the machine or structure takes place in a short time, the
load is known as dynamic.
The examples of dynamic loads are suddenly applied load, impact
load and repeated variable load.
Suddenly applied loads are transmitted instantaneously in their
total magnitude. An example of this type of loading is the force of
pressure of the wheels of a locomotive when it enters a bridge.
Impact loads appear when there is a sharp change in the velocity of
adjoining elements of a structure, for example the impact of drop
hammer during pile driving.
The repeated variable loads act on the elements of structures for a
considerable number of limes. For example, repeated steam pressure,
alternately stretching and coinpressing the piston rod and the connect­
ing rod of the steam engine. In a number of cases the load represents
a combination of dynamic loads of different nature.
We shall first of all study the resistance of materials to static loads;
the selection of material and cross-sectional area for each element of
the structure does not present many difficulties in this case.
In Chapters 29-31 we shall discuss the action of dynamic loads in a
number of instances which occur as often as static loads; they require
careful study because their effect on the elements of structures differs
from that of static loads, and the material also resists them in a differ­
ent manner.
Concluding the classification of forces acting on the elements of
structures, let us consider the action of parts which support these
elements; the forces acting on these supports are known as the reaction
forces—they are unknown quantities and are determined from the
condition that each element of the structure must remain in equilib­
rium under the action of all the external forces applied to il and the
reaction forces.
Ck I) Introduction. 21

§ 3. Deformations and Stresses

In theoretical mechanics (statics) we study the equilibrium of a
perfectly rigid body; this concept of material in statics is sufficient to
determine the conditions in which the body will remain in equilibrium
under the action of external forces applied to it. However, this rough
and approximate concept of the properties of materials does not hold
good in strength of materials; here we must take into account the fact
that there does not exist a perfectly rigid body.
The elements of a structure, as well as the structure as a whole,
change their dimensions and shape to some extent under the action of
external forces and are Iiable to complete failure in the end. This change
in shape and size is called deformation.
The magnitude and nature of the deformation depend uponthestruc-
ture of the material used. All materials may be divided into two
groups: crystalline and amorphous.
Crystalline materials consist of a very large number of extremely smalt
crystals. Each of these is a system of atoms arranged very close to each
other in regular rows. These rows form the so-called crystalline lattice.
In amorphous materials the atoms are not arranged in a particular
order. Thev are held in equilibrium by the forces of interaction. The
deformation of bodies takes place due to change in the location of
atoms, i.e. due to their getting closer or farther.
Deformations are divided into elastic and plastic. Elastic deformation
disappears when the force causing the deformation is removed; in
this case, the body completely regains its initial shape and dimensions.
This deformation occurs due to elastic distortion in the crystalline
lattice. It has been experimentally observed that the elastic deforma­
tion continues till the forces being applied do not exceed a certain limit.
If, however, the external force exceeds this particular limit, the
body fails to regain completely its initial shape and size after the
force is removed; the difference in size which thus remains is called
the plastic (residual) deformation. In crystalline materials, this defor­
mation is caused by the irreversible displacement of one layer of
crystalline lattice with respect to the other. After the removal of exter­
nal forces the displaced layers of atoms retain their position.
In deformation, the displacement of atoms under the action of exter­
nal forces is accompanied by a change in the forces of interaction
between the atoms, i.e. the forces of attraction and repulsion.
Additional internal forces accompanying the deformation appear in
the elements of structures under the action of external forces. These
internal forces resist the external forces and try to prevent them from
breaking the element, changing its shape or separating one part from
the other. They try to regain the initial shape and size of a deformed
part of the structure. In order to assess the effect of the external forces
on the deformed element, we must know how to measure and calculate
22 Introduction. Tension and Compression \Parl I

the interatomic forces that appear as a result of the deformation caused

by the action of external forces.
In strength of materials this is achieved by the method of sections,
which we shall try to explain by the following example. Lei us imagine
a bar (Fig. 1), which is subjected to the action of two equal and oppo­
site forces, P, and let us imagine that the bar is cut in two parts /
/ m I
— -fei---------- -Ci- f

Fig. 1

and II by a plane mn. Under the action of forces P both halves of the
bar tend to go apart, but are held together owing to the forces of inte­
raction between the atoms located on both sides of plane mn. The
resultant of the forces of interaction is the internal force transmitted
through mn from one half of the bar to the other, and vice versa. The
internal force of interaction per unit area around any point of section
mn is called the stress at the point of the given section. The stresses
acting from part II on part I and from part I on part II are equal
in magnitude according to the law of action and reaction.
A number of planes dividing the bar in two parts in different ways
can be drawn through a single point of the bar. The magnitude and
direction of the stresses transmitted through the given point from one
part of the bar to the other will depend upon how the plane cuts the
bar.
Thus, it is wrong to speak of stresses without indicating the plane
through which they are being transmitted. Therefore we speak about
“the stress on a particular area in a particular plane”. Since stress is a
force per unit area, it is measured in kgf.'cm*, kgf/mni*, tf/cma, tf/ms,
N/m4, etc.
In future, we shall denote stress by letters p, a, and x; letter p is
used for stresses applied to certain area in any plane inclined at an
arbitrary angle, <j denotes stress at right angles lo the plane, i.e. normal
stress, and t denotes stress in the plane, i.e. shearing stress.
The stress at any point is the measure of internal forces which appear
in the material owing to its deformation under the action of external
forces. The force transmitted from part I of the bar to part II (see
Fig. 1) holds part II in equilibrium, i.e. counterbalances thesvstem of
external forces acting on part II. This force may be expressed in terms
of the stress to be determined: if we consider an elementary area dA
in the plane of cutting, then the elementary force acting on this ele­
mentary area will be p dA, where p is the stress at the point around
which the elementary area is located. The sum of these elementary
forces gives us the total force transmitted through the particular plane.
Ch. 1\ Introduction 23

Thus, to determine the stresses, it is necessary to imagine the ele-

ment to be cut in two parts and write down the conditions of equilib­
rium for the system of forces acting on one of the cutoff parts; this
system includes the external forces applied to the part of the bar under
consideration and also the force transmitted through the given plane
and expressed in terms of stresses sought. This is the method of sec­
tions, which we shall constantly apply in future.
Let us point out, that, in strength of materials, Ihe term "stress’*
is verv often used instead of the expression “internal forces of interac­
tion between parts of the bar”; therefore in future when we mention
“uniform and non-uniforin distribution of stresses over the section”
and “force as the sum of stresses”, we must bear in mind that these
expressions are to a certain degree conventional. For example, to deter­
mine the force one cannot sum up the stresses at various points; as
mentioned above, it is necessary to find at each point of section the
elementary force which is transmitted through an elementary area dA
and then sum up ail these values. Recapitulating what has been writ­
ten above, we come to the conclusion that when an external force is
applied to an element of structure, the latter gets deformed and the
deformation is accompanied by stresses in the element.
Strength of materials studies, on the one hand, the relation between
the external forces and, on the other hand, the deformations and
stresses due to them. This enables the engineer to solve the important
problem of selecting a bar of proper dimensions and appropriate mate­
rial to resist the external forces. In the next section we shall give an
outline of the solution to this problem.

§ 4. Scheme of a Solution of the Fundamental Problem of Strength

of Materials
While selecting the size and material for an element of the structure
we must provide for a certain safety factor against its failure and plas­
tic deformation. The element should be designed so that the maximum
stresses that occur during its operation should always be less than the
stresses at which the material fails or undergoes plastic deformation.
The stress at which the material fails is called the ultimate (tensile)
strength ; we shall denote it with the same letter as stress but with
subscript u. The stress beyond which the material deforms insigni­
ficantly and only up to a predetermined value is known as the elastic
limit* -These quantities are known as the mechanical characteristics
of resistance of materials to failure and plastic deformations. To ensure
the smooth functioning of the structure without a risk of failure, we
must see to it that the element is only subjected to stresses which are
less than its ultimate strength.
* Ultimate strength and elastic limit will he more precisely explained in § 10-
 24 Introduction. Tension and Compression [Part I

The permissible stress is denoted by the same letter but is put in

square brackets; it is related to the ultimate strength pu by the fol­
lowing expression:
[P] aa£5E
k

where &is t he safety factor wh ich shows how many times the permissible
stress is less than the ultimate (tensile) strength. The value of this
factor varies from 1.7-1.8 to 8-10 and depends upon the operating con­
ditions of the structure. It will be discussed in greater detail in §§ 16
and 17.
Denoting by pmax the maximum stress that appears in the designed
element under the action of external forces, we may write the basic
condition, which the size and material of the element must satisfy, as
follows:
Pmax ^ [ P ] (1 •1)
This is the strength condition, which states that the actual stress must
be not greater than the permissible.
Now we may compile the plan for solving the problems of strength
of materials as follows.
(1) Ascertain the magnitude and nature of all the external forces,
including the reactions, acting on the element under consideration.
(2) Select an appropriate material that is most suitable in the
working conditions of the element (structure) and the nature of loading;
determine the permissible stress.
(3) Set the cross-sectional area of the element in numerical or algeb­
raic form, and calculate the maximum actual stress pmax which deve­
lops in it.
(4) Write down the strength condition pmax^ fp] and with the help
of it calculate the cross-sectional area of the element or check whether
the set value is sufficient.
The plan of solution of problems in strength of materials is sometimes
altered; in some structures Ihe safety factor for the whole structure
is found to be greater than that for the material in the point of great­
est stress. If the limiting lifting capacity of the material is exhausted
at this point, this does not necessarily mean that the limiting lifting
capacity of the whole structure has also been reached. In such cases,
the strength condition p ^ ^ l p l is replaced by the strength condition
for the structure as a whole:

here P is the load on the structure, Ppcr is its permissible value, and
P u is the limiting force which the structure can withstand without brea-
Ch. ;j Introduction 25

king down. Thus, the design based on permissible stresses is replaced

by the design based on permissible loads.
In this case, it is necessary to:
(1) ascertain the magnitude and nature of all the external forces,
acting on the given element;
(2) select the appropriate material that is most suitable in the
working conditions of the structure and also takes into considera­
tion the nature of loading; determine the safety factor;
(3) set the cross-sectional area of the elements of structure in nume­
rical or algebraic form, and calculate the maximum permissible
load P ^ r\
(4) write clown the strength condition P^Pper and with its help
calculate the cross-sectional area of the elements of structure or
check whether the set dimensions are sufficient.
In a number of cases, as we shall see later (§ 150), both methods give
similar results.
In general, we shall be using the conventional method of design
based on permissible stresses; however, along with this, the method
of design based on permissible loads will be explained, especially in
cases where the two methods give dissimilar results.
In the majority of cases the strength condition must be supplement­
ed by stability and rigidity tests. The first test ensures that the
elements of structure must not change their predetermined type of equil­
ibrium, and the second test sets limits to the deformations of elements.
While solving problems on strength of materials, we have to take
the help of theoretical mechanics and experimental techniques. The
determination of external forces is based on equations of statics; in
statically indeterminate structures, it is essential to determine the
deformation of the material. This, as shown in § 18, is possible only
if we have reliable experimental data on the relation between deforma­
tions and forces or stresses.
To estimate the permissible stresses we must know the ultimate
strength of the material and its other mechanical properties. This
information can also be obtained by a study of the properties of mate­
rial in special material testing laboratories. Finally, to determine actu­
al stresses we seek the help of not only mathematical analysis and mech­
anics but also the available experimental data. Thus strength of ma­
terials consists of two methods of solving a problem: analytical, based
on mathematics and mechanics, and experimental. Both these methods
are closely interrelated.
Strength of materials should not be considered a branch of science
which deals only with theoretical determination of stresses in some
homogeneous elastic body. The problems studied in strength of mate­
rials can be solved only if we have sufficient experimental data on the
mechanical properties of real materials, keeping in mind their struc­
ture, methods of fabrication and machining. Therefore, we have paid
20 Introduction. Tension and Compression I Pari I

considerable attention to this aspect in our book. Experiments play

an important role in the understanding of a subject and must be car­
ried out by the students along with their theoretical studies. These
experiments, worked out on the basis of facilities and equipment
available in the strength testing laboratories, have been presented in a
separate manual.*
Although at the very outset strength of materials was identified
with the necessity for solving a number of purely practical problems,
its further development was more on the theoretical side, resulting at
times in discrepancies between the outcome of experimental investi-
P-*" ( q ^ — *-P
la}

gat ions and their practical application. Laboratory research went along
a special path, chiefly to set the acceptable standards for various types
of materials. Now strength of materials studies real materials in
accordance with their operation in structures accentuated by intensive
experimental and theoretical investigations for solving developing
day-to-day practical problems. These problems, for example, are those
connected with the study of the strength of new materials, conditions
pertaining to their failure, determination of stresses not only within
the limits of elasticity but also beyond them, etc.
• N. M. Belyaev, Laboratory Experiments in Strength of Materials, Qostekhiz-
dat, 1U51 (in Russian).
Ch. 1] Stress and Strain W ithin Elastic Limit 27

§ 5. Types of Deformations

Having accepted a general method of solving problems of strength

of materials, we may now go over to studying individual problems.
These may be divided into a number of groups depending upon the
type of deformations.
The common types of deformations (Fig. 2) are: (1) tension or com­
pression, (o) and (/;), as in chains, ropes, cables, bars of trusses working
under tension or compression, columns; (2) shearing (c)t as in bolts
and rivets; (3) torsion (d), as in shafts; (4) bending (e), as in beams of
all types. These four types of deformations are called simple deforma­
tions.
The operation of elements in structures is generally more complex;
they are subjected to two or more types of deformations simultaneously,
for example, tension or compression and bending, bending and torsion,
etc. These are cases of the so-called composite deformation. For each of
the abovementioned types of deformations, we shall find out methods
for determining the stresses, selecting the material and cross-section­
al area of the elements and determining the magnitude of deformation.
To make it easy for the reader to understand, initially we shall
consider only (hose elements of structures and machines which are in
the form of prismatic bars with a straight axis. A body which has a uni­
form cross-sectional area ail along its length may be considered a
prismatic bar. The centres of gravity of all the sections of the body lie
on one straight line, which is called the axis of the bar. Later on we
shall also consider bars with a non-uniform cross-sectional aiea and
curved axis.

CHAPTER 2

Stress and Strain in Tension

and Compression Within the Elastic Limit.
Selection of Cross-sectional Area

§ 6. Determining the Stresses in Planes Perpendicular to the Axis

of the Bar

We shall start the study of strength of materials with the simplest

case of tension or compression of a prismatic bar.
Axial tension or compression of such a bar is its deformation under
the action of two equal and opposite forces applied at the end faces of
the bar along its axis. If these forces are directed outwards, the bar is
2ft Introduction. Tension and Compression \P art I

said to be under tension (Fig. 3a); in the opposite case, under compres­
sion (Fig. 3l>).
According to the general method of solving problems of strength of
materials, we must first determine the magnitude of the external forces
P stretching (compressing) the bar. The value of force P can usually
be determined by considering the interaction of the bar with the other
elements of the structure.

la) m
Fig. 3

As a simple example, consider a round steel coupler threaded at the

ends and loaded by axial tensile forces P = 25 tf (Fig. 4). Our task is to
select the cross-sectional area of the shaft which provides sufficient
strength. It is required to find out the stresses due to forces F, deter-

--------- —
Fig. 4

mine the permissible stress and select the cross-sectional area in such a
way that the actual stress does not exceed its maximum permissible
value.
To determine the stresses, it is essential to select the planes by which
the bar is to be cut into two parts. Strength should be checked in the
critical section, i.e. in the section through which the maximum stress
is transmitted. We shall first derive formulas for determining stresses
in a plane perpendicular to the axis of the bar, and later on in inclined
planes too; we will thus be in a position to find the critical section.
.7

Fig. 5 Fig. 6

Let us take a stretched bar and cut it in two parts by a plane nrn
(Fig. 5), perpendicular to the axis of the bar. Let us discard the second
part; lo retain the equilibrium of the first part, we must replace the
discarded part by the forces transmitted through section mn (Fig. 6).
a . 2\' Stress and Struin Within Elastic Limit 29

The equivalent forces must balance force P. Therefore they must

compose a resultant force N equal in magnitude to force P and directed
along the axis in the opposite direction (Fig. 6). This resultant N
is the force acting in the bar.
In future the resultant of internal elastic forces, transferred from
one part to the other across the imaginary section, will be called normal
or axial force. However, since the cutoff portion of the bar must remain
in equilibrium under the action of the normal force and the external
forces acting on it, the normal force may also be calculated through
the external forces. It is numerically equal to the resultant of external
forces applied to the part of the bar under consideration and acts
in the opposite direction. If the normal force acts inwards into the
part under consideration, the bar is said to be compressed; if it acts
in the opposite direction, the bar is said to be in tension.
Thus, the conditions of equilibrium of the remaining portion of the
bar only give us the magnitude of the resultant of the internal forces
transmitted through section mn, its direction and point of application.
They, however, do not give us any idea of how the stresses are distri­
buted over the section, i.e. what forces are being transmitted through
various unit areas of the section. Let us point out that to ascertain the
maximum danger of failure of a material, it is essential to determine
the maximum stress and also the unit area of the critical section
through which it is transmitted.
Experiments on tensile loading of bars of various materials reveal
that if the forces are directed along the axis sufficiently accurately,
then the elongations of lines drawn on the surface of the’ bar parallel
to the axis are equal. This gives rise to the hypothesis or uniform
distribution of stresses over the section. Only at the faces of the bar,
where force P is directly transmitted to it, the distribution of stresses
over different parts of the section is not uniform. The portions to which
force P is applied directly gel overloaded; but just a small distance
aw'ay from the point of application of the force the material starts
behaving more uniformly and stress distribution over the section
perpendicular to the axis becomes uniform. These stresses are directed
parallel to force P, i.e. perpendicular to the section; therefore they
are called normal stresses and denoted by the letter o. Since they are
distributed uniformly over the section, N=oA\ on the other hand,
N —P. Hence
£
o —A ( 2. 1)

This formula enables us to determine stress cr if the tensile force and

the cross-sectional area are known. On the other hand, if ure know the
maximum permissible normal stress, this formula helps us to find the
required cross-sectional area A.
30 Introduction. Tension and Compression I Part /

§ 7. Permissible Stresses.
Selecting the Cross-sectional Area
To ascertain the permissible stress limit for proper functioning
of a bar of the given material, we must experimentally establish the
relation between the strength of the bar and the stresses that appear
in it. For this, it is essential to prepare a specimen (usually of a round
or rectangular cross section) of the given material, clamp its ends in a
machine for tensile loading and gradually increase the tensile force P.
The specimen will stretch and ultimately break down.
Let Pu be the maximum load which the specimen can sustain before
rupture. The normal stress due to this load is

and is called the ultimate (tensile) strength of the material under tension.
It is usually expressed in the units kgf/mm* or kgf cm*.
As pointed out earlier in § 4, the maximum permissible normal
stress l<rl is several times less than the ultimate strength ou; the per­
missible stress is obtained by dividing the ultimate strength by the
safety factor k. The value of k depends upon a number of factors, which
shall be discussed in detail later on (§ 16). At any rate, the value of
the safety factor must ensure not only the normal working of the ele­
ment, i.e. working without failure, but also prevent the formation of
plastic deformations which may afTect the working of the machine or
structure. The safety factor depends upon the material of the element,
nature of the forces acting on the element, economic conditions and a
number of other factors.
In view of the importance of properly selecting the safety factor and
the permissible stress, these quantities have been standardized Tor a
large number of structures and machines, and must be strictly followed
by the designers. Hence, the permissible stress |<xl may be considered
in each case a known quantity. Therefore, to determine the cross-
sectional area of a stretched bar one may, using formula (2.1), write
down the strength condition', this condition states that under the action
of a force P , the actuat stress in a stretched bar must not exceed the
permissible stress lal:
( 2 -2)

From this condition the minimum cross-sectional area of the bar

may be determined as

With the help of formula (2.3), one can select the cross-sectional area
of the bar.
Ch. 2) Stress and Strain Within Elastic Limit 31

Sometimes the cross-sectional area is preset. Then, from formula

(2.3) we can find the permissible load
P<A[a] (2.4)
Returning to the design of the wagon coupler (§ 6, Fig. 4), it Is
required to select the material and the permissible stress. The coupler
is made of steel with an ultimate strength of about 50 kgf/mm*. The
material is selected such that the coupler is not too heavy, this condi­
tion being fulfilled only by using a high-strength material. At the same
time, the material should have good resistance to shocks and impacts
A steel of a very high ultimate strength cannot be used because it is
brittle.
The coupler should not only withstand fracture but also resist any
noticeable plastic deformation to prevent jamming of the coupler
thread. The elastic limit for the selected steel is approximately 0.6
times its ultimate strength cr«. We shall see later that the stress under
sudden loading is nearly twice its value under static loading, i.e. its
value as determined Under laboratory conditions. The permissible
stress should therefore not exceed
0.5x0.6afl = 0.3(TB
Hence, the safely factor

Therefore, in this case we may take the permissible stress

[oj = Y' = 0.3a„ = 5 0 x 0 .3 = 15 kgf/mm* = 1500 kgf/cma

The required cross-sectional area at P=25 tf is

. ^ _P__ 25 000
H ^ [ o \ ~ 1500
16.7 cras

The diameter of the coupler d is computed from the condition

nd*
4
A 5 s 16.7
wherefrom
Y -4 .5 5 cm * 4.5cm

The calculated diameter corresponds to Ihe base of the thread with

the minimum cross-sectional area. When the cross-sectional area of
the bar is decreased in a particular place, for example due to a bolt or a
rivet hole, a circular cut or a groove (threading), it is essential to de­
termine the minimum cross-sectional area, called net area and denoted
22 Introduction. Tension and Compression I P.art I

by /4nct or An. The cross-sectional area without weakening is called the

gross area and denoted by /lgt0„ or A Rr Having computed the net
area A„, we can obtain the gross area A„ from design considerations.
The formulas derived above are valid for tension. They can be used
for compression as well without any changes. The difference will be
in the direction of normal stresses and the magnitude of ^p erm issib le
stress Iql. The compression of bars is more complex in that the bar may
become unstable, i.e. it may suddenly bend. Designing for stability
will be discussed in Part VIII.
Figure 7 shows normal distribution in a section perpendicular to the
axis of the bar for tension and compression. For a number of materials
(e.g. steel) the permissible stress value is the same in tension and com-
Tension
P
IS r
Compression
— r- $ er S P

Fig. 7

pression (for short bars, i.e. bars in which the length does not exceed
five times the diameter of cross section). In other materials (e.g. cast
iron) the permissible stress is different in tension and compression,
depending upon the ultimate strength for the recorded deformations.
In a number of cases, compressive stresses are transmitted from one
element of construction to another through a comparatively small area
of contact between them. This type of stress is generally called the
bearing, or contact stress. Stress distribution around the area of contact
is very complex and can be analyzed only by methods of the theory
of elasticity. Usually, in simple designing, these stresses are considered
as compressive stresses and a special permissible stress limit is fixed.
Later on the question of selecting permissible stresses in special cases
will be dealt with in greater details.

§ 8 . Deformations Under Tension and Compression.

Hooke's Law
To have complete idea about the working of a stretched or compressed
element, it is essential to know ways of calculating the change in its
dimensions. The corresponding laws can be obtained only on the basis
of experiments with a stretched or compressed specimen of the given
material; these experiments also help to study the strength of the ma­
terial and determine its ultimate strength and other characteristics
(§ 10).
Ch. 2} Stress and Strain Within Elastic Limit 33

These experiments are conducted in the laboratory on special ma­

chines which deform the specimen till it breaks down and measure the
force required for this purpose.
Simultaneously, the deformation of the specimen is measured* with
the help of sufficiently accurate measuring instruments — strain
gauges (tensometers). The testing machines are capable of applying a
sufficiently large load on the specimen and accurately measure the
same. Whole parts of structures (columns, portions of walls) can be
tested for compression-on presses having a capacity of up to 5000 tf.
Tension test can be conducted in- the laboratory on machines which are
capable of exerting a tensile load of up to 1500 tf. However, in a ma­
jority of the laboratories machines of considerably less capacity (from
5 to 100 tf for tension test and 200 to 500 tf for compression test) are
employed.
A detailed description of these machines and measuring instruments,
particularly of the well-known Gagarin Press, is available in the book
Laboratory Experiment in Strength of Materials, and also in special
manuals, on mechanical testing of materials. With the help of these
machines and measuring instruments one can establish how the mate­
rial specimen will change its dimensions under tension or compression.
Experiments enable us to conclude that up to a certain limit of
loading, the elongation is directly proportional to the tensile force P
and length of the specimen I but inversely proportional to the cross-
sectional area A. Denoting by M the elongation of the specimen due
to force P, we may write down the following relation between these
quantities:
A< = £ (2.5)

where E is the proportionality factor which depends upon the material.

Quantity Al is called the absolute elongation of the bar due to force P.
Formula (2.5) is called Hooke's law after the scientist who founded
the law of proportionality in 1660.
Relation (2.5) may be presented in a different form. Let us divide
both sides of the relation by /, the initial length of the bar:
M p_
l ~ EA

The ratio 7 of the absolute elongation to the initial length I iscalled

the relative elongation (strain)-, it is denoted by the letter e.
Relative elongation is a dimensionless quantity, as it is the ratio
between two lengths Al and / and is numerically equal to the elongation
pf a unit length of the bar. Replacing 7 by e and - j by the normal
34 Introduction. Tension and Compression {Part I

stress a, we get another expression for Hooke’s law:

s= 4 (2 .6 )

or
o = e£ (2.7)
Thus, the normal stress under tension or compression is directly
proportional to the relative elongation or shortening of the bar.
Proportionality factor £ , which links the normal stress with the
relative elongation, is called the modulus of elasticity of the material
under tension (compression). The greater the modulus of elasticity of a
material, the less the bar is stretched (compressed) provided all other
conditions (length, cross-sectional area, force P) remain unchanged.
Thus, in physical interpretation, the modulus of elasticity characteri­
zes the resistance of a material to elastic deformation under tension
(compression).
Since relative elongation e is dimensionless quantity, it follows from
formula (2 .7) that the modulus of elasticity has the same units as stress
a, i.e. it is expressed in units of force divided by area.
It should be noted that the modulus of elasticity £ does not remain
constant even for one material, but varies slightly. In some materials
the modulus of elasticity has the same value under tension and com­
pression (steel, copper), in other materials it has different values for
each of these deformations. In general this difference is ignored in de­
signing, and for a vast majority of materials a single value of E is
accepted both for tension and compression.
It should be borne in mind that Hooke’s law has been represented by
a formula which sums up the experimental data only approximately;
it cannot therefore be considered an accurate relation.
In all materials the deformation under tension or compression more
or less deviates from Hooke’s law. In some materials (most of the
metals) this deviation is negligible and it may be assumed that there
is exact proportionality between deformation and load; in other mate­
rials (cast iron, stone, concrete) the deviation Is considerably greater.
However, for practical purposes we may ignore the small deviation
from formulas (2 .5) and (2 .6) and use them as such in determining
deformation of the bar.
The mean values of the modulus of elasticity E for a number of
materials are given in Table I.
From formula (2.5) it is evident that the greater its denominator the
less is the elongation (pliability) or, in other words, the greater is the
rigidity of a bar. Therefore, the denominator of formula (2.5), the quan­
tity EA, is called the rigidity of the bar under tension or compression.
We see that the rigidity of a bar under tension or compression depends,
on the one hand, upon the material (modulus of elasticity E) and, on
Ch. 2\ Stress and SI tain Within Elastic Limit 35

Table 1
Modulus of Elasticity and Lateral Deformation Coefficient (Poisson’s Ratio)

Modulus of Coefficient of
Mnterln) elasticity £ lateral deforma-
(10* kgf.cm*) lion n

Iron grey, white 1.15-1.60 0.23-0.27

Carbon steel 2.0-2.1 0.24-0.28
Alloy steel 2.1 0.25-0.30
Rolled copper 1.1 0.31-0.34
Rolled phosphor bronze 115 0.32-0.35
Cold*dravt^i brass 0.91-0.99 0.32-0.42
Rolled naval brass 1.0 0.36
Rolled manganese bronze i.l 0.35
Rolled aluminium 0.69 0.32-0.36
Rolled zinc 0.84 0.27
Lead 0.17 0.42..
Glass 0.56 0.25
Granite, limestone, marble 0.42-0 56
Sandstone 0.18
(from granite 0.09-0.1 0.16-0.34
Masonry: I from limestone 0.06
(from brick 0.027-0.030
100 kgf/cm 0.146-0.196
Concrete having ultim ate 150 kgf/cm
strength' 200 kgf/cm
{ 0.164-0.214 )■ 0.16-0.18
Timber along the fibres 0 . 1-0.12
Timber across the fibres 0.005-0.01
r-rc 0.04 )
Ice • at temperatures < —3BC 0.07 } rsO.36
( ^~5BC 0.10 j
and below
Rubber 0.00008 0.47
Bakelite 0.02-0.03
Celluloid 0.0174-0.0193 0.39
Textoltte 0.06-0.1
Laminated Bakelite insulation 0.1-0.17
Rigid polyvinyl chloride (PVC) 0.040 0.22-0.3
Caprolan 0.02-0.023 0.28-0.34
High-pressure polyethylene 0.002-0.0025 0.40-0.46
Phenoplast 0.15-0.20 0.22-0.27
Polycarbonate 0.022-0.024 0.24-0.28
Plexiglas 0.028

* S N IP 1 1 -S 7 -7 5 (S N IP sta n d s for C o n str u c tio n S p e c ific a tio n s an d R e e u la tlo n s [In th e

USSRJ).
36 introduction. Tension and Compression [Part /

the other hand, upon its cross-sectional area A. Sometimes it is more

FA
convenient to use the term relative rigidity ---, i.e. the ratio of rigidity
to the length of the bar.
Formulas (2.5) and (2 .6) enable us to determine the elongation or
shortening in the bar of a structure under tension or compression.
Conversely, knowing the elongation, dimensions and the material of
the bar one may calculate the normal stresses acting in it. Thus, normal
stress can be determined by two methods. If the tensile or compressive
force P is known, a is calculated from the formula (2.1):

If the external force is not known but the elongation of the bar can be
measured, a is determined from formula (2.7):
<j=e£
The relative elongation may be calculated according to the following
formula if the total elongation A/ for a length I of the bar can be mea­
sured:

We shall show later that the second method has to be employed very
often to determine stresses in a number of cases.

§ 9. Lateral Deformation Coefficient.

Poisson’s Ratio
Apart from longitudinal deformation, the bars working under ten-
sick or compression are also subjected to lateral dcfcrmalion.
Experiments show that under tension (Fig. 8 ) the length of a bar
increases by A/, whereas its width decreases by A6 =(i>—h ). The re­
lative elongation

and the relative lateral deformation

Ab
«i b

In compression, the shortening of lh(! bar is its longitudinal deforma­

tion ahd the increase in its cross-sectional area is the lateral deforma­
tion. Ft has been experimentally proved that for a majority of the ma­
terials £i is from 3 to 4 times less than e.
Ch. 2] Slress and Strain Within Clastic Limit 37

The modulus of the ratio of the relative lateral deformation e, to

the relative longitudinal deformation e is called the coefficient of lateral
deformation, or Poisson's ratio p:
ll ( 2. 8)
K
Like the modulus of elasticity £ , Poisson's ratio p is also characte­
ristic of elastic properties of materials. For materials which have
identical elastic properties in all directions, these properties can be
completely characterized by constants £ and p. Such materials are
called isotropic. With sufficient accu­
racy as far as practical application
is concerned, we may consider
steel and other metals, most of the
stones, concrete, rubber and non-
laminate plastics as belonging to
the group of isotropic materials.
In addition to the isotropic mate­
rials, we also have anisotropic ma­
terials, i.e. materials having dissi­
milar properties in different direc­
tions. To this group of materials
belong wood, laminate plastics,
some of the stones, cloth, etc. A
single value of £ and p cannot char­
acterize their elastic properties; it Fig. 8
is essential to have a number of
values of these constants in various directions.
For numerical determination of p, it is essential to measure simulta­
neously the longitudinal and lateral deformation of a bar under tension
or compression. Generally, these deformations are measured in stretch­
ing a specimen in the form of a long and wide plate (metals), or
for a prismatic specimen (stone) under compression.
The values of the coefficient of lateral deformation of various ma­
terials are given in Table 1 for deformations within the clastic limits.
Knowing the value of p, we can calculate the change in the volume
of the specimen under tension or compression. The length of the de­
formed specimen is / ( I +&). The cross-sectional area of the deformed
specimen is A ( 1—ep)2. The volume of the deformed specimen is
Vt = Al (1 + e) ( I — pe)a - V (1 + e) (1 — pe)*
where V is the initial volume.
Since e is a negligibly small quantity up to the limit of proportiona­
lity, we may ignore its square. Then volume V\ becomes
V , - V [ l + » ( 1- 2 |»)]
38 Introduction. Tension and Compression I Part 1

The relative increase in volume (volume strain) is

~ ^ = e(l — 2 ^)

If Poisson’s ratio p=0.5, there is no change in the volume due to

deformation. However, since p<0.5 for a majority of the materials,
tension is accompanied by an increase and compression by a decrease
in the volume. For rubber p«0.5, therefore there is almost no change
in Its volume when it is stretched.
The lateral deformation that accompanies the longitudinal deforma­
tion has great practical significance. More light will be thrown on this
aspect in the succeeding discussion.
Let us consider the following example of applying the methods and
formulas derived above.
Example. A load of Q=4 tf is suspended from bracket ABC, consist­
ing of a wooden rod AC and an iron pull rod AB (Fig. 9). Pull rod AB

has a round section and rod AC a square section. Find diameter d of

rod AB and sides a of the square section of rod AC if the permissible
stress for wood is le . 1=25 kgf/cm2, for steel l<r+l=900 kgf/cm* ([a_]
is the permissible stress under compression, Io+l is the permissible
stress under tension); determine the vertical and horizontal displace­
ments of point A. The length of rod AC is U=l m.
Forces A\ and AT* in rods AB and AC can be determined from the
equilibrium condition of hinge A, at which the given force Q and the
unknown forces N, and are applied.
By plotting the equilibrium triangle for these forces (Fig. 10), we get
AT,= = 2 Q = 8 tf
A = Q cot 30* = Q1/3 = 6.93 tf
Ch. 2] Stress and Strain Within Elastic Limit 39

The required cross-sectional areas of rods AB and AC are

. _ A', 8000__o oQ nrnz
^ ■ [ o f +] == 900 — 8 ^ 9 cm
a — >6930
- ^ 2 077
7 7 cm*a

Diameter of rod AB is
d= *(/ ^ | ^ = 3.34 cm « 3 . 4 cm

Side of the square section of rod AC is

a= V ~A i = Y 277 — 16.6cm « 17 cm
Both the values have been rounded—for the steel rod to the nearest
mm, and for the wooden rod to the nearest cm.
To determine the displacement / of point A, we disconnect the rods
and represent them by their new lengths BA t and CAit Increasing and

decreasing their initial lengths by b li—A A i and &l2= A A s, respective­

ly, without changing their direction (Fig. 11(a)). The new position of
point A can be located by bringing together the deformed rods by rotat­
ing them about points B and C. Points A t and A 2 will move along
arcs A tAa and A 2A 3t which due to their small length may be conside­
red as straight lines, perpendicular to BAi and CA2. The horizontal
displacement of point A will be
/ a s AAt = A/j
40 Introduction. Tension and Compression IP a rt i

and the vertical displacement (Fig. 11 (&))

f | = A sA a = "1"^ 4^3
The segment
AtAA— AAz —AfjSina
and

But
AxA4 = A^Aj -f AtAt = A/t cos a - f A/a
Therefore
A^A 3 = (A/, cos a + A./4) t#?1- A/ 1cosa a —&lt cos «
sin a
Consequently,
/, - /!,/!. + ^ / l , - A/,sin a + ^ c°s‘ sin
° +a41»eosct- *'■+*'! M»«
'* * ■ * '« * 1 sin a

Deformation of the rods is determined by the formulas

a/ Ajlj 0930X 100 <i 4 i rt_i _
A,> = s 3 ; = T w r 7 r = 2 -4 x 1 0 cm
m ox ioox2 = 5.07x I 0 ’ a cm
A' 1i = t-iAi
rr
2 x! 0« v H 4 ^ X ^ T

Hence the horizontal displacement of poipt A is/a= 0.24 mm, and the
vertical displacement is
.• i .i
r __ a\/i + A / s Cos a
0.507rf-0.24X
^
^s—
2 , ___
' I --------ilTTa------- --------- O ---------=5 .43 mm
Total displacement AAa is
f - V T \W i = 1 1.43* + 0.24* = 1.45 mm

CHAPTER 3
Experimental Study of Tension and Compression
in Various Materials and the Basis
of Selecting the Permissible Stresses
§ 10. Tension Test Diagram.
Mechanical Properties of Materials
In the previous chapter, while determining the cross-sectional area
and deformation, we came across a number of quantities which chara­
cterize a material not only within the limit of proportionality (modulus
Ch. 3\ Experimental Study of Tension and Compression 41

of elasticity, limit of proportionality) but also beyond it up to its

complete breakdown (ultimate strength). To have a good idea about
the mechanical properties of materials under tension or compression, it
is essential to study ex peri mentally the phenomena that accompany
these processes.
By the difference in their mechanical properties under simple tension
or compression at room temperature, the materials may be classified
as brittle and ductile. The brittle materials break down under a very
small residual deformation. The failure in case of ductile materials
occurs after a considerable residual deformation. Cast iron, stone and
concrete, are examples of brittle materials. The low-carbon steels and
copper belong to the group of ductile materials.
Let us examine the behaviour of both types of material when subject­
ed to tension till failure. A prismatic specimen of round or rectangular
section is prepared, th e working portion of the specimen is calibrated
in centimetres ch* fractions of centimetre , to be able to ascertain
the change in its length after the experiment. The specimen is placed
on the testing machine and its ends are.clamped. By straining the spe­
cimen axially it is stretched with a load, which increases gradually
without shocks or impacts. A number of successive load values are
applied, and the. corresponding increases in the length / marked on
the specimen are measured.
The experimental results can best be represented in the form of a
tension test diagram; a majority of thetesling machines have an attach­
ment which automatically plots this diagram when the specimen is
stretched. In this diagram, load P is plotted along the vertical axis and
elongation A/ along the horizontal axis.
The tension test diagram for a specimen from ductile material, e.g
low-carbon steel, is of the pattern shown in Fig. 12. The first part of the
diagram up to the point A corresponding to the limit of proportionali­
ty is a straight line. Ordinate OAi is the value of the tensile force that
corresponds to the limit of proportionality i.e. the maximum stress
which, if exceeded, results in deviation from Hooke’s law; for low-
carbon steel op is approximately equal to 2000 kgf'cm*. This stress is
determined from formula (2 . 1) in which the original value -of the cross-
sectional area A is used. This stress is known as conditional stress.
In future, no special mention will be made when the original cross-
sectional area is used. The word conditional will also be dropped.
When the tensile force is increased beyond ordinate O A the defor­
mation starts increasing more rapidly than the force—the diagram takes
a curved shape bulging outwards. Then we notice a sharp change in the
behaviour of the material; at a certain value of the tensile force OCt
the material begins to “flow”. Almost no force is required to further
deform the body. A horizontal (or almost horizontal) plateau is ob­
tained on the diagram. The stress at Which the material starts to flow,
i.e. at which the deformation increases at an almost constant load) is
42 Introduction. Tension and Compression IPari I

called the yield stress ov. For the material under consideration a„ is
approximately equal to 2400 kgf/cm9.
During the flow of metal the Luder's flow lines appear on the surface
of the material more or less distinctly (Fig. 13). These lines are caused
by relative displacement of the material particles when considerable
plastic deformation of the specimen takes place.
After the yielding zone the material again starts resisting further
tensile strain and to elongate it by a length A/ the force should be
increased. Point D of the diagram corresponds to the maximum load.

Fig 12 Fig 13
At this instant there is again a sharp change in the behaviour or
the material. Up to this point, the whole bar was being deformed;
each unit length of the specimen elongated almost equally. Similarly
there was a uniform decrease in the cross-sectional area of the specimen.
From the instant the load achieves the value ODu the. deformation
gets concentrated in a certain part of the specimen. A small portion of
the specimen around this spot is from now on subjected to the maximum
stress. This results in a localized reduction of the cross section, and a
"neck” is formed (Fig. 14).
As a result of the decrease in the cross-sectional area of the deformed
portion, a continuously decreasing load is required to further elongate
the specimen. Finally at a load OKi the specimen breaks down.
If we stop the experiment at a load less than OAx and unload the
specimen, then the relation between the force and deformation will be
represented by the same straight line as during loading up to OA.
The deformation disappears when the force is removed, implying there­
by that the deformation was elastic.
If jye start unloading the specimen from a point Z on the diagram*
Ch. 31 Experimental Study of Tension and Compression 43

lying between points C and £>, then unloading will take place along
line ZOi, which is almost parallel to the line OA. The specimen in
this case will not regain its initial dimensions, segment 0%0t will
represent the elastic deformation, which, as in the previous case, chan­
ges in direct proportion with the load at constant modulus of elasticity.
Segment 0 0 x will represent the residual deformation and segment OOa
the total deformation at a load OZ\. We may find a load OBx below
which only elastic deformations occur. The corresponding point B
on the diagram usually lies a little above but very close to point A,
which represents the limit of proportionality. The stress which if
exceeded results in very small (of
the order of 0.001-0.03%) residual
deformations is called the elastic li­
mit <v On the tension lest diagram
(Fig. 12) the load causing this stress
is represented by the ordinate OB,.
Points A and B are so close to each
other that generally the limit of pro­
portionality and the elastic limit
are considered to be the same. There­
fore although it is commonly said
that a material follows Hooke’s law
lill it reaches the elastic limit, it
would be more precise to say till it
reaches the limit of proportionality.
The maximum tensile force stretch­ Fig. 14
ing the specimen is represented
by the ordinate OD,; it is commonly referred to as the crushing load,
because it is essential to apply this load for rupture to begin; the
ultimate breakdown occurs at a load represented on the diagram by the
ordinate of point K. The stress caused by the maximum load is called
the ultimate strength or ultimate resistance o„. The ultimate strength,
obtained as a ratio of the maximum load to the initial cross-sectional
area of the specimen, characterizes the force required to crush the spe­
cimen of the given material under tension; for low-carbon steel it
reaches 4000 kgf'cm3.
While studying the tension test diagram, we marked on it a number
of ordinates representing loads connected with various mechanical
properties of the material. Table 2 contains a summary of these loads
and their corresponding characteristics (stresses) with*lhelr notations.
Any of the required stress can be obtained by dividingthecorresponding
load by the initial cross-sectional area of the specimen.
All the mechanical properties (limit of proportionality and elastic
limit, yield stress, and ultimate strength) characterize the ability of
a material to resist the tensile forces tending to deform and crush a
specimen .made from it.
44 Introduction. Tendon end Compression [Part I

Table 2
Mechanical Properties of Materials

C orresponding stress and

Load it s n o ta tio n

Load corresponding to the end of straight line Limit of proportionality <Jp

OAi
Load corresponding to the beginning of residual Elastic lim it at
deformations OBi
Load corresponding to the flow of the material Yield stress oJt
(increase in deformation at constant load OCj)
Maximum toad ODt Ultimate strength or ultimate
resistance au

The jc-coord mates of the diagram characterize another property of

the material, namely, the ability to deform to a certain degree before
breaking down.
Segment 0 sOi (Fig. 12) gives the value of elastic deformation at the
time of breakdown, which disappears as soon as the breakdown occurs.
Its length O0n=A/, is the residual deformation of the specimen of
length I after its breakdown. The greater the measured length of the
specimen and the greater the pliability of the material, tne greater
will be this residual deformation.
The ratio of elongation A/ 0 to the initial length I is a measure of the
plasticity of the material, i.e. its ability to undergo considerable
deformations before breaking down.
This ratio expressed in per cents is denoted by 6 and is called the
residual relative elongation of the specimen after breakdown and for the
commonly used grades of steel varies from 8 to 28%. Thus,
8 = ^ 1 x 100

It must be noted that the residual relative elongation of the specimen

depends to a large extent upon its shape and chiefly upon the ratio of
its length to its cross-sectional area. Therefore, in’laboratory experi­
ments the residual elongation after breakdown is not measured over
the full length of the specimen, but only over its certain part called the
reduced length. In round specimens the reduced length is generally taken
equal to 1(5/; sometimes it is taken as 5d. In rectangular specimens the
reduced length is selected in such a way that the ratio of length and
cross-sectional area of a round specimen having the same cross-sectional
area A as the rectangular specimen remains the same. For example,
the reduced length of a rectangular sped men corresponding to the length
lOd of a round specimen should be 11.3^ A. The specimens ar^prepared
Ch. S I, Experimental Study of Tension and Compression 45

such that the length between the heads somewhat exceeds the estima­
ted length.
The pliability of a material under tension or compression can also be
ascertained by another quantity called the permanent relative reduction
(of area). After the maximum load is reached, a “neck” starts forming
in a particular section of the bar, and
at the place of failure the cross-sec­
tional area of the specimen is gene­
rally less than its initial value (Fig.
14). Let us denote the initial cross-
sectional area by A 0t and the area of
the section at which the specimen
breaks down by Ail the quantity
y^A < rpAlx 100

(in per cents) is called the relative

reduction after breakdown. The
greater is this quantity, the more
pliable is the material.
Finally, the tension test diag­
ram shown in Fig. 12 enables us to
study one more mechanical property of materials related to their re­
sistance to impact loading.* The greater is the amount of work required
to break the specimen, the higher is its resistance to impact load­
ing. Therefore, the amount of work done in stretching the specimen
up to the elastic limit or crushing point may be taken as a characteris­
tic of the resistance of material to suddenly applied loads. This work
is represented by the area of the tension test diagram (Fig. 12).
Let us consider the part of the diagram which is within the limits of
applicability of Hooke’s law' (Fig. 15). When the specimen fixed atone
end is stretched by applying a gradually increasing force P at the other
end, the displacement of this end is equal to the gradually increasing
elongation A /= |^ ; this relation is expressed by the straight line OB.
An elongation A/ (segment 0Bt in Fig. 15) corresponds to a particu­
lar value of the force P (segment BtBJ. If we increase the force by dP,
the elongation increases by dA/ and the tensile force having an average
value of P + ifdP will perform the w'ork

d-W = (/> + j d P JdAi = P d A /+ y dP dM

0 For greater details see § 179.

46 Introduction. Tension and Compression [Part I

Neglecting the second-order term - ^dPd^l , we get

dW = P d \l
Graphically work dW is expressed by the area of the shaded rectangle
of height P and base dA/.
Considering the gradual increase of force P as a number of successive
elementary additions of loads dP, we find that the work done by the
external forces in gradually stretching the specimen is the sum of the
areas of the elementary rectangles (Fig. 15). When the load P is in­
creased continuously, i.e. dP and dA/ are infinitesimal quantities, for
particular values of P and A/, thissum may be obtained as the area
of triangle OBtB* equal to
^B & O B .^^P M

Thus, the work performed in elastic deformation of a bar by AI

may be expressed by the formula
W=-?PM (3.1)

and graphically represented by the corresponding part of the tension

test diagram.
The same argument holds good for the whole of the tension test diag­
ram (Fig. 12). The area of the digram represents the total work W,
expended in breaking a specimen of length I and cross-sectional area A.
To obtain the quantity which is a characteristic of a material and
not the specimen, we divide work W by the volume of the specimen.
The ratio is called the specific work of elastic deformation under
tension.
Similarly, we may determine the total specific work rifi
this
is the work, required to break the specimen. The greater this quantity,
the more reliably the material withstands shock and suddenly applied
loads.
We have seen above that the specimen material continues to experi­
ence elastic deformation in accordance with Hooke’s law even after the
yield stress has been passed; in this case are added residual deforma­
tions. This is observed while unloading the specimen after loading it
beyond the yield stress (point Z on the tension test diagram in Fig. 12).
If we now start stretching the specimen after unloading it, the
loading diagram will be represented by almost the same unloading line
0\Z parallel to OA, and beyond point Z, by the same curve ZDK as
prior to unloading. Hence, if we compare tension test diagram OCZDK
of a specimen not experiencing unloading with diagram 0%ZDK of a
Ch. 3\ Experimental Study of Tension and Compression 47

specimen of the same material, which has been preliminarily loaded up

to point Z and ihen loaded back to point 0 U we see that the limit of
proportionality increases lo reach the stress up to which the specimen
has been preliminarily loaded, whereas the plastic deformation decrea­
ses by 0 0 ], i.e. by the residual deformation incurred during prelimi­
nary loading.
This increase in the limit of proportionality and decrease in plastic
deformation due to preliminary loading beyond yield stress and sub­
sequent unloading is called cold hardening. Under cold hardening, cor­
responding portion of the tension test diagram is, so to say, cut off,
resulting in a decrease in the total specific work wt. In facl, cold har­
dening is much more complex than the simple process by which it has
been explained here. In particular, if the specimen is allowed to “rest”
and reloaded only one-two hours after unloading, the corresponding
part ZDK of the tension test diagram passes a little higher than with
the absence of “rest".

§ 11. Stress-strain Diagram

The tension test diagram shown in Fig. 12 illustrates the behaviour
of a material for a specimen of the given dimensions; therefore, to get
a curve characteristic of the behaviour of the material irrespeciive of
the dimensions of the specimen, the tension test diagram is slightly
modified.
The ordinates of the curve in Fig. 12 depicting loads are divided by
the initial (before the start of experiment) cross-sectional area of the
specimen /lo, and the abscissas A/ are divided by the estimated length /.
Then in the new diagram we plot along the vertical axis

and along the horizontal axis

Such a diagram, shown in Fig. 16, is called the stress-strain diagram

for the given material under tension. It is similar to the tension test
diagram in Fig. 12. In this diagram all the stresses that characterize
the mechanical properties of the material are marked: limit of propor­
tionality o;j, yield stress o,„ and ultimate strength o„.
If we consider a portion of the diagram OA, up to the limit of pro­
portionality, then for a certain stress or and its corresponding relative
elongation e, the area of triangle OAB (Fig. 17) equal to will repre­
sent the specific work in stretching the material to stress a. We knosv
48 Introduction. Tension and Compression 1Part I

that
at _ PM
~5~2Al
W
Knowing that e= -^, one may write down the expression for specific
work of deformation within the elastic limits as follows:
es tfl (3 .2 )
- ir ­
T 2E
By analogy, the total area of the diagram show in Fig. 16 represents
the total specific work wt at the moment of breakdown of a specimen of
the given material. This quantity may be expressed as the product of

Fig. 16 Fig. 17

the length 6 by the maximum ordinate au and a coefficient n which

represents the ratio of the area of the diagram to that of a rectangle
having sides 6 and ou:
te>f=»iorB8 (3 .3 )
Thus, the total specific work at rupture depends to a certain degree
upon the product of the ultimate strength and the strain after the
rupture. Therefore, very often, the ability of a material to Withstand
shocks is judged by the product <ju6 .
From the diagram in Fig. 17, it is evident that
ta n a = —
8
= £
Hence, graphically the modulus of elasticity £ is represented by the
slope of the straight portion of the diagram.

§ 12. True Stress-strain Diagram

The stress-strain diagram for tension shown in Fig. 16 may be
considered as characterizing the properties of the given material under
tension.
Or. 5] Experimental Study of Tension and Compression 49

However, this diagram is only a conditional characteristic of the

mechanical properties of the material. In the initial stages of the test,
the cross-sectional area of the specimen almost remains constant, but
beginning from the yield stress a noticeable reduction takes place,
which is initially uniform over the entire length of the specimen, and
after crossing the ultimate strength it becomes localized. Therefore,
bevond the ultimate strength the ordinates of thecurve shown in Fig. 16
represent conditional stresses calculated for the initial cross-sectional
area and not the real one.
Similarly, until the ultimate strength is reached the abscissas in
Fig. 16 depend only upon the ability of material to elongate. However,
once the neck is formed, the relative elongation also becomes dependent
upon the dimensions of the specimen (its length and diameter) and
thus is no more a characteristic of the material only. Therefore, to
obtain a more precise diagram characterizing the properties of the ma­
terial, the true stress-strain diagram is plotted. It illustrates the rela­
tion between stress and strain in the section of rupture.
To plot the true stress-strain diagram it is essential to register the
tensile force at various moments and at the same lime measure the
cross-sectional area of the specimen in the narrowest place.
Let the true stress be denoted by a and the true cross-sectional area
in the narrowest section by A, then
JP (3.4)
a
A

When deformation is large, the original length of the specimen also

changes considerably. Consequently, the true elongation e must be
related to the actual length of the bar at the given instant of test and
may be calculated by the formula

(3.5)

where U is the original length of the specimen, and h its length at

the time of measurement. When true elongation is large in magnitude,
it is denoted by e instead of e.
Let us establish the relationship between true and conditional strains
and true and conditional stresses.
When the specimen deforms uniformly along its length

Finelly
e«=ln(l + e ) (3.6)
3—3810
50 Introduction, Tendon and Compression IPart I

where
At
e
to
is the conditional strain.
Formula (3.6) cannot be used in case of non-uniform deformation
because it is difficult to measure A/ for computing e.
It is known that the specimen volume does not change under non-
uniform deformation beginning from the moment of neck formation.
This is known as the law of constancy of volume and may be expressed as
AJ0= X l
where Ao is the original cross-sectional area. It ensues that
A>*. = ( 4 , - A / l ) ( / 0 + A/)
after dividing by AM
i A 0~—AA
Ao TT
or
(1 — ^>)(1 + e ) « l , where '• i
I
wherefrom

Upon substituting the last expression in formula (3.6), we finally ob­

tain
« = ln r = « (3.7)

It should be noted that ij? is determined in the narrowest part of the

neck.
In order to obtain the relationship between true and conditional
stresses it should be recalled that
P = cA0= aA
where cr is conditional stress, i.e. stress related to the original cross-
sectional area. Further,

and
Ch. 3] Experimental Study of Tension and Compression 51

Considering the relationship between e and t|\ earlier obtained for

conditions of uniform deformation, we obtain
a = c ( l+ e ) (3.8)
Under conditions of non-uniform deformation, beginning from the
moment of neck formation, true stress a is found directly from formula
(3 .4 ) as it is meaningless to determine conditional stresses in this state
of the specimen because of the large difference between A and A o.

The true stress-strain diagram is shown in Fig. 18. However, for

practical use this diagram is somewhat simplified. It is considered that
Oyf&Oy and a small portion of the curve just preceding rupture is
ignored. The diagram is then plotted as shown in Fig. 19.
p9 _
Yield stress 0 ^ = -^ . The true ultimate strength au is calculated from
the formula (3.8).
The true rupture stress is found from formula (3.4), i.e.

The true uniform elongation is determined from formula (3.6) i.e.

eu= ln (l+ e ), where e is the conditional strain at the moment when the
neck begins to form.
Finally, the toted true rupture strain is found from formula (3.7),
where t|> is computed for the cross section of rupture:

e tu p = i

It is evident from the diagram of Figs. 18 and 19 that the stress o

increases right up to the moment of rupture, rapidly at first but com­
paratively slowly after the maximum (stress oB) is reached. At the
3*
52 Introduction. Tension and Compression [Part I

moment of rupture the stress corresponding to the actual cross-sec­

tional area is more than the ultimate strength obtained by theconven-
tional method.
However, it would be erroneous to use the latter value for calculat­
ing the maximum load which the bar can withstand before breaking
down, which is very important from the practical point of view. This
is clear from the tension test diagram in Fig. 12. The maximum load
that the specimen withstands corresponds not to the moment of break­
down but to an earlier moment—the magnitude of this load is charac­
terized by the ultimate strength for the specimen of a given cross-
sectional area. The actual stress increase in this case is due to the
sharp reduction in the working cross-sectionai area of the specimen,
i.e. due to its rupture.
We may set a number of mechanical properties using the true stress-
strain diagram. They were enumerated (marked by italics) when the
plotting of true stress-strain tension test diagram was explained.
The ordinates of the true stress-strain diagram show the ability of
material to resist plastic deformation.
To increase the plastic (residual) deformation, wc must subject the
material to a continuously increasing stress; the greater the plastic
deformation of the material, the greater is its resistance to such a tie-
formation. This is known as strengthening. The ability of a material
to strengthening is judged by the steepness of the true stress-strain dia­
gram, i.e. by tan a.
The difference of true total and uniform elongation is characteristic
of the ability of material to deform locally (at the neck) and is known
as local elongation.

§ 13. Stress-strain Diagram for Ductile and Brittle Materials

In the preceding sections, we have discussed the physical aspect of
the process in which a specimen of ductile material, such as low-
carbon steel, is subjected to tension. Stress-strain diagrams similar
to the one shown in Fig. 16 are obtained for other ductile materials
capable of plastic deformation.
Some (special) grades of steel, copper and bronze do not have Ihe
yielding zone. There is a smooth transition of the straight-line portion
of the diagram into the curved portion. As an example, the stress-strain
diagrams for cast steel (a), bronze (b), nickel sleel (c), and manganese
steel (d) are shown in Fig. 20 .
For the materials which do not have a yielding zone, the yield stress
is conditionally taken as the stress for which the residual deformation
is the same as with a yielding zone. The residual relative elongation in
this case is usually taken as 0 .2 %.
Brittle materials are characterized by the breakdown even at small
deformations. When a specimen from a typical brittle material, such
Ch. 3\ Experimental Study of Tension and Compression 53

as cast iron, is stretched, inconsiderable deformation is observed right

up to the moment of rupture. The specimen breaks down suddenly.
The relative elongation and relative reduction in area are found to be

very small. The stress-strain diagram of cast iron under tension is

given in Fig. 21. It should be noted that in Fig. 21 the horizontal
scale of the diagram is approximately 40 times more, and the vertical
scale is approximately 6 times more
than the corresponding scales in
Fig. 20.
As a rule, brittle materials have
poor resistance to tension; their ul­
timate strength is less than that of
the ductile materials.
The relation between stress and
strain when stretching brittle ma­
terials does not concur well with
Hooke’s law; even at low stresses
we get a slightly curved line instead
of the straight line on the diag­
ram, i.e. a strictly linear propor­
tionality between the force or stress
and the corresponding deformation
is absent.
Therefore, the modulus of elasticity £ , which is equal to the
slope of the diagram (see § 11) cannot be considered a constant quan­
tity for brittle materials; it changes depending upon the stress for
54 Introduction. Tension and Compression [Part /

which the deformation is to be calculated. As the stress increases, the

modulus oi elasticity increases or decreases depending upon the direc­
tion in which the curve is bulging—upwards or downwards.
However, the deviation from Hooke’s law is insignificant for the
stress range in which the materials generally function in structures.
Therefore, in practice, the curved portion of the diagram (Fig. 22) is
replaced by the corresponding chord, and the modulus of elasticity £
is considered constant. This is permissible, the more so because for
different specimens the mechanical properties of brittle materials
change in a greater range than those of ductile materials; hence, there
is no sense in using a very accurate expression for the relation between
stress and strain.

§ 14. Rupture In Compression of Ductile and Brittle Materials.

Compression Test Diagram

Specimens in the shape of a cube or a cylinder whose height is just a

little more than its diameter are used in studying the strength of ma­
terials under compression. In longer specimens it is difficult to avoid
bending.
The size of the specimens varies for different materials and fluctuates
(for the cube edge) from 2 cm (wood) to 20-30 cm (concrete).
Under compression at stresses below the limit of proportionality or
yield stress, a specimen from ductile
material behaves as under tension.
The limit of proportionality (as also
the yield stress for steel) and the
modulus of elasticity are almost
equal under tension and compression
for ductile materials.
After passing the limit of pro­
portionality, noticeable residual de­
formations appear resulting in a
shortening of the specimen and an
increase in its diameter. The lateral
deformations of the specimen at the
ends are hindered due to friction be­
tween the faces of the specimen and
Fig. 23 the bearing plates of the press; the
specimen acquires the shape of a
barrel (Fig. 23).
As the cross-sectional area of Ihe specimen increases, it requires a
greater force for further deformation: the specimen continues to com­
press and ultimately becomes oblate. The stress which may be said to
be analogous to ultimate strength in tension is not observed.
Ch. 5} Experimental Study of Tension and Compression 65

A typical stress-strain compression diagram for a ductile material

(low-carbon steel) is shown in Fig. 24. As under tension, cold harden­
ing takes place under compression too.

As under tension, the brittle materials, such as stone, cast iron,

and concrete, fail after a small deformation under compression. Figure
25 shows the stress-strain diagram of a stone specimen under compres­
sion (a granite cube I0 X 10 X 10 cm). Figure 26 shows the stress-strain
diagram for a cast iron specimen under compression. Here also it

Fig. 27 Fig. 28

should be noted that the scales of diagrams in Figs. 25 and 26, especial­
ly the horizontal ones, are much larger than the scale of the diagram
in Fig. 24.
50 Introduction. Tension and Compression I Part I

The nature of rupture in a stone specimen is shown in Fig. 27; the

crushed specimen represents truncated pyramids joined by their
smaller bases. This form of rupture is due to the friction force between
the specimen and the bearing plates of the press. If we remove this
friction, for example, by greasing the specimen faces with paraffin,
the nature of rupture will be different: the stone will break into parts
with cracks running parallel to the direction of the compressive force
(Fig. 28). The crushing load for such a cube will be less than for a cube
tested by the common method, without greasing. Therefore, the ulti­
mate strength in compression is to a considerable extent a conditional

Fig. 29 Fig. 30

characteristic of the strength of material. This must be taken into

consideration when fixing the safety factor.
It has been observed that when a prismatic specimen made of stone
or concrete is compressed slowly, the rupture starts with the appearance
of lengthwise cracks parallel to the direction of the force. Therefore,
we may say that the material of the specimen under compression rup­
tures apparently due to the failure of certain portions.
The nature of rupture for cast iron is close to that observed in case of
stone. Figure 29 shows a cylindrical cast-iron specimen crushed by
axial compression. It must be noted that the resistance of brittle ma­
terials to compression is much greater than their resistance to tension.
Compression of a timber specimen gives sharply differing results
depending upon the direction of compression with respect to the fibres;
timber is an anisotropic material, i.e. it has different properties in
different directions. The ultimate strength of timber compressed along
the fibres is about 10 times more than when it is compressed across the
fibres, whereas the deformation is much less. Figure 30 shows the com­
pression test diagramfor a timber cube tested along and across the fibres.
Table 3 contains data on ultimate strength under tension and compres­
sion for most important materials.
 [Purl I // Ch 2fi ] Design for Ppntii^ihlc Lnadi ■ih'i
III,'ll stress In statically indeterminate beams the formation of one ductile hin.ee
•ached. In is not enough for full utilization of their bending capacity; it is essen­
itral axis; tial that at least one more ductile hinge be formed. We shall explain
1 rough the tins with the help of an example.

* L *1 7
—
h,
.tiu

sect inn by
sum ol the

c(|ii;h por-
mi brought
ressed ;ind
■refore, the
lie sect inn.

e see that

Let ns consider a two-span continuous beam of uniform section

i\\er halves
(F:g. 'i77(u)). Its bending moment diagram for work within the elastic
i condition limits (Fig. .377(/>)) is the difference of the bending moment diagrams
loi lorce P ;md support moment M , - Graphic subtr.aetion ol
(L’fi.fj)
the dmgrams is shown In dotted lines. The resultant bending moment
diagram is hatched. The maximum stressed sections are the section ol
presence of
application of the load with a moment •>n
the middle support with moment .VI, — PI I.:
(TTPL When the load
is increased, stresses in the beam become equal to the \ ield stress n,,
first of all in the top and bottom Livers of the section under load I\,
nut ion of a and limy be expressed b\ the relation
\ dctiniii- l /y (T wherefrom I
m i ll,,.,
a I if
58 Introduction. Tension and Compression [Part /

The amount of work required to crush ductile materials is greater

than that required for brittle materials. Therefore, ductile mate­
rials are more suitable for structures designed to absorb the maximum
possible kinetic energy of impact without failure.
The brittle materials fail easily under impacts just because their
specific work of deformation is very small. Due to their small deforma­
tion up to stresses close to the ultimate strength, the same brittle mate­
rials are sometimes capable of bearing far greater stresses than the duc­
tile materials provided deformation is under the action of a placid,
gradually increasing compressive force.
The second distinguishing feature between these materials is that
in the initial stages of deformation, the ductile materials may be consi­
dered to behave identically under tension and compression. The re­
sistance of an overwhelming majority of the brittle materials to tension
is considerably lower than their resistance to compression. This restricts
the field of application of brittle materials or requires that special mea­
sures be taken to ensure their safe working under tension as, for exam­
ple, in reinforcement of concrete elements, working under tension,
with steel.
A sharp difference is observed in the behaviour of ductile and brittle
materials with respect to the so-called local stresses, which are distribut­
ed over a comparatively small portion of the cross section of the ele­
ment but the magnitude of which exceeds the average or nominal
value, calculated from common formulas. Local stresses will be discus­
sed in detail in § 186.
Since we do not observe any considerable deformation in brittle
materials almost up to the moment of failure, the non-uniform stress
distribution shown above remains unchanged under tension as well as
compression right until the ultimate strength is reached. Due to this, a
weakened bar of brittle materia! with local stresses will fail or crack
p
at a much lower value of the average normal stress o=-^- as compared
to a similar bar without local stresses. Thus, we may say that local
stresses greatly reduce the strength cf brittle materials.
The ductile materials are affected by local stresses to a much lower
degree. The role of ductility as regards local stresses is to level them
to some extent. The mechanism behind this levelling will be discussed
in Chapter 31.
We have given a very simplified picture of the working of a bar wilh
a non-uniform distribution of stresses. Actually, levelling out of stres­
ses is hindered not only by strain hardening, but also by the change in
the stressed state at the location of stress concentration, its transition
from a linear stressed state to a three-dimensional stressed state. This
compound stressed state will be discussed later in Chapter 6 .
There is one more factor which stipulates the selection of one or the
other type of material for practical purposes. Often, while assembling a
Cti, 5] Experimental Study of Tension and Compression 59

structure, it is necessary to bend or to straighten a bent element.

Since the brittle materials are capable of withstanding only very small
deformations, such operations on them usually give rise to cracks. The
ductile materials, capable of taking considerable deformations without
rupture, can be bent and straightened without any dufficulty.
Thus, brittle materials have poor resistance to tension and impacts,
are very sensitive to local stresses and cannot bear change in the
shape of elements made from them.
The ductile materials are free from these drawbacks; therefore ducti­
lity is one of the most important and desirable property in materials.
The points in favour of brittle materials are that they are usually
cheaper and often have a high ultimate strength under compression;
this property may be utilized for work under placid loading.
Thus, we "see that ductile and brittle materials have exceedingly
different and contrasting properties as far as their strength undo*
tension and compression is concerned. However, this difference in pro­
perties is only relative. A brittle material may acquire the properties
of a ductile material, and vice versa. Both brittleness and ductility
depend upon the treatment of the material, stressed state and tempera­
ture. Stone, which is conventionally a brittle material under compres­
sion, may be made to deform like a ductile material; in some experi­
ments this was achieved by pressing a cylindrical specimen not only at
its faces but also on its side surface. On the other hand, mild steel,
conventionally a ductile material, may under certain conditions,
e.g. low temperature, behave exactly like a brittle material.
Hence the properties “brittleness" and “ductility", which we assign
to a material on the basis of compression and tension tests, are related
to the materials behaviour only at ordinary temperatures and for the
given kinds of deformation. In general, a brittle material may change
into a ductile material, and vice versa. Hence it would be more precise
to speak not of “brittle” and “ductile” materials but of brittle and pla­
stic states of materials.
It must be noted that a comparatively small increase in the ductility
of a brittle material (even up to 2% relative elongation before break­
down) enables its use in a number of cases which are otherwise preclud­
ed for brittle materials (in machine parts). Therefore, research work
on improving the ductility of brittle materials such as concrete and
cast iron demands the maximum possible attention.

§ 16. Considerations .in Selection of Safety Factor

A. In the preceding sections, we discussed the methods of computing
stresses, determining the mechanical properties of materials under ten­
sion and compression, and gave recommendations for selecting one or
the other type of materials (ductile or brittle) depending upon the work­
ing conditions.
60 Introduction. Tension and Compression [Part I

However, the information given till now is not sufficient to find out
the permissible stresses suitable for different types of loading. The va­
lues of all the mechanical properties of materials (ultimate strength,
relative elongation, limit of proportionality, etc.) are obtained from
laboratory experiments under static loading, i.e. when the load increa­
ses gradually without impacts, shocks and change of sign. Similarly,
the formulas correlating normal stress a with the tensile or compressive
force P have been derived for static loading. It was assumed that the
external forces and stresses acting on the cut-off portion of the bar
balance each other. In practice, however, we often come across dynamic
and systematically changing loads.
As compared to the static load, the suddenly applied load has a two­
fold effect; on the one hand, the brittle and ductile materials react
differently to the dynamic action of the load and, on the other hand, the
stresses are also different. This problem will be discussed in greater
details in the chapters on dynamic loading. Here we shall pay atten­
tion only to the fact that stresses are generally higher under a dynamic
load than under a static load of the same magnitude. This statement is
confirmed by experimental results and may also be proved theoreti­
cally, as has been done in Part IX.
The ratio of stress <r«j due to dynamic action of the load to stress a
due to static action of the same load is called the coefficient of dynamic
response and denoted by Ka-

The coefficient of dynamic response depends upon the type of dyna­

mic loading and has a very large value in a number of cases.
B. The strength of materials under loads systematically changing their
magnitude or magnitude and sign is much different from their strength
under static and impact loads.
If, for example, we alternately subject a steel bar to a large number of
tensions and compressions, we shall observe that after a definite number
of such changes in stresses, the bar in some cases cracks and then rup­
tures at a stress considerably lower than its ultimate strength. Even for
plastic materials the plastic deformation of the specimen before break­
down under similar loading is very small: a brittle fracture takes
place.
The failure of materials under a variable load at stresses lower than
the ultimate strength is called fatigue. This name does not reflect the
physical nature of the phenomenon, but it has become such a customary
term that it is used to this day.
Experiments show that under alternative tension or compression a
decrease in the acting force results in an increase in the number of
alterations of this force required to break the specimen. Each material
has a maximum normal stress cr at which the specimen can withstand
Ch. 3\ Experimental Studs/ of Tension and Compression. 01

practically an unlimited number of alterations of the force without

breaking down. This stress in denoted by orc and is called the endurance
limit or the fatisuc limit. The element will not fait until stresses in it
do not exceed Inis limit, irrespective of the number of alterations of
the stresses.
Thus, in systematically varying loads, it is essential to specify
another mechanical property of the material, namely, endurance
limit; it determines the resistance of the material to alternating stres­
ses. The fatigue of materials for various types of loads will be studied
in greater detail in Chapter 31.
All that has been stated above must be taken into account when
selecting the permissible stresses in tension or compression or, which is
the same, when determining the safety factor k from the formula
(see §§ 4 and 7)
[«] (39)

The safety factor should be. so selected that the normal stresses acting
on the whole section do not exceed the elastic limit (or yield stress)
of the material, otherwise the bar will get plastically deformed; under
a varying load the normal stresses should not exceed the endurance li­
mit, which is usually lower than the yield stress.
It should be taken into consideration that the stresses are generally
higher under impact loading. Since the stresses in this case are also
usually determined by assuming the load to be static, the dynamic
action of the load must be accounted for by a corresponding increase in
the safety factor.
C. As far as Ihe local stresses are concerned (see § 15), it is possible to
reconcile to their exceeding the elastic limit or yield stress in the case
of ductile materials provided the alternating load is absent. In this
case plastic deformation occurs over an extremely small portion of the
section and does not affect the working of the construction. Due to
plastic deformations the local stresses stop increasing and partially
approach the normal stressses in the remaining portion of the section.
The brittle materials do not have this property (see § 15): in their case
a higher safety factor has to be taken, the more so because their strength
under impact loading is lower than that of ductile materials.
Under an alternating load, when we have to reckon with the possibili­
ty of developing cracks due to fatigue, it is very essential to lake into
consideration the local stresses, which seriously affect the selection of
safety factor of ductile materials. For the crack due to fatigue to appear,
the actual stresses in a particular section must exceed the endurance
limit. Since the local stresses are greater than the stresses elsewhere
(acting over a larger portion of the section), the chances of the crack
appearing are due to namely the local stresses exceeding the endurance
limit. As the dimensions of the section are computed from considera-
62 Introduction. Tension and Compression I P art /

tions of the maximum general stresses from the formula

< w = !< M = T
the safety factor selected for the general permissible stresses should
ensure that the local stresses do not exceed the endurance limit. This
requires considerable increase of the safety factor k as compared to
its value under static loading.
In the case of ductile materials, when the endurance limit exceeds
the yield stress, the local stresses may be ignored as yielding reduces
the possibility of their spreading, playing the role of a buffer.
For brittle materials, which do not have a yield plateau, the danger
of fatigue cracks appearing under variable loading is more pronounced,
and this requires that the corresponding safety must be increased in
comparison with that under static loading.
Thus, since the choice of the safety factor depends upon the proper­
ties of the material and the method of applying the external forces, its
value is generally greater for brittle materials than for ductile ones;
similarly, a higher value of the safety factor has to be taken for dynamic
and varying loads as compared to static loads.
D. A number of other factors have to be taken into account when
selecting the permissible stresses. The magnitudes of forces required
for computations are not known exactly; the mechanical properties of
materials frequently deviate considerably from their known values;
the methods of computation and our knowledge of the interaction be­
tween different parts of structures are usually simplified and approxi­
mate. The safety factor must cover all these unavoidable inaccuracies
of computation and design.
The less the homogeneity of material, the poorer is our knowledge of
the forces acting on it, the more simplified is our presentation of the
interaction between various elements of the structure, and the greater
has to be the safety factor. In operation, machine parts wear out; there­
fore, in a number of cases a “wear factor” has to be provided for.
Similarly corrosion and rotting have to be taken into account in the
design of metal and wooden structures.
On the other hand, in certain machines (aeroplanes), the safety factor
has to be reduced to the lower possible value to ensure minimum weight.
Hence, proper selection of the permissible stresses is a highly compli­
cated problem, connected with the method of computation, investiga­
tion of properties of the material and a large number of other conside­
rations including economic ones. A particular value of the permissible
stress determines the consumption of the given material and ways of
its use in the structure; this value determines the life of the structure
and the field of application of the various materials.
In a large number of structures the standard values of permissible
stresses are defined by the standards, and the engineer should only be
Ch. 3] Experimental Study of Tension and Compression 63

able to properly apply them. However, In exceptional cases, say, for

example, during war time, the engineer has to abandon the standard
values; he may then follow the general considerations, laid down in
this section and in Chapter 31.
E. Summing up all that has been stated above, we may formulate the
following main points.
The safety factor should be selected in a way so as to provide a
definite reserve against the appearance of the so-called critical stale
of the material, which may endanger the working of the machine.
Under static or impact dynamic loading, this slate is characterized
in ductile materials by the appearance of large plastic deformations
(yielding), and in brittle materials by the appearance of cracks pre­
ceding ultimate failure. Under repeatedly varying loads the critical
state of material is characterized by the appearance and development
of fatigue cracks. We shall denote the stress corresponding to the
start of critical state by ou. This stress is
cf„ (yield stress) when the ductile material
begins to yield
o_ <*/i (ultimate strength) when the brittle
0 ~ material ruptures (cracks)
ae (endurance limit) when the fatigue
crack appears
Hence, formula (3.9) may now be written more precisely and replaced
by the three formulas depending upon the nature of critical state:

where k y, ku and k e are the corresponding safety factors. The three

formulas may be generalised in the form

w-%
Here o° implies either av, or cru, or ae, and k implies either k,Jt or kut
or k e.
However, formula (3 9) still retains its practical importance. As
the yield stress and endurance limit are to a certain extent related to
the iiltimate strength, the safety factor for all the critical states may
be expressed in terms of the ultimate strength.
These in general are the basis considerations essential for properly
evaluating the permissible stresses.
F. Passing over to the considerations in assigning the value of the sa­
fety factor, we shall give some very brief instructions. The non-homo­
geneity of the material, inaccuracy in force determination, error of
computation, i.e. the common factors are accounted for by the main
64 Introduction, Tension and Compression [Part I

safety factor ka. For ductile materials (steel) it is taken as k,,—k0=*

= 1.4-1.6 , for brittle materials and wood ku= k 9 —2.5-3. Other factors,
such as the dynamic nature of the forces, alteration of their action
and the effect of local stresses are taken accounted for by additional
coefficients, by which the main safety factor is multiplied.
It should be borne in mind that the permissible stress lol obtained
according to formula (3 .9 ) should be compared to the actual stresses in
the part of the structure without considering the dynamic action of
the force and other additional factors.
If only the general points are considered while assigning the safety
factor, i.e. the overall safety factor is taken equal to the main safety
factor, the dynamic action of forces and the local stresses are taken
account of, as far as possible, in the value of the actual stress, multip­
lying the main stress under static loading by the coefficients of dvnamic
loading and stress concentrations. It is not difficult to see tnat the
results in both the cases will be identical.
Table 4 contains approximate values of the overall safety factor
with respect to the ultimate strength for various types of materials
and loads including the factors accounting for the dynamic nature
of loading and local stresses.
Table 4
Safety Factors

Load ini; Type of material

Static toad /D uctile 2.4-2.6

\ Brit tie 3.0-9.0
Impact load Ductile 2.8-5.0
Varying load (tension and compression of Ductile (steel) 5.0-15.0
equal magnitude)

The table is only of a tentative nature; it gives an idea about the

change in the safety factor depending upon circumstances. Numerous
aspects affecting the safety factor under impact and alternating loads
will be discussed in greater details in Chapters 30 and 31.

§ 17. Permissible Stresses Under Tension

and Compression for Various Materials
In the preceding section we tried to elucidate the numerous factors
which affect the safety factor and consequently the value of permissible
stress. Tn Table 5 are given the tentative values of permissible stresses
under tension and compression for some important materials used in
engineering and machine building. The table has been compiled on the
basis of present Soviet standards.
Ch. 3\ Experimental Study of Tension and Compression 65

Table 5
Tentative Values of the Permissible Stresses for Some Commonly Used
Materials

Permissible stress
(kgf/cm1)
Material
... . Under com-
Under tension presslon

Gray cast iron 280-800 1200-1500

Low-carbon steel IGOO-2000 —
Structural carbon steel used in machine building 600-2500
Structural alloy steel used in machine building'' 1000*4000 and higher
Copper 300-1200
Brass 700-1400
Bronze 600-1200
Aluminium 300-800
Aluminium bronze 800-1200
Duraluminium 800-1500
Textolite 300-400
Laminated Bakelite insulation 500-700
Bake!ite impregnated veneer 400-500
Pine along the fibres 70-100 100-120
Pine across the fibres — 15-20
Oak along the fibres 90-130 130-150
Oak across the fibres — 20-35
Stonework up to 3 4-40
Brickwork up to 2 6-25
Concrete 1-7 10-90

The materials enumerated in the Table 5 must satisfy the require­

ments and norms (of strength, ductility, production process, chemical
composition, etc.) of the corresponding standars. It does not cover all
materials, nor the diverse conditions in which they work. In each de­
sign problem the permissible stresses should be specified in accordance
with the official technical specifications and design standars for the
given structure, and, in their absence, on the basis of factors discussed
in the preceding section.
 PART II
Complicated Cases
of Tension and Compression

CHAPTER 4

Design of Statically Indeterminate Systems

for Permissible Stresses

§ 18. Statically Indeterminate Systems

Our ability to calculate the deformation of bars under tension and
compression enables us to determine the changes in the shape and size
of parts of structures under the action of external forces. Usually these
deformations are so small that they seem devoid of any practical im­
portance.
However, in a number of structures, it is impossible to check the
strength and determine the cross-sectional area of the various elements
without the knowledge of deformation; these structures are known as
statically indeterminate systems', finding the forces acting in the ele­
ments of these structures in a statically indeterminate problem.
In all the examples which we have considered till now, the tensile
or compressive forces acting on the bar were determined from static
conditions of a solid body.
In case of weight Q suspended from two bars (Fig. 31), AB and AC,
we find tensile forces Ni and N* stretching the bars from the equilib­
rium conditions of point A . Three forcesappl ied to point A must satisfy
two equations of equilibrium: the sum of the projections of these
forces on the two coordinate axes must be zero. Thus, we see that the
number of unknown quantities (two) is equal to the number of
equations (two), therefore forces Mi and Ms may be determined
from these equations. This is statically determinate problem.
The conditions will be different if weight Q is suspended from three
bars (Fig. 32). In this case point A is in equilibrium under the action
of four forces: Q, M3 , Mi, and M3 , three of the forces being unknown.
The number of equations remains the same, i.e. two. Hence, the num­
ber of unknown quantities exceeds by one the number of equations,
the structure is one degree indeterminate, and the problem cannot
be solved with the help of static equations only.
Ch. 4) Design of Statically Indeterminate Systems 67

The additional equation required for the solution of the problem

can be compiled using the ideas gained in passing over from the theo­
retical mechanics to the strength of materials. We must take account
for the ddorinability of material. One more equation can be found
in studying the deformations of the structure. It turns out that it is
always possible to find as many additional equations as is required
to complete the number of static equations so that the number of
equations be equal to the number of unknown quantities.

Fig. 31

The extra equations are formed on the basis of the common principle;
they should express the coruiilions of joint deformations of the system.
Any structure deforms in such a way that there are no ruptures
of the bars, their disconnection or any unforeseen relative displacement
of one part of the structure with respect to the other. This in brief
is the principle of joint deformation of the elements of a system.
The general method of solving statically indeterminate systems is
as follows. First of all we must decide what are the forces to be deter­
mined, then write down all the static equations of a solid body, and
finally derive the required number of extra equations to find the un­
known forces.
A course of the solution of the problem is shown for the particular
example (Fig. 32). Suppose the side bars of equal cross-sectional areas
are made of steel, whereas the middle bar is made of copper. The length
of the middle bar is /s and that of the side bars, /*. Suppose the per­
missible stress for steel is la,l and for copper l<jfl. It is required to
determine the safe dimensions of the cross sections of these bars under
the action of suspended weight Q.
First of all we shall determine the forces acting on each of the three
bars. Since there are hinges at points A , B, C and D, all the three bars
can be subjected to only axial forces. Let us consider these forces to
be tensile. In order to determine these forces, we must consider the
equilibrium of point A to which the only known force Q is applied.
A scheme of the forces acting on point A and the location of coordinate
G8 Complicated Cases of Tension and Compression [Part It

axes are given in Fig. 33. Let us equate to zero the sum of projections
of the forces acting on point A on the coordinate axes:
ALsina— Af,sma —0
Q — N3—Aft cos a — A/4 c o s a = 0
From the first equation we gel Aft=A/a; replacing Ar2 by M in the se­
cond equation, we obtain
N :i-\-2 N i cosa= Q (4-1)
Now we have one equation with two unknowns.
To obtain the extra equation we must study the deformation of the
structure. All the three bars will elongate under the action of force
Q, and point A will descend. Since N x and N a are equal and bars / and 2

are of the same material, elongations A/( and A/* will be equal if the
bars are of equal length, point A will descend vertically downwards.
Let us denote the elongation of the third bar by A/s.
The elongation of all the three bars is joined, i.e. the bars remain
hinged at point A after deformation. To find the new position of this
point, we assume the bars to be disconnected and plot on the diagram
(Fig. 34) the new lengths of the side bars CCa and BB* by increasing
their initial lengths by At,—AB* and M S=AC*. The new position of
point A is obtained bv rotating the elongated bars CCt and BBa about
points D and C. Points B3 and Ct will coincide at point A ,, moving
along the arcs C*At and B*Ai which due to the small deformation
may be considered as straight lines perpendicular to CC2 and BBS,
respectively.
The new position of the side bars BAi and CA i is shown by dotted
lines. Since the end of the middle bar is also fastened to the hinge,
it will also come to point Ai, and elongation At3 will be equal to A A t.
According to Hooke’s law, the elongations Alt, A/a, and A/a of all
the three bars will be directly proportional to the tensile forces slret-
Ch. 41 Design of Statically Indeterminate Systems <19

chirtg them. After finding the relation between these elongations from
the figure, we shall obtain the extra equation correlating the unknown
forces in the bars. From triangle AiABs, we have
ABi -=AAl cosa or A/, = Afa cosa (4.2)
Let us express &U and A/3 in terms of the forces A'i and Na. This
is possible only if we know the cross*sectional area of the bars. Here
we must state a very important feature of the statically indeterminate
systems: to determine the forces acting in bars we must know before­
hand either the cross-sectional area of these bars or their ratio.
Let Ai and A 9 be the cross-sectional areas of the bars; let us denote
the modulus of elasticity of steel by E g and that of copper by £ c.
Then
Ali Nili . A/ , = EcAsi (4.3)
EsAt ’

Putting these values of Alt and Al 9 in equation (4.2), we get

Nxh N3 I3 cosa

It is evident from triangle ABD (Fig. 34) that

lt = /j cos a
Therefore,
A \= N , ^ c o s ' a (4.4)

Thus, by examining the joint deformation of the system, we have

obtained an extra equation correlating Arf and Nu.
Joint deformation takes place in statically determinate structures
too, but there it does not impose any constraints on force distribution.
Only one system of forces satisfying the. equilibrium conditions is
possible in this case. Since the number of unknowns is equal to the
number of static equations in such structures, the deformation is
compatible with the conditions of joint deformation. For example,
the forces acting in the bars can be fully determined from the equilib­
rium conditions of point A. Both the bars may elongate under the ac­
tion of these forces without getting disconnected, and the condition
of joint deformation is automatically fulfilled.
On the contrary, in statically indeterminate structures, there can
be any number of force systems satisfying the equilibrium conditions,
because the number of unknowns is greater than the number of equa­
tions. From all the possible combinations of forces, the combination
which actually occurs is the one that corresponds to the condition of
joint deformation.
70 Complicated Cases of Tendon and Compression [Pari U

In the statically indeterminate system (Fig. 32), the location of

point A after deformation combines the elongations of all the three
bars. For the condition of joint deformation to be satisfied it is essen*
tial that the elongations should be in a definite ratio. This condition
gives us the extra equation (4.4) required for determining the unknown
force.
Continuing the solution of the problem we put the value of Ni
from (4.4) into Eq. (4.1) and obtain
W,+ 2 ^ , 1 ^ 'os’ a = Q

whencefrcm
Na (4.5)
l + 2 -^rCOS3a
EcA3
and from (4.4)
Q jrircos*a
______
N,= - AT, (4.6)
1+2 cos3 a

It is evident from the formulas obtained that the value of N depends

not upon the absolute values of the cross-sectional areas A and moduli
of elasticity E , but upon their ratio. By setting different values of
the ratio n— 4 l we obtain various combinations of the forces Mlt
Nt, and Ns.
Knowing the forces and the permissible stress we can find Ai and
A% from the conditions

Calculating Ai from the first condition and knowing the selected

ratio n = 4Ati» we can find n
This value can be checked up
by seeing whether it satisfies the second condition of (4.7); if not,
the value of At is found from this condition, and A t is determined by
the formula
A l = rtA3 (4.8)
Thus, in a statically indeterminate structure with a given load we
may obtain a number of different modifications of force distribution
between the bars by changing the ratio of their cross-sectional areas.
Let us take a numerical example for greater clarification.
Let Q=4 tf; a =30°; [aj=1000 kgf/cm* £,=2X10® kgf/cm\
[acl=600 kgf/cm*; Ee— 1 X 10* kgf/cms.
Ch. 41 Design of Statically Indeterminate Systems 71

For preliminary calculation let us assume an arbitrary value of

Then
4 xf p F cosi3°a
1.67 tf
1+ 2 x r | w cos330<>

N l.U tf
1+ 2X?¥4?=8 co8530<»
fxKii
From strength condition we obtain
„ Nt 1670
1 — “ 1000 “
1.67 cm*

As we have assumed A i=A a, then Aa=1.67 cm8

Let us check whether these dimensions will satisfy the strength
condition for the middle bar:
^ 11^ = 667 kgf/cma> 600kgf/cm*

The assumed value of A a is not enough; it should be

noo
600
= 1.85 cm*

To maintain the condition A i= A 3 which formed the basis of our

calculation, we must lake A1= A 3=1.85 cm* instead of the required
value of 1.67 cm2 obtained from the first condition. In this way we
shall have an additional reserve in the side bars.
If we wish to avoid this extra reserve and take
A, = Aa= 1.67 cm8, As = 1.83 cm* (4.9)
then forces Ni, Afa, and N 3 will change immediately; the ratio A tfA t
will no longer equal 1, as assumed earlier, but will be0.9. In formula
(4.5) the denominator becomes less and Na increases; in formula (4.6)
the decrease in the value of the denominator will be less as compared
to the numerator, therefore ATi and jV2 will decrease.
By decreasing the cross-sectional area of the side bars as compared
to that of the middle bar, we reduce the forces acting on the side bars
and increase the forces acting on the middle bar.
This reflects the general law which governs the force distribution
between the elements of all statically indeterminate systems: the
forces are distributed in accordance with the rigidity of the bars; the
72 Complicated Cases of Tension and Compression [Part f t

greater the cross*secttonal area of a given bar, the greater is the share
of total force that it takes, and vice versa.
If we approximate the areas A t and A , to zero, then forces Mi and
Mi will tend to zero and M* to Q. If, on the other hand, we decrease
A,, then Na will decrease, whereas Mi and M2 will increase.
For a value of n=0.9 [formula (4.8)1, Afi=N2=1.60 tf, Af3=1.20 tf.
This requires >4a=2.0 cm* and ^ = 1 . 8 cm* instead of 1.6 cm* as found
from the strength condition for steel bars.
Had we assumed ^4»=1 .6 cm* for the side bars, the ratio n would
have been reduced again, and the middle bar would have again got
overloaded. Thus we should again be reconciled with the reserve in
side bars. From formula (4.4) it follows that a particular ratio, //=
A
= t J , which ensures that the stresses in all the bars are equal to the
permissible stresses, is possible only for a definite value of the angle a.
Indeed, had we determined the areas A exactly in accordance with
the permissible stresses, we would have got the relations
(4.10)
Putting these values in (4.4), we get
A ilo s] ~ A §\o€j § ^ W a (4.11)

wherefrom we have
cos* a —[ocJ4£7
s- (4.12)

i.e. in order to select the cross-sectional area without excessive reserve

for any value of the ratio n i t is essential that cos a should satisfy con­
dition (4.12).

Table 6

Results ot Calculations for Various Values of ft

At*At (cm*)
- A< N,=Nt
<«> «6 (cm*)
required assumed

0 .8 1 .5 6 1 .3 0 1 .5 6 1 .7 4 2 .1 7
0 .9 1.0)0 1 .2 0 1 .6 0 1 .8 0 2 .0 0
1 .0 1 .6 7 1.I t 1 .6 7 1 .8 5 1 .8 5
1 .2 1 .7 5 0 .9 7 1 .7 5 1 .9 4 1 .6 2
1 .5 1 .8 3 0 .8 2 1 .8 3 2 .0 6 1 .3 7
Ciu 4\ Design of Statically Indeterminate Systems 73

In our numerical example we obtain

lpooxiyto 8 hence a = 24°
GOOX2XIO* ’
Since in the given structure a=30°, then for an arbitrary n we shall
either have to give excess reserve in one group of bars, or overload the
other group. The value of n itself should be selected from economic con­
siderations. Table 6 contains the values of different quantities for
various values of a. Knowing the cross-sectional areas, lengths and
materials of the bars, we can select a combination which is economi­
cally most effective as far as the cost of material is concerned.

§ 19. The Effect of Manufacturing Inaccuracies

on the Forces Acting in the Elements of Statically Indeterminate
Structures
In the preceding sections we established the main features of the
working and design of statically indeterminate systems.
1. The extra equations required to calculate the forces may be ob­
tained only from the condition of joint deformation of the system.
2. The force distribution between the elements of statically inde­
terminate structures depends upon the ratio of their cross-sectional
areas, moduli of elasticity and
lengths.
3. The more rigid an element,
i.e. the smaller its length and the
greater its cross-sectional area and
modulus of elasticity, the greater
will be the share of force that it
will take.
In this section we shall study
another property of statically inde­
terminate structures which is of
great practical importance.
It is impessibie to manufacture
parts of structures with absolute
accuracy; small manufacturing er- FlS- 33
rors and inaccuracies must always
be taken into account. In a stati­
cally determinate structure, these
inaccuracies cannot give rise to stresses in the system. Thus, for exam­
ple, if bar AB (Fig. 31) is made a little shorter than it should be accord­
ing to the drawing, all that will happen is a slight distortion of tri­
angle CAB. In the absence of force Q the forces in bars AB and AC
will be equal to zero irrespective of the manufacturing accuracy of
the bar length.
74 Complicated Cases of Tendon and Compression [Part I I

The statically indeterminate structure shown in Fig. 32 will behave

in an entirely different manner. Let the manufactured length of the
bar be less than the required by A A 0=& (Fig. 35). To join the end of
the middle bar A 0 with the ends A of the side bars somewhere at point
A lt it is necessary to stretch the middle bar by M a=AoAi, and to
compress the side bars by k l 1= :A B a= A C 1. Drawing through points
C* and B t perpendiculars to the initial positions of the side bars as
explained in § 18, we get the point of junction Ax of the ends of all
the three bars. From the figure we may write down the condition of
joint deformation:
AnA = A 6 Al -\-A1A
or
6 = A/3 + (4.13)
cos a

Since there are no external forces and Nx is a compressive force

whereas N 9 is tensile, the equilibrium condition (4.1) takes the form
N 9 —2A/’1c o s a = 0 (4.14)
Replacing in (4.13) Ah and A/s by their values
A/j = Nik and
at N 9k N sk cos a
A/’ = f p r 3 —
and solving equations (4.13) and (4.14), we have
f)£cAg mr _
EeAs f> ' 1 2 cos a
(4.15)
2EsAx cos3 a j

The plus sign before the values of Ni and Nasigni lies that our assump­
tions about the directions of these forces are correct.
It should be pointed out that in formula (4.3) the length of the middle
bar can be replaced by I* and not by l 9 —6 , because 6 is an infinitesimal
as compared to /3. This simplification can always be applied when the
manufacturing inaccuracies are being considered.
The above computations reveal that the manufacturing inaccuracies
will give rise to stresses in the bars even if there are no external forces
acting on the structure. Hence, the possibility of the so-called initial
stresses is also an important property of the statically indeterminate
structures.
If all the three bars are of the same material and have the same cross-
sectional area, then under weight Q (Fig. 32) the tensile force in the
middle bar wilt be greater than in the side bars (4.4). The manufactur­
ing inaccuracy gives rise to an additional tensile force in the middle
bar and to compressive forces in thesidebars. In this particular example
Ch. 4) Design of Statically Indeterminate Systems 75

the initial stresses increase the non-uniformity in the working of

the bars and are therefore harmful.
Had the middle bar been longer by 6 , the initial stresses would have
had opposite signs and would have levelled off to some extent the non-
uniformity in force distribution between the middle and the side bars

(a)

under the action of weight Q. In this case, the particular property of

the statically indeterminate systems discussed above would have
helped in- better working of the structure.
Another example of the expedient use of initial stresses is putting
on the tyre on the wheels of a rolling stock. The wheel consists of
two parts: the central cast portion and the steel tyre which is put on
6

It (Fig. 36 (a) and (&)). The tyre is fastened to the central portion by
means of special fixtures; besides, its internal diameter d3 is made a
little less than dt. Usually this difference is of the order of — dt—
1 n
approximately §555 di- Before slipping the tyre on the central portion,
it is heated so that its internal diameter becomes greater than the dia­
meter of the central portion; the fitted tyre contracts upon cooling
land presses the central portion. A tensile force At appears in the tvre,
and a reaction p, between the tyre and the central portion (Fig. 36 (b)).
76 Complicated Cases of Tension and Compression [Part l l

If we cut the tyre across the diameter (Fig. 37), the two Forces
N must balance the total pressure on the internal surface of the
cutoff portion of the tyre. Let us write down the equilibrium
condition by projecting all the forces on the /pax is (Fig. 37). A pres­
sure p ds acts upon the element of length ds of the lyre; its projection
on the //-axis is equal to —pds sin a = —p-y sin a da, because ds=*
= 4*0 da. The equilibrium condition takes the form
n
2N — C p y s in a d a = 0, or 2N— p -y fsin a < 2 a = 0
ahc o

whencefrom
2N—p d = 0 and /V= y , or p= ^
Thus we have one static equation for two unknowns N and p; this
is a statically indeterminate problem. The unknown forces can be
determined only by considering the joint deformation of the structure.
The tension in the lyre and the compression in the central portion
should be such that they level the difference between Ihe diameters
di and d2. Neglecting the deformation of the central portion due to
its much greater mass as compared to that of the tyre, we find that
the levelling of the difference in diameters takes place chiefly due to
elongation of the tyre. If this difference is y of the lyre diameter,
then the relative elongation &R of the diameter and, consequently,
of the whole tyre will also be
The relative elongation of the tyre under force N is e 4v=j% , where
A is the cross-sectional area of the tyre. Equating the values ert=»
=e.v, we obtain an extra equation
JV = 1
EA n ’ ti
whence
2EA
P= n r (4.16)

N E
The stress in the tyre is cr— .
In formula (4.16) d may be replaced (instead of the original diameter
d2) by dt, because the difference in the diameters is infinitely small.
Let us consider a numerical example (the tyre of a freight wagon
13 cm broad and 7.5 cm thick). Let d=r/|=900 mm; E—
Ch 4) Design of Statically Indeterminate Systems 17

**2xl0‘ kgf/cm*; A = 7 .5 x 13=97.5 cm*. Then we have

2 X10*
1000
= 2000 kgf/cm4

N = 2X 5 = 195 000 kgf = 195 tf

2X2X I0*X97.5
P 90Xl(Kk) = 4330 kgf/cm

§ 20. Tension and Compression in Bars Made

of Heterogeneous Materials
This type of bars belongs to the group of statically indeterminate.
As an example, we shall discuss how to determine the dimensions of a
composite column (Fig. 38) under the action of compressive forces
P. The column consists of a round steel bar of diameter d* and is

Fig. 38

located inside a bronze jacket of external diameter db and wall thick­

ness I.
Let us introduce the following notations:
Ao— cross-sectional area of the bronze pipe;
A s— cross-sectional area of the steel bar;
lo«l> £*— permissible stresses and moduli of elasticity of
bronze and steel, respectively.
The required dimensions of the bar should be such that enable
it to withstand load P.
Let us find stresses Oj, and <r„due to load P over areas Ai and A„
respectively, and write down the strength condition.
The bar is axially compressed by forces P applied at the centre of
gravity of the section through rigid slabs 5 whose deformations are
considered negligible (Fig. 38). The part Pi of the compressive forces
is transmitted through the bronze jacket, and part P a, through the
central steel bar (Fig. 39). We have only one equation of statics to
determine these two forces which give rise to stresses in the steel bar
and bronze jacket:
P ,+ P „ = P (4.17)
Thik is a statically indeterminate problem. The second equation
is obtained from the condition of joint deformation according to
78 Complicated Cases of Tension and Compression [Pari II

which both the bronze jacket and steel bar of the column (Fig. 39)
must shorten by the same length A/, since the top and the bottom
planes of both coincide. From Hooke’s law we have
P bi PJ
Af = Eb (4.18)
EsAs

This is the second equation correlating Pb and P3. From (4.18) we

find
P, AbEb
Substituting this value of P t in (4.17), we get

and
p Eg
p T>-rb p (4.19)
Ph= P .=
T T Z s il I I A& Ej>
!■+AAbiMEbi I + T ,I T
p
crh = ' (4.20)
E, ’
Ab+As-g A,+ Abf -
b
The distribution of forces between the elements of statically inde­
terminate structures depends upon the ratios of their cross-sectional

Fig. 39

areas and moduli of elasticity. From equation (4.18), taking into

consideration that

we find that the ratio of the stresses in bronze and steel depends only
upon the ratio of their moduli of elasticity:
Ch. 4\ Design of Statically Indeterminate Systems 79

and the stresses are directly proportional to the moduli of elasticity.

Assuming that under compression £ , —2X10' kgf/cm2 and £ ?,= 1x
XiO« kgf/cm2 {Table 1), it is obvious that stresses in steel bar will
always be two times higher than the stresses in bronze jacket. The per­
missible stresses for steel are usually three times greater than permis­
sible stresses for bronze. Therefore, if the stresses in the bronze jacket
are equal to the permissible stress for bronze, the stresses in the steel
bar will be smaller than the permissible stress for steel. I-lence the
dimensions of the column are obtained from strength condition of
bronze jacket under compression:
P
°b (4.21)
Ah+ A s J *

Let £ —25 tf. The ratio A J A h of the cross-sectional areas is usually

selected from design considerations. Let A J A b~2, and the permissible
stress [o&]=500 kgf/cm2. Equation (4.21) will then be written as
__25 000 <*5og
^ ( ! + 2 X2) S^ OUU’
wherefrom
Ab 10 cm4 and A, = 2 x 1 0 = 2 0 cm*
The diameter of the steel bar is calculated from the condition
> As, wherefrom dt = j / ^ = j / = 5.05cm « 51 mm
The di mensions of the bronze jacket section can be found if we assume
a particular value of wall thickness t from design considerations. Let
/= 5 mm=0.5 cm. Now, applying the approximate formula for a ring,
we have
Ab < n dbt, wherefrom db> ^ = 3I 4W 5 = ^-48 cm ~ ^ mm
The deformations of such structures are calculated according to the
general principles. Since the steel and bronze portions of the column
shorten by the same amount.it is immaterial which of the formulas
in equation (4.18) is employed for calculating A/.

§ 21. Stresses Due to Temperature Change

In statically indeterminate systems, stresses without any external
loading occur not only due to the inaccuracy of manufacturing and
assembling, but also due to a change in temperature.
Considerable stresses of this type may arise in rails welded into a
continuous line. The rails are subjected to tensile or compressive stres-
so Complicated Cases of Tension and Compression [Part f l

ses when the temperature changes with respect to that at which they
were welded. The problem may be schematically expressed as follows:
we have a restrained bar whose both ends have been rigidly fixed at a
temperature find the stresses arising when the temperature changes
to U (Pig- 40). The length of the bar is f, cross-sectional area is 4
and modulus of elasticity is E.
Let us ascertain the forces which will act on the bar when the tem­
perature rises from U to U- The bar will tend to elongate and push apart

— I -
Fig. 40

the supports A and B. The supports will resist this with reactions
directed as shown in the figure. These forces will cause the bar to be
compressed.
These forces cannot be found from static conditions, because all
that we come to know from the single equilibrium condition is that
the reactions at points A and B are equal in magnitude and opposite
to each other. The value ol the reaction P remains unknown, and hence
the structure may be considered statically indeterminate.
The additional equation can be written from the consideration that
length I of the restrained bar remains unchanged in spite of the change
in temperature. This implies that shortening A/P due to force P is
equal in its absolute value to the temperature elongation Ai, which
the bar would have experienced had the end A been fixed and end B
free to move. Hence
A/f—A/p = 0 (4.22)
This is the condition of joint deformation; it shows that the length
of the bar remains constant despite the temperature change, since it
does not tear away from the fixed supports.
Since
A/P = ~ j and A/t = «/(<,— it)

where a is the linear thermal expansion coefficient of the bar material,

we have
a / ( / 8 /,)
and
i = a = a £ ( ( ,- /.) (4.23)
 Ch 4 1 Design of Statically Indeterminate Systems 81

i.e. the stress due to temperature change in a restrained bar of uniform

cross-sectional area depends only upon the modulus of elasticity of the
material, its linear expansion coefficient and the temperature difference
and not upon its length or the cross-sectional area.
Force P may be calculated from the expression
P = a E A [tt - Q

In this example, if /*> /„ stress a will be compressive, because the

direction of reaction P inside the bar has been considered positive.
If we follow the generally accepted convention of writing the compres­
sive stresses with a minus sign, and the tensile stresses with a plus
sign, then formula (4.23) should be
written in the following manner to
automatically give the proper sign:
<j—a E (tt — tt)
If the cross-sectional area of the bar
is not constant along its length or if it
is made from different materials, or if
the supports permit a slight change in
length, or if all these conditions take
place simultaneously, the method of
determining thermal stresses somewhat
changes although basically it remains
the same.
The variability in the cross-sectional
area and the use of different materials
must be taken into account when calcu­
lating A/; it is determined as the sum
of elongations calculated separately for
each portion. The possibility of the
bar to slightly change its length is re­
flected in the equation of joint deforma­
tion (4.22); the difference of deforma­
tions caused by the temperature change and the forces is in this case
not equal to zero, but equal to the length, by which the bar is free to
elongate.
A steel bar consisting of two parts of length /i= 40 cm and fa= 60 cm
and cross-sectional area /41=10 cm* and Aa= 20 cm*, respectively,
has one end rigidly fixed whereas the other end misses the support
by Ao^O.S mm (Fig. 41). Find the stresses in both the parts if the tem­
perature increases by o0°C, a=125X 10“7.
^ Increase in temperature causes elongation of the bar by Al t, and
the compression from the support reaction P results in its shortening
by'Alp. The difference of these two deformations (in absolute value)
4-3910
82 Complicated Cases of Tension and Compression [Part I!

is Ao (see Fig. 41):

- a/ , - a/ „ = a 0
This is the condition of joint deformation. The respective values of
A lt and Alp are
A/t = a / ( / i + / a), ild ll
/i/ij J
Therefore
M il
/|4 #J A,
whencefrom
D _ jaf (fi + /a) ^ A ,1 EAj _ [125X10-?X 100 x 6 0 —0.03]X2Xl08Xl0

4 +&] ^ R l]
= 9300 kgf

Stress in the upper portion is

9300 n o A i - r ,
o . = - ^P- = — = 930 kgf/cm*.

Stress in the lower portion is

0" = ^ - = ^ ? = 4 6 5kgf/cra*

(both the stresses are compressive).

Had there been no gap A0, the force as well as the stresses would
have increased 1.92 times.

§ 22. Simultaneous Account for Various Factors

Sometimes, in statically indeterminate systems, we have to consider
simultaneously the effect of external forces, change of temperature
and manufacturing inaccuracies. The problem can be solved in two
ways: first method-simultaneous account for all the factors. In this
case the equation of joint deformation must contain terms reflecting
the effect of all the factors (load, temperature and manufacturing inac­
curacies). The forces and stresses obtained as a result of such a compu­
tation are final.
In the second method, we compute separately the forces and stres­
ses due to the load, temperature and manufacturing inaccuracies. In
other words, a number of separate problems are solved, each problem
taking into account only one factor. The final forces and stresses are
obtained as the algebraic sum of the values obtained from each of the
solutions. The second method is often simpler and more convenient,
Ch. 4] Design of Statically Indeterminate Systems 83

although it calls for more calculation. I t is known as the method of

cumulative action of forces. This method is valid because of the appli­
cability of the principle of superposition of forces. When deformations
are small in magnitude, the deformation caused by a force or a group of
forces either does not affect the deformation due to another force or
group of forces, or the effect is so small (less by an order) that it may

Fig. 42 Fig. 43

be neglected. This principle is not applicable for extremely flexible

or highly deformable structures like long thin bars, membranes,
rubber parts, etc.
We shall solve the following example to illustrate the technique
of simultaneous accounting of various factors.
Three parallel vertical rods of equal length 1=2 m support a rigid
beam AB to which a force P —4 tf (Fig. 42) is applied. The distance
between the rods and from the middle rod to the point of application
of force P are a= 1.5m, b= 1 m and c=0.25 m, respectively. The middle
bar is shorter than its design length by 6=0.2 mm. Data about the
bars are given in Table 7.
Table 7

A B a
No. of rod Material (cm*) (kgf/cm*)

1 Conner 2 IX io* 17x10”*

2 Steel 1 2X10* 13x10"*
3 Steel 3 2X10* I3X10-*

During operation the temperature of the structure may rise up by

by 2(fC.
Find the stresses in each of the three rods.
Let us suppose that forces Nu iV1? and Ar3 in all the rods are tensile.
The reactions at the supporting points, equal to them, are shown in
4*
 $4
Complicated Cases of Tension and Compression [Part U
Fig. 42. As the forces are parallel, we can write down only two equilib­
rium equations. The first is the sum of the projections of forces on
the vertical axis and has the following form:
+ =0 {4.24)
For the second equation let us take the sum of moments of all forces
with respect to the point of support of the second rod:
Nla— Nab+Pc=Q (*■25)
These two equations are insufficient to determine three unknown
forces. We must consider the deformations. Figure 43 shows a sketch
diagram of the structure, with the assumption that all the three rods
are subjected to tensile forces. From this diagram we may write down
the following condition of joint deformation:

— A/, — A a (4.26)

The values of the deformations entering into the equation (taking

into account the temperature change) will be as follows:
_n± 4 -cc j- a 8/ A/
EiAi
4 '» = 2%L+ a >l i ‘

Putting these values in equation (4.26) we get

NJ At,I
E$A9 - « ,( At ■ E,A| ■aJAt a~b
NJ
(4.27)
(-a*/ At ■ a, lM—t>

Solving equations (4.24), (4.25), and (4.27) simultaneously, we

determine Nu Nt, and Ata. Their values are
Nt = 792 kgf, N2= 1020 kgf, and JV, = 2188 kgf
Had our assumption about the direction of the forces been wrong
for any of the rods, the value of that force would have been obtained
with a negative sign.
Let us now determine the corresponding stresses:
in the first rod <j1= ^ = ^ = 396 kgf/cm*
in the second rod 09 = ^ — 1020 kgf cm*

in the third rod oa= ^ = ^ ^ = 7 2 9 k g f/c m *

Ch. 4) Design of Statically Indeterminate Systems £5

The problem could also have been solved by considering separately

the effect of load, temperature and manufacturing inaccuracies and
subsequently adding the stresses. The result would obviously have been
the same.

§ 23. More Complicated Cases of Statically

Indeterminate Structures
In all the statically indeterminate systems that we have considered
till now, the number of unknown forces exceeded by one the number
of static equations. They were all first-order statically indeterminate
problems; one of the unknown forces may be considered as a redundant
unknown, which cannot be determined from the static equations.

az
1

ali
i
60SW#
Fig. 44

There may be cases when the number of these redundant unknowns

is greater; in such cases it becomes necessary to write down an equal
number of extra equations from the conditions of joint deformation
of the system. The structure shown in Fig. 44 may be taken as an
example: a very rigid bar is hinged to a fixed support suspended
from three rods and loaded with force P.
We may write down three static equations for bar AB. The number
of unknowns is however five: forces in the three rods and the horizontal
and vertical components of the reaction at hinge A.
The extra equations can be written by considering the deformation
of the system. Since we are considering bar AB to be very rigid, its
deformation may be ignored. Remaining straight, it will occupy posi­
tion ABt- From the similarity of triangles we may find the relation
between Alu A a n d A/,; this will give us two extra equations, name­
ly:
Ali <*x and Atj (lj
A1$ O) Al» ~ a9
cos 45°

Further solution is the same as in the example discussed above (§ 18).

86 Complicated Cases of Tension and Compression [Part H

CHAPTER S
Account for Dead Weight in Tension
and Compression. Design of Flexible Strings
§ 24. Selecting the Cross-sectional Area with the Account for the
Dead Weight (in Tension and Compression)
Till now, in determining the external forces stretching or compress­
ing the elements of structures, we ignored the dead weight of these
elements. The question arises: Does not this simplification introduce
a considerable error in the computations? Let us therefore determine
the stresses and deformations of a stretched or compressed bar ac­
counting for its dead weight.
Let a vertical bar (Fig. 45 (a)) be fixed at its upper end and load p
suspended from its lower end. The length of the bar is /, its cross-sec­

Pf
(b)
Fig. 45

tional area is A , modulus of elasticity E and specific weight y. Let

us calculate the stresses in section M N located at a distance x from
the free end of the bar.
Cut the bar through section M N and separate the lower part of
length x loaded by force P and its own dead weight yAx (Fig. 45 (b)).
These two forces are balanced by stresses acting on face M N from cut-
oil portion. These stresses will be normal, uniformly distributed and
directed outwards of the portion of bar under consideration, i.e. they
will be tensile. The magnitude of these stresses will be
(5.1)
Thus, when the dead weight is accounted for, the normal stresses
are found to be not constant along the length of the bar. The most
stressed and hence the critical section will be the uppermost section
Ch. 5] Dead Weight. Design of Flexible Strings 87

for which x has Hie maximum value equal (o /; stress in this section
will be
(5-2)

It is this section which must satisfy the strength condition

(5.3)

Herefrom the required cross-sectional area may be calculated as

(5.4)
A > W = ^l
The only difference between this formula and the one for deter*
mining the cross-sectional area of a stretched bar without account for
the dead weight is that quantity yl is subtracted from the permissible
stress.
Let us calculate the stresses for both the cases to evaluate the im­
portance of this correction. Consider a mild-steel bar 10 m long having
ta l= 1600 kgf/cm1and the quantity v*=7.85 X 10"3X 10*=7.85 kgf/cm2.
Thus, for a mild-steel bar the correction in the cross-sectional area
7 85
will be i.e. approximately 0.5%. Let us now consider a brick
column also 10 m long, which has [o]=12 kgf/cm1 and the quantity
y/= 1.8x 10“SX 103= 1.8 kgf/cma. Therefore, the correction for the
brick column will be i.e. 15%.
It is obvious that the effect of deadweight in tension and compres­
sion may be neglected if the bar (column) is not very long or if it is
not made from a low-strength material (brick, stone) with a great
weight. The dead weight has to be considered when designing long
elevator ropes, various types of long rods and high stone structures
(beacon towers, supports of bridge trusses, etc.).
In such cases it becomes necessary to determine the most expedient
shape of the element. If we select the cross-sectional area of a rod
(Fig. 45) according to formula (5.4) and take it uniform along the whole
length, the material of the element will be poorly utilized: the nor­
mal stress will reach the permissible limit only in the uppermost sec­
tion. In all other sections we will have margin of stress and conse­
quently excessive material. Therefore, it is desirable to design the
element in such a way that the normal stresses are the same in all
its sections (perpendicular to the axis).
Such an element is classified as the bar of uniform strength under
tension and compression. The element will have minimum weight if
the stresses are equal to the permissible stress.
88 Complicated Cases of Tenstrn and Compression [Part l l

Let us consider a long bar subjected to compression by the force P

and its own weight (Fig. 46). The nearer a section is to the base, the
greater is the force causing stresses in the section and therefore the
greater must be the dimensions of the section area. The bar will have
a shape continuously widening downwards. Cross-sectional area A
will change along the height depending upon the value of x t i.e. A —
=/(*)•
Let us establish a relation between the cross-sectional area of a sec­
tion and its distance x from the top end.
The cross-sectional area of the top
end A q is determined from the
P strength condition:

or A' = w

where [or] is the permissible stress

under compression; stresses in all
other sections must also be equal to

Let us take two infinitely close

sections at a distance x from the top
rig. 46 end to elucidate the variation of
cross-sectional area with the height
of the section. Let the distance between the sections be dx. Let us de­
note the area of the upper section by A (jc) and the area of the adjoining
section by A (x)+dA (*). Increment dA (x) of the area between the two
sections must bear the weight yA (x) dx of the element of the bar en­
closed between these two sections. Since it should cause a stress equal
to the permissible stress lal on the area dA (x), we may determine the
increment of area dA ( jc) from the condition

(S'*)
wherefrom

A[x)

Integrating both sides, we get

In /I(* )+ C = j2 j* (5.6)
At x= 0 the area A (.v)=j40; putting this value In equation (5.6),
we have
ln A i+ C ^ O , or C=*— In A,
Ch 5] Dead Weight. Design of Flexible Strings 89

Therefore,

£ g -& ‘

and

A{x) = A 0eW * (5 7 )
If the cross-sectional area changes exactly according to this law,
the lateral faces of the bar have a curved shape (Fig. 46), which com­
plicates the machining operation and increases production cost. For
this reason usually a shape approxi­
mately corresponding to the shape
of a uniform-strength bar is em­ ’1
ployed, for example a truncated h
pyramid with plane faces. h
The above computation is only -f-
approximate. We had assumed that k
only normal stresses are transmitted
through the whole section of the -f
uniform-strength bar; actually, near
the edges, the stresses are directed
along the tangent to the lateral sur­
face.
In long ropes or stretched rods
the shape of a uniform-strength
bar is obtained approximately by Fig- 47
dividing the bar lengthwise into a
number of parts, the cross-sectional area remaining constant over each
separate section (Fig. 47)—a so-called step bar is obtained.
For a given length the cross-sectional areas 4 ,, A s, . . . are deter­
miner! in the following manner. The cross-sectional area of the first
portion from the bottom will, according to formula (5.4), be
A - P -
' M -Y h
The cross-sectional area of the second portion can be determined
by considering it to be loaded by external force P and the weight of
the first potion y A J t
A — ^
4 " |CFj —

The cross-sectional area of the third portion is determined by adding

the weight of both the first and second portions to force P. The cross-
sectional area of all other portions can be determined in identical man-
90 Complicated Cases of Tension and Compression [Part I I

ner. Let us consider the following example to compare the effectiveness

of using uniform-strength bars, step bars and bars of constant section.
A support with height h —4 2 m is subjected to compression by axial
force P=400 t f . Assuming the unit weight of the laying as 2.2 tf/m3,
and the permissible stress under compression as 12 kgf/cm3, compare
the volume of laying for
support of constant section;
support made of three prismatic parts of equal length;
support of uniform strength under compression.
We shall carry out the calculations in tons (force) and metres.
For the first case the cross-sectional area is
A_ _ MO _ ij c _j
la } -A y “ 120 - 4 2 X 2 .2 “

The volume
V = Ah = 14.5 x42 « 610 m3
In the second case, the area of the upper portion is
400
. , h 120— 14X2.2 = 4.48 m*
'* 1—3 *
The cross-sectional area of the second portion is

P + Yi4,T 4004-2.2x4.48X14 *
h--------------120— 1*4 X2.2---------6 04 m
l° l ~ 3 ?
The cross-sectional area of the third portion is

t _ P + ' f'/,1T + ' l’/ l j T 400+ 2.2 x4.46X14 + 2.2 x 6.04X14

= 8 .12 m3
‘ (O l-A y I i o — 14X2.2
3

The total volume of the laying is

V = (At + A a + A , ) i = (4.48+6.04 -f 8 .12) 14 = 261 m»

The same result may be obtained from the condition that the force
at the bottom of the third portion, equal to P+G (where G is the total
weight of the support), is simultaneously equal to IcrlAa; therefore
1/ 6 |<Tl Aa— P n e !t ,
Ch 5] Dead Weight. Design of Flexible Strings 91

In the case of support of uniform strength under compression* the

cross-sectional area of the upper face is
_P _400
A, l<r] “ 120 = 3.33 m*

The area of the bottom section is

V . 2 2X *2
AA= A <
fi«» = 3.33e 120 = 3.33e° 77 =7.15 m*
The weight of the support of uniform strength G is determined from
the condition
P + G = [o]Ah
wherefrom
G = [<j] Ah— P = 120x7.15—400 = 460 tf
The volume of the support is

| / = T = ^ =209nia
which is 20% less than the volume of the step support and approxi­
mately three times less than that of the support of constant section.

§ 25. Deformations Due to Dead Weight

In determining the effect of dead weight on deformation under ten­
sion and compression we must take into account that the relative
elongation of various portions of the bar will vary just as stress a(x).
To calculate the total elongation of the bar of constant section let us
first determine the elongation of an infinitely small portion of length
dx, which is located at distance x from the end of the rod (Fig. 48).
The absolute elongation of this portion (equation (2.5)) is
A* . e ^ * _ * [ '+ ,,]
The total elongation of the bar

Al = j& dx = j % [-x+ H=-E?-+-!£

As for deformation in uniform-strength bars, the relative elongation
is the same over the whole length because the normal stresses are
the same in all the sections and equal to the permissible stress [<j 1:
92 Complicated Cases of Tension and Compression IPart 1!

The absolute elongation of a bar of length I will be

f<T|/__ pi
M —d =
E ~ EAo

(the notations correspond to Fig. 46).

The deformation ol step bars should be determined by parts, calcu­
lating the deformation separately for each prismatic portion. The de­
formation of each portion has to be
determined by considering not only
its dead weight, but also the weight
T of portions which affects its defor­
mation in addition to the external
force. The total deformation is ob­
tained as the sum of deformations of
r separate portions.
f
z
i § 26. Flexible Cables
pu A. In engineering practice, we come
across one more type of a stretched
element in which the dead weight
plays an important part in determin­
Fig 4B ing its strength. These are the
so-called flexible cables. This term
covers the flexible elements in electric transmission lines, cableways,
suspension bridges and other structures.
Let us consider (Fig. 49) a flexible cable of constant section loaded
by its own weight and suspended Iron! two supports at different heights.
The cable sags along curve AOB under its own weight. The horizontal
projection of the distance between the supports (points of fixation) is
called the span and is denoted by I.
As the cable is of a constant section, its weight must be distributed
uniformly over its length. Generally, the sag of the cable is small
as compared to its span, and there is little difference (not more than
10%) between the length of curve AOB and its chord AB. In this case
we may consider with a sufficient degree of accuracy that the weight
of the cable is distributed uniformly not over its length, but over the
horizontal projection of its length, i.e. along the span I. We shall
study only this type of flexible cables. Let us assume that the intensity
of the load uniformly distributed along the cable span is q. This load,
having dimensionality iforcet/llengthl, may be not only due to the
weight of the cable per unit span but also the weight of ice or any
other load also distributed uniformly. This assumption about the law
of load distribution considerably simplifies the calculations, but si­
multaneously renders them approximate. In exact calculations (load
C ft. 5| Dead Weight. Design of Flexible Strings 93

distribution along the curve) the sag curve is a catenary, whereas in

approximate calculations it is found to be a quadratic parabola.
Let us take the lowest point of sag 0 as the origin of coordinates
(Fig. 49); its position, which is as yet unknown, obviously depends
upon the magnitude of q, upon the ratio of the length of cable along
the curve to the span and also upon the relative location of the sup­
ports. Evidently, tangent to the curve at point 0 is horizontal. Let
us direct the #-axis to the right along this tangent.
m

Let us cut a part of the cable by two sections—one passing through

the origin of coordinates and the other at a distance x from it (section
nm). Since the cable is flexible, i.e. capable of resisting only tension,
the discarded portion can act on the remaining portion only in the
form of a iorce directed along the tangent to the sag curve at the point
of section. Any other direction of the force is ruled out.
Figure 50 depicts the cut-out portion of the cable with Ihe forces
acting on it. The uniformly distributed load of intensity*/ is directed
vertically downwards. The action of the left discarded portion (ho­
rizontal force H) is directed to the left because the cable is working
under tension. The action of the right discarded portion, force T, is
directed to the right along the tangent to the sag curve at this point.
Let us write down the equation of equilibrium for the cut-out por­
tion of the cable. Let us take the sum of the moments of all forces
about the point of application of force T and equate it to zero. Pro­
ceeding from the approximation introduced earlier, we consider that
the resultant of distributed load of intensity q is qx and that it acts
at the midpoint of segment x (Fig. 50). We get
Hy— q x j = 0, wherefrom y=zff (58)
It follows from this equation that the sag curve is a parabola. When
both the supports are at the same level, In this case / is
called the sag. It can be easily determined from equation (5.8) that
due to symmetry the lowest point of the cable is at the middle of the
span, and Substituting the values ofA = 6= -jand y= f in
94 Complicated Cases of Tension and Compression [Part H

equation (5.8), we get

/-& (8.9)

From this formula we determine the value of H:

H=s%f (5.10)

The quantity H is called the horizontal tension of the cable.

Thus, if load intensity g and tensile force H are known, we can de­
termine sag f from formula (5.9). If q and / are given, then the tensile
force H may be determined by formula (5.10). The relation between
these quantities and the length of wire s along the sag curve may be
established with the help of the well-known approximate mathemat­
ical formula *
s » / ( l + -§ £ .) (5.11)

Let us return to Fig. 50 and write down one more equilibrium

condition for the cut-out portion of the cable, namely, let us equate
to zero the sum of projections of all the forces on the x-axis:
— H + T cosot = 0
From this equation we find 7\ the tensile force at an arbitrary point:
H
cos a
(5.12)

It is evident from equation (5.12) that force T increases from the

lowest point of the cable towards the supports and is maximum at
the suspension points, where the tangent to the sag curve makes the
maximum angle with the horizontal axis. This angle is small when the
sag is not considerable, therefore we may consider with sufficient ac­
curacy for practical purposes that the cable is subjected to the action
of a constant force equal in magnitude to horizontal tension H. The
strength design of a cable is generally carried out for this value. If,
however, it is essential to design for the maximum force at the supports,

• Element of curve length d s ^ d x ^ / ' ' from formulas (5.8) and

(5.10) It follows that Therefore

After integrating from jc=0 to x=U2 and multiplying by 2, we obtain formula (5.11).
Cft. $| Dead Weight. Design of Flexible Strings 95

ils value for a symmetrical cable is determined in the following manner.

The vertical components of support reactions have the same value
equal to half of the total load on the cable, i.e. The horizontal
components are equal to force H which Is determined by formula
(5.10). Total reactions of the supports are obtained as the geometrical
sum of these components:

r « -
If the cross-sectional area is denoted by A, the strength condition
for a flexible cable may be written as
a

Replacing H by its value from formula (5.10), we get

Sag f can be determined from this formula provided /, q, A and

lal are'known. The solution is much simpler if q is considered to ac­
count for the dead weight of the cable only; then q=yA, where y is
the unit weight of cable material, and
x yAi* _ yf*
' — 8 i4 [ a T “ 8 [ a J
i.e. cross-sectional area A does not affect the value of /.
B. If the suspension points are at different levels, we find /i and / a
by putting a and x= b in equation (5.8):
f —QJL (5.13)
H
From the second expression we determine tension
qb* (5.14)
H=

Dividing the first expression by the second one, we find

or a = ± b /%
h “ **
Keeping in view that bA-a—U we get

b ± b Y J T = l or b =
96 Complicated Cases of Tension and Compression [Part II
Substituting this value for b in formula (5.14), we finally determine *
II (5.15)
2 (V f* ± Vfi)*
Two signs in the denominator indicate that the cable may have two
main shapes of sagging. The first mode, corresponding to the smaller
value of n (plus sign before the second root), gives us the peak of the
parabola between the cable supports (Fig. 49 and the dashed curve
A O iB in Fig. 51). At the higher value of tensile force H (minus sign
before the second root) the peak of the parabola will be located to the
11

left of support A (solid curve 0 2AB in Fig. 51). We get the second mode
of the curve.
A third shape (intermediate between the two main) of sag is also
possible; it corresponds to the condition / j = 0 . In this case the origin
of coordinates 0 3 coincides with point A. One or the other shape will
be obtained depending upon the ratio between the length of cable along
sag curve AOB (Fig. 49) and chord A B.
If sags / 1 and f a are not known for a cable hanging from supports
at different heights but tension H is known, then the values of a and
b as well as sags f t and fs can be easily determined.
The difference h in the level of supports (Figs. 49 and 51) is
A= W »
Let us substitute the values of /i and f* from equation (5.13) in the above
expression and transform it keeping in mind that a + 6 = /:
h =■§7j + a )~ -ifij(b —a)

• The formula for // In this form was first obtained by Prof. I. Ya. Shtaermau
(iVatika i Tekhnika, Odessa Polylechnieal Institute Journal, 1925).
Ck. S\ Dead Weight. Design of Flexible Strings 97

wherefrom
2Hh
b— a =
ql
and since a+b**l,
„ l Hh , i_ I . Hh
a ~ 2 ql
and fr=3f + *^7-

It should be noted that for a> 0 the first shape of sag will occur
(Fig. 51), at c< 0 , the second shape of sag, and at a= 0, the third shape.
Putting the values of a and b in expressions (5.13), we get the values
of ft and /,:
f h
'l T
and
f q r - .H /r - .h
'* *lx8H + 2qii ^‘ T
C. Let us now see what will happen to a symmetrical cable covering
a span I if its temperature increases to U and the load intensity to qt
(say, for example, due to ice-covering), the initial temperature and
load intensity being U and qu respectively. We assume that in the ini­
tial condition either sag f> or tension H is known. (Knowing one of
these two quantities we can always determine the other from formula
<5.10).l
While calculating the deformation of the cable, which is considerably
smaller quantity as compared to the cable length, we make two assump­
tions: the length of the cable is equal to its span, and tensile force is
constant and equal to H. These assumptions give a small error in gently
sloping cables.
In this case the cable elongation due to the increase in temperature
will be
A S i* a ( / . — JO / (5.16)

where a is the linear thermal expansion coefficient of the cable mate­

rial.
The cable elongates when the temperature increases. This will re­
sult in an increase in its sag and, consequently, in accordance with
formula (5.10) decrease in its tension. On the other hand, from the
same formula (5.10) it is evident that tensile force will increase due to
increase in load. Let us assume that the final effect is the tightening
of the cable. Then, according to Hooke’s law, the elongation of the
cable due to increase in tension will be
(H.>— H ,) l
(5.17)
— zr1—
98 Complicated Cases of Tension and Compression [Part I I
If H2< H u As2 will be negative. When the temperature decreases,
ASi is negative.
Thus, the length of the cable in its second condition will be the sum
of its length in the first condition and the deformations due to the in­
crease in temperature and tensile force:
s8 = Sj “1~AS|-f-ASj (5.18)
The change in the length of the cable will also cause change in its
sag. Instead erf it will become^.
Let us now substitute for St and s2 in equation (5.18) their expres­
sions from formula (5.11), and for the deformations Asx and As8 their
values from formulas (5.16) and (5.17). Then equation (5.18) takes the
following form:

Replace/i and ft by their values from formula (5.19):

and

After certain transformations, equation (5.19) may be written in

the form

« l = [ ^— (5. 20)

Having determined tension Hz from equation (5.20), we can find f 2

from formula (5.9).
If the transition from the first condition to the second one occurs
only due to a change of temperature without any change in the load,
then in equation (5.20) load intensity qt is replaced by qt. If the tran­
sition occurs only due to a change in the load intensity without a
change of temperature, the middle term in the square bracket is
equated to zero.
Obviously, equation (5.20) is also valid for decrease in temperature
and reduction in load intensity.
When the sag is not small compared to the length of span, the for­
mulas derived above strictly speaking are not valid, because the actual
sag curve—catenary—will differ appreciably from the parabola ob­
tained by assuming uniform load distribution over the span and not
over the length of the cable, what in reality takes place.
Accurate calculations reveal that the errors in the value of H are
as follows: for the error does not exceed 0.3%, for '
the error reaches 1.3%, and for j = y the error is slightly more
than 5%.
Ch. 6\ Compound Stress. Stress and Strain 99
C H A PT E R 6

Compound Stressed State. Stress and Strain

§ 27. Stresses Along Inclined Sections Under Axial Tension or Com1
pression (Uniaxial Stress)
In the preceding sections, while testing the strength of a stretched
or compressed bar, we determined stresses only in a section perpendi­
cular to its axis. However, proper evaluation of the critical stresses
in the bar is possible only if we know its state completely; this requires
the ability to calculate stresses not only in sections perpendicular to
the axis.
Let us calculate stresses acting in an arbitrarily inclined section.
Let us consider a prismatic bar stretched by forces/* (Fig. 52). Suppose
Pk
A )£*•
\
\

p\< PM
Fig. 52

the bar is cut into two portions / and II by plane mn forming anglea
with cross section mk perpendicular to the axis. The normals to these
sections also form the same angle.
Let us assume that angle a is positive if mk coincides with mn when
rotated counterclockwise. We shall call normal OA directed outwards
with respect to the cut-off portion of the bar the outer normal to sec­
tion mn. Let us denote the cross-sectional area mk by A 0 and the area
of section mn by A a.
To determine the stresses transmitted through the given section from
the upper portion (/) to the lower portion (II), we imagine the upper
portion to be removed and its action on the lower portion replaced by
stresses pa. To maintain the equilibrium of the lower portion, stresses
pa must compensate for force P and must be directed parallel to the
axis of the bar. It is e vid e n t that the stresses are not perpendicular to
the plane on which they are acting. Their value will also differ from
that in section mk.
100 Complicated Cases of Tension and Compression IPart II

Assuming that at a sufficient distance from the point of application

of external forces P stresses p* are uniformly distributed over section
mn, we find

A
But since A a=*-----,
w cos n.
P cosa
P « = —no
7— = cosa
p
where a ,= j* is the normal stress in section mk perpendicular to the
direction of the tensile force.
The magnitude of stresses p« changes with angle a. In order that
we may have tostudy only one and the same type of stresses irrespec­
tive of angle a, we resolve stresses p« into two components: one in
plane mn and the other in a plane perpendicular to it (Fig. 53). Thus,

Pa

P t
Fig. 55

stress pa acting at point A of plane mn may be replaced by two mutual­

ly perpendicular stresses: normal stress rrB and shearing stress t*.
The magnitude of these stresses will depend upon angle a which the
normal to the section forms with the direction of the tensile force.
From Fig. 53 we have
C7a 33 Pa, cos a = a„ eosaa (6.1)
T« = PaSina=o#sinacosa = y a 0sin2a (6.2)
Let us lay down the following conditions as regards the signs of
stresses a and t a. Tensile stresses oa, i.e. stresses coinciding with the
direction of the outer normal will be considered positive; normal stres-
a i. 6\ Compound Stress. Stress and Strain 101

ses in the opposite direction, i.e. compressive stresses, will be con­

sidered negative.
We will consider the shearing stresses positive if their direction is
such that the outer normal has to be rotated clockwise to make it co­
incide with them. The reverse direction of t« will be considered nega­
tive.
Figure 54 shows the accepted convention as regards the signs of
a , a , and t .
We always have only two types of stresses acting at every point of
the cutting plane irrespective of its angle of inclination a: normal and
shearing.
Figure 55 shows these stresses acting on a thin layer of the material
(hatched in the figure) cut out of the stretched bar by two parallel
sections /-/ and 2-2. Each of the planes experiences normal tensile
stresses oa as well as shearing stresses t a which make sections 1-1 and
2-2 shear one parallel to the other.
It means that the two types of stresses correspond to two types of
deformations: lengthwise deformation (elongation or shortening) and
shear. Corresponding to these two types of deformations we have two
inodes of failure of the material: by breaking away and by shearing.
To check the strength of the material, it is essential to determine the
maximum values of att and ra depending upon the location of plane rnn.
It follows from formulas (6.1) and (6.2) that oa reaches its maximum
value when cos4 a is equal to unity, i.e. a= 0 . The maximum value
of Ta is obtained when sin 2 a = l, i.e. when 2a=90g or a = 45°. The
maximum values of oa and t a will be
maxoa =<r() = -^ -, maxxa = ^ (6.3)

Hence, the maximum normal stresses are acting in sections perpendi­

cular to the axis of the bar; the maximum shearing stresses act in
sections forming an angle of 45° with the axis of the bar and are half
of the maximum value of the normal stresses.
A logical question that arises is for which of these stresses should
the bar be tested, which of these stresses plavs the decisive role in the
failure of material. These points will be discussed in detail in Chapter 7.

§ 28. Concept of Principal Stresses.

Types of Stresses of Materials
In the preceding chapters we got acquainted with the behaviour
of materials under axial (or, as it is often called, simple) tension or
compression. However, there may be cases in practice when the ele­
ment is subjected to tension or compression in two or three directions
under the action of external forces, i.e. it finds itself in a composite
stressed state.
102 Complicated Cases of Tension and Compression tPart l l

In § 27 we showed that even under simple tension two types of stres­

ses are possible: normal a and shearing t. It follows from formulas
(6.1) and (6.2) that in sections perpendicular to the axis of the
stretched bar (a=0), wehaveonly normal stresses (t= 0), and In sections
parallel to its axis (a—90°), we
have neither normal nor shearing
stresses (cr=0 and t =0).
The planes in which the shearing
stresses are. totally absent are called
principal planes; the normal stresses
acting in these planes are called
principal stresses.
It has been proved In the theory
of elasticity that three mutually per­
pendicular principal planes through
which three principal (normal)
stresses are transferred can be drawn
through an arbitrary point of a
stressed body. Two of them have ex­
Fig. 56
treme values: one is the maximum
normal stress, the other is the mini­
mum normal stress; the third prin­
cipal stress is intermediate between the above two. In every point of
a stressed body we carl isolate an elementary cube whose faces are the
principal planes. The cube material is stretched or compressed by
three mutually perpendicular principal stresses which are transmitted
through the principal planes (Fig. 56).
In the case of simple tension (§ 27) one principal plane at every point
is perpendicular to the bar axis (a—O0), and the other two are parallel
to it (a=90°). Since the normal stress is not zero (Ca^O) in the first
principal plane and in the other two it vanishes, it may be concluded
that in simple tension and compression out of the three principal stres­
ses only one is not equal to zero at any point of the bar; this principal
stress is parallel to the tensile force and the bar axis. This stress of the
material is called uniaxial. The element isolated from the bar is de­
formed in only one direction.
There are cases when the cubic element of the material is subjected
to tension or compression in two mutually perpendicular directions
or even in all three directions (Fig. 56). When two principal stresses are
not equal to zero, the material is said to be in biaxial (plane) stress.
When all the three principal stresses are not equal to zero in the given
point, this pertains to the most general case of stress distribution in the
material, the iriaxial (volumetric) stress; the elementary cube is sub­
jected to tension or compression in all three mutually perpendicular
directions.
Jn future we shall denote the principal stresses by ait a, and <j8.
Ch. 6 ] Compound Stress. Stress and Strain 103

The order of numbering the principal stresses will be set in such a

way that Oi represents the maximum stress in algebraic value, and
<ts the minimum one. The compressive stresses will be taken, as before,
negative Therefore, if, for example, the principal stresses have the
values of +1000 kgf/cm1, —600 kgf/cm1, and +400 kgf/cm1 the num­
eration should be
<ji = + 1000 kgf/cm1, <ia« + 400 kgf/cm1, o4= — 600 kgf/cm1
then the condition will be satisfied.
Thus, we distinguish three kinds of stressed states:
1. triaxial stress, when all the three principal stresses are not equal
to zero (for example, tension or compression in three mutually perpen­
dicular directions);
2. biaxial stress, when one principal stress is equal to zero (ten­
sion or com pression in two directions);
3. uniaxial stress, when two principal stresses are equal to zero
(tension or compression in one direction).
In § 27 we studied the stress distribution in uniaxial stressed state;
below we give examples of planar and volumetric stressed states ex­
plaining how stresses are distributed in different planes in these cases.

§ 29. Examples of Biaxial and Triaxial Stresses.

Design of a Cylindrical Reservoir
A. As an example of a composite stressed state we shall consider the
stresses in the material of a thin-walled cylindrical reservoir which is
filled with gas, steam or water at pressure of q atm, i.e. q kgf/cm1. The

side walls and the bottom of the reservoir are subjected to a uniformly
distributed pressure q. The dead weight of the fluid in the reservoir is
ignored.
The pressure on the bottom will tend to break the cylindrical portion
across the cross section; on the other hand, the pressure on the side
walls will tend to burst the reservoir along the generatrix of the cy­
linder. Thus, if we isolate rectangular element A BCD from the cylin­
drical portion of the resevoir, this element will be subjected to tension
104 Complicated Cases of Tension and Compression [Part / /

in two directions: by stresses cr' in sections perpendicular to the gene­

ratrix and stresses a" in sections along the generatrix (Fig. 57).
The method of sections will be employed for calculating stresses
o' and or". Suppose the internal diameter of the reservoir is D, and
the thickness of its walls is/. Weshall consider / to be small as compared
lo £> (/< |).
Let ns imagine the reservoir (Fig. 57) cut along the plane 3nd con­
sider the equilibrium of the cutoff pari, for instance, the right one

(b)
rtg. 58

(Fig. 58 («)). The resultant of the forces acting on the bottom and
stretching the cylindrical portion of the reservoir along the generatrix is

The area of the ring (a thin strip of thickness / and approxi mate length
ji D)upon which this force ads is
A » InD
Hence normal stress in this section is:
rrZ>-

A IjiD 41
Ch. Cl Compound Stress. Stress and Strain 105

Stresses a" in sections parallel to the cylinder generatrix will be

found by isolating a ring at some distance from the reservoir bottom
cut by sections mn and m V at a distance a from each other and by
considering a diametrical section of this ring (Fig. 58 (6)). The diametri­
cal surface of the gas (or fluid) experiences a pressure? which has a re­
sultant Pi^aDa. The area of the diametrical section (two walls)
which bears this pressure in Ai=2ta, and stresses in the walls are
qDa qD
'l t T = - W
These stresses are two times greater than stresses n' acting in the
ring section. s
Since there are no shearing stresses in the ring and diametrical sec­
tions, the sections A and A\ qualify as principal planes, and stresses
o' and o" as principal stresses. The third principal stress o‘" ——q
acting on the reservoir wall in the radial direction is negligibly small
sis compared to o' and a"; it may therefore be considered equal to zero.
Consequently, element A BCD cut out of the reservoir wall (Fig. 57)
is subjected to plane stress (biaxial tension). In accordance with the
accepted numeration, the principal stresses are

. 0s = TT and °3 = 0 M

Biaxial state of stress also occurs in spherical, conical and other

thin-walled vessels, plates, various types of shells, etc.
B. The example of a triaxial stress is transfer of pressure from the balls
to the race in a ball-bearing or from the wheels of a rolling stock to
the rails.
As the contact between the rail head and the tyre may be looked
upon as that between two cylinders of different diameters and crosswise
generatrices, these surfaces must touch each other at a point. The nor­
mal stresses arising at the point of contact when pressure is transmitted
from one body to the other are known as contact stresses.
When force is transmitted, the materials of the tyre and the rail
get deformed around the point of contact, and pressure is transferred
through a contact surface of elliptical shape. The area of contact de­
pends upon the pressure and the radii of the contacting surfaces. If
we cut out a small cube (for example, with sides of 1 mm) of the rail
material at the centre of the contacting surface, and if the faces of
this cube are parallel and perpendicular to the rail axis (Fig. 59),
then the stresses acting on the faces will be normal compressive *.
Thus (Fig. 59(6)) we have three mutually perpendicular planes loaded
by principal stresses o', o", and a The emergence of lateral stresses

* These stresses are calculated in the theory of elasticity.

 IOC Complicated Cases of Tension and Compression [Part I I

a" and o '" can be explained as follows: under the action of stress a '
perpendicular to the plane of transmission of pressure, the cube ma­
terial tends to expand laterally, and this results in reactions o* and
o '" from the rail material surrounding the cube, that hinder transverse
deformation.
The computed values of these stresses* show that they actually
attain high values. Thus, for example, the values of o', or*, o '" at the
contact between the locomotive runner and rail are
ct' = --110 kgf/mm*, o' — 90 kgf/mm*, o ' " = — 80kgf/mm*

By applying the convention of numeration of principal stresses to

this example, we get

ai = *=— 80 kgf/mm*, o2 =*<f =* — 90 kgf/mm*,

o3 =o'=*r—n o kgf/mm*

In the given example all the three principal stresses are negative.
This is a case of triaxial compression. An example of triaxial tension
ts the yielding of material at the neck in a specimen subjected to ten*

Fig 59

siofi. We also often come across cases of composite triaxial stressed

state in which the principal stresses have opposite signs: o (!> 0
and oa< 0 (for instance, in the wall of a thick-walled boiler).
The triaxial stressed state is the most general stale of stress at a
point; the biaxial and uniaxial stressed states are the particular cases
when one or two of the three principal stresses become equal to zero.

* N. M- Belyaev, Compulation of maximum design stresses in compression of

contacting bodies’*, in Proceedings of the Leningrad Institute of Railroad Engineers,
issues 99 and 102 (1929).
Ch. 61 Compound Stress. Stress and Strain 107

§ 30. Stresses in a Biaxial Stressed State

It is essential to determine the maximum normal and shearing stres­

ses to check the strength of a material in biaxial or triaxiai stress.
Let us begin with the biaxial stress. Let us assume that principal
stresses <7< and cr» are acting on the side faces of a right-angled paralle­
lepiped (Fig. 60). Both these stresses are tensile. There are no stresses
on the front faces of the element; therefore the third principal stress
is zero. If one of the principal stresses a*, aa or both are compressive,

Fig. 60

then their values in the succeeding formulas must be taken with a mi­
nus sign, and the numeration should be altered in accordance with the
order given in § 28. Thus, if one of the principal stresses is tensile and
the otner is compressive, then the first will have to be numbered o*
and the second o3; if both of the stresses are compressive, then the stress
having lower absolute value will have to be numbered o8. and the
greater o3.
Our aim is to determine the maximum normal and shearing stresses
in sections perpendicular to the front faces.
Let us draw a section the normal to which forms angle a t with di­
rection / (Fig. 60). The same normal forms angle a* with direction //.
This section will be subjected to both normal stress oc and shearing
stress t«, which depend upon o* and a,. Their values can beobtained
by studying the action of at and <Ta separately and summing up the
results. The fraction of the normal stress caused by Oi may be expressed
according to formula (61) as ot coss oti; the other fraction of oa,
caused by stress <j8, may be written according to the same formula as
108 Complicated Cases of Tension, and Comprission IPart 11

ori cos* <x2. The total normal stress then becomes

<ja = a, cos* a, + <jacos* a a = a, cos* a , -j- o4 cos* (a, -f 90c)
or
aa = a, cos* a, -f <j2 sin4a, (6.5)
By similar reasoning and with the help of formula (6.2) we may
find the shearing stresses in the given section:

= y [c>sin ^a >+ a«s,n = y 1°1 i><n2a, + o„sin 2 (a, + 90')]

or
^ = 2 ^ sin 3a* (6 .6 )

In these formulas angle aj has been measured from the direction

of axis I (stress a,) up to the normal to the given section by rotating
counterclockwise. We shall follow the rules laid down earlier in § 27
in choosing proper signs for ow and xa as well as for angles a t and a*.

In future, in formulas giving the values of oa and xa we shall denote

a, by a, always measuring it from the maximum (algebraic) principal
stress in the anticlockwise direction.
Employing formulas (6.5) and (6 .6) which give the stresses in section
a-a (Fig. 61), we can easily determine the stresses in perpendicular
section b-b which has normal n$ forming angle (^a-f-9041 with the
Ch. 6] Compound Stress. Stress and Strain 109

direction of the maximum principal stress:

On—<j, cos* p+c* sin* ft —o, cos* (a -f 90°) -f os sin* (a + 9(f)
<T|)= or, sin* a -f or*cos* a

T? = ^ ^ s i n 2 p = 2 i ^ s i n ( 2 a 4 - 1 8 0 g) 1
Tp = " '- g — sin 2 a J

The formulas derived above clarify the properties of stresses acting

in mutually perpendicular planes. For normal stresses we have
oft = o, cos* a -f o8sin* a
or„= Oj sin* a cr, cos* a
Summing up, we get
0 a -t-ff0 = a , + o a=coo$t (6.7)
i.e. the sum of normal stresses in two mutually perpendicular planes
is constant and equal to the sum of the principal stresses.
For the shearing stresses, by comparing (6 .6) and (6 .6 '), we get
t * - — *« (6 .8)
Hence, the shearing stresses in two mutually perpendicular planes
are equal in magnitude but opposite in sign. This property is generally
called the law of complementary shearing stresses, this law being valid
in all cases in which shearing stresses are acting.
The system of stresses oa, oftr Ta, t p depicted in Fig. 61 acts on the
faces of an elementary parallelepiped turned through angle a with
respect to the directions of principal stresses cr, and a*. The pair of
shearing stresses that tends to rotate the element in the clockwise
direction will beconsidered positive. In Fig. 61 these stresses are denot­
ed by Tft. It should be noted that this rule for choosing the sign for x
coincides with the convention already decided upon ($ 27).
It is evident from formulas (6.5) and (6 .6) that the normal and
shearing stresses in a plane depend upon its inclination.
Let us study expression (6.5) for maximum to determine the ma­
ximum normal stress. By diiferentiating with respect to a and
equating the first derivative to zero, we get
—_ 2 at cosa sm a + 2a* sin a cosa = 0
or
“ — (<*« — <>i) sin 2a - 0 (69)
UO Complicated Cases of Tension and Compression IPart I!

A comparison of expressions (6.9) and (6 .6) reveals that the condition

Tor maximum of ora is the same as obtained by equating to zero the
shearing stresses in the corresponding planes. From the same expression
it follows that oa^ a x cos*a-f-ors sin2 a is maximum either for a = 0
or for Since Oi>Oa. then
max oa = <r1 (ata = 0)
mi noa — ( at a=90°)
i.e. the maximum and minimum normal stresses at the given point
are the principal stresses at and o3 acting in mutually perpendicular
planes free of shearing stresses.
It is evident from formula (6 .6) that the maximum shearing stress is
m axta = <r<~ » - (a ts in 2 a = !, i.e. at a = 45°) (6.10)
Hence, the maximum shearing stress is half of the difference of the
principal stresses and acts in planes inclined at 45° to the principal
ones and perpendicular to the plane of the diagram. In planes parallel
to oT|, the maximum shearing stress is
max t 2 = ( 6 . 10 ')

§ 31. Graphic Determination of Stresses

(Mohr’s Circle)
The calculation of oa and ta from formulas (6.5) and (6 .6 ) may be
replaced by graphic determination (Fig. 62).
Let us take a rectangular coordinate system with axes a and t.
Theo-axls directed to the right is taken positive. On the cr-axis we plot

segments 0/1 and OB representing in a certain scale the numerical values

of a t and <Ti(it is convenient to draw theo-axis parallel to the maximum
principal stress a,).
Ch. 51 Compound Stress. Stress and Strain HI

In Fig. G2 both these stresses are considered tensile and are laid off
on the o-axis in the positive direction. Had one or both of the stresses
been compressive, we would have laid them off in the opposite direc­
tion. Taking segment AB as the diameter, we draw a circle with the
centre at C, which is called the stress circle (Mohr’s circle). To deter­
mine normal stress oa and shearing stress x« in a plane the normal to
which makes angle a with the maximum principal stress a u we must
draw a central angle 2a at point C, plotting its positive value from the
a-axis counterclockwise. Point D of the stress circle will correspond
to the required plane; its coordinates OK and DK will be equal to
oa andxa, respectively. This can be easily proved. From the diagram,
the radius of the stress circle is
CD = A C = BC = ^ =,

From the right-angled triangle KDC we have

D K = CD sin 2 a = sin 2a = xa

Further
OfC=OB + 5C + C A r-oi + ^ ^ + 2 i ^ ? c o s 2a
=o3 (i 4 . cos 2 a) —o„_}_2 izJZ?2 cos*a
= o 4-f <Jt cos* a —a 3 cos* a = cr1 cos* a-j-o 3 sin* a = o a
Thus, the coordinates of points on the circle determine the stresses.
The values of oB are measured by the segments along the o-axis. Po­
sitive values of crB are plotted in the positive direction of the o-axis.
The values of xa are measured by the segments parallel to the x-axis.
Positive values of xa are directed upwards, because according to the
convention decided upon by us, the values of a between 0 and 90° cor­
respond to positive values of xa; this is also obvious from the formula
xa = g |~ CT2sin 2a

in which the maximum principal stress is taken as o<.

Having determined stresses oa and xa from the stress circle, let us
represent them on the diagram of the cutoff element, taking care
of their signs (Fig. 62). Let us recapitulate that we have decided to
plot angle a specifying the location of the outer normal to the cutting
plane always from the line of action of the maximum (algebraic) prin­
cipal stress. Let us therefore bring the direction of the maximum prin­
cipal stress Ot in line with the o-axis on the stress circle. Then line BD
inclined at an angle a to the o-axis will be parallel to the normal to
the cutting plane, i.e. parallel to oa. Line BM will be parallel to xa.
112 Complicated Cases of Tension and Compression | Part II

As is qiear from Fig. 62, the maximum shearing stress is equal to

segment CD9, i.e. the radius of the stress circle:
0| —0*
max x« =
angle 2a corresponding to this condition is 90° and, consequently,
a=45°. In the stress circle max t a is represented by the ordinate CDo
whose abscissa is i.e. in the plane where the
normal stress has an average value.
It is similarly clear from Fig. 62 that the maximum normal stress
is represented by segment OA which is equal to Oi, and the minimum,
by segment OB equal to <r3.11 ensues that the normal stress in any plane
at an angle a must be between the principal stresses <ri and o8.
Thus, knowing the principal stresses at a point of a body in biaxial
stress we can find the stresses and their directions in any other plane
passing through this point with the help of Mohr’s circle.

Let, for example, the principal stresses at some point of the material
be Oi=300 kgf/cm* and o*=>—700 kgf.'cm*. We shall find the normal
and shearing stresses in a plane inclined at a ——30° to the direction
of <Ti. The construction is shown in Fig. 63. For the chosen scale the
stresses were found to be oa=50 kgf/cm* and xu= —430 kgf/cm*.
Their directions are shown in Fig. 63 on the right.
If the principal stresses <Ti and aa are known, then with the help of
the stres^ circle we can determine the stresses in two mutually perpen­
dicular sections a-a and b-b the normals to which (Fig. 64) make
angles a pnd P, respectively, with the direction of the maximum prin­
cipal stress ox.
Ch. 6} Compound Stress. Stress and Strain 113

Let us plot angle 2a at point C of the stress circle (Fig. 64). Point
Da will correspond to section a-a, and segments DaK* and OK* will
represent the respective shearing and normal stresses in the plane.
To determine the stresses in section b~b we must plot angle 2 B,
i.e. add 180° to angle 2a. All that is required for that is to extend ra­
dius CDa; point Dp will correspond to section b-b.

Stresses t p and <rp are represented by segments Dpffp and OK$, res*
peetively. It is clear from the diagram that t p=3 —xa and
<*a+ = a, + oa = const
The stresses acting on the faces of the element cut by planes a and b
are shown in Fig. 64 on the right.
By bringing in line the direction of the maximum (algebraically)
principal stress or with the a*axis on the stress circle (Fig. 64), we

Fig. 66

find that line BD* joining the extreme left point of the circle with
point Da is parallel to stress oa, and line BD&is parallel to stress <rp.
The arrows are put in accordance with the signs obtained.
Figure 65 shows how to construct Mohr’s circle when both of the
principal stresses are compressive.
5 - miq
114 Complicated Cases of Tension and Compression IPart I I

§ 32. Determination ot the Principal Stresses with the

Help of the Sticss Circle

Sometimes it is required to solve a problem opposite to the one dis­

cussed in the preceding section, i.e. determine the principal stresses
if the stresses cr?, x«, erg, and tg are known. The easiest way of doing
that is by plotting Mohr’s circle.
Assume that tfie normal and shearing stresses in two mutually per­
pendicular planes having normals n x and n v are known (Fig. 66). Let

Fig. 66

us denote the normal stresses in the vertical plane (nx) by tra, and in
the horizontal piane, by erg since they make certain angles a and (3
(JS^a-HIO*) with the principal stresses <Tj as yet unknown. The shear­
ing stresses are correspondingly denoted by xa and xg; according to
the law of complementary shearing stresses, xa= —xp. For the sake
of definiteness while constructing Mohr’s circle let us assume that
aa><Tp>0 j and Ta> 0 .
Let us plot stresses oa, <Tg, xa, and xg using the coordinate system
of the required stress circle (Fig. 66 ):

Oa~0Kat ^H~OKfl, Ta = /Ca^a

= KvPo.^ — KpDg

Since points Da and Dp corresponding to mutually perpendicular

sections nfust lie at the opposite ends of the circle diameter, the point
of intersection of line Da Dg with the cr-axis will give centre C of the
stress circile. Circumscribing a circle of radius CDa or CDg around
centre C, we get segments OA and OB on the o-axis which represent
the principal slresses: OA=aj and OB=aa.
CKk\ Compound Stress. Stress and Strain i 15

The direction of ca is represented on the stress circle by BDa which

is inclined at a positive angle a to the cr-axis. Consequently, angle a
should be plotted in the anticlockwise direction moving from point A
towards Da in order to pass over from line Oi to line ca in the circle.
In our example, we assume the direction of oa to be known. This means
that in order to represent the direction of ax in the diagram of the ele­
ment under consideration, we must plot angle a in the opposite direc­
tion from aa, i.e. in the clockwise^direction. The relative disposition
of stresses Oi and cr* shown on the stress circle by OA and BDa must be
retained in the diagram of the element as well.
We may also show on the stress circle the true direction of principal
stress <T| as coinciding with the direction obtained on the diagram of
the element by the method explained above. For this from *he extreme
left point B of the circle, we must plot an angle a in the clockwise
direction from the o-axis which is parallel to <ya, in other words, point
D 0 should be brought down to D*. Line BD’a coincides pn direction
with stress Oi, and cr2 will be directed perpendicular to it. While repre­
senting the principal stresses (in our example o, and or.) it is essential
to take care of their signs obtained by plotting the circle, and also fol­
low the rule of numeration of the principal stresses.
Let us point out that in the problems on biaxial stress discussed here,
the third principal stress is zero. Therefore, if both principal stresses
obtained from the stress circle are positive (Fig. 66), then the higher
one will be oi and the lower a 2; if one of the stresses is positive and the
other negative, then the former will be <sr, and the latter a3; finally,
if both stresses are negative, then the one with the greater absolute
value will be o8 and one with the smaller absolute value o*.
Angle a may be determined by the formulas (Fig. 66)
\

(6 . 11)
0i —Op

The minus sign is used because for positive values of oa and xa angle a
(the angle of rotation of plane cr« to the principal direction) is mea­
sured in the clockwise direction.
From Fig. 66 , we can get the formulas for calculating the principal
stresses in biaxial stress; they are represented by segments OA and
OB. From the diagram we have
OA = OC+C A and OB = OC— CB
Further
OC=2«+£&, C K a -C K ^ — ^-
116 Complicated Cases of Tension and Compression [Pari l l

The radii of the stress circle CA*=*CB are equal to CDK=CD$ which
may be found from the following expression:

CA~CB=* CDa = V CKl + K M » *[/r<£2L ^ i ! +

= —OTp)*-j- 4t£
Therefore,

= i [(aa + a ,) ± Y i f l n - » , ) * + 4 ^ ] (6.12)

In practice we often come across the cases of biaxial stress when

ofl= 0 . For these cases the formulas for principal stresses will take the
form

’j - t k i R + 5 ? . ] (6.13)

Here the minimum principal stress is denoted by <j« because it is nega­

tive (the quantity under the radical sign is greater than cra).
The angle of inclination of the first principal stress to the o-axis
is determined by formulas which are a corollary of (6 . 11):
ta n 2a = — -^ 2-
Oa (6.14)
ta n a = — — /
ai
Given below are examples on determining the principal stresses
with the help of stress circle.
Suppose we know the stresses at a given point of the material, acting
in two mutually perpendicular planes:
or* = 400 kgf/cm*, Ta = — 300 kgf/cm3
Op = — 200 kgf/cm3, tp = 300 kgf/cm3
Figure 67 shows Mohr’s circle constructed for these data. The prin­
cipal stresses are
<t, = 530 kgf/cm*, a, = — 330 kgf/cms (o„ = 0)
and the angle between aa and o, is a = 22 °.
In another example
oa —1000 kgf/cm3, xa = 400 kgf/cm1
Op = 0 , = — 400 kgf/cm3
Ch. 61 Compound Sirens, Stress and Strain 117

The plotting of the stress circle is shown in Fig. 68 from which iI

ensues that Oi=1140 kgf.cm* and a*——140 kgf/cm1. On the front
face that lies in the plane of the figure, o 4—0 .

For both examples the reader is advised to calculate the principal

stresses according to formula (6 . 12) and compare the analytical values
with those obtained by graphic construction.

§ 83. Stresses in Triaxial Stressed State

In the general case of a state of triaxial stress, normal as well as
shearing, stresses act on the faces of an elementary cube cut out of the
material of a body (Fig. 69). In accordance with the law of complemen­
tary shearing stresses, xxt=*xzx, and x»z=%zy*. The set of
six stresses ox, a, and xXit, xXI, xyx completely describes the state
of stress at a point and is known as the stress tensor.
It is established in the theory of elasticity that around any point
of stressed material we can always isolate an elementary cube in which
no shearing stresses act on the faces by rotation of planes. In this case
the stress tensor is determined by three principal stresses au ait and o3.
* The subscripts on t should be deciphered as follows: the first subscript de­
notes the plane in which they act (direction of the normal to the plane), the sccon 1
subscript denotes the direction of shearing stress (along which axis t is acting).
118 Complicated Cases of Tension and Compression [Part II
In particular, when 0 i=<xa“ 0 «—o (uniform triaxial tension or compres­
sion), the stress tensor is known as spherical.
Suppose we have a cubic element cut from the body. The faces of
the cube are subjected to principal stresses Oi, o 2 and o8 (Fig. 70). Our
aim is to determine the normal and shearing stresses in any inclined
plane cutting the given cube, provided Oi>o£>o£>0.

Fig. 69

First we shall determine these stresses in planes parallel to one of

the principal stresses, for example c a. This plane is hatched in
Fig. 70 (a).
We have seen earlier (§30) that the principal stress parallel to a
given plane gives rise to neither normal nor shearing stresses in it.

Fig. 70

Therefore, stresses in the planes under consideration will depend only

upon ot and cr8—we will again deal with the biaxial stress. Points on
the stress circle drawn for the principal stresses ot and os (Fig. 71) will
correspond to these planes.
Ch. 6] Compound Stress. Stress and Strain 119

Identically, stresses in planes parallel to a3 (Fig. 70 (b)) will be

represented by the coordinates of the points of the stress circle const­
ructed for stresses Oi and os. In planes parallel to a, the stresses will
be represented by points of the stress circle constructed for a , and cr,
(Fig. 70 (c)).

Thus, coordinates of the points on three stress circles (Fig. 71) rep­
resent the normal and shearing stresses in sections of the cube which
are parallel to one of the principal stresses.
As for the planes cutting all the three axes of principal stresses, it
has been proved in the theory of elasticity that stresses a„ and xa
are represented by coordinates of points D in the hatched area of
Fig. 71.
The values of these stresses may be calculated by the following for­
mulas:
an ■*<*i cos* cty 4 - at cos4 a,+ < js cos* a, (6.15)
— cos* cos* a , -f o? cos4 a 0 —or* * (6.16)
Here a*, a fl and cta are angles which the normal n to the plane makes
wdth the directions of principal stresses oa and <rt , respectively.
It is clear from Fig. 71 that in triaxial stress the maximum and mi­
nimum normal stresses are equal to the maximum and minimum prin­
cipal stresses, respectively.
The maximum shearing stress is equal to the radius of the largest
circle and, consequently, half of the difference of the maximum and
minimum principal stresses, ft acts in planes inclined at 45 ° to the
direction of these principal stresses, the normal stresses in these planes
being equal to half of the sum of the maximum and minimum prin­
cipal stresses ( o t ^ a ^ a ) .
Thus, in the most general case of the stressed state of a material,
When all the three principal stresses are nonzero at the given point
120 Complicated Cases of Tension and Compression [Puri II

we have
m axa^O p m inan = a „ maxTfl= 2 i ^ l (6.17)

In planes parallel to one of the principal stresses and inclined at

45° to the other two, the shearing stresses will be maxx=x lt3 according
to formula (6.17), and further
_ _°i—0* _ _ <*a—a3 (6,17')
*i. a £ * ^a. a = 2

The stresses x*, „ xu », and r,, 3 are sometimes called the principal
shearing stresses. '
For checking the strength of material in compound stressed state
(see Chapter 7) it is of interest to know the stresses in the octahedral

Fig. 72

planey the normal to which makes equal angles with the directions of
all the three principal stresses (Fig. 72). Bearing In mind that
cos* a t + cos* a 4-f cos* 03 = 1
and when the angles are equal ( a ,= a 2= a 3= a ), 3 cos2a = l , orcos3 a =
=1/3, from formulas (6.15) and (6.16) we obtain

*oct = j (<7i + oa-f-a8) = (6-18)

(<h— <**)*+ (<T,—<r3)* + («!—<Ta)* (6.19)

Using expression (6.17) for the principal shearing stresses, we get

4 K li..+ (6.19')
a i. 6] Compound Stress. Stress and Strain 121

It is evident from expressions (6.18) and (6.19) that the normal octa­
hedral stress is equal to the arithmetic mean of the three principal sires*
ses, whereas the octahedral shearing stress is proportional to the geo­
metric sum of the principal shearing stresses.
An expression of the type (6.19) will be used in Chapter 7 under the
name of stress intensity, which also characterizes the stresses in a ma­
terial:

^ = y = V (o, —a*)14 - (0 a—0 a)a + (0 , —0 3)* (6 .20 )

It can be easily seen that when 0 „= 0 #=O, i.e. in the case of simple
uniaxial tension, intensity ot= au

§ 34. Deformations in the Compound Stress

When testing the strength of an element (Fig. 56) whose faces are
subjected to stresses 0 *, at> and ait it becomes essential to determine
the corresponding deformations. Let us number the edge parallel to
principal stress 0 , as first, and those parallel to principal stresses o#
and 0 a as second and third. Let us now determine the relative longitu­
dinal deformations of the element along these edges by considering
the effect of each stress separately and then summing up the results.
Under stress or, the element will get elongated in the direction of
the first edge, and the relative elongation is

The first edge, however, is simultaneously the lateral dimension

for stresses a2 and o3; therefore, the element undergoes relative shorten­
ing in the direction of the first edge due to stress oa and stress o*,
which is equal to (see § 9)
* Of iff Om
«! = — = “ *F“T

The total relative deformation in the direction of the first edge may
be written as

e, = el + e i+ el’' = -^-—p p -
122 Complicated Cases of Tension and Compression [Part II

Similar expressions may be written for deformations in the other

two directions, and we finally get

( 6. 21)
K tT + T t )
_ °3
■ > •(* + * )
If some of the stresses Ou o», cr3 are compressive, their numerical
values should be put in formulas (6 .21 ) with a minus sign.
Now from (6.21) we can easily get expressions for tension or compres­
sion in two directions by putting one of the principal stresses equal
to zero. For example, for tiie case shown in Fig. 60, we have
_ *1
-'i— T *l - r
— (6 .21 a)
'2 ■p t
e, = —p- -f* £
Let us calculate the change in Ihe volume of a rectangular paralle­
lepiped having edges of a, b and c, if it is under triaxial stress. Its
volume before deformation is Vd=abc. After deformation, due to elon­
gation of its edges its volume becomes
Vi =• (a + Aa) (b -f A6 ) (c *f Ac)
or, neglecting the product of small deformations,
Vv= cibc -f abXc -f- qcA6 + bc&a = V„ (1 -f- ex+ e±4 -s,)
The relative change in volume is
«v; ( . 6 22)

Replacing the sum of relative elongations by some mean

e, -i et ~ z x
Btni*.ni 3

we express the relative change in volume as

zY = 3emcan (6.22')
Replacing in (6.22) the values of ei, e3) and e3 from (6.21), we get
Sy = et + ' (<*1 + “T(6.23)
Cn. 6] Compound Stress. Stress and Strain 123

It is evident from (6.23) that if Poisson’s ratio |ul is equal to j , the

relative change in volume is zero. We have already obtained this re­
sult for uniaxial stress in § 9. It is clear from the same formula that
if the sum of the three principal stresses is equal to zero, there will be
no change of volume within the limits of elastic deformation.
It should be noted that formulas (6.21), (6.22), and (6.23) can also
be used for an arbitrarily orientated element of the material the faces
of which experience both the normal and shearing stresses (Fig. 69).
For this all that is required is to replace c Jt a 2, and c» by normal stres­
ses ox, ov, and <r?, and &i, e*, en by e„, and e*. It will be shown later
(§ 36) that shearing stiesses change neither the linear dimensions of
the element nor its volume.
Let us return to-formula (6.23) from which it is obvious that the
change in volume depends only on the sum of the principal stresses
and not on their ratio. This means that the volume will change by the
same value if the cube’s faces are subjected to equal mean stresses

0 me«n 3

The relative change in volume may in this case be expressed as

‘7 = ^ ^ . <6.23')

£
The quantity _2 . is called the bulk modulus. Introducing
this notation in formula (6.23), we obtain
_ _ffmean + <*a (6.24)
K 3K

^mean = K&y ~ 3/Cemean (6.24')

Formulas (6.24) and (6.24') describe the general Hooke’s law similar
to Hooke’s law for uniaxial tension. It is evident from these formulas
that if equal mean stresses

u m ea n 3

forming the spherical stress tensor are applied to the cube’s faces, all
the edges experience identical strain
_ °mean (6.25)
m ean 3/ f
124 Complicated Cases of Tension and Compression | Part 11

In this case the change in volume of the cube is not accompanied by a

change of its shape—the cube remains a cube, but the dimensions of
the new cube are different. Therefore, if we are interested in problems
related to the change in volume and shape under compound stress, it is
convenient to represent each of the principal stresses as a sum of two
stresses:
c i • - ^inean + <T2 —<*mc«n+ = ^mean +
The given stress tensor formed by the principal stresses 0 1? cr2, and
trs consists of two terms: the spherical tensor (made up of equal stresses
Gthm
t- ____cZ~smwn
7 K
t'v fa a& I 6 mean ^T ^m eo n i Gf&uttaa

T
Gfitcn 7tfneon
T
Fig. 73

^mtan) and 3 supplementary tensor known as the stress deviaior (Fig. 73)
which represents a system of normal stresses
= ^mean* = ^ “ "^mean* 0 ^= 03 —Onican
It can be easily seen that the sum of these supplementary stresses
is equal to zero. Obviously, aH -oJ+a;—0^ + 03+ 03—3oroean= 0,
therefore they do not cause any change in volume (§ 34 ). The stress
deviator (Fig. 73 on the right) is only responsible for the change
of shape.
We shall return to the problem of the change in volume and shape
later while discussing problems of strength of materials in compound
stressed state (Chapter 7).

§ 35. Potential Energy of Elastic Deformation

in Compound Stress
Potential energy of deformation is the energy accumulated by the
material as a result of elastic deformation caused by external forces.
To calculate the potential energy accumulated by an elastic system,
we may use the law of conservation of energy.
Let us first consider the case of simple tension (Fig. 74). If we load
a bar statically by gradually suspending small loads A/>, then after
each addition the suspended load comes down and its potential energy
decreases, whereas the potential energy of deformation of the stretched
bar increases.
Ch. 6] Compound Stress. Stress and Strain 125

When the load increases slowly and gradually, the velocity of displa­
cement of the iree end of the bar is very small. Therefore, we may ne­
glect the inertia of the moving mass and, consequently, assume that
the deformation is not accompanied by any change in the kinetic ener­
gy of the system.
Under these conditions the potential energy of the lowering load is
transformed into the potential energy of elastic deformation of the bar
(we neglect the dissipation of ener­
gy due to thermal and electromag­
netic processes accompanying the
elastic deformation). Thus an elastic
system under static loading maybe
considered as a machine transform­
ing one form of potential energy into
another.
As the potential energy lost by the
load is equal to the work accomp­
lished by it in lowering, the problem
of determining the potential ener­
gy of deformation comes to calculat­
ing the work done by the external
forces. In § 10 we obtained expres­
sion (3.1) for the work done by the Fig. 74
external forces in simple tension:
v -s #

This implies that the potential energy of tension is also

y = = <6 2 «>
a/ Pi
since &1=y a '
The potential energy accumulated by a unit volume of material is
<0 <T8
u = w= (6.27)

Let us now pass over to the determination of potential energy accumu­

lated in a unit volume of a material which is in a compound (planar or
volumetric) stress. Making use of the principle of superposition of
forces and assuming that the principal stresses increase gradually, we
can determine the potential energy as the sum of the energies accumu­
lated by a unit volume of the material under the action of each of the
principal stresses Oi, o2, and o3 according to (6.27)
W=W "TT"*fliCj i OaSa i
§ * 2"
126 Complicated Cases of Tension and Compression [Part 11

where e*, e* and e3 are strains calculated from formulas (6.21). The spe­
cific energy of deformation will be

or after multiplication

u = i t°»'+'a*+ + ai<V)J (6.28)

Hence, the total energy of deformation accumulated in a unit volume
of the material (a cube with edges of unit length) may be calculated
from formula (6.28). It may be considered as consisting of two parts:
( 1) uv due to the volumetric change in the cube under consideration
(i.e. uniform change of all its dimensions without any change in its
shape) and (2) u,h due to the change in its shape (i.e. energy' spent in
transforming the cube into a parallelepiped).
This division of the potential energy in two parts facilitates thestudy
of strength of materials in volumetric stress (Chapter 7).
Let us calculate the values of both the components of the specific
potential energy. It had been shown earlier (§ 34) that when the edges
of the cube deform uniformly, i.e. when there is a change in the volume
only, the relative elongation of each edge of the cube may be calculated
from formula (6.25):
p _^tntan
°moan — 3

where 0 mean—-
meBB
* 3 , and the bulk modulusK- E
— ------ J' - 3 ( l - 2 p )
The specific energy due to the change in volume will be
, . ’>o ~mcBiiM
°m e«n8mint;aii
ean ^itiean
°itiea n (Olj
(O 0 3)a
Ulyy - S §-----~ ~ 2 K ~ --------- JSR-----
OF
_ I - 2n
it \f ™
6£
(o r , + (6.29)

The potential energy corresponding to the change in shape of the

isolated element may now be calculated as the difference
=*u—«v 5^[o?+ 0^+<ri—2p(o,oa+o,0 $+ aao,)J
-----(a*+ + Cj)s
After simplifying, we get
«.h= fa*+ *a+ 0 ! — <r,oa— <
y1oi —oaot) (6.30)
Ch. 6) Compound Stress. Stress and Strain 127

Formula (6.30) may also be expressed through octahedral stresses

(6.19) by writing the expression in brackets as the difference of squares:

Miill' 6E + (Oi-O.)*] = ^ (6.30')

p
In simple tension, when O i= a= ^-, o2= 0, and o3= 0, the specific
potential energy corresponding to the volumetric change in the elemen­
tary cube is
_ < ! - 2n) 0* (6.31)
6E
and due to change in shape
(6.32)

Obviously, the sum of the two will give the total specific energy of
tension:
tt = M„ +i u8h= - §F

§ 36. Pure Shear. Stresses and Strains. Hooke’s Law.

Potential Energy

A. While dealing with compound state of stress (§ 33) it was noticed

that like in simple tension or compression (§ 27) planes inclined to the
direction of principal stresses experience normal .stresses that result
” in elongation (shortening) as we! las shearing stresses which correspond
to shear.
In studying shear deformation it is desirable to find planes in which
only shearing stresses act, i.e. planes that are free of normal stresses.
An analysis of formulas (6.5) and (6.6) reveals that in biaxial stress
under certain conditions (oc=45° and <r,-H73=0) the normal stresses
in the inclined plane vanish (<Ja=0 p=O); only shearing stresses Ta=
•*rmx act in this plane.
The stressed state in which only shearing stresses act on the faces
of an element of a material is known as pure shear.
Consider a cubic element with a front face abed (Fig. 75). We apply
equal shearing stresses t to the faces perpendicular to the front face
(recall that t a ——t p). The front face experiences neither normal nor
shearing stresses; it is, therefore, a principal plane in which the prin­
cipal stress Is zero. The two other principal stresses can be found by
solving the reverse problem: we determine them through known stresses
acting in two mutually perpendicular planes (§ 32). Let us use Mohr’s
128 Complicated Cases of Tension and Cotnpression I Part il

circle for solving the problem with the following data:

on the vertical face oa = 0 , t a = x
on the horizontal face Op = 0 , xfi —— x
Since the normal stresses laid off on the o-axis are equal to zero, from
point 0 (Fig. 76(6)) we plot segment ODa—xa—x upwards and seg­
ment 02 >0=Tpcs—T downwards. As points Da and Dp lie at the end

/ zz:7
1
V _______

7
Fig. 75 Fig. 76

points of the diameter of Mohr’s circle, its radius is equal to 0Da=T.

The segments OA and OB cut by the circle on the cr-axis are also equal
to the radius and determine the magnitudes of the principal stresses:
O A = a t = x, OB = o3= — t, d a = -0 (6.33)

Identical results are obtained if we put oa=<r6SS!0 and r a —r in for­

mulas (6 . 12).
The direction of principal stress is shown on Mohr’s circle by the
line flDp which makes an angle of 45° with the normal to plane be,
A similar conclusion ensues from formula (6 . 11). The element cut out
of the material around the same point by the principal planes (Fig.
76(a)) is stretched by stresses along diagonal bd and compressed by
Ch. 6) Compound Stress. Stress and Strain 129

stresses a# along diagonal ac. This can also be proved by considering

the equilibrium conditions of a part of the cube cut out by a diagonal
plane (Fig. 77).
Thus, pure shear is equivalent to a combination of two equal prin­
cipal stresses—one of them tensile and the other compressive (the
third equal to zero). In other words this is a particular case of biaxial
stress when <r£= —a3. Planes inctined at 45° to the direction of princip­
al stresses experience only shearing stresses which subject the ele­

ment to shear. At the same time, the material of this element is

stretched and compressed in the direction of principal stresses.
It should be noted that shear is always accompanied by tension
(compression), and vice versa.
B. We shall now consider deFormations in pure shear. Let a cubic
element of the material be in a state of equilibrium in pure shear
(Fig. 78). If we fix the face AB of this element, then the shearing stres­
ses will displace the face CD parallel to AB by a distance DDi=CC^=
=As called the absolute displacement. The element A BCD gets warped
and the right angles transform into acute or obtuse angles changing
by a value y. This angle is called the relative shear or angle of shear, and
serves as a measure of distortion (warping) of the angles of the rectan­
gular element. Since in structures we usually come across only elastic
deformations, this angle is extremely small.
The magnitude of the angle of shear is connected with the absolute
displacement and distance a between the planes AB and CD:

? = ta n v = -— (6.34)

i.e. the angle of shear is equal to the absolute displacement divided

by the distance between the shearing planes; it is expressed in radians.
It can be shown that the angle of shear is directly proportional to
shearing stress t . Thus the angle of shear numerically defines the shear
deformation,
190 Complicated Cases of Tension and Compression [Fort 11

Let us study Fig. 78 to establish the relation between x and y. Due

to warping of the given element, diagonal AD gets elongated.This elon
gation may on the one hand be related to the acting stresses and on the
other, to the angle of shear; comb in
ing the two relations we can estab
lish the dependence between x and y
From Fig. 78, we can obtain the
absolute elongation of the diagonal
by cutting the new diagonal ADi by
an arc with a centre A and radius
AD We get a right-angled triangle
DDxDi in which arm DDt represents
the absolute elongation As and arm
DsPt represents the elongation A/
o f the diagonal. The angle at point
Di may be taken as 45° due to the
small value of deformation. Then
A/ = As cos 45a

The relative elongation of the diagonal is

where Therefore
c = ~ cos 45° sin 45°
a
Since — = v, and cos 45° sin 45°=0.5, we get
a '
e = .L (6.35)

On the other hand, the relative elongation of the diagonal caused

by the principal stresses o ,= t and o3= —t (Fig. 76 (a)) may be expres­
sed by formula (6 .21 ):
e = fci=-£— M■7f=a'2r (l "HO

Putting this value of e in formula (6.34), wo get

- g - o + i* > - 4 v
wherefrom

T“ 2 ( I T S V (636)
Ch. 6\ Compound Stress. Stress and Strain 131

Thus, angle of shear v and shearing stress x are directly proportional

to each other, i.e. in shear the stress and corresponding strain are re­
lated by Hooke’s law.
£
Denoting the proportionality factor correlating t and y
by G, we get
t= G y (6.37)
where
E
G (6.38)
2(H-n)
Quantify G is called the modulus of elasticity in shear, or shear modu­
lus, and expression (6.37) is Hooke’s law for shear. We see that it is
completely identical to Hooke’s*law for tension (ar=£e). Shear mo­
dulus G, 1ike E, has the di mensions of stress.
Since in formula (6.38) for the shear modulus only two of the three
elastic constants E, p, and G are independent, the third may be ex­
pressed through the first two. However, it can also be determined di­
rectly from experiments on torsion of round bars (Chapter 9).
The absolute displacement depends not only upon shearing stress
but also upon the dimensions of the isolated element. Let us denote by
A the area of the faces on which the shearing stresses are acting; the
distance between the parallel faces is denoted by a (Fig. 78), and the
force acting along these faces, which is a resultant of stresses t (with
the assumption that shearing stresses t are uniformly distributed over
area A), by Q=*t>4. Substituting t and y in equation (6.37), we obtain
iL = ~-G , wherefrom (6.39)

Absolute displacement is directly proportional to the shearing force

and the distance between the sheared planes and inversely proportional
to the cross-sectional areas of the sheared planes and the shear modulus,
i.e. we have a formula which expresses Hooke’s law for shear that is
identical to the formula for absolute elongation under tension:

. With the help of expression (6.39) we can also calculate the potenti­
al energy of shear through the work done by force Q. Considering that
force Q is applied statically, gradually increasing from zero to a finite
value, we can express the work done by this force in affecting a dis­
placement As as
i r = 4 QAs
132 Complicated Cases of Tension and Compression I Part 11

Substituting As from equation (6.39), we get

., Q}a T*i4a
U = 'm = - w ~ (6.40)

Dividing by volume V—aA, we find the potential energy in pure

shear as
U= y - = l g (6.41)

The same result could have been obtained from formula (6.28),
§ 35, by considering pure shear as a compound stressed state with
principal stresses o 2= 0 , and <r,=>—t.
In should be noted that in pure shear the potential energy is spent
only on changing the shape, as the change in the volume in shear is
zero. This becomes clear from formula (6.23) if it is taken into account
that in pure shear the sum of principal stresses is equal to zero.

CHAPTER 7

Strength of Materials in Compound Stress

§ 37. Resistance to Failure. Rupture and Shear
Some problems related to the strength of the elements of structures
under uniaxial loading were discussed in §§ 16 and 17. It is well known
that among other conditions, the design of a structure must also satisfy
the strength condition which requires that maximum stress in each
part of a machine or structure must not exceed the permissible stress
that constitutes a certain fraction of the failing stress. In order to select
the permissible stress it is essential to study the behaviour of material
during its deformation from the moment the load is applied right up to
failure. The latter is also required for other purposes, for instance, for
controlling the plastic deformation processes (wire drawing, stamping,
rolling, forging, metal cutting, pressing of laminated plastics and
other materials).
We do not meet any difficulty in experimentally investigating the
behaviour of materials in uniaxial tension or compression with ma­
chines commonly installed in material testing laboratories. The tension
or compression test diagrams obtained as a result of these experiments
give a clear idea about the resistance of a given material to elastic, or
plastic deformation and enable us to determine mechanical character­
istics like yield stress and ultimate strength which are so important for
assessing the strength of material and specifying permissible stress.
The behaviour of material under loading depends upon its properties
and the state oi stress. In some cases strain remains more or less pro-
Ch. 7] Strength of Materials in Compound Stress 133

portional to stress right up to failure; failure occurs without any plas­

tic deformation (Fig. 25). In other cases elastic deformation is succeeded
by plastic deformation of considerable magnitude that ends in failure
(Figs. 16 and IS). A continuously increasing plastic deformation may
not necessarily lead to failure (Fig. 24).
The first diagram (Fig. 25) describes the behaviour of a brittle materi­
al in uniaxial tension or compression, fn this case failure should be con­
sidered as thecritical state of the material, and the ultimate strength
as the failure stress. Under tension failure occurs in a section perpen­
dicular to the tensile force, and under compression (with regular lubri­
cation of the specimen faces that come in contact with the press plates)
in sections parallel to the direction of compressive force (Fig. 28).
In both cases faiture takes place through separation of material partic­
les from one another, i.e. through rupture. In the case of tension, rup­
ture can be caused both by the maximum normal tensile stress and
the maximum elongation in the direction of action of the tensiie force,
in the case of compression, failure may be considered to occur due to
considerable tension in the direction perpendicular to the compressive
force. It is noteworthy that under compression brittle materials often
fail in sections that are inclined with respect to the direction of the
compressive force. It may therefore be assumed that failure is more
complicated in nature than described above and the cause of failure
are normal as well as shearing stresses acting on these inclined planes
(see § 40B for a more detailed discussion).
The second diagram shows the behaviour of ductile materials under
uniaxial tension (Fig. 18 depicts the true stress-strain diagram for
tension). The critical states in this case may be the beginning of yield­
ing, neck formation, and rupture. The corresponding failure stresses
will be yield stress, ultimate strength and true stress at rupture. The
appearance of shear lines (Luder’s lines) after permanent plastic
deformation (Fig. 13) and failure of specimens in planes inclined at
n/4 to the direction of tensile force (§ 27) enable us to consider lhat
the starting and growth of plastic deformation and the final failure
occur due to slip and shear under the action of maximum shearing
stresses. Such a failure is known as failure due lo shear.
The third diagram describes the behaviour of a ductile material under
compression when plastic deformation does not lead to failure (Fig. 24
shows the compression lest diagram). The beginning of yielding should
be considered as the critical state, and yield stress, which does not
differ much from yield stress under tension, as the failure stress. In
this case plastic deformation begins and develops due to shear under
the action of shearing stresses.
The two different concepts of failure of materials discussed above,
namely ( 1) failure in the form of rupture due to elongation or mainly
under the action of normal tensile stresses and (2) failure as a shear
under the influence of shearing stresses, have been known for a long
134 Complicated Cases of Tension and Compression [Part II

time. These concepts led to two types of resistance of materials to fai­

lure: resistance to rupture and resistance to shear.
Till recent past it was considered that every material possessed
only one type of resistance to failure—either resistance to rupture or
resistance to shear. Such a one-sided concept of failure prevented a
general solution to the problem of strength of materials from being
found.
A few years ago a new concept that has a sound experimental support
was put forward in the Soviet Union. According to this concept every
material depending upon the working conditions may fail both due to
rupture and shear and may therefore possess resistance to both types of
failure. This new approach to failure helped us to clarify the concept
of failure. Therefore, at the present stage of the science of strength of
materials only the new approach should be considered correct.
The possibility of failure of materials due to rupture, supported by
experimental evidence was not subjected to any doubts till now. On
the contrary, many scientists tend to explain all cases of failure by
the rupture phenomena.
The nature of failure due to rupture depends both on the type of
material and the state of the stress. In principle it is possible that fai­
lure may occur in some cases on account of brittle rupture without any
plastic deformation and in other cases due to ductile rupture accompa­
nied by the plastic deformation of more or less considerable magnitude.
Thus, for instance, it is experimentally established that some grades
of bronze and aluminium alloys are capable of failure due to rupture
even after undergoing permanent set of about 2 0 %.
The resistance to rupture is best studied by the testing of brittle
non-metallic materials (glass, plastics, concrete, and stone).
It is extremely difficult to study the resistance to rupture of ductile
materials because during testing it is, as a rule, impossible to avoid
the stage of plastic deformation and hence the shearing stresses of a
considerable magnitude. On account of the fact that ductile materials
have a much lower resistance to shearing stresses (shear) as compared
to their resistance to rupture, it is difficult to achieve rupture of these
materials by conventional tests because failure due to shear takes place
earlier. Therefore in order to determine the resistance to rupture the
test conditions (type of stressed state, temperature, rate of deformation)
should be altered so that the resistance to shear improves considerably
without any change in the resistance to rupture.
Available experimental data enable us to consider that resistance
to rupture does not depend much upon the rate of deformation and test
temperature. It therefore follows that by conducting dynamic tests
at low temperatures we can find, with certain approximation, the re­
sistance to rupture in normal conditions.
Numerous experimental investigations reveal that the resistance to
rupture of brittle materials is constant for different types of loading.
Ch 7] Strength, of Materials in Compound Stress 135

However, we do not have sufficient data to be able to come to a similar

conclusion for ductile materials. Some experimental studies point out
that resistance to rupture depends upon cold hardening—It increases
with the degree of cold hardening.
Failure due to shear is more complicated than rupture because it is
usually preceded by considerable plastic deformations which result in
redistribution of stresses and other complications. The existence of
this type of failure, caused mainly by shearing stresses, is confirmed
by a number of experimental data.
The failure of materials under tension accompanied by neck forma­
tion, shear, torsion and bending usually occurs along planes close to
the planes of maximum shearing stresses. Although it is not always
possible to conclude about the type of failure {rupture or shear) merely
from the angle of rupture, in a number of cases the location of the plane
of failure and the appearance of the breakdown surface can be decisive
factors in this respect. Thus, for instance, if failure under torsion oc­
curs in planes perpendicular to the bar axis, it is undoubtedly caused
by shearing stresses because in this case the surface of breakdown plane
is completely free of normal stresses.
It is much more difficult to differentiate between failures due to
rupture and shear when the body is under a compound stress. Still in a
number of cases of complex loading it was established that shearing
stresses played a major role in many instances of failure, which were
earlier considered obvious examples of failure due to rupture.
In ductile materials shear occurring without preceding permanent set,
usually of a considerable magnitude, is highly improbable, because
failure due to shear takes place due to shearing stresses, which also play
the major role in plastic deformation of materials. At least it has not
been possible till now to practically achieve such failure in metals
although some of them (for example, compressed magnesium and its
alloys) fail due to shear after small plastic deformation (5-15%).
This is known as the so-called “brittle shear".
Experimental data show that resistance to shear practically does
not depend upon the type of stressed state for pure metals (copper,
aluminium, iron) and some alloys. It is also established that it depends
upon the rate of deformation and temperature to a much greater extent
than the resistance to rupture. Resistance to shear increases with
increase in rate of deformation and reduction in temperature.
The assumption about materials having resistance to both types of
failure is confirmed by experiments on failure of cold-short metals
and some brittle materials. For one and the same material the magni­
tudes of resistance to rupture and shear are different: for ductHe mate­
rials usually r^COnjp on the contrary, for brittle materials ^ > 0 ^ .
The laws governing xgh and tr^p may differ depending upon the changes
in composition of material and its machining and heat treatment.
 136 Complicated Cases of Tension and Compression [Part i f

The above discussion about the resistance of materials to failure

may serve as a basis for strength test in simple and compound stales of
stress. The application of the resistance characteristics fs discussed
in succeeding sections. The considerable growth of research on failure
of materials in recent years is fully reflected in the book Fundamentals
of the Mechanics of Failure by L. M. Kachanov, Nauka, Moscow, 1974.

§ 38. Strength Theories

As has been already slated, in the case of uniaxial loading it is
not difficult to find the breakdown stress which is used as a basis for
designating permissible stresses.
It is much more difficult to find the breakdown stress in compound
stressed stale which is in general characterized by the three different
principal stresses. Experiments show that the breakdown state of an
element of structure (yield, rupture) depends upon the nature of stres-

^ 1 1
s* Known
&2 ct >a2>^i
pL Known:

% I o
GfO
~A Strength
condition?

F it;. 7 9
sed slate, i.e. upon the ratio between the three principal stresses.
Since the number of various possible ratios between the principal
stresses is infinitely large, there exist a corresponding infinite number
of potential stales of failure of the structure element. Hence, for each
new ratio between the principal stresses it is necessary to experimen­
tally find the permissible stresses anew. It should be borne in mind that
it is much more difficult to conduct tests in compound stressed state as
compared to simple tension or compression; these tests are more time
consuming and expensive, and, as a rule, require special accessories
to the machines available in laboratories.
Therefore, it is necessary to find ways of expressing the strength
condition under compound stress in terms of a,, and <r„ obtained from
experiments for the uniaxial stress.
Thus, in the general case, when all the three principal stresses are
nonzero, the strength of the material is tested according to the fol­
lowing plan:
( 1) the three principal stresses cCXiiXJs are calculated;
(2 ) the material is selected;
c/i. 7] Strength of Materials in Compound Stress 137

{3) the critical stresses on=a,j or or°=<ju and the permissible stresses
are determined experimentally for the given material under simple
tension or compression.
I t is required to write down the strength condition for the compound
stress knowing cr(, oSt and <r3 and retaining the same safetv factor k
(Fig. 79).
The above problem can be solved only on the basis of the assumption
(hypothesis) about the type of function relating the strength of mate­
rial to the value and sign of the principal stresses, and the factor that
causes the critical state.
These factors may be numerous. As a matter of fact, even in simple
tension of a bar of ductile material we may put the question: what is
the cause of yielding?
We may assume that yielding starts when the maximum normal
stresses in the bar reach the yield point au. However, one may as well
look at the problem from a different point of view and assume that
yielding starts when the maximum elongation of the material reaches a
certain limit. One may also assume that large plastic deformations
begin to occur when the maximum shearing stresses achieve a certain
value.
Thus, we can put forward a number of hypotheses and on their basis
formulate various theories erf strength. We shall see later that in simple
tension or compression (in uniaxial stress) the results obtained by the
strength tests are the same irrespective of the hypothesis used. This
is so because the strength test is based directly upon experimental data.
The matters will be very much different in compound stress. In
the succeeding sections we shall show how the strength condition chan­
ges depending upon the accepted theory. One or the other theory is
selected for practical application only after it has been experimentally
verified for the compound stressed state.
Whichever strength hypothesis we choose, it can be expressed analy­
tically as some function of principal stresses
<D(o,, o8, <r3) —const = C (7.1)
In this form the strength theory expresses the condition of constancy
(irrespective of the nature of stressed state) of the set of principal
stresses that has one or the other physical interpretation. At the same
time, equation (7.1) also describes some limiting surface in three-
dimensional space of the principal stresses. Thus, for example, if
C=<Jy or C = ou, the corresponding limiting surface is the surface which
determines the conditions under which yielding or failure of material
takes place.
Before we begin to expound various theories of strength, let us take
note that the critical state for ductile materials (appearance of large
plastic deformations) as well as brittle materials (appearance of cracks)
lies at the boundary of application of Hooke's law (with known approxi*
138 Complicated Cases of Tension, and Compression [Part 11

mat ion sufficient for practical purposes). This enables us to use the
formulas which have been derived in the preceding sections and which
are valid only within the limits of application of Hooke’s law for cal­
culations relating to the strength test.
Our earlier discussion about the resistance of materials to rupture
and shear emphasizes the need to distinguish between the strength
theories for materials that fail due to rupture and the theories in which
failure due to shear is considered the breakdown state. These theories
are dealt with in §§ 39 and 40 separately.

§ 39. Theories of Brittle Failure

(Theories of Rupture)
As has been already stated, failure in the form of rupture may be
considered to occur either due to maximum normal tensile stress or
due to maximum elastic elongation.
A. The assumption that failure is related to the maximum tensile
stresses was put forward as early as the seventeenth century and sub­
sequently supported by G. Lamd (1833) and W. J. M. Rankine (1856).
At present the theory in which the maximum tensile stress is taken as
the strength criterion is known as the theory of maximum tensile stresses
or the first strength theory.
If then the stress <Tj will be the maximum tensile stress
tfwDx- According to the first strength theory, failure will occur irrespec­
tive of the stressed state when
^tnax = = ^rup
where orw„ is the resistance to rupture which is constant for a given
material. For many brittle materials ornp is equal to the stress
at the moment of failure under tensile loading. The safe state will obvi­
ously correspond to the condition

°m3x = cri < ^ T L= M < (7.2)

where fcl, is the permissible stress in tension. Equation (7.2) repre­

sents the strength condition according to the first strength theory.
It is applicable only when Oi>0.
This theory is confirmed by tensile tests of brittle materials such
as stone, brick, concrete, glass, and porcelain. In the case of compound
stress the theory often comes into conflict with experimental data be­
cause it does not take into account the other two principal stresses
upon which the strength of material depends in many cases.
B. The idea that brittle failure is connected not with the maximum
tensile stress but with maximum strain was first expressed by French
scientists Ed. Mariotte (in 1686) and C. M. L. Navier (in 1826) and later
supported by other French scientists, J . V. Ponceiet (1839) and
Ch. 7| Strength of Materials in Compound Stress 139

B. Saint-Veuant (1837). The strength theory based upon this supposi­

tion is known as the theory of maximum strain, or the second strength,
theory. According to this theory failure occurs irrespective of the
state of stress when maximum elastic strain £max becomes equal to a
certain value efup which is constant for the given material. In general

8n.a x = 8 , = -g - I ° i — ^

whereas in simple tension e=o/£; it is obvious that etup=aIUV/E.

In the compound stress, failure will occur when

The stressed state may be considered safe if in this expression CFra„

is replaced by [oJt. The strength condition in the second strength
theory may be written as
+ (7.3)

Thus in the theory of maximum strain, the permissible stress under

tension is compared not to one of the principal stresses but to a combi­
nation of all of them, called the reduced stress and determined by the
formula

This hypothesis is also not supported by some experiments on the

strength of ductile materials. If it were true for ductile materials,
then the specimen stretched in two or three directions should be stron­
ger than the specimen stretched in only one direction; this is not
confirmed by experiments. This hypothesis is similarly not confirmed
for uniform bulk compression.
For brittle materials, the theory of maximum strain generally gives
results which match well with the available experimental data. Ex­
pression (7.3) may be applied if at—p.(cr#-{-O3) > 0 . Application of the
second strength theory for the case of compression enables us to satisfac­
torily explain the reasons behind the failure of brittle materials along
planes parallel to the direction of compressive force and also explain
more, or less correctly why the strength of brittle materials under
compression is considerably higher than their strength in tension
(in tension emax==-^=&riip and a?=£efUP, whereas in compression

SjBax®8—j r 0?” 8™ and I °?l= 7 W ,,e‘7 times gfeater)- However,

the second strength theory is also confirmed mainly by experiments
on brittle materials only.
Both theories discussed above are theories of rupture; none of them
is universal, i.e. valid in all the cases of failure due to rupture.
J40 Complicated Cases of Tension and Compression IPart II

Sometimes the first theory conforms better to experimental data, some­

times the second. For a solid uniform body the second theory appears
to be more logical and well founded than the first one.

§ 40. Theories of Ductile Failure

(Theories of Shear)
A. The fact that shear lines appear on the specimen surface during
plastic deformation and that under tension ductile materials fail along
the planes of maximum shearing stresses enables us to accept these
stresses as the criterion of strength. This idea was first proposed by the
French physicist Ch. A. Coulomb in 1773 and supported by the experi­
ments of H. Tresca (1868), J . J . Guest (1900) and others. The strength
theory based upon this assumption came to be known as the theory of
maximum shearing stresses or the third strength theory. According to
this theory the critical state of material (in the form of yield or failure)
occurs, irrespective of the stressed state, when the maximum shearing
stress V j, becomes equal to a certain value t 0 which is constant for
the given material, i.e.

or
Tmix — T?J — %\x

where xu is the yield stress in shear and Tsh is the maximum shearing
stress when the material fails due to shear. The safe functioning of
material is obviously governed by the strength condition

(7.4)

In compound stress Tmax= ( 0 ,—<j3)/2. If we assume, following this

theory, that permissible stress [xl does not depend upon the type of
stressed slate, we shall find its value from experiments on simple ten­
sion in which failure occurs as a result of shear. In this case 0 3=O and
Tnuxa*Y' ^ si1"655 ^ the right-hand side of the Jast expression Is
raised to permissible stress k l, the left-hand side of the same expres­
sion will represent the permissible value of shearing stress r; thus,
Substituting now the values of Tfflax and I t J in expression
(7.4), we obtain

or
0 .- 0 , < ["] (7.5)
Ch. 7] Strength of Materials in Compound Stress 141

Thus, for strength cheek according to this theory the permissible

stress in tension or compression is compared not with the maximum
normal stress, but with the difference between the maximum and mi­
nimum normal (principal) stresses. The reduced stress in this case is
0 ^ = 01 o3

The advantage of the theory of maximum shearing stresses lies in

its simplicity and the linearity of the strength condition, as in the
first and second theories. It is well supported by experiments on ductile
materials that have equal resistance to tension and compression, and
also by experiments on bulk compression. This theory usually ensures
sound dimensions of the designed elements of structures; sometimes
the dimensions are even slightly on the higher side.
The drawback of the theory of maximum shearing stresses, which is
seen immediately, is that it completely ignores the effect of the average
principal stress on the working of the material. It implies that for
constant maximum normal stress tri and minimum normal stress o#,
we may vary cr« in any way without changing working conditions of
the material as long as it is less thanoi and greater than o,. This state­
ment is quite dubious, and experiments reveal that a* does have an
effect upon the strength of materials. The theory also underestimates
the danger of failure of elements subjected to approximately equal
tensile stresses in the three principal axes. To this may be added that,
according to this theory, the stressed states in cubic elements isolated
near inclined planes (Fig. 54 (a) and (b)) must be identical from the
point of view of failure if shearing stresses t « in these planes are equal
to each other. As t a increases the yielding and failure in the material
in these elements begin simultaneously. Experiments show that for
materials having higher resistance under compression as compared to
tension, case (a) in which the normal stresses in the plane of shearing
stresses are tensile is more dangerous than case (b), when the normal
stresses in the plane of ra are compressive. As the shearing stress xa
increases, the material of the element will begin to yield or rupture
earlier in case (a) than in case (b). Thus, the strength of material is
influenced not only by the shearing stress but also by the normal
stress acting on the same plane. This factor is taken into account by
Mohr’s theory (1900) which is discussed below.
B. The breakdown conditions p ? = Ty or = r sta discussed
above should be looked upon in a broader aspect than as mere interpre­
tation of the theory of maximum shearing stresses. According to these
formulas, it can be considered that critical state is determined only by
the maximum and minimum principal stresses. Experiments do not
fully confirm this hypothesis; however, the maximum possible error
due to ignoring medium principal stress <r2 does not exceed 15% and in
a majority of cases is considerably smaller. Therefore while writing
 142 Complicated Cases of Tension and Compression [Part I I

the strength conditions it is permissible to restrict ourselves to studying

the effect on strength only of the maximum and minimum principal
stresses.
It is common knowledge that various cases of stress can be graphically
represented by Mohr’s stress circles. Figure 80 depicts a number of
such circles: circle 1 represents simple tension: a a= a 8= 0 ;

Fig. 80 Fig. 8 !

circle 2, simple compression: Oj=<t2= 0 , <*3^ 0 ; circle 3, pure shear:

—°>i <*1= 0 . The stress circles constructed for principal stress
values corresponding to the critical state of materials will be called
limiting stress circles. The limiting stress circles corresponding to the

Fig. 82

state of stress depicted in Fig. 80 are shown in Fig. 81. The diameter
of the limiting circle which depicts the critical state in simple tension
is the ultimate strength in tension; in the case of simple compres~
sjon the ultimate circle diameter is ultimate strength in compres­
sion; and m the case of pure shear the limiting circle diameter is equal
to
Ch. 71 Strength of Materials in Compound Stress 143

O. Mohr postulated that all the limiting stress circles constructed

from arbitrary centres can be inscribed into a smooth curve, the envelope
of the family of limiting stress circles, which is tangent to all of them
(Fig. 81). The envelope intersects the or-axis at a certain point H, which
corresponds to uniform triaxial tension (if <j,=o 2= a 3> the stress circle
becomes a point). The envelope is open on the opposite side because
the failure of material under uniform triaxial compression is impossi­
ble. Plotting the envelope can be simplified by considering it, in the
first approximation, as a straight line tangent to the limiting circles

of tension and compression. If the ultimate strengths under tension

and compression are equal, the envelope branches remain parallel to
the c-axis over a large distance (Fig. 82). In this case Mohr’s theory
coincides with the theory of maximum shearing stresses.
By reducing the diameters of all limiting circles k times, where k
is the safety factor, we obtain a family of circles which represents the
permissible stressed states instead of the limiting stresses (Fig. 83).
In Fig. 83 segment OA (the diameter of circle /), represents the permis­
sible stress under simple tension lo]„ segment OB (the diameter of
circle 2 ) represents permissible stress under simple compression !olc.
The intermediate circle 3 with centre at 0, touches the envelope CtCiH
at point C3 and represents a stressed state with principal stresses
and <r3.
From the similarity of triangles OiO*D4 and 0 i0 ^ )3 it ensues that
OjDj 0 |0 j OjCj—O jC j OOi+OOj
or 0 2Ca— O A “ 0 0 , + OOj
By substituting corresponding stresses in place of segments, we obtain
<*i—o3—[<rh M<—(<h-+<*>)
l^lc—Mr Mi + l<Mc
144 Ompiiatted Cases of Tension and Compression [Part t i

After some transformations we get the strength condition according

to Mohr’s theory:
ai— } ^ - <3f3= < y ,-p o r,^[0 j/. (7.6)

The same condition can be derived without using stress circles*

if it is kept in mind that shear (according to Mohr) leading to failure
occurs in that (breakdown) plane which has the most unfavourable,
combination of normal and shearing stresses. The condition restricting
the value of a particular reduced shearing stress in the breakdown
plane may be written as
Vd = M + /<T<[T]rod (7.7)
where | t | and a are stresses in the breakdown plane (the sign of normal
stress is taken into account) and f is the coefficient of friction. The
location of the plane of maximum reduced shearing stress (in Fig. 83
this plane corresponds to point C3) is determined by angle a which
this plane makes with the plane of principal stress a,: tan 2 a = y =
= taiTp * 11 can be sbown (Fig. 83) that

a" d

when /= 0 strength condition (7.7) changes into the similar condition

of the third strength theory: if o X ) (tension), condition (7.7) is satis­
fied only when the value of lx| is reduced as compared to the value for
o=0; if o < 0 (compression), condition (7.7) is satisfied even with a
higher value of Ir Jas compared to the value for o = 0 . These conclusions
are supported by experiments discussed earlier at the end of §40A.
It can be easily seen that the strength condition (7 .6) according to
Mohr’s theory coincides with the strength condition according to the
theory of maximum shearing stresses if p=* 1, i.e. foj,= f 0 jc. If the
permissible stress under tension is very small (brittle materials),
i.e. if It can be considered that f d ,= 0 and p—0, Mohr’s theory
changes into the theory of maximum normal stresses. In biaxial stress,
when <ja= 0 and Mohr’s theory coincides with the theory of
maximum strain. Thus, to a certain extent Mohr’s theory generalizes
the first three strength theories; it correctly describes plastic deforma­
tion and failure due to shear of materials having different resistances
to tension and compression. All experiments that verify the first and
third theories and some experiments verifying the second theory also
support Mohr’s theory, it undoubtedly represents a forward step as

• See S. I. Druzhinin and Yu. I. Yagn, Strength of Materlah, "Kubuch',

1933 (in Russian).
Ch. 7| Strength of Materials in Compound Stress H5

compared to the first three theories. Yet it cannot be considered uni­

versal, since in a number ot cases it does not correctly reflect the nature
of failure due to rupture, and like the third strength theory it does not
take into account the intermediate principal stress.
B. A number of authors suggested that the appearance of the critical
state in materials depends not upon the magnitude of deformations
and stresses separately but upon their combination and other factors
like the potential energy or the numerically equal to it specific work
of deformation. The amount of this work is expressed in terms of
all the three principal stresses.
By the end of last century (1§85) Italian scientist F. Beltrami pro­
posed that the total potential energy of deformation per unit volume
of the material should be taken as the criterion of pliability and
strenglh of materials. On the basis of this hypothesis the condition
expressing the approach of the critical state may be written as
u~u°
where a0 is the potential energy accumulated in a unit volume of the
material when yield or rupture sets in.
This hypothesis was not confirmed by experiments and is only of
historical importance. It, however, formed the foundation upon which
the new energy theory of strength was built; the latter generally gives
results matching well with the experiments.
Considering the fact that plastic deformation takes place without
any change in volume, F. Huber in 1904, R. Mises in 1913 and H, Hen-
cky in 1924 proposed that instead of total potential energy of deforma­
tion only that part of the energy which was spent on changing'the shape
of a body should be accepted as the strength criterion. According
to this hypothesis, irrespective of the stress yielding or rupture of the
material starts when the potential energy of distortion per unit volume,
reaches a certain limiting (critical) value for the given
material, i.e.
h = *& (7-8)
where =«*»,.* or Klh=“*n. w
It is known that in compound stressed state (see formula (6.30'),
§ 35)
—' ^£“[(^1— + (a t —<*,)*+ (<*3—°i)i] =* ^2£ ^^^
and in uniaxial tension

If we accept, as already stated, that the critical value of the poten­

tial energy of distortion (for example, corresponding to the beginning
6 -8310
146 Complicated Cases of Tension and Compression [Pari It

of yielding of material) does not depend upon the type of stressed state,
we can consider that in uniaxial as well as in any other type of stress

Substituting the expressions for ush and h in equation (7 .8), divid­

ing both sides by 2 ^ ^ and extracting from them square roots, we
get the following expression which determines the beginning of criti­
cal state:

y j V <T(t — + (ora—<rs)a+ (o4— cr,)* = tnct=atJ

It can be easily noticed that this equation represents the condition
of constancy of stress (or constancy of the octahedral shearing stress).
The strength condition according to this theory, known as the theory
of potential energy of distortion or the fourth strength theory, may be
written as
■— V ( 0 + (0,-0,)*+(0,- ff,)»
~ a T.ct ^ = [ffJ (7-9)

The theory of potential energy of distortion is well supported by

experiments on ductile materials, but fails when applied to brittle
materials. This is natural because it is the theory of octahedral or me­
dium shearing stresses unlike the third theory, which is the theory of
maximum shearing stresses. The fourth strength theory takes into
account all the three principal stresses and is therefore more complete
than the theory of maximum shearing stresses. Unlike the first three
strength theories and Mohr’s theory, tne fourth strength theory is non­
linear, which somewhat complicates its practical application.
Keeping in mind that the resistance of materials to plastic deforma­
tion is to some extent affected by the mean normal stress omeB„, the
condition expressing the onset of yielding according to the theory of
potential energy of distortion may be written more precisely as
+ (7-10)
where A and B are constants that depend upon the properties of a
material. The strength condition may then be written in the form
- y j Vt o + fa—<*.)*+(<*3—*i)J
+ -y-(Ol-f Oj + O a X S , fo] (7.11)
Ch. 71 Strength of Materials in Compound Stress 147

This expression can apparently be employed for checking the strength

of parts of machines and structures made not only of ductile but also
of some brittle materials. Unfortunately the possibility of applying
this condition to brittle materials has not been studied sufficiently
till now.

§ 41. Reduced Stresses According to Different

Strength Theories

In conclusion of our discussion of strength theories, we may write

the strength condition in triaxial stress as follows:
*«*<[*] <7-l2>
where <rrad is the reduced stress and Tol is the permissible stress in
simple tension or compression. The reduced stress o.e<j may be interp­
reted as the tensile stress in uniaxial loading equivalent to the compo­
und stressed state under consideration as far as the danger of failure is
concerned. f t ,
The expressions for cred according to different theories are as fol­
lows:
Wed “ ^max ~
Wo<l = ^®max = U|—fl (0*
Ofoi = 1=3^1 —
“®3* Wed pOa
o™ ,=-j£=-V (<rI- c .) 1+(<JJ- f f .) a+ (9 ,-o ,)’

With a number of theories at his disposal for assessing the strength of

parts from brittle and ductile materials, an engineer must choose in
each particular case the most suitable theory proceeding from the actu­
al properties of material. It is difficult to make the proper choice be­
cause of the fact that in compound stress the division of materials
into ductile and brittle is conditional. A material having good ductility
under simple tension and compression may behave like a brittle mate­
rial in compound stress and fail without undergoing large plastic defor­
mation. On the other hand, a material that shows brittle in uniaxial
loading may behave as a ductile material when subjected to other
types of stress. Hence, ductility and brittleness of materials depend
upon the condition in which the given structure functions. Therefore
it is more correct not to speak of brittle and ductile materials but of
brittle and ductile states of materials.
The main factors that affect brittleness and ductility are temperature
(low temperature increases brittleness, high temperature as a rule
improves ductility), rate of deformation (in case of fast dynamic load­
ing brittleness increases, whereas ductility is retained when loading is
6*
 148 Complicated Cases of Tension and Compression fPari / /

static and gradual), type of stress (states of stress close to uniform tri-
axial tension are known as “tough” and they lead to higher brittleness;
on the contrary stressed states close to uniform triaxial compression
are known as “soft” and improve ductility).
At present many materials can be made to acquire brittle or ductile
state by different means. If a material can deform and fail both as brit­
tle and ductile, then, as was earlier stated, it also has two characteris­
tics of resistance to failure that are determined experi men tally: resis­
tance to rupture and resistance to shear. The resistance to rupture
<jrw is found as the maximum normal tensile stress required for causing
rupture crmi,x= o 1 (first strength theory) or the reduced normal stress
which is the product of maximum strain and modulus of
elasticity £ , i.e. o^d=^i — (second strength theory). The
resistance to shear is determined by the maximum shearing stress
when failure occurs due to shear Tsh=t?ll8x= i.( o I- ( r 3)« (third strength
theory), by the limiting value of stress a® at the moment of failure
(fourth strength theory) and the limiting value of reduced stress
) 'n the case of shear failure (Mohr’s theory).
In the light of above, while designing, for instance, the elements of
structures from mild steel, a ductile material in certain conditions
(static loading, room temperature, uniaxial stress), it is not always
possible to apply the third or fourth strength theories without taking
into account the actual working conditions of the structures; similarly,
while designing parts from concrete—a brittle material under the
afore-mentioned conditions, the first theory is not always applicable.
The problem of applyingone o- the other strength theory’can be solved
to the first approximation with the help of the so-called mechanical
state diagram proposed by Prof. Ya. B. Fridman on the basis of re­
search on the strength of materials carried out bv Prof. N. N. Davi-
denkov and his followers.*
As an example consider the transmission of pressure from the locomo­
tive wheel to the rail (Fig. 59). The elementary cube with edges of 1 mm
cut at the centre of the area through which the pressure from the wheel
is being transmitted to the rail is subjected to compressive principal
stresses: o ,= —80 kgf/mm8; <r,=—90 kgf-'mm*; cr3= — )10 kgf/mm2.
We shall calculate by the third and fourth strength theories the reduced
stress which should be compared with the permissible stress. Accord­
ing to the theory of maximum shearing stresses

°red ~ CTi—o3= —80-)- 110 = 30 kgf/mm2

* F«>r instance, see N. M. Belyaev, Strength of Materials, JO-14 editions. $ 252

(in Russian). *
Ch. 71 Strength of Materials in Compound Stress 149

According to the distortion energy theory

o!«v, = i V/ (—80 + 9 0 ) - - r ( —90 + 1 10)* + (— 110 + 80)=“

=26.4 kgf/mm9
Since the yield stress for commercial rail steel is approximately
40 kgf/mm* and the elastic limit is nearly 30 kgf/mm*, the computed
principal stresses are within the permissible limits. This is confirmed
by the behaviour of rail steel in exploitation.
Finally, it should be noted that all the preceding discussions of the
strength theories pertain to the materials which may be sufficiently
accurately considered as isotropic. The formulas derived above are not
appiicabfe for anisotropic materials. For example, in the case of timber
the direction of force with respect to fibres has to be taken into account.

§ 42. Permissible Stresses in Pure Shear

The permissible shearing stress in pure shear could, it seems, be
determined as in uniaxial tension or compression, i.e. by experimen­
tally establishing the critical stress (corresponding to yielding or
rupture) and dividing it by the factor of safety. There are, however,
some practical difficulties in applying such a method. It is very diffi­
cult to simulate pure shear in laboratory conditions; the working of
bolts and riveted joints is complicated due to the presence of normal
stresses. In the case of torsion * of solid bars of round and other cross
sections the stressed state is not uniform in the whole volume of the
bar. Moreover, the plastic deformation preceding failure is accompanied
by redistribution of stresses, which complicates the determination of
critical stress. When thin-walled bars are subjected to torsion, the
bar walls can easily loose stability. In the light of all these considera­
tions the permissible stresses in torsibn and pure shear are chosen on
the basis of one or the other strength theory depending upon the per­
missible tensile stress that can be determined more reliably.
Keeping in mind that in pure shear Oi = t, crs= 0, and <j 3-=—x
(see formulas (6.33) in §36), we can establish relationships between
fa and fcrl according to the different strength theories.
After substituting CT!=t in the first strength theory condition
<|or!f, it may be written as x < [cl<, wherefrom
M '- f e l* (7.13)
After substituting in the second strength theory condition

* The malarial of a bar subjected to torsion experiences pure shear (sec Chap­
ter 9).
150 Complicated Cases of Tension and Compression IPart I I

the values of principal stresses in pure shear, we get

H (— *0 “ (1 + 1*> < [<*]*
or
fob
l +H
wherefrom
_ [oh
Mn t+ft (7.14)
After substituting in the third strength theory condition
*i—
the values of <*i and or», it takes the form
x — (— t ) = 2 t ^ [ o ]
or
* < i r 0J
i.e.
(7.15)
The strength condition according to Mohr’s theory is crj—pcr8<
< lo lf; in the case of pure shear we get t — p ( — t ) — ( 1 + p ) t < ( < j ] , ,
wherefrom or

<716>
In Fig. 83 permissible stress Ix]M is represented by segment OF.
Applying the fourth strength theory, we find
Y ~ V (T-0)«+<0+T)>+(T+T)>=V5T<M
i.e.

* < y = - and [,J1V“ T T (717)

Expressions (7.15) and (7.17) should be used in the design of the
elements of structures from ductile materials that have equal resis­
tance to tension and compression. The difference between It ]111 and
lxl,v is about 15%. Expression (7.16) must be used in the case of ma­
terials that have unequal resistance to tension and compression. Ex­
pression (7.13) is used rarely. It is desirable to use expression (7.14)
only for brittle materials; however, it is also used in the design of parts
working in shear (bolts, and rivets). Since n«0.3 for steel, then
W n = -n 3 W . = (0.75-0.8)[a],
 PART III
Shear and Torsion

CHAPTER 6
Practical Methods of Design on Shear

§ 43. Design of Riveted and Bolted Joints

While studying the stresses acting in inclined planes (§ 27) we saw
that even in simple tension or compression two parts of a bar cut by
an inclined plane tend not only to separate from each other but to
shear along the sectioning plane. This is due to the fact that both nor­
mal and shearing stresses act in the plane. We came across these, types
of deformations—tension or compression and shear—while discussing
compound stress and, in particular, in pure shear {§ 36).
In practice a number of parts of structures work mainly under shear
due to which strength test for shearing stresses acquires major impor­
tance. The simplest examples of such parts are bolted and riveted joints.
In many fields rivets have been replaced by welding; however, riveted
joints are still widely used for joining all types of metal structures:
rafters, bridge trusses, cranes, for joining plates in boilers, ships,
reservoirs, etc.
To make a riveted joint, holes are drilled or pressed in both plates.
A red-hot rivet with one head is placed in these holes and its other end
is riveted by strokes from a special hammer or by pressure from a hy-
draulic press (riveting machine) to make the second head. Small rivets
(having diameter less than 8 mm) are deformed in a cold state (in
aviation structures).
A- Let us take the simplest riveted joint to study the working of riv­
ets (Fig. 84). Six rivets placed in two rows join two plates by a lapped
joint. Under the action of forces P these plates tend to shift over one
another, this being hampered by the rivets to which forces P are trans­
ferred.*
While checking the strength of rivets we shall stick to the established
order of solving problems of strength of materials. Two equal and
* The resistance due to friction is not taken into account.
152 Shear and Torsion 1Part I I I

opposite forces are transferred to each rivet: one acting from the first
plate and the other, from the second plate. It has been experimentally
shown that some of the rivets in a row carry greater load than the others.
However, at the moment of breakdown, the forces acting on various
rivets more or less level out due to plastic deformation. Therefore, it
is generally accepted that all the rivets work under similar conditions.

Fig. 84 Fig. 85

Thus if there are n rivets in the joint shown in Fig. 84, then each of
the rivets will be subjected to two equal and opposite forces
(Fig. 85). These forces are transmitted to the rivet through the pressure
of the corresponding plate on the semcylindrical surface of the shank.
Forces Pt tend to shear the rivet along plane mk which is the parting
plane of the plates.
To determine the stresses acting in this plane let us imagine the rivet
shank to be cut by section mk and the lower portion removed (Fig. 85).
The internal forces which are transferred through this section from the
lower portion to the upper one will balance force P u i.e. they will
act parallel to it in the cutting plane and will give a resultant force /V
Therefore, the stresses appearing in this section and acting tangential­
ly to it will be called shearing stresses t. Generally it is assumed that
they are distributed uniformly over the whole section. If the rivet
shank has diameter d, then the stress per unit area of the section will be

Denoting the permissible stress in shear by frl, we may write the

strength condition of the rivet under shear as follows:

T= - T = - S i i - < M (8 . 1)
* T
i.e. the actual shearing stress x acting in the rivet material should not
exceed the permissible shearing stress (see § 42).
Clu 5] Practical Methods of Design on Shear 153

From this condition we can determine the required diameter of the

rivets if their number is known, and vice versa. Usually the diameter
of the rivet shank d is given in accordance with the thickness t of the
parts to be joined (generally d&2t) and the required number of rivets
n is determined from the relation
p
(8 . 1')

The denominator of this formula represents the force which each of the
rivets can withstand safely.
While deriving formula (8.1) one more inaccuracy had been allowed.
Actually forces Pi acting on the rivet are not directed along a straight
line but constitute a force couple.
This couple is balanced by another
force couple, formed by the reaction
of the riveted plates on the rivet
head (Fig. 86 ) and gives rise to nor­
mal stresses acting in planemk.
Besides these normal stresses,
section mk is subjected to normal
stresses from another source: during
cooling the rivet shank tends to
shorten which is hindered by the
stop of the rivet heads by the plates.
This, on the one hand, leads to tight­
ening of the plates by the rivets
giving rise to forces of friction be­
tween them, and on the other hand
causes considerable normal stresses
in the sections of the rivet shank.
These stresses are not very harmful. The rivets are made of steel posses­
sing sufficient ductility; therefore, even if the normal stresses attain
the yield point we can only expect some plastic deformalion (elonga­
tion) of the rivet shank, which will reduce the friction between the
plates. The rivets will however continue to work on shear as designed.
These normal stresses are therefore not taken into consideration while
designing riveted joints.
Expression (8.1) has been derived for single-shear riveted joints.
If the joint is lapped by two cover plates (Fig. 87), each rivet experi­
ences shear in two planes—mk and g f (Fig. 88). Such rivets are known
as double shear rivets. If n rivets are required to transmit force P from
one plate to the cover plates, then force acting on one rivet is Pi—
—P'n. The area of shear ant* the shearing stresses in
154 Shear and Torsion IPart I I I

tions m k and g f (Fig. 88) are

t ___ £ _
2nd2
n 4
The strength condition for a double-shear rivet may be written as
t < [t], wherefrom —- £ — (8 .2 )

Hence, in a joint having two shear planes, the number of rivets

required according to the strength condition against shear is two times
less than that required in single shear (formula (8 . 1)).

Fig. 87 Fig. 88

In the case of multiple-shear rivets that are sometimes used in metal

structures, the shear area of each rivet is A6h— and the strength
condition is
(8.3)

where k is the number of shear planes.

However, the observance of strength conditions for shear alone does
not always ensure that the riveted joint is sufficiently strong. The
joint will be spoiled if the hole walls or the rivet shank get crushed along
the semicylindrical contact surface when the force is being transmitted
from the plate to the rivet. Therefore, in order to ensure reliable work­
ing of the riveted joint it is essential to check the rivets (or plates)
against crushing.
Figure 89 presents an approximate picture of transmission of pressure
to the rivet shank. The distribution of this pressure over the cylindri­
cal surface is not known; it depends to a large extent upon the conditi­
ons of manufacturing the structure. It is assumed that the non-uniform
pressure transmitted to the semicylindrical surface of a rivet is distrib­
uted uniformly over the diametral plane BC of the rivet. The stress
Ch. 8) Practical Methods of Design on Shear 155

in this diametral plane is found to be equal to the maximum bearing

stress ab at point A of the rivet surface (Fig. 89).
This conditional bearing stress can be calculated by dividing the
force acting on each rivet with the area of the diametral section BCC'B'
(Fig. 89). This area is a rectangle one side of which is the rivet diameter
and the other the thickness of the plate through which the pressure is
transmitted to the rivet.
a
The pressure on each rivet is —, therefore
_ _ P
b ntd
The strength condition for bearing will be

(84)

where lorhl is the permissible bearing stress. From this formula the
required number of rivets may be determined as
n td |<r6| (8.5)

The permissible bearing stress is generally taken from 2 to 2.5 times

greater than the permissible stress under tension or compression to],
because the test for bearing strength is actually a simplified test of

C
Fig. 89

strength for contact stresses. Expressions (8.4) and (8.5) are equally
valid for single-shear and double-shear rivets.
We shall illustrate with an example how to calculate the required
number of rivets. Let us compare two types of a riveted joint, one
lap-joint with single-shear rivets (Fig. 84) and the other with double­
shear rivets (Fig. 87). Let P = 48 000 kgf, t= \ cm, [rl=1000 kgf/cm2
and l<r&l=2400 kgf/cm*. The thickness of cover plates h is always more
than 0.5 t.
 156 Shear and Torsion [Part 111
(a) For a lap joint we have
according to the strength condition for shear (8 . 1)
48000
3 .1 4 x 2 4 15
1000

according to the bearing strength condition (8 .5)

.x P 48000
I IX 2 X 2400-
Number of rivets required is 15.
(b) For a butt joint with two cover plates according to strength
condition for shear (8 .2)
8

according to bearing strength condition (8 .5 )

10

We should use 10 rivets (on each side of the joint).

We see that the number of single-shear rivets was determined by
the strength condition for shear, whereas that of double-shear rivets
by the bearing strength condition.
B. The presence of rivets introduces certain changes in the methods of
checking the tensile or compressive strength of the plates themselves.
The critical section of each plate (Fig. 90) is the section which passes
through the rivet holes. The effective width of the plate is minimum
in this section; it is said that the section is weakened by the rivet holes.
If b is the total width of the plate, then we get the following strength
condition:
7<53ST<W (8 .6 )
where rn is the number of holes in the section (in our case there are two
of them).
Knowing the plate thickness l> we can find its width b from the above
condition. The area of the weakened section ( b — m d )t is called the net
area, whereas the area of the full section b t is called the gross area.
The account of the effect of the rivet holes on the strength of riveted
plates is generally accepted but is rather conditional. Actually, consid­
erable local stresses arise over the contour of the plate, at the ends
of the diameter perpendicular to the direction of tension. These local
stresses in the material may reach the yield point and cause plastic
deformations, though in a small volume of the plate material.
Ch. $J Practical Methods of Design on S f m r 157

These local stresses are potentially capable of causing cracks only

in a material having low fatigue limit when it is subjected to a variable
load (§ 16). However, in the usual working conditions of riveted joints
this danger may be ruled out. To avoid failure of riveted plates due to
rivets the latter should be located at a certain distance from each other
and from the edge of the plates. The location of rivets in the top view
is conditioned not only by the strength and lightness of the joint but
also by manufacturing considerations.
C. Like rivets, link bolts of lugs and the common bolted joints experi­
ence shearing and bearing stresses, therefore their design does not differ
from that of riveted joints.
A somewhat different method is employed for designing high-strength
bolted joints which have found wide application in recent years, es­

pecially in bridge building. These bolts are used instead of rivets and
are tightened by means of torque wrenches to a very high values of
tensile forces, which ensure such tight pressing of the joined elements
that the frictional forces at the interface are able, to bear all the forces
transmitted through the joint. The high-strength bolts experience
neither shear nor bearing strain.
The basic idea behind the design of these bolts lies in providing the
equilibrium between external forces P transmitted through the joint
and the frictional forces that develop between the joined elements. If
we denote the tensile force for one bolt by N and the coefficient of
friction by we get the following strength condition for the joint:
P=aNfn (8.7)
Here n is the number of bolts and a is a coefficient that accounts for
possible deviations of N and f from their nominal values (a < l). The
required number of high-strength bolts is calculated from equation
(8.7):
( 8-8)
In accordance with the existing standards for steel bridges, we
assume
A f-0.6<jaA
156 Shear and Torsion [Part HI
where A is the area of bolt section weakened by thread and o,t is the
ultimate strength of bolt material which is not less than 12 000 kgf/cni*.
Depending upon the grade of steel, the following values are used: a —
=0.78 and /=0.4-0.45.

§ 44. Design of Welded Joints

In manufacturing metal structures electric-arc welding is often em­

ployed. It was invented at the end of the XIX century by Russian engi­
neers N. N. Benardos (1882) and N. G. Slavyanov (1888) and ulti­
mately found wide application throughout the world.

Electrode holder a
^ /w v y r
Electrode
— d.c Pouter
— sourse
L/vwm Welded metal

Fig. 91 Fig. 92

In electric-arc welding by Slavyanov’s method the electrode mate­

rial (steel) melts under the heat of the electric arc and fills the joint
of the elements to be welded, which are also heated to the fusion point
by the electric arc. As a result, upon cooling the molten metal forms a
weld which rigidly joins the elements (Fig. 91).
A thick protective coating is applied to the electrode to shield the
molten metal from the harmful influence of the surrounding atmo­
sphere. When the electrode melts, the protective coating forms a large
amount of slag and gases which isolate the molten metal from the
surrounding atmosphere. This ensures high quality of the weld metal,
which may otherwise have very poor mechanical properties due to
atmospheric oxygen and nitrogen (if the electrode is not coated or if
the coaling is thin).
At present manual arc welding is used mainly in joints requiring
relatively short welds, for example, in welding steel trusses, tacking
of angles, etc. Structures requiring long welds (such as mass produced
welded beams, ship bodies and gas holders) are welded by automatic
arc welding under a flux layer (Fig. 92). In automatic arc welding the
electrode wire rolled into a coil is fed to the joint at a certain distance
from the weld, thus ensuring constant arc length. The carriage with
the electrode moves along the joint at a rate which is determined by the
welding conditions. The arc and weld are protected from atmospheric
oxygen and nitrogen by a layer of flux (granulated slag of special com-
Ch. Practical Methods of Design on Shear 169

position) which also melts in the arc flame, forming a brittle, easily
removable skin.
Structures made of aluminium alloys, that have won wide popularity
in recent years, are welded by argon-arc welding using an infusible
tungsten electrode and an aluminium welding rod. The distinguishing
feature of argon-arc welding is that the arc and molten metal are pro­
tected from the atmospheric undesirable impurities by an argon jet.
Besides arc welding, resistance spot welding is employed in some
cases when thin metal sheets have to be welded (for example, welding
of thin plating and thin profiles). In spot welding the parts to be joined
are placed between tightly pressed to them copper electrodes through
which electric current is passed. The metal around the points of contact
gets heated up to a temperature which is sufficient to ensure welding
of the elements.
If the joint design, the electrode material and the welding method
are properly selected, the welded joint is found to be in no way inferior
to the riveted joint under static as well as dynamic loading (including
impact and alternating ones). In addition, electric-arc welding has
a number of advantages over riveting, the most important of which
are lower labour consumption and the absence of weakening of the
section of the elements due to rivet holes. This gives considerable sav­
ing of resources and metal besides the economy due to greater compact­
ness of the joint. The economic gains from electric-arc welding and
the fact that it simplifies the structures have in the last few years
led to gradual replacement of riveted joints by welded.
The welded joints, like the riveted joints, are designed on the
assumption that the stresses are uniformly distributed in the weld
section. The design is closely connected to the welding method; in
particular, this is reflected in the permissible stresses, which are se­
lected for the particular weld material in accordance with the welding
method (manual or automatic welding) and also the thickness and com­
position of the electrode coating.
According to existing standards, the permissible stress for weld
material is taken the same as for the base metal in the case of automat­
ic arc welding under a flux layer and manual arc welding with top-
quality electrodes. For welding with common electrodes the permissi­
ble stresses are reduced by 10 %.
The gauge length of the weld is often assumed to be 10 mm less
than the actual length to account for poor fusion at the beginning and
crater formation at the end of the weld and also to take into considera­
tion the difference in structure of the base and weld melafs.
Let us discuss the design methods for some types of welded joints.
The butt joint is the simplest and most reliable of all joints. It is
obtained by filling the gap between the end faces of the elements to
be welded with fillet metal. The butt joint, depending upon the thick­
ness of the elements is made according to one of the methods shown in
160 Shear and Torsion IPart (I!

Fig. 93. The strength test is done for tension or compression according
to the following formula
—
U< K I (8.9)

Here lt~ A m is the nominal effective cross-sectional area of the weld

having gauge length of the joint l= b and weld height h equal to the
thickness of the plates t.
Taking into account the possibility of poor fusion at the ends the
weld length is taken as l= b—10 nun and the weld has a different
strength as compared to the base metal. It should be noted that with

an appropriate quality of welding the strength of the butt joint is not

less than that of the base metal even under impact loading.*
In order to achieve greater joint strength, it is sometimes made in
the form of a cross-shaped joint with the help of a plate which is welded
by means of fillet welds (Fig. 94). Similar welds are employed in lap
and butt joints which are made with the help of cover plates.
The fillet welds laid perpendicular to the direction of force are called
edge transverse fillet welds, whereas those laid parallel to the force
acting on the lap joint are known as side or side fillet welds.
The fillet weld does not have a very definite shape of section (Fig.
95(a)). In theoretical calculations of strength the weld section is con­
sidered to be an isosceles triangle (shown by dotted lines) of height h **
(Fig. 95(6)).

* Ya. I. Kipnis, D. I. Navrotskii. Investigation of Strength of Welded Joints

Under Impact, Transzheldorizdat, 1056 (In Russian).
•* Sometimes transverse fillet wolds are made concave with height h <0.7/.
The cathetus of the wetd may be even less than the thickness of the plate.
Ch. 8} Practical Methods of Design on Shear 161

The joints made with the edge (end-lap) welds are shown in Figs. 94
and 96. These welds fail in the weakest section AB, as established
experimentally.
I t is clear from Fig. 95(b) that the total stress acting in section AB
may be resolved into normal and shearing components. As the resistance

(tt) (8)
Fig. 95

of steel to shear is lower than that to tension, the transverse fillet welds
are designed for shear assuming that the shearing stresses are distri­
buted uniformly over the area of section AB. Keeping in mind that

Fig. 96

force P acting on the joint is taken by two end-lap welds (Fig. 96),
the upper and the lower ones, we get
T = -£ -
* 2/tw
As the area of the weld section is Aw=hl=*tl cos 45c« 0 .7 tl, and the
gauge length is l= b , the strength condition may be written as

T«=T 177^fTJ (8.10)

The cross-shaped butt joint depicted in Fig. 94 is designed similarly.
162 Shear and Torsion I Pari I I I

Actually, the weld material is subject to compound loading, the

distribution of stresses in section AB being non-uniform. A study of
the welds by the methods of the theory of elasticity, which has a
sound experimental support, reveals that there is a high stress concent­
ration at the corners of welds.
Apart from this, due to shrinkage of the joints in the welding zone
during cooling, additional stresses occur not only in the weld material
but also in the base metal, thus subjecting it to a compound stress.

This factor may result in lower ductility of the weld metal thus ma­
king the joint (with transverse fillet welds) less reliable, especially
under impact or alternating loads, as compared to butt joints without
cover plates.
A joint with side (longitudinal) fillet weld is shown in Fig. 97(a).
The weld shown in Fig. 97(6) fails over a considerable length of the
joint due to shearing of the weld metal parallel to the weld in the weak­
est section AB. The strength condition for two symmetrically placed
welds may be written as

Ttt,— 2 x 0.7// (8 -H )

The number of welds doubles if two overlapping plates cover the

joint, and the strength condition takes the form

~ 4 x 0 7 /7 ^ (8 .1 2 )

The required gauge length / of the side fillet welds is generally calcu­
lated from formulas (8.11) and (8.12). The actual length of each weld
is taken as /0=^-HO mm.
Experiments show that the side fillet welds fail in a way similar
lo the failure of ductile materials with large permanent deformation.
Th is makes the working of side fillet welds more favourable as compared
to that of end-lap (transverse fillet) Welds. However, it should be borne
Ch. 8] Practiced Methods of Design on Shear 163

in mind that there is high stress concentration at the ends of side fillet
welds too.
In designing welded joints greater reliability of the joint is sought
to be achieved by using, instead of a butt joint, or in addition to it,
overlapping cover plates, which are welded by side fillet welds or edge
welds or both. As was earlier pointed out, under impact and alternat­
ing loads such “strengthening” of the butt joint may do greater harm
than benefit.
In design of a combined joint using end-lap welds and side fillet
welds simultaneously, it is considered that the joint resistance is the
sum of the resistances of individual welds, i.e. P= Pe+ P a, where
the resistance of the edge weld for a gauge length le is Po=0,7 tle hrJ ,
and the resistance of the side welds is P8—2x0.7 il, Irwl; here/„=&,
where b is the width of the cover plate. By substituting these values
we get
P -( Q .7 « #+ 1.4«,)[ tJ (8.13)
The length of the side fillet weld /, can be determined if the length
of the edge weld is known. If cover plates are used on both sides, the
number of welds doubles, i.e. the right-hand side of formula (8.13)
should be doubled.
The wide application of electric-arc welding in metal structures has
led to the development of various types of welded joints, the design
and analysis of which are discussed in special literature *.
The methods discussed in this chaptef on the design of riveted and
welded joints for permissible stresses are accepted in machine building,
ship building, aircraft building, etc. A fundamentally new method
of design for limiting state (Chapter 25) is applied*in the Soviet
Union for defining engineering structures (civil and industrial build­
ings, bridges, tunnels, etc.). This method, however, does not differ
much from the design for permissible stresses.
The joints in timber structures (grooves, keys, etc.) working under
shear and bearing are also designed by the limiting-state method.
The distinguishing feature of timber is its anisotropy due to which it
has different shearing and bearing strength depending upon the angle
between the direction of force acting on the element and the direction
of fibres. Timber has higher shearing and bearing strength along the
fibres than across the fibres or in an inclined direction; this is taken
into account by means of coefficients. The design and analysis of these
joints is available in special literature **.
* See, for example, G. A. Nikolaev, S. A. Kurkin, and V. A. Vinokurov, Ana­
lysis, Design and Preparation of Welded Structures, Vysshaya Shkola, 1971 (in Rus­
sian).
•* See, for example, A. P. Pavlov, Timber Structures, Goslesizdat, 1959 (In
Russian). See also Building Structures, edited by G. Ovechkin, Gosstrofizdat, 1975
(in Russian).
 1<A Shear and Torsion IPart I I I

CH A PTER 9

Torsion.
Strength and Rigidity of Twisted Bars
§ 45. Torque
The results obtained during the study of shear enable us to pass
over to the study of strength under torsion. In practice we come across
torsion very often; the examples of rods working under torsion are
axes of a rotating wheel, transmission shafts, elements of three-di­
mensional mechanisms, springs and even an ordinary ward key.

We shall first study torsion in round shafts. Let us imagine (Fig. 98)
shaft CF on which two pulleys, / and //, are fitted tightly. The shaft
is supported by bearings C, D, E, and F*. Pulley I rotates the shaft
with the help of a belt drive from an electric motor. Pulley / / trans­
mits this rotation to the machine tool through another belt drive.
Pulley I is acted upon from the tight and slack sides of the bell by
pulling forces 7\ and / lt respectively, which lie in a plane perpendi­
cular to the shaft axis. Similarly pulling forces T a and U act on pulley
II and transmit to it the resistance offered by the machine tool. On
the one hand, these forces exert pressure on the bearings (in the same
way as the d?ad weight of the pulley) and, on the other hand, they con­
stitute force couples lying in a plane perpendicular to the shaft axis.
Denoting the radius of any of the two pulleys by R and keeping
in mind that tension (T) of the tight side is greater than tension (i)
of the slack side, we can write the f o l l o w i n g equation of moments with

* This arrangement of the bearings has been decided upon so that the bending
of the shaft may become negligible.
Ck. Pj Torsion. Strength and Rigidity of Bars JC5

respect to the centre of circle (Fig. 98):

M >=TR — tR = {T— i)R (9.1)
Thus forces T , and li form a torque (7\—tx) R x which twists the
shaft in one direction (shown by arrow), whereas the resistance of
machine tool gives a torque (T* —/*) Rt which is directed oppositely.
For uniform operation of the machine all the forces acting on the
shaft must be in equilibrium; the torque ( 7 \ — tx)R x should all the
time be balanced by the resisting torque (7%— /s) R it i.e.
(Ti - M /?, = ( 7 * ,- U) R 3= M (9.1')
There always exists an equilibrium between the torque transmitted
from the motor to the machine tool through the shaft and the reactive
torque on the shaft due to the resistance offered by the machine tool,
irrespective of the type of transmission employed (V-belt, tooth gearing,
friction gearing, etc.).
The portion of the shaft between the centres of the pulleys is sub­
jected to the action of two equal and opposite force couples, acting in

Fig. 99

parallel planes, that rotate one with respect to the other: the shaft gets
twisted. Thus torsion is caused by force couples lying in planes per­
pendicular to the shaft axis.
We shall employ the method of sections in order to investigate the
internal forces acting in cross sections of the shaft under the action
of. these force couples. Let us consider, for example, the part of the
shaft which is located to the left of section inn (Fig. 99). It ensues
from the conditions of equilibrium of the part under consideration
that the internal forces must result in a moment M t~ M x that balances
the external moment, i.e. acts in the opposite direction. Similarly, if
we consider the equilibrium of the part to the right of section mn
we ffnd that in the same section the internal forces create a moment
The moment of internal forces acting in an arbitrary section of the
shaft subjected to torsion that tends to rotate this section about the
shaft axis is called torque or twisting moment. The magnitude and di­
rection of torque depend upon the magnitude of the external moments
acting on the length of shaft under consideration.
166 Shear and Torsion [Part I I I

It is easier to determine the signs of torques through the directions

of external moments. Torque M t will be considered positive if the ex­
ternal moment acts in the anticlockwise direction when seen from the
side of the section; in Fig. 99, Af ,> 0 .
This sign convention for M t corresponds to the direction of internal
forces that are transmitted from the part of the shaft under conside­
ration to the other part, for instance, from left to right.
In the above case there were only two pulleys on the shaft, which
transmitted to it equal and opposite torques (9.17); this resulted in
torsion of the portion of the shaft between the pulleys by the torque
There are more complex situations when a number of pulleys are
mounted on the shaft, one of them being the driving pulley and the
rest driven. Each pulley transmits its torque to the shaft, and if the
shaft is running uniformly, the sum of all the moments acting on the
shaft must be zero.
Figure 100 shows a shaft which is acted upon by torsional moments
Af», Afj, Af.,, Mi; torque M| acts in one direction (from the driving
pulley), and M3, M3 and M4 in the opposite direction (from the driven
pulleys). For uniform rotation of the shaft
—Afj-f-A4a-j-M5-f-M4= 0 (a)
The torque will have different values in different portions of the
shaft. The portion of the shaft to the left of section /*/ will be in
equilibrium under the action of torsional moment M 4 and the torque
in section /•/. Thus, Af, for this section will be equal to M4 with
a minus sign because M4 acts in the clockwise direction when seen
from the side of section. Therefore,

Similarly, if we consider the portion of the shaft located to the left

of section 2-2 (Fig. 100), we find that the moment of internal forces
in this section is M t = —Af4+Af,. We would have obtained the same
value of torque in section 2-2 if we considered the equilibrium of the
portion of the shaft to the right of section 2-2. In this case the expression
for torque would have been M t = M t + M 9\ moreover, according to
condition (a)
Af8-j-Af3= Af,— M4
Finally, for section 3-3, considering the right portion of the shaft,
we get M t = M 9 or Af,a=Afi—M s—M4. From the expressions for
M t which are given here it is evident that the torque in an arbitrary
section of the shaft is numerically equal to the algebraic sum of the
moments of external forces acting either to the left or right of this
section.
Ch. 9] Torsion. Strength and Rigidity of Bars 167

The value of the torque in different portions of the shaft may be rep­
resented graphically by plotting the so-called torsional moment diag­
ram. For this, the x-axis is plotted below the shaft drawing and the
ordinates representing the value of torsional moments in the parti­
cular section are laid off from it with proper signs (positive upwards).

Fig. 100

The torque diagram is plotted in the form of rectangles because within

the limits of a particular portion, the value of the torque Mf does
not depend upon the position of the section between the pulleys.
Suppose inFig. 100 Af i=600 kgf- in, Afa=300 kgf • m, Ma= 100 kgf- m,
and Af4=200 kgf - m. The distribution of torsional moments along the
length of the shaft is shown in Fig. 100.

§ 46. Calculation of Torques Transmitted to the Shaft

To find the torques acting on a shaft we must know the moments,

transmitted to it by all the pulleys. These moments may be determined
if we know the number of revolutions of the shaft and the power trans­
mitted by the pulley. Let a force couple having moment M be acting
on the pulley (Fig. 101). We can imagine the couple to be consisting
of two forces P applied at the contour of the pulley. Upon rotation
the couple performs work; the magnitude of this work per unit time
is equal to the power transmitted by the pulley.
Let us calculate the work done by the couple when the pulley is ro­
tating. As the pulley revolves through an angle a, each force of the
couple covers a distance Rat, where R is the radius of the pulley. The
total work done by the couple of forces will be
W = 2P R a= M a
 168 Shear and Torsion I Port HI
Thus, the work done by a force couple when it is revolved through
an angle a is equal to the moment of the couple multiplied by the ancle
of rotation (in radians).

If the shaft Completesm revolutions per unit time, then the work done
will be W=2nmM. On the other hand, work per unit time is the power
/V. Therefore, the torque may be expressed through the given values
of power and number of revolutions per unit lime of the shaft:
M (9.2)
If the power is given in h.p., then N = L h. p. or N = 75 L kgf- m/sec
and if the speed is m =n r.p.m., or per second n/60, then
75 X 60 x L
m
M —---- .. = 2250/,
— 716.2-^-kgf-m (9.3)
2nn an
The power may also be given in kilowatts, N —K kW. As 1 kW is
approximately equal to 102 kgf* m/sec, we get
M« !L kgf. m = 973.6 £ kgf •m (9.4)
For given L or K we calculate the moment transmitted by each pulley
from formulas (9.3) and (9.4), plot the twisting moment diagram and
find the critical section in which M t= M ttm9X.

§ 47. Determining Stresses in a Round Shaft Under

Torsion
Having plotted the A/fdiagram, we cin find in anv section of the
s' aft the twisting moment made up of the moments of internal forces
acting in this section. Let us try to determine these internal forces
and the corresponding stresses in the section. For solving this problem
we shall use the results of experimental research given below.
C/t. 9] Torsion. Strength and Rigidity of Bars 169

A. Experiments show that when a round shaft is twisted by a couple

of M (Fig. 102), the following points are observed.
All the generating lines revolve through an angle y, and squares
drawn on the shaft surface warp changing into rhomb, i.e. they are
subjected to shear.
Each cross section revolves w.r.t. the other about the shaft axis
through an angle called the angle of twist. The value of this angle is

Fig. 102

directly proportional to the torque and the distance between the sec­
tions.
The end face remains a plane and the contours of all the sections
remain undistorted (circles remain circles). Radii marked on the end
face remain straight lines even after deformation.
The distance, between adjacent sections practically does not change,
i.e. sections /-/ and 2-2, while turning with respect to one another
through angle A<p retain their relative distance Ax.
Thus, the experiments show that a bar in torsion represents a system
of rigid discs mounted centrally on a common axis 0 ,0 3. Upon defor­
mation all these discs turn w.r.t. one another without changing their
shape, size and relative distance.
The experimental observations enumerated above give us the basis
for formulating the following hypotheses:
1. all cross sections remain planes;
2. radii on the sections remain straight lines;
3. distances between the sections remain unchanged.
The applicability of these hypotheses is further supported by the
fact that the formulas obtained on their basis give results which agree
well with those obtained experimentally.
B. Let us now pass over to determining stresses in sections perpendi­
cular to the shaft axis. Let us imagine (Fig. 103) the twisted shaft
0,0* to be cut in two portions / and I I by a section /-/ perpendicular
to the shaft axis and located at a distance x from section 0,. Let us
remove portion II and consider portion I. This part must remain in
equilibrium under the. action of external moment A! applied in sec­
tion Oi and torque M t acting in section 1-1. The equilibrium condition
170 Shear and Torsion [Part I U

of the cutoff portion may be expressed as

Mt = M
According to its definition, torque M* is the moment of internal for­
ces that replace the action of the removed portion of the shaft. In
order to be able to create moment M t, the internal forces in the section
and the corresponding stresses must be tangential to the section and
perpendicular to the radii *. For calculating the moments of these

Fig. 103

elementary forces and their sum let us consider an arbitrary

point at a*distance p from the centre of the circle and isolate an ele­
mentary area dA around it (Fig. 103). The force actingon the elementa­
ry area will be dP — dA , where t p is the shearing stress at the given
point. The moment of this force about point 0 is
dAft =
Considering area dA to be infinitely small, we can find the sum of the
moments of all the forces as a definite integral over the area of the
section:
2 Aft - S *pP dA
A
or, since 2Aft = M #,
$TppdA=M* (9.5)
A

* If w e assume that t Is not perpendicular to radius, then it must have a compo­

nent along the radius, which, according to the law of complementary shearing stres­
ses, must give rise to shearing stresses along the cylinder generatrices, including
the ones on the external surface of the shaft which is free of aU stresses (see Fig. 122).
Ch. 9\ Torsion . Strength and Rigidity of Bars 171

However, we cannot yet find x from the above equation as we do

not know how the shearing stresses are distributed over the section.
C, It is not possible to determine the stresses in section 1-1 with

Fig. 104

the help of static equations only. It is a statically indeterminate prob­

lem, and to solve it completely we must take into account the defor­
mation of the shaft shown in Figs. 102 and 104.
Let us isolate (Fig. 104) on the surface of the shaft, prior to deforma­
tion, a rectangle ABDC by two adjacent generating lines ab and cd
•and two portions of sections 1-1 and 2-2.

After deformation both sections /-/ and 2-2 turn about the fixed
end through angles <p* (section 1-1) and <p*-fd<p (section 2-2). On
the basis of the accepted hypotheses, we can say that both sections
will remain planes, radii 0 J3, CM, OlC and 0*D will remain straight
172 Shear and Torsion [Part I t!

lines and distance, dx between sections 1-1 and 2-2 will remain un­
changed. Under these conditions the whole element ABDCOtOt will
be displaced and warped because its right face which lies in section
2-2 turns through dep w.r.t. the left face which lies in section /-/
Rectangle ABDC occupies the posi­
tion which is shown in Fig. 104 by
hatched lines. The warped element
AiByDxCiOxOi is shown in Fig. 105;
on the same figure is shown the form
of the element if it had remained
undistorted, i.e. if its left and
right faces both had turned
through the same angle.
The warping caused by unequal
turning of sections 1-1 and 2-2
transforms right angles of rectangle
ABDC into acute and obtuse angles;
the material of the element exper­
iences shear (Figs. 102 and 104).
The magnitude of this deformation
is characterized by the angle of
distortion or the angle of shear. On the shaft surface in rectangle
this angle is equal to BAtBx\ it is denoted in Fig. 105 by y.
We already know that shear is accompanied by appearance of
shearing stresses in the faces of the warped element {§ 36).
Figure 105 depicts the stresses acting on an elementary area fl,D,Oa
enclosed between the right face (section 2-2) and the horizontal sur­
face of the element AiBiOtOi. Their value may be expressed through
the angle of shear y which characterizes the warping of the rectangle
AyBiDiCi by formula (6.37):
r= y G

As the absolute displacement of the element on the shaft surface

liU* A.
is BB'~rd<p, at the angle of shear V ~ j -q = r~£i> the stress around
point Bi will be
T/»= Gy = rG —

Let us now determine the stress t p at another point of section Lit

which is at a distance (> from the centre (Fig. 105). For this we must
find the angle of shear of the material at point L \. In Fig. 105 the
angle of shear or the angle of warping LKU is denoted by y,». It will
be less than the angle of shear y at the shaft surface. By the same
Ch. Torsion. Strength and Rigidity of Bars 173

reasoning as in determining y. we find that 7<»=p^and get

Tp = pG "j~ (9.6)

The angle of shear and the shearing stress at any point in the cross
section of the twisted shaft are directly proportional to the distance
p of this point from the centre of the section. Graphically, the variation
of the shearing stresses may be depicted by a straight line (Fig. 106).
The shearing stresses t are maximum at points lying on the edge of the
section and zero at the centre.
Thus, we have established the law of distribution of shearing stres­
ses in cross sections of a twisted shaft.
D. The shearing stress may now be determined from equation (9.5).
Replacing tp by its value from equation (9.6) and taking the quantity
G g (which is constant when integrating over area) out of the integ­
ral sign, we get
G g Jp W -A f,
A
($ p 3d/4, i.e. the sum of the products of elementary areas into the
A
squares of their distances from point 0 , is called the polar moment
of inertia and denoted by Jp. Consequently,

From the equation we can find the twisting angle per unit length of
the rod
dq> Mt
dx GJ p
(9.7)

Substituting ^ into (9.6), we get

(9.8)

The shearing stress is maximum at points of the section which lie

at the rod surface, i.e. when p=tpm.,x= r:
r __ MfPmax M ir
(9.9)
''max 7
Jp ~7Jp
The formulas for may be written in another form:
_ M fP max ____ M_t______M t
(9.10)
174 Shear and Torsion [Part I N

The ratio ■—- — Wp is called the section modulus', as the mo-

Pmax
ment of inert ia Jp is expressed in units of length to the fourth power,
the section modulus Wp is measured in units of length to the third
power.
The quantities Jp and Wp are geometrical characteristics of the sec­
tion under torsion, i.e. they show how shape and size of the section
influence the torsional resistance of the rod. As described later {§48),
their values are determined through the rod’s diameter.

§ 48. Determination of Polar Moments of Inertia

and Section Moduli of a Shaft Section

To determine Jp= ^p%dA we isolate a circular ring between radii

A
p and p+dp (Fig. 107). Now in this ring we isolate an elementary area
dA. We sum up the products p*di4 for the ring and then sum up the

values obtained from all the.rings into which the section may be divid­
ed. As all the elementary areas in a particular ring are located at a
fixed distance from the centre, p, we have
2 p * d A = p*'£dA
In a ring 2 dA is the area of a thin circular strip; 2gL4,=2jip dp and
therefore pa2<14=2np3dp. Summing these quantities for the whole
of the section, we get
r r
J p = ^ 2np» dp = 2 n J p» dp =* ^
b o
or expressing r through the diameter
nd*
's F ‘ O.ld* (9.11)
Ch. 91 Torsion. Strength and Rigidity of Bars 175

This is the polar moment of inertia of a circular section. The section

modulus of a circular section under torsion will be
it/ 4 nr9 nd*
Wp T t?r « 0.2d* <9.12)
Pmax 1 7
It is clear from formula (9.8) that shearing stress r is not large at
points of thesection close to the centre (where p is small). The twisting
moment is balanced, chiefly, by stresses acting in the section near its
surface; the material of the central portion of the shaft experiences
low stresses and does not contribute much to the resistance to torsion.
Therefore the shafts of large diameters are sometimes made hollow
to make them lighter and cheaper (Fig. 108). In this manner we re­
move the central portion of the shaft, which is incidentally the weak­
est portion of the forging, affected most by harmful inclusions.
We shall determine the moment of inertia and section modulus of
such a tubular section. Let us denote the outer radius by R and the
inner radius by r. Then substituting dA=2np dp, we get
R

(9.13)
W 4— *4) * (D*— d•)
The section modulus is
m - Jp r«) n {D * -d * )
? Pmax 2# 160
(9.14)

If we assume the ratio dlD=a, or #=aZ>, wc get

nD4
^P ~ 32 ^ (9.13')
nD*
^ P ~ is (1 a<) (9.140

If the thickness of the tubular section is small (t<,0.\R), then de-

noting the mean radius of the pipe by and keeping in mind
that R— r= l, we obtain

J p - 7 (R* “ ■
r4) - y (*“■+r*)(R + r) (R - r) = £ (*• + r*) 2r01

Replacing R by r0+ j a n d r by r%— j and neglecting the square

of the thickness t we get
J p « 2nr%t (9.15)
176 Shear and Torsion [Part I I I

Similarly the section modulus may be found as

Wp*s2nr\l (9.16)
These approximate formulas are very convenient for practical cal­
culations.
It is obvious that for each cross section the polar moment of inertia
and the section modulus have a single definite value which depends
upon the dimensions of the shaft section.

§ 49. Strength Condition in Torsion

Knowing the section modulus we can determine max t from formula
(9.10).
According to the strength condition the maximum shearing stress
must not exceed the permissible stress, i.e.
max t *=
Wp
< ; [tJ (9.17)

From this formula we can determine the section modulus for a known
twisting moment and the assumed permissible stress, and then from
the determined section modulus we can calculate the required radius
or diameter of the shaft.
As explained earlier (§ 42), the permissible stress [xl should be taken
0.5 to 0.6 of the principal permissible tensile stress, as in the case of
pure shear. In practice the value of It] for mild steel varies from 200 to
1000 kgf/cm*, and for carbon steel from 300 to 1200 kgf/cm4, depending
upon the type of load (static, alternating, impact) and the magnitude
of local stresses that occur in <the keyways, bosses and other places
where the shape of the shaft section changes.

§ 50. Deformations in Torsion. Rigidity Condition

We had seen in § 47 that the torsional deformation of a cylindrical

bar is distinguished by relative rotation of adjacent sections. The angle
of rotation of one section with respect to another was called the angle
of torsion <p. For sections located at distance dx from one another we
had obtained expression (9.7)

If the distance between the sections is I, the angle of torsion is

(9.18')
Ch. 9\ Torsion. Strength and Rigidity of liars 177

Usually the torque is constant within the limits of a particular por­

tion; therefore integrating with respect to at, we obtain
Mft
GJp
(9.18)

Formula (9.18) has complete analogy with the corresponding formula

for tension and compression and expresses Hooke’s law for torsion.
It is evident from formula (9.18) that the greater the torsional rigidity
GJj„ the smaller is the angle of twist <p (for a given M t). Thus, Jp
reflects the effect of the dimensions of the cross section on the deforma-
bility of the rod under torsion, and G the effect of elastic properties.
If the shaft is mounted by a number of pulleys which divide it into
portions subjected to different twisting moments M t, then formula
(9.18) enables us to calculate the angle of twist of one end w.r.t. the
other for all the portions.
Computation of the angle of torsion has practical importance:
firstly it is required for determining the reactions of support of twisted
shafts in statically indeterminate systems; however, this is a rare case.
Secondly, we must know the angle of torsion to check the rigidity of
the shaft.
The maximum permissible limits of angle <p, which should not be
exceeded to ensure safe working of the machine, have been established
experimentally. These limits are as follows: under normal conditions
[<pl—0.3° per unit length of the shaft; under alternating loads l<p]=
=0.25°; under suddenly changing (impact) loads Upl=0.15°. Sometimes
under normal working conditions we take l<p!=l° per 20 times the
shaft’s diameter *.
Hence, the shaft dimensions should be calculated not only from the
strength condition (9.17) but also the rigidity condition

This condition is often the most important when designing long

shafts. We shall explain the rigidity check with the help of the follow­
ing example.
Suppose a shaft transmits A7=150 hp at n=60 r.p.m. It is required
to select shaft diameter from the strength condition and check it for
rigidity if the permissible stress [xl—600 kgf/cma and the. permissible
angle of torsion per metre length of the shaft is hpJ=0.3°. Shear modu­
lus G=8xlCfi kgf/cm*

• For some time now the permissible angle of torsion f<f] Is taken up to 2®and
more per metre length of the shaft depending upon its functioning. Thus, for instance,
angles of torsion of up to 2.5° per unit length are permitted for automobile cardan
shafts.
7 -8 8 1 0
178 Shear and Torsion {Part I I I

The moment transmitted by the shaft may be calculated from exp*

ression (9.3);
M = 7 1 6 .2 - = 716.2 1800k g f m = !8x 10*kgf-cm

From strength condition (9.17) the section modulus may be found as

M, M 18X 10* o,-,a
Wp> lt|
= - ^ - = -<550-=300
As fFp«0.2d3 Iformula (9.!2)J, the shaft diameter will be

d > j / o = l l -45<:,n
Let us take d = ! 1.5 cm and check the section for rigidity. According
to formula (9.11) the polar moment of inertia may be calculated as
J p « 0. Id* = 0.1 X (11.5)* = 1745 cm*
The angle of torsion per metre (or 100 cm) length of the shaft is
calculated from formula (9.18):

^ = ^ = ^ T r a = 0 l2 9 rad = 0-78° > W

We see that although the strength condition is satisfied, the shaft
diameter should be increased to improve rigidity and it should be cal­
culated from expression (9.19):
M tl

wherefrom, by substituting the value of permissible angle of torsion

in radians l< f'l= 0 .3 ~ , we obtain
•1 / ISXHXXI80
V 0. I x 8 x I0BX 0.3 —
14.56 cm

Hence, the shaft’s diameter should be taken d=14.6 cm to ensure re­

quired rigidity.

§ 51. Stresses Under Torsion in a Section Inclined

to the Shaft Axis
While studying the stresses in a round shaft under torsion (§ 47),
we saw that sheafing stress t acts at every point of a section perpendi­
cular to the shaft axis. According to the law of complementary shearing
stresses, a similar stress (Fig. 105) will act in the faces of the cutoff
element lengthwise. These stresses will also be maximum at the sur­
face and will be zero at points on the axis.
Ch. 91 Torsion. Strength and Rigidity of Bars 179

Thus, if we cut the twisted shaft by a diametral plane (Fig. 109),

the points on straight line AB perpendicular to the shaft axis will ex­
perience shearing stresses which change linearly. There will be no
normal stresses in these planes.
Normal stresses act only, in in­
clined planes and are maximum
in planes which are inclined at
45° to the shaft axis.
Actually, element ABCD cut
near the shaft surface (Fig. 110)
experiences only shearing stresses
on its side faces. It is in similar
conditions as the element abed in
Fig. 75, i.e. in conditions of pure
shear. Therefore, there are no
shearing stresses in planes inclined at 45° to the shaft axis; these
are the principal planes which are subjected to tensile and compres­
sive stresses oi and o3 equal to x at each point (see Fig. 105).

m (b)
Fig. n o

The value of these stresses varies from point to point in direct pro­
portion to their distance from the centre and is equal to x. Brittle ma­
terials like cast iron fail in torsion due to rupture in an inclined sec-

Fig. i l l

tion BC (Figs. 110 and 111), i.e. in a section where the tensile stresses
are maximum.
Knowing the magnitude and direction of the principal stresses at
any point, we can determine the normal and shearing stresses in any
7*
180 Shear and Torsion [Part III

inclined plane by Ihe stress circle or with the help of formulas (6.5)
and (6.6). As the absolute values of the maximum normal and shearing
stresses are equal and the permissible shearing stress is less than nor­
mal one, we can limit the strength check in torsion, as in pure shear, to
analyzing shearing stresses only.

§ 52. Potential Energy of Torsion

Previously, when we were studying tension, it was shown (§ 35)
that when an elastic system deforms, it accumulates energy called the
potential energy of deformation.
This phenomenon occurs in torsion as well. If we twist an elastic
rod within the limit of elasticity, then, when the external forces are
removed, it will untwist and perform work at the expense of the po­
tential energy it had accumulated during
deformation. Neglecting the irreversible
losses (healing, internal friction, etc.), we
can consider that the work done by the
internal forces, which is determined by the
amount of accumulated potential energy
£/, is equal to the work W done by the
external forces.
Suppose we have a shaft one end of
which is fixed and the other is loaded by
a force couple which creates a moment
that gradually increases from zero to a
finite value M. The increase in M will
result in a corresponding increase of <p which is related to M t by
equation (9.18):

If we plot the angle of twist (deformation) along theje-axis and the

corresponding values of the twisting moment along the y-axis, then the
relation between the two will be represented by an inclined straight
line OA (Fig. 112). By the same reasoning as employed in calculating
the work done by a tensile force P, we find that the work done by the
force couple M may be expressed through the area of triangle OAB:
(9.20)
The constant 1/2 in formula (9.20) is due to the fact that the moment
M has not been applied in its full magnitude at once but increased
gradually, ‘'statically” from zero to its finite value.
Replacing <p by its value from equation (9.18) and keeping in mind
Ch. 91 Torsion. Strength and Rigidity of Bars 181

that U=W, we get the expression for potential energy in torsion:

mU
U 2QJp (9.21)
The potential energy may also be expressed through deformation
if we replace the torque in formula (9.20) by its value from formula
(9.18):
GJ,
= p
Then
_gj_l
~ 21 r (9.22)
It is evident from formulas (9.21) and (9.22) that the potential ener­
gy of torsion, as of tension, is a function of the square of force or de­
formation.
§ 53. Stress and Strain in Close-coiied Helical Springs
Tension and compression helical springs are used in wagons, valves
and other parts of mechanisms. When designing such springs, we must
know how to calculate their maximum stress (for strength check)
and deformation, their elongation or compression. This is essential
because the load on the spring is controlled by deforming it more or
less under tension or compression.
Hence, we must determine the relation between deformation and the
force, acting on the spring provided its dimensions are known. It
will be seen that the spring material experiences torsion when it is
stretched or compressed.
We shall restrict our discussios only to close-coiled helical springs,
i:e. springs in which the distance between adjacent coils (pitch) is
small as compared to the diameter. If this condition is satisfied the
coil inclination may be ignored and it may be assumed that any arbit­
rary cross section of the spring is parallel to forces P acting along
the spring axis and either stretching or compressing it (Fig. 113).
Let us introduce the following notations: radius of the spring helix
R\ diameter of the spring wire d=2r> number of turns in the spring
n and shear modulus of the spring material G.
To determine the internal forces and stresses acting in the spring
section when it is stretched (or compressed), let us cut one of the coils
by a plane passing through the spring axis and consider the equilibrium
of one of the cutoff portions, for example, the lower onev(Figs. 113(b)
and 114). External force P acting on this portion in The downward
direction is balanced by an upward acting internal force P%=*P which
lies in the plane of the section and is transmitted through this section
from the upper portion to the lower one.
Since forces P and Pi form a couple with a moment M =PR that
182 Shear and Torsion |Part I I I

rotates the portion of spring under consideration in the anticlockwise

direction, it can be balanced only by a moment M t=PR of the inter­
nal forces lying in the plane of the section and acting in the clockwise
direction. Since internal forces P\ and their moment Af*, which replace
the action of the upper portion of the spring on the lower, lie in the

Fig. 113 Fig. 114

plane of section, they are made up of shearing stresses. Shearing force
Pi= P is formed from elementary shearing forces dPi=Tt cM that
prevent the section from shearing downwards (Fig. 115 (a)). If the
distribution of shearing stresses over the section area is assumed to
be uniform, then force Pi may be expressed as Pi —TlA, wherefrom
shearing stress
<9 - 2 3 >

The torque M t that prevents the section from rotating (Fig. 114)
is related to shearing stresses t 3 in torsion by the formula

Both systems of stresses t * and t „ that appear in the spring section

when it is subjected to a tensile force P are depicted in Figs. 115(a)
and (b). At each point of the section stresses T i and t 2 are summed geo­
metrically as their directions coincide only along radius AO.
As the shearing stresses due to torsion are maximum at the periphery
of the section Iformula (9.10)1, i.e.
- _ Mt _ 2 PR
”*ni«K W ^—1CFT
Ch. 9] Torsion. Strength and Rigidity of Bars 18J

point. A on the internal edge of the contour will be the critical point
since here stresvsesxi and Ta add up arithmetically. Hence, the maximum
total shearing stress in the spring section is
T _ p . a>*_ r w\ (9.25)
m ax nra 1 nrt \ ' r)

From strength condition rwax< It ], or

(9.25')

As in a majority of cases the second term inside the brackets is con­

siderably greater than unity, the first term is usually ignored, i.e.
the stresses due to pure shear are neglected and only stresses due to
torsion by moment PR are consi­
dered. Therefore

(9-26)

The elongation of the spring’s

axis under tension, which is de­
noted by K can be very easily cal­
culated on the basis of this appro­
ximation.
Let us cut from the spring a
segment of length ds by two ad­
jacent sections CO} and C02 pas­
sing through the spring axis (Fig.
116). As we select the sections ve­
ry close to each other, it may be
assumed that before deformation
the radii R drawn from the
spring’s axis to the centres of the
sections lie in the same plane and Fig. IIS
form a triangle OiCOt.
After deformation due to torsion of the segment ds, the second section
turns w.r.t. the first by an angle Consequently, radius
0 £ turns w.r.t. radius OiC by the same angle dtp and point C occupies
the position Ci, which means that the end of the spring moves down by

If we consider that all similar elements ds deform in an identical

manner, then the total distance by which the Iowa* spring end moves
184 Shear and Torsion [Part III

down, i.e. its elongation, may be expressed as the sum of dk:

X= GJp
(9.27)

/
Here /= ^ ds is the total length of the spring wire and is the rela*
A ^
tive angle of torsion of the ends of the spring wire, which is determined
under the assumption that the spring wire has been straightened.

F ig . 1 1 6
Neglecting the inclination of the spring coils to the horizontal and
assuming the number of turns to be n we get the total length of the
spring wire equal to
I = 2 siR n
Therefore
(9.28)

A similar formula can be derived by comparing the work done by

external forces W = ^ P l with the potential energy of torsion V =
Af2/
— j-\ the reader is advised to do this independently.
Denoting the permissible elongation (or compression) of the spring
by [XI, we can write the following rigidity condition:
*= ^ < [> •1 (9.2»')

Formulas (9.25') and (9.28') enable us to check the strength and find
the deformation of the spring.
The greater the permissible shearing stress [xl, the more flexible
is the spring and the greater wifi be its compression under a particu­
lar load P, because it may be manufactured from a thinner wire. The
wagon springs must be sufficiently flexible, therefore they are made
from tempered steel with a high elastic limit. The permissible shearing
Ch. 9] Torsion. Strength and Rigidity of Bars 186

stress may be up to 40 kgf/mm* and sometimes as high as 80 kgf/mm*.

The permissible stress for chromium-vanadium steel in tension springs
is taken up to 70 kgf'mm* at r—6 to 8 mm. The permissible shearing
stress for phosphor-bronze is |t!=13 kgf/mm® at G=4400 kgf/mm* and
r up to 8 ram.
These values of permissible stress are valid only under static loading;
under alternating loads they reduce by about 1/3, and for springs work­
ing non-stop (valve springs) by about 2/3. In these cases an important
factor is the development of fatigue cracks (see § 16). In addition, the
valve springs often work at high temperatures; this also requires a re­
duction in the permissible stresses.
In practice, when designing springs according to formula (9.25),
we introduce a correction factor ft which, apart from shearing, takes
into account other factors (bending of the spring wire, longitudinal
deformation, etc.) that were not considered above. The greater the ratio
•g , i.e. the greater the torsional rigidity of the spring, the greater the
value of factor ft.
Formula (9.25), which in addition to torsion accounts for shear due
to force P, is replaced by

(9.29)
2

The value of correction factor ft may be taken from Table 8.

Table 6
Correction Coefficients for Designing Springs

Rir 4 5 6 7 8 9 10 It 12 15

k 1.42 1.31 1.25 1.21 1.18 1.16 1.14 1.12 1.11 1.09

In design of springs, the known quantity sometimes is not the force P

which stretches or compresses the spring but energy T which it must
absorb. As in tension or compression of a rod, the potential energy of
deformation U of the spring is measured in terms of the work done by
the external forces.
As P and ?* arc linearly dependent upon each other Iformula (9.28)1,
the potential energy of deformation of the spring may be written as
,, I 2Pt R3n
186 Shear and Torsion IPart H i

From formula (9.26) we have

Tnr*
PR-

Putting this value in equation (9.30), we obtain

4G

As 2nRn is the length of spring wire and nr2 its area of cross section,
U (9.31)
W

Here V is the volume of the spring. Keeping in mind that U =T, we

can write formula (9.31) as
4GT
(9.32)
:rF F
Thus, by assuming the limit ing value of stress T=rr] we can calculate
the volume of the spring required to absorb energy T = U such that
the permissible stress ItI is not exceeded. The compression of the spring
under |x| should be checked; it should be such
that the gap between the spring turns is not
completely eliminated.
As an illustration we shall calculate the ma­
ximum stress and elongalion of the cylindrical
spring shown in Fig. 113, if spring radius
R=lQO mm, spring wire diameter d=20 mm,
number of turns n=10 and tensile force
P=220 kgf. Shear modulus 6'=8.5 x lO^kgf/em2.
The stresses will be calculated with the help
of formula (9.29). As the ratio — =10, the
correction factor 6=1.14 (Table 8) and

T - l .! 4 ? m ^ = 1 5 9 2 k g f / c r a 1

The elongation (or compression) of the spring

according to formula (9.28) is
Fig. 117 4PR? 4 x 220xl0!'x i a
Gr* 8.5x !06XI4
= 10.4mm

In addition to cylindrical coil springs conical springs (Fig. 117)

are also used in engineering practice. The radii of the top and bottom
turns in Fig. 117 are denoted by R t and R t respectively; the average
Ch. 9\ Torsion. Strength and Rigidity of Bars 187

radius R may be calculated by the formula

where n is the number of turns and a is the angle formed by the radius
under consideration with the top radius Ri and measured along the
spring turns.
The strength of conical spring is checked with the help of formulas
(9.25) or (9.26) by replacing R with its maximum value R». To deter­
mine X, as in the previous case, we must add the elementary deforma­
tions
dX = ^Rrfs

M t—PR is now a variable quantity. Therefore

(9.33)
Sometimes springs are manufactured not from a round wire but a
rectangular-section wire; for such springs the formulas given in § 54
(Table 9) have to be used to calculate the stresses and deformations.

§ 54. Torsion in Rods of Non-circular Section

In engineering practice we often come across rods of non-circular
section subjected to torsion; these include rolled and thin-walled rods.
Under torsion the cross sections of such rods do not remain planes, they
warp. As depicted on the example of a rectangular section, points of the
section do not remain on the plane (some get displaced inwards, others
outwards) and the section undergoes plane skift (warping) (Fig. 118).
A. When a rod of uniform section is twisted by force couples applied
at its free ends, all cross sections of the rod undergo equal plane shift.
Therefore the distance between equally displaced points on adjacent
sections does not change, i.e. the lengths of longitudinal fibres remain
unchanged. This means that the cross sections of the rod are free of
normal stresses when plane shift of its sections is uniform.
Torsion is known as pure or free when cross sections of the twisted
rod are free of normal stresses. It should be noted that free torsion
is possible only under conditions of unconstrained (free) plane shift
of all sections. In pure torsion the magnitude and distribution of
shearing stresses are the same in all cross sections.
If plane shift of a single cross section of the twisted rod of non-
circular section is constrained (for example, by the conditions of fixa­
tion or loading), torsion can no more be considered free: it will be ac-
 S f m r and Torsion \Part I I I

companied by change in length of longitudinal fibres and normal

stresses wil 1begin to act in cross sections. In this case the shearing stres­
ses have different magnitude in different sections: they are made up
of shearing stresses of pure torsion and additional shearing stresses
connected with the non-uniform
plane shift of the rod along its
length. Torsion with constrained
plane shift is known as constrai­
ned torsion.
Figures 119 and 120 depict the
states of pure and constrained
shear in an I-section. Figure 119
shows the deformation of I-sec­
tion with free ends to which force
couples with moments M„ are
applied, i.e. an 1-section subject­
ed to pure torsion. Figure 120
depicts the deformation of the
same I-section when same mo­
ments Mo are applied to its ends.
In this case, however, one end of
the rod is rigidly fixed: therefore
the fixed end remains plane, its warping is completely constrained and
hinders free plane shift of adjacent sections. Torsion may be considered
free only at the free right end of the rod. Hence, this is a case of const­
rained torsion or, as it is also known, bending torsion (the l-seclion
flanges, like elements of thin-walled sections in general bend when the
section is subjected to torsion).
Tin' problem of constrained torsion was first formulated and solved
by Prof. S. P. Timoshenko in 1905 **. However, these problems drew
the attention of engineers and research workers only towards the end
of twenties in connection with the developments in aircraft industry
and introduction of thin-wailed structures in civil engineering. Soviet
scientists contributed much to the theory of design of thin-walled
structures and shells, in particular Prof. V. Z. Vlasov who put for­
ward the general theory of design of thin-walled open-profile rods
(1939**). This theory further developed in subsequent years and
along with the theory of shell design grew into an independent branch
of mechanics of structures, which is widely covered in literature.
The theory of constrained torsion is to a certain extent based on
the theory of pure torsion of rods of non-circular sections; some of
the results of the theory are given below.

• Proceedings of St. Petersburg Polytechnicol Institute, Vol. 4, 1905.

• • V. Z. Vlasov, Thin-walled Elastic Rods, Stroiizdat, 1940 (in Russian). Also
seeN. M. Belyaev, Strength of Materials, Nauka, 1965 (in Russian).
Cft. 9] Torsion. Strength and Rigidity of Bars

B. Since torsion of rods of non-circular sections is accompanied by

warping of the sections, one of the basic hypotheses of strength of
materials—the hypothesis of plane sections—becomes inapplicable.
The problem of torsion of such rods requires more complicated mathe­
matical analysis and can be solved only by the methods of the theory
of elasticity.

(b)
Fig. 119

The first theoretical investigation of pure torsion in rods of non-

circular sections was carried out by Saint-Venant in 1864; he also
presented a number of solutions of particular problems (torsion of
rods of rectangular and elliptical sections). Solutions to many problems
on free torsion of rods, including rods of very complex profiles, have

lb)
Fig. 120

been found by now on the basis of the general method of design of

such rods developed by Saint-Venant. However, in spile of the com­
plexity involved in solving these problems by the theory of elasticity,
their results can be presented in a simple and convenient form for
practical use. The formulas for maximum shearing stresses and strains
are presented in the form of expressions which are completely identi­
cal to the formulas for maximum shearing stresses and angle of twist
 190 Shear and Torsion [Part I U

of round bars in torsion:

‘max up (9.34)
M,l
(9.35)

In these formulas J t and Wt are geometrical characteristics of the

section (similar to J p and \Vp for round sections and having same
units) which arc conditionalty called the torsional moment of inertia
(Jt) and the section modulus in torsion ( Wt). For circular and ring
sections J t= J p and Wt^ W 9.
For some sections these geometrical characteristics have been de­
termined in closed form: for instance, for an ellipse
t _ na1^3
(9.36)

Here a and b are the major and minor axes of the ellipse, respecti­
vely. The shearing stress diagram for the elliptical section is shown
in Fig. 121. Along the profile of the section the stresses form a conti-

Fl«. 121

nuous flux tangent to Ihe profile, and attain maximum value at the
end of the minor axis (Troax= ^ - j ; at the end of the major axisxa=
_ b tJ

Identical shearing stress fluxes are directed along closed curves

shown bv dotted lines (shearing stress trajectories). The magnitude
of shearing stress x increases, as we move from the centre of the ellipse
towards its periphery, in direct proportion to the distance (Fig. 121).
Ch 91 Torsion. Strength and Rigidity uf Bars 191

It should be noted that when a rod of an arbitrary profile is sub­

jected to torsion, the shearing stresses at the contour should be tan­
gent to the section in accordance with the law of complementary shear­
ing stresses. If the possibility of stress component perpendicular
to the periphery is conceded, this will imply that complementary
shearing stresses act on the side surface free of all stresses (Fig. 122).
For the same reason shearing stress t= 0 at the corners. We can see
this in the example of a rectangular
section (Fig. 122; top left corner).
For a rectangular section with
sides b and /i, the geometrical char­
acteristics depend upon the ratio
between the sides ana are expressed
by the following formulas:
J t = ab* and Wt = fto» (9.37)
or
J t = * Lhb*y W t ^ Q Lhb* (9.37')
where
b
‘ k a and 0, =-r?>
a.
Fig. 122
The distribution of shearing stres­
ses over a rectangular section is
shown in Fig. 122. Along each side the shearing stress t varies accord­
ing to a parabolic law and attains maximum value at the middle
of the longer side ; at the middle of the shorter side x=
= Y W - and at the corners t= 0.
Table 9 contains the values of coefficients a, p and y.

Table 9

Coefficients for Designing Rectangular Bars Under Torsion

h/b a 0 y h/b a 0 V

1.0 0.141 0.208 1.000 3.0 0.790 0.801 0.753

1.2 0.199 0.263 0.935 4.0 1.123 1.128 0.745
1.5 0.294 0.346 0.859 5.0 1.455 1.455 0.744
i .75 0.375 0.418 0.820 6.0 1.789 1.789 0.743
2.0 0.457 0.493 0.795 8 .0 2.456 2.456 0.742
2.5 0.622 0.645 0.766 10.0 3.123 3.123 0.742
192 Shear and Torsion [Part l / I

Table 10
Data on Torsion of Non-circular Sections

e. , Moment of Inertia Ji Section modulus

Shape of the section (cm*> Wf (cm*) Remarks

h i _h% =m>l
bi —77
x m* a<l
(1—a4) Wt = n r (1—a*)m
10 (/«*-1-0 Mi.
Tmax — Wt

At tfie mi ddle
of longer side
Wu —2h0b,fij , M*
r (a- 2V o « i T' - T J T
At the middle
of shorter side
Mf
X‘ &n

4 - > 0.5
* ( s - 4 - ‘) d
■ x M,
‘max —
8 ( » - 4 ('°-7)

At the base of
the tfroovc
m •p _ Mi
‘max —

a and fl from tfie tabic depending upon the ratio d;D.

d/D 0.0 0.05 0.10 0.20 0.40 0.60 0.80 1.0
a 1.57 0.80 0.81 0.82 0.76 O.fifi 0.52 0.38
P 1.57 1.51* 1.56 1.40 1.22 0.02 0.63 0.38
Ch. 9\ Torsion. Strength and Rigidity of Bars 193

According to Table 9, for narrow rectangular sections

coefficients a and p^3.123, and a t and {Jj are approximately equal
to -3- (from 0.312 toO.333). In accordance with formula (9.37 ) for such
rectangular sections we obtain
J , = \hV>, r , = ±Ai>> (9.38)

Table 10 contains formulas for the geometrical characteristics of

more complicated profiles and maximum shearing stresses in torsion.
If we are studying torsion in a rod of complex profile which may be
divided into a number of elements, then for such a section
A s=Ai + A i + " ,5S2 '^ In
where n = l , 2, 3, . . . are the numbers of the elementary parts into
which the section is divided.
As the angle of twist is the same for the complete section and all
its parts, we have
(fl _ Mil Mi,t __Mt„l
* GJt GJtl GJln
the torque is distributed over different portions of the section in direct
proportion to their rigidity:
= = ............

Correspondingly the maximum shearing stress in each of the n

portions of the section is
t A ft ( Jtn\
,n T^TW/J /ilw'/J
The maximum value of x occurs in the element for which is
maximum. Hence "

In addition to Table 10, we give here the formulas for J t and x

for sections composed of narrow and long rectangles, for example,
L-, T«, I- and IJ-shaped sections.
For such sections we may take
(9.38')
194 Shear and Torsion IP a r t IN

where 6 is the shorter and h the longer side of the rectangles into which
the section may be divided.
Coefficient t| depends upon the shape of the section and has the
following values:
for an L-section r| = 1.00
for an 1-section i] — 1.20
for a T-section r ) = l .l 5
for a U-section r| = 1.12
Angle <p is expressed as before by the formula
<P GJt

The maximum shearing stress is expected to occur in the broadest

of the rectangles into which the given section has been divided. It
may be calculated by the following formula:
_MtSmax (9.39)
max —

where 6 nuix is the maximum thickness among all the portions.

We may use the formulas for round sections in the analysis on tor­
sion of pipes with a non-circular section and small thickness of the
wall. According to formula (9.16) the section modulus of a thin-
walled ring is
W p ='2nr\t=- 2 A
where A 0 is the area of the circle bounded by the midline of the ring,
and t its thickness. Assuming that the shearing stresses are distributed
uniformly over the ring section, we get

This formula may be employed for the analysis of thin-walled rods

of non-circular closed sections.
The angle of twist may be determined by the formula
re _ Milr° _ xl
' GJ pTb (/To

Multiplying and dividing by 2 rcr0= S , we get

<9+1>
where S is the length of the centre line cf the pipe section, and A 0
is the area bounded by the midline of the given closed section.
 PART IV
Bending. Strength of Beams

C H A P T E R 10

Internal Forces in Bending.

Shearing Force and Bending-moment Diagrams

§ 55. Fundamental Concepts of Deformation in Bending.

Construction of Beam Supports
A prismatic bar with a straight axis bends if it is acted upon by
forces perpendicular to its axis and lying in a plane passing through
the axis.
A bar working under bending is usually called a beam. It has been
shown experimentally that under the action of such forces the beam’s
axis takes the form of a curve, and the beam bends. Figure 123 shows

a system of forces bending a rectangular beam; the forces act in the

plane of symmetry of the beam. If the plane of action of the forces
differs from the plane of symmetry then in addition to bending the
beam is subjected to torsion.
Beams are the most commonly used element of structures and ma­
chines; they withstand the pressure of other elements of the structure
(for example, forces P „ P a, and P„ in Fig. 123) and transfer it to the
parts supporting them (for example, forces PA and P 6 in Fig. 123).
Thus, the beam experiences the forces applied to it and the reac­
tions of the supports. Both kinds of the forces must be known to
enable us to solve the problem on strength of beams under bending.
IOC Bending. Strength of Beams \ Bart IV

The external applied forces may be calculated if we know the parts

of the structure supported by the beam. These forces are classified
as concentrated P (tf, kgf, N), force couples with moment M (tf* m,
kgf-m, N*m), and loads uniformly and non-uniformly distributed
over the length of the beam.
The uniformly distributed loads are measured by their intensity qf
i.e. load per unit length of the beam and are expressed in tf/m, kgf/m
or N/m.
The intensity of non-uniformly distributed loads varies along the
length of the beam and is denoted by q (x). In this case q (x) is the load

(a) safe

Fig. 124

per unit length of the beam at the given point *. In other words*
q(x) is equal to the limit of the ratio of load acting over a length of
dx near the particular point to the length dx.
A few examples of beams are given in Fig. 124(a), (b) and (c). The
first one is a joist loaded by a uniformly distributed force^ = 2 0 0 kgf/m;
the second beam is a dike support loaded by a triangular force (waler
pressure) of intensity qix) varying from 0 to <70= 1200 kgf/m; the third
one is the main beam of a bridge, which takes the forces exerted by
the engine wheels.
The wagon axle is a beam supported by the wheels and subjected
to the pressure of the axle box; beams in the aeroplane wings are bent
due to air pressure.
For the time being we shall study only the beams which satisfy
the following two limitations;
( 1) the beam section must have at least one axis of symmetry
(Fig. 125);
(2 ) all external forces must lie in the plane of symmetry of the beam.
The reactions of the supports, which balance the external forces
applied to the beam, must also, obviously, lie in the same plane.

* Notation ?(*) shows that the intensity of load in this case is a function of x.
Ch. 10] Internal Forces. Q- and M-diagrams 197

To determine the reactions of the beam supports we must study

their construction. The supports generally belong to one of the follow­
ing three types:
(a) fixed hinged support;
(b) movable hinged support;
fc) rigidly fixed support.
The fixed hinged support is schematically shown by point A in
Fig. 126. It allows the supported section of the beam to revolve freely

Fig. 125

round a hinge mounted at the centre of gravity A of the supported

section, but does not permit linear displacement of this end of the
beam. The resistance of such a support is expressed by the reaction
which is transmitted from the support to the beam end through the
hinge and which lies in the plane of acting forces.

F»g- 126 Fig. 127

We know only the point of application of the reaction—the hinge—

as it is the only point at which the beam and the support come into
contact, but we know neither the magnitude of reaction nor its di­
rection. Therefore, we must always replace the support by two compo­
nents: Ha , parallel to the beam axis, and A, perpendicular to it.
From this reasoning a fixed hinged support gives two reactions
(A and HA) of unknown magnitude.
A movable hinged support permits, besides rotation, free displace­
ment in ihe relevant direction (Fig. 126, point B). Hence, this support
only hinders displacement perpendicular to the particular direction.
Therefore, the reaction of such a support passes through the centre
of the hinge and is directed at right angles to the line of free displace­
ment of the support (usually the beam axis). Thus a hinged movable
support gives only one unknown reaction B.
198 Bending. Strength of Beams [Part IV

Finally, a rigidly fixed support hinders all types of displacements

of the beam end in the plane of action of the forces. It may be obtained
from a fixed hinged support by removing the hinge (Fig. 127).
By removing the hinge we prevent rotation of the beam end, i.e. we
introduce a new reaction that prevents such a rotation. This reaction
is created by a force couple. Therefore a rigidly fixed end of the beam
gives three unknown reactions: component HA parallel to the beam
axis, component A perpendicular to it, and the bearing moment M A.
iA <C

m
8

0
Ha I
m
Fig 128

A beam may rest on a number of supports cf the types explained

above. Figure 127, for example, shows a beam with a rigidly fixed
end; in Fig. 126 the beam is supported at one end by a fixed hinged
support and at the other end, by a movable hinged support;
in Fig. 128 (a) the same beam is supported additionally at the centre
by a movable hinged support; the beam in Fig. 128(b) is rigidly
fixed at one end and supported by a movable hinged support at one
of the intermediate sections.
In all these figures we have depicted the reactioas of the supports
of a particular construction, which may arise under the action of
external forces; the forces have not been shown in the figures.
To determine the unknown reactions we shall first use the static
equations expressing the condition that under the action of the forces
applied to it and the reactions the beam as a whole remains in equi­
librium. As all the forces lie in a single plane, we may write down
three static equations. Thus, the problem of determining the reactions
from conditions of statics is determinate if the number of unknown
reactions is not more than three.
Hence, the beams with the construction of supports that gives three
reactions (Figs. 126 and 127) are statically determinate. Multiple-
support beams with intermediate hinges also belong to the group
of statically determinate beams. These beams may be classified into
the basic statically determinate beams (A -f and 2-3) and the suspended
Ch. 10| Internal Forces. Q- and M ‘diagrams 190

statically determinate (1-2 and 3-D), which are supported by the former
through hinges (Fig. 129).
All other beams belong to the group of statically Indeterminate;
they will be analyzed later in special chapters.
The construction of the supports is, in fact, sometimes very much
different from the construction shown in Figs. 126 and 127. Therefore,
before we start analyzing a beam we must first study the design of

1 I
m i

!!!!!!!
3
Fig. 129

its supports and establish to which group of supports shown in Figs. 126
and 127 do they belong.
As the deformation of the beams is usually very small and stresses
are within the elastic limit, we must find out whether the support
permits even a small rotation or displacement. If this is so, it is suf-

Fig. 130

ficient to consider the support hinged or movable. If the end of a me­

tallic or wooden beam is fixed in a brick wall to a small depth, then
a little rotation of this end is quite possible, and therefore the end
should be considered as hinged.
Thus before determining the support reactions, we must represent
the supports by schematic diagram replacing the actual construction
by an equivalent approximate drawing. Thus, for Instance, the wagon
200 Bending. Strength of Beams [Part IV

axle (Fig. 130fo)) that experiences pressure P of the wagon body and
transmits it to the rails may be looked upon as a beam loaded by for­
ces P at points A and B and resting on hinged supports C and D of
which one may be considered movable (Fig. 130(b)). This schematic
drawing describes the actual working of the wagon axle with some
approximation because the supporting sections may rotate under
bending load and the distance between points C and D may also
slightly change.
We shall employ the three equations of equilibrium to determine
the support reactions in statically determinate beams. The axis of
the beam is assumed as the x-axis and the centre of one of the hinges
as the centre of coordinates; the y-axis is directed vertically upwards
(it is assumed that the beam is horizontal).
To determine the horizontal component of the reaction we equate
to zero the sum of projections of all the forces on the .v-axis. The ver­
tical components of the reactions and the support moment are deter­
mined by equating to zero the sum of moments of all forces about
any two points of the beam, usually about the centres of gravity of
the supported sections of the beam. The sum of projections of all
the forces on the y-axis should be equaled to zero to check the correct­
ness of calculations; this condition must become an identity when
the values already obtained are substituted in it.
In beams with intermediate hinges, we first study the equilibrium
of the suspended beams as beams on two supports and find their reac­
tions. These reactions must balance the forces transmitted from the
suspended beams to the base beam through the hinges. Knowing
the forces, we can determine the reactions of the base beam (see § 59).

§ 56. Nature of Stresses in a Beam.

Bending Moment and Shearing Force
Selection of the design scheme and determination of the support
reactions completes the first part of the problem of beam analysis—
determination of the external forces acting on the beam.
We can now proceed with finding the stresses in beam sections;
this will be the next step in solving the problems on bending. For
discussion, let us consider a hinged beam (Fig. 131) which is loaded
by forces Pu Pi, and P0. For the given system of forces the horizontal
reaction HA is zero, the reactions A and B are determined from the
equations of moments; thus all the external forces can be determined.
Before determining the stresses we must find the critical section
of the beam through which the maximum stresses are transferred.
This can be achieved by deriving formulas which enable us to deter­
mine the stresses in any section (for example, inclined section 3-3).
Once we have derived these formulas, we shall be able to determine
the critical section as well as the maximum stresses.
Ch. W\ Internal Forces. Q- and M-diagrams 201

Let us start by determining stresses in a plane perpendicular to

the axis of the beam, then in a plane parallel to the axis, and finally
in any plane. Let us take a section /-/ perpendicular to the axis of
the beam with its centre of gravity 0 t at a distance x from the left

support. To determine the stresses in this section we remove one por­

tion of the beam and replace its action on the remaining portion by
the unknown stress. For convenience of calculation, the equilibrium
of that portion of the beam should be considered to which less number
of forces are applied; in the example under consideration, the left
portion. This portion must maintain
equilibrium under the action of
external and internal forces acting
on it.
The only external force—force
A acting upwards—is applied to
the left of section 1-1 (the weight of
the beam is neglected). This force
can be balanced only by the inter­
nal force Q=A (or Q = F a+ JV —B=
=A) which is transmitted from the
right cut-out portion of the beam (*)
and acts vertically downwards x _ f ** cr
along the tangent to the section A
(Fig. 132(a)). Since forces A and Q
lying in the vertical plane form Fig. 132
a* couple with moment M = A x in
the clockwise direction, the section
must experience internal forces which also result in a moment of
the same magnitude M = A x acting in the anticlockwise direction.
Only normal stresses acting in the section are capable of creating this
moment which retains the left portion of the beam in equilibrium.
Hence, the internal forces in section /■/ that replace the action of
the removed right portion of the beam on the left are: force Q—A
parallel to the external forces and made up of shearing stresses acting
in the beam cross section; a force couple of moment M = A x that acts
202 Bending, Strength of Beams [Part IV

in the plane of action of external forces and is made up of normal

stresses.
This means that the section of the beam under consideration ex­
periences shearing as well as normal stresses (Fig. 132(6)) that add up
into internal force factors Q and M which together balance the system
of external forces acting on the portion of beam being considered. It
goes without saying that force factors Q and M of the same magnitude
but acting in the opposite direction are transmitted across section
/-/ from the left portion of the beam to the right, and they balance
the external forces applied to the right portion.
2 2

To determine the stresses acting in various sections of the beam,

wo must learn to determine the magnitudes and directions of internal
forces acting in an arbitrary section of the beam by expressing them
through external forces. Let us consider, for example, an arbitrary
section 2-2 (Fig. 131) and find the internal forces transmitted from
the left portion of the beam to the right. In order to do this we
remove the left portion and transfer the forces acting on it to the right
portion—to the centre of gravity of section (point 0 *). In the proc­
ess of transfer, the forces acting in a plane are reduced to a resultant
force acting at the centre of forces and a force couple. Hence, the forces
transferred from the left portion to the right must be applied at point
0 2 in the form of force factors (Fig. 133): force
Q = A — Pl (10.1)
force couple with moment
M = Ax — P ,(x —a) (10.2)
Assuming that A > P Jt we direct force Q upwards and moment M—
clockwise. Identical internal force factors Q and M acting in the oppo­
site direction are transferred from the right portion of the beam to
the left (Fig. 133).
It is clear from the above discussion that in any cross section of
the beam the internal forces can be reduced to force Q and force couple
Ch. 10] Internal Forces. Q* and M'diagrams 203

of moment M , which together replace the action of one cutoff portion

of the beam on the other.
Force Q, the resultant of elementary shearing forces acting in the
beam section, is known as the lateral or shearing force. This force has
the tendency to shear the section under consideration with respect
to an adjoining section (Figs. 133 and 134). It is evident from equation
( 10 . 1) that the shearing force in each cross section is calculated as
the sum of projections on the normal to the beam axis of all external
forces acting to the right or left of the section. When all the forces
acting on the beam are perpendicular to its axis, the shearing force
Q>0

Fig. 134 F‘g- 135

may be calculated as the algebraic sum of forces acting on the portion

of the beam the equilibrium of which is being considered.
The moment of internal force couple made up of elementary normal
stresses acting in the beam's cross section is known as bending moment.
The bending moment tends to rotate the section under consideration
with respect to an adjacent section, which leads to deformation of
the beam axis, i.e. bending (Fig. 135).
It is evident from Eq. (10.2) that the bending moment in an arbit­
rary section of the beam is equal to the algebraic sum of moments of
all external forces acting to one side of the section about central axis
y that is normal to the beam axis.
Let us establish the sign convention for Q and M. As is shown in
Fig. 133, the internal force factors Q and M act in opposite directions
depending on whether the section under consideration belongs to the
left portion or the right. This circumstance should be taken into ac­
count when dealing with the sign convention in order to get identical
values of Q and M not only in magnitude but also with the same sign
irrespective of whether we consider the forces acting on the left cut­
out portion or the right one.
 204 Bending. Strength of Beams [Part IV

In accordance with the above (for a horizontal beam) we shall con­

sider shearing force Q positive if the external forces to the left of the
section tinder consideration act upwards or the forces to the right of
the section act downwards. In other words, Q>0 if the resultant of
external forces acting to the leH of the section is directed upwards;
for forces acting to the right of the section Q> 0 if their resultant is
directed downwards. According to this convention the direction of
Q coincides with the direction of shearing stresses t which constitute
the shearing force (Fig. 134).
The bending moment will be considered positive if the algebraic
sum of moments of forces applied to the left of the section gives a re­
sulting moment acting in the clockwise direction; or if forces applied
to the right of the section give an anticlockwise resulting moment
(Fig. 135). Hence, for the left cutoff portion the bending moment due
to each individual force is considered positive if the moment of this
force w.r.t. the centre of gravity of the section is clockwise; on the other
hand, A4<0 if the force gives an anticlockwise moment w.r.t. the
centre of gravity of the section. If the right cutoff portion of the beam
is considered, the convention is just the reverse.*
The accepted sign convention for M is related to the nature of de­
formation of the beam: if the bending moment is positive, the beam
bends with its convex surface down, if the bending moment is nega­
tive, with its convex surface up (Fig. 135). In the section where M.
passes through zero the beam axis has an inflection point; the beam
axis remains straight in the segments where A4 r=0 .
We have seen that the expressions for shearing force and bending
moment are different in different sections of the beam (/-/ and 2-2).
By the very definition of internal force factors it is obvious that the
shearing stresses are maximum in the section where Q~Qv a whereas
the normal stresses are maximum in the section where M = M max.
Therefore, for checking the strength of beams we must find those
sections in which shearing force and bending moment are maximum.
The search for these critical sections is greatly facilitated by plotting
of bending-moment and shearing-force diagrams, i.e. diagrams that
show how bending moment M and shearing force Q vary in different
sections of the beam when they are plotted as a function of x.
Thus, the shearing force Q ( a * ) and bending moment M (x) are func­
tions of x. In future for brevity's sake we shall denote these quantities
by Q and M, and use the notation (x) only when we want to empha­
size that Q and Af are variable quantities which depend upon x.
While plotting the diagrams, Ihe ordinates which, to a certain scale,
represent the value of the bending moment or the shearing force, are

Some writers rotate the signs for Q and At with the direction of the coordinate
axes, which in some cases (for example, compound bending of bars with broken
axis) simplifies the sign convention.
Ch. 101 Internal Forces. Q- and M-diagrams 205

laid off under the given section from the x-axis parallel to the axis
of the beam. Positive ordinates of the Q- and Af-diagrams will be laid
off upwards and negative downwards. Some books recommend plott­
ing of the bending-moment diagram on the convex side of the bent
beam, the positive ordinates downwards and negative upwards.
However, this is merely a matter of liking, which is not significant.
It may become easier to plot these diagrams if we are able to estab­
lish some relation between the values of bending moment and shearing
force in an arbitrary section and also the relation of Q and M w'ith
Ihe forces acting on the beam.
In the next section we shall explain how to correlate the external
forces, the shearing force and the bending moment.
§ 57. Differential Relation Between the Intensity
of a Continuous Load, Shearing Force and
Bending Moment
ll was shown in §56 that for equilibrium of the cut-out portion of
a beam it is essential to apply in the section force factors Q and M
that replace the action of the removed portion on the portion under
consideration. Hence, if we cut from the beam (Fig. I3o) an element
of infinitely small length dx, it
must remain in equilibrium un­
der the continuous load of inten­
sity if (which may be considered
constant over the length dx), the
forces Q and Qt and moments M
and Mi, which represent the ac­
tion on the element of the left and
right cutoff * portions, respective­
ly. Let us note that Q,=QH dQ
and M i—M+dM because the in­
crement of these quantities in
transition from section mn to an
infinitely close section m»rti is
also an infinitely small quantity.
The conditions of equilibrium
of Ihe isolated element may be
written as
£ F = 0 , Q + q d x -iQ + d Q ) = 0
2 M , =*0, M + Q d x + g d x d{ ~ ( M + dM) = 0
From the first equation we get
qdx—dQ = 0
• No concentrated force or moment acts over the element dx.
206 Bending. Strength of Beams [Part IV

wherefrom
§ -1 CO-3)

i.e. the derivative of the shearing force w.r.t. the abscissa of the sec­
tion is equal to the intensity of the continuous load in the same sec­
tion.
From the second equation, neglecting the infinitesimals of the
second order, we get
Q d x -d M = 0 or ^ = Q (10.4)

i.e. the derivative of the bending moment w.r.t. the abscissa of the
section is equal to the shearing force in the same section.
Differentiating both sides of Eq. (10.4), we get
d*M dQ d*M (10.5)
dx* = dx dx2 ~ q
I.e. the second derivative of the bending moment w.r.t. the abscissa
is equal to the intensity of the continuous load. If q is directed down­
wards, equations (10.5) become
d*M dQ
and - f c ~ - q

By integrating formulas (10.3) and (10.4), we get

X

Q(*) = $ q{x)dx + Q (0) (10.30

0
X

'M(x) = S) Q (x)dx+ M (0) (10.4')

o
The arbitrary constants Q(0) and M (0) are concentrated force and
moment (if they exist) in the beginning of the segment. These formu­
las are convenient to use while plotting diagrams for non-uniform
loading q=q(x). In the geometrical sense each integral represents
X X

area: ^ q(x) dx=(oq is the load area (see §59) and J Q(*)dx=a>Q
O 0
is the area of Q-diagram over length x. Formulas (10.3') and (10.4')
may be written in the form
Q W - » f + Q (0 ) (10.30
jM(x)=( 0q + M(0) (10.40
The relations obtained above may be usedin plotting the diagrams
for Q and M , especially if we consider that the derivative of a function
Ch. 10] Internal Forces. Q- and M-diagrams 207

geometrically represents the slope of the tangent to the curve at the

given point. In other words, the shearing force in a section may be
regarded as the slope of the tangent to the bending-moment diagram
at the point corresponding to the given section. Therefore, it should
be borne in mind that if the x-axis is directed from right to left, then
~ = —Q, because the slope of the tangent to the curve changes its
sign if the direction of the axis is reversed.
It follows from Eq. (10.3) that in the section where intensity of
the load q=Q, the shearing force Q=Qmax or Q=Qmin» because if
q = ^ = 0 then the tangent to the shearing force diagram must be
parallel to the x-axis. By the same reasoning we come to another
more important conclusion from Eq. (10.4): the bending moment is
maximum (or minimum) in the section where = 0 , i.e. where
the shearing force passes through zero.
Although Eq. (10.4) enables us to get Q as the first derivative of Af,
it should be determined independently when plotting the shearing
force diagram, and Eq. (10.4) should be employed only for checking
its value. Similarly, for checking whether we are plotting the bending-
moment diagram correctly, we can use formula (10.4') according to
which the ordinate of Af-diagram in an arbitrary section is equal lo
the area of Q-diagram to one side of the section or differs from it by
a value equal to the concentrated bending moment Af (0), if the latter
enters in the expression for Af (x). Likewise Eq. (10.5) may also be
used to check the correctness of plotting the Af-diagram, because the
direction of convexity of the bending-moment diagram is determined
by the sign of the second derivative of Af. Instructions on checking
the correctness of plotting the shearing-force and bending-moment
diagrams will be given below (§ 60).

§ 58. Plotting Bending-moment and Shearing-force

Diagrams *
Example, Plot the bending-moment and shearing-force diagrams
for a simply supported beam loaded wiih force P (Fig. 137).
To calculate Af and Q in any section of the beam, it is first of alt
necessary to find the reactions. The assumed directions of the reac­
tions A, H ai, and B is shown in Fig. 137.
By equating to zero the sum of the projections of all forces on the
axis of the beam we get
^ =0

* A number of examples on plotting Q- and M-diagrams are given in problem

books. Sec N. M. Belyaev, Problems in Strength of Materials, Pergamon Press, 1966,
208 Bending. Strength of Beams [Part IV

This result could have been predicted beforehand, because all the
forces acting on the beam are perpendicular to its axis.
By taking the sum of the moments of all the forces about point Bf
we get
2 M b=0, + A t— P b= 0
or

Similarly
2 . ^ = 0, ~~Bl + P a= 0
or
b = + £
To check the correctness of the results obtained, we take the sum
of the projections of all forces on the vertical y-axis:
A — P + B = 0 or A + B = P
Substituting the values of the reactions found above, we get
Pb L Pa P(a±b) n
I ’ I ~~ I
which is in accordance with the condition of equilibrium. Such a
check is always desirable, because an error in determining the reac­
tions will inevitably lead to errors
in plotting the bending-moment
and shearing-force diagrams.
The expressions giving the val­
ues of shearing force and bending
moment in any section may be ob­
tained by taking an arbitrary sec­
tion 1-1 between A and C at a dis­
tance from A. Let us take note
that the expression “taking a sec­
tion” includes not only marking of
the section on the drawing but also
giving its distance from tne select­
ed origin of coordinates. The cen­
tre of gravity of the section is de­
noted by Oi.
It is more convenient to consider
the left cutoff portion to determine
the shearing force Q in the section, because the left portion is acted
upon by a less number of forces (only force A). Considering the por­
tion of the beam to the left of section Ox and projecting the forces
acting on it on a plane perpendicular to Its axis, we get the expression
Ch. 10] Internal Forces. Q- and M-diagrams 209

for shearing force <2t in the section at a distance x± from support A:

= + (10.6)
The shearing force in a section having abscissa x, does not depend
upon this distance. Thus, as long as xt varies from 0 to a, the shearing
force remains constant, and its diagram in this portion is represented
by a straight line FsDt parallel to the axis of abscissa A 2B3 {"Fig. 137).
Expression (10.6) for Q, holds good as long as the section does not
go beyond point C, i.e. till If XjX i , the left portion of the
beam will experience two forces A and P\ consequently, the sum of
the projections of forces acting on the left cutoff portion will change.
To find the shearing force in the second portion, we shall have to
take another section between points B and C with centre of gravity
at Oj. Its distance x2 will be measured from the right support B.
It will be convenient for us in this case to consider the equilibrium
of the right portion of the beam as it is acted upon by only one force B.
Considering the right cutoff portion of the beam, we get the expres­
sion for shearing force in section 2-2:

- T (10.7)

The minus sign shows that force B acting on the right cutoff portion
is directed upwards.
It is obvious that if we had considered the left cutoff portion, we
would have obtained the same expression for Qt:
Q3= A — P = — B (since A-\-B — P)
Expression (10.7) is valid for any value of x2not exceeding the lim­
its of the portion BCy i.e. for O < X i0 ; this expression also shows
that Qs does not depend upon x2.
The shearing-force diagram over the length of the second portion
is a straight line EgG* parallel to the x-axis. I t has a discontinuity—
a jump at the point of application of force Px. At this point the shear­
ing force passes through zero and is not equal to D tE ^ P . In a
section immediately to the left of point C
Q= + ™

in a section to the right of point C

n—
Q ----- tPa

Let us note that the absolute value of the jump is equal to the con­
centrated force P acting in this section.
8 —3310
210 Bending. Strength of Beams [Part IV

Such a shape of the shearing-force diagram {Fig. 137) is possible

only if we consider the concentrated force P acting at a single point C.
Actually pressure P is transferred to the beam through a very small
area (Fig. 138). Therefore, in this area the shearing force changes
gradually from to — ^ . p a s ­
sing through zero in the process.
The maximum absolute value of
the shearing force in this example
will be {if a>b)

I Qm« I = T
All sections of portion CB of the
beam are prone to failure due to
shearing stresses.
In plotting the bending-moment
diagram we shall use the same
sections l- i (with the origin of
coordinates at point A) for the left
portion of the beam and 2-2 (with
the origin of coordinates at point B) for the right portion of the beam.
Considering the left portion, we determine the moment in section
/- / as the sum of the moments of forces acting on it about the centre
of gravity of the section Qt:

M ^ A x ^ j-X i ( 10.8)

Afi is a linear function in x ,. Therefore, if we move the section, i.e.

change x t, then Mi varies linearly. Expression ( 10.8) for Af, holds
good as long as the section does not go beyond point C, i.e. till 0 <
As soon as x, becomes greater than a, the left portion of the beam
starts experiencing two forces: A and P , and formula (10.8) no more
holds good. As the diagram is a straight line, it is sufficient to give
two values to *, to obtain the two points required for plotting the line.
At x,= 0, we get Afi=0—this is the ordinate under section A. Simi-
larly at X i= a we get Mt=-\-^-j- ; this is the ordinate under section C.
Laying off upwards from the x-axis (positive moment) the segment
CiDu which expresses to a certain scale the ordinate and join­
ing points Di and A t by a straight line, we get the first portion of the
bending-moment diagram. To plot the diagram for the second p6r-
tion, we write down the expression for the moment about point Ot
Ch. 10] Jitternot Forces, Q- and M-diagrams 211

of forces acting on the right cutoff portion of the beam:

M a= B x2^ ^ Xi (10.9)

In this portion also the moment is positive, because we consider

the right portion, and force B rotates it about point Ot in the anti­
clockwise direction. Expression (10.9) represents the equation of a
straight line and holds good for 0<x*<&. At xz—bt Afa* = + ^ p and
at xa—0, Afa= 0.
Thus, the second portion of the bending-moment diagram is repre­
sented by the straight lineDjBj. The bending moment is positive over
the whole length of the beam and is maximum in section C, the point
of application of force P, where it
is equal to

Mmax = -7" (1CU°)

The maximum normal stresses will
act in this section. £
At a<=b^ (the force acts in 2
the middle of the beam) we get

<w«.x= - r (1010')
In any cross section of the beam
taken between the end points'
A-C and C-B the values of Q and Fig. 139
Af are graphically represented by
the ordinates of the corresponding diagrams shown in Fig. 137 by
vertical hatching.
We shall study a few more examples on plotting bending-moment
and shearing-force diagrams for beams subjected to various types of
loading.
Let us plot the Af- and Q-diagrams for the beam shown in Fig. 139,
loaded by a continuous uniformly distributed force of intensity q
(expressed in kgf/m, tf/m, N/m, etc.).
Tt is essential to determine the support reactions before we start
solving the problem.
The reaction HA is zero, reactions A and B are equal from symmetry;
each of them is equal to half of the total load on the beam:
212 Bending. Strength of Beams [Part IV

Let us take a section 0 at a distance x from the left end of the beam.
We shall consider the equilibrium of the left-hand portion to deter­
mine Q and M. It is acted upon by reaction A and load q uniformly
distributed over length x.
We must take the sum of all the forces acting on the left cutoff
portion to determine the shearing force in section 0. To the left of
the section is force A — ^ directed upwards, and the resultant of
the uniformly distributed load over the length x , equal to qx and
directed downwards. Therefore,
Q = A — q x ^ ^ - — qx
The shearing force varies with x linearly, and the line may be plotted
by taking two values of the variable x : at *= 0, Q= y and at x= l,
Q =— y • The shearing-force diagram is shown in Fig. 139; Qmax= y .
To plot the bending-moment diagram we take the sum of the mo­
ments of the same forces acting on the portion of beam under conside­
ration about point 0. Keeping in mind that resultant qx acts in the
middle of the segment of length x, with an arm of lengthy about point
0, we get
M -+ A x-qx$-$x-
This equation of moments is valid for determining the bending mo­
ment in any section of the beam.
In this case the bending moment depends upon the square of ab­
scissa x; therefore, the diagram is of the shape of a square parabola.
To plot the curve we need at least three or four points lying on it;
at x = 0 M —0

a t*=T
at * “ T m = + t (* —t ) = + T
at x = I M =Q
The bending-moment diagram is of the shape shown in Fig. 139.
To determine M we find abscissa x» of the corresponding section
by equating the first derivative of M w.r.t. x to zero:
dM ql 2qxa n
dx 2 2
wherefrom
*. = |- and +^ (10.11)
Ch. 10J Internal Forces. Q- and At-rftegrams 213

The maximum bending moment occurs at the middle of the span,

i.e. in the section where Q—0. This is a check of the relation between
M (x) and <?{*) established above (in § 57).
Let us consider one- more example—beam AB rigidly fixed at one
end (Fig. 140). Such a beam is usually known as cantilever. Since the
right end of the beam is free, it is essential to consider the forces act­
ing to the right of the section while plotting the bending-moment
and shearing-force diagrams. In this
case it is not necessary to deter­ *| (
mine the support reactions of the 4 If B
rigid constraint (at the left end of C H
the beam). Force P divides the beam
into two parts: AC of length a
- a --- »
and CB of length b. i\1
If we consider section l-l at a dis­ w f Ill]^ __ !It
tance Xi from the free end, we ob­
serve that there are no external for­
£
ces to the right of the section and W2* *"T*IT3
therefore it is free of internal for­ 18^ Tmllyiii!!1■
ces. For all values of Xi between 0 C
and xt=b B J
QL= 0 and Mt = 0 <»> Fig. 140
The distance to s e c tio n s in por­
tion AC will be laid off from the
point of application of force P. To the right of the section we will
have force P acting upwards and a uniformly distributed force of in­
tensity q acting downwards over a length x 2 and having a resultant
gxt. The internal forces in section 2-2 will be
0.2 — — P + qx2 <b>
= Px2 <4 (c)
As the abscissa is varied from xs=0 to x2=a, shearing force Qa
changes according to linear law and bending moment Afa according
to a parabolic law with a maximum in the section where 0 * = * ^ = 0.
i.e. in the section where as is evident from equation (b).
Let us now plot the Q- and M-diagrams. In order to calculate the
ordinates in a general form we shall assume a particular ratio between
P and q (this can always be done when the quantities are known
numerically). Suppose, for instance, that P— g
It is evident from equation (a) that in the first portion of length b
the ordinates of both Q- and A4*diagrams are equal to aero, and the
214 Bending. Strength of Beams [Part IV

diagrams coincide with the x-axes. In the second portion of length

AC=a (0<*a<a) from equations (b) and (c) we have
at xa= 0 , = _pr ~—~ 3qa ,
nVa—
. P a 0 —0 M -W a r qa*
at x8« ^ ’ 3 » 3 3 2x9 13
qa*
at x%*=a. Qa = — P + q a = jq a , “X
The Q- and Af-diagrams are depicted in Fig. 140. It is clear from
the diagrams that the absolute maximum values of Q and M occur
at the fixed end and are

§ 59. Plotting Bending-moment and Shearing-force

Diagrams for More Complicated Loads
Having studied the characteristics of bending-moment and shearing-
force diagrams and the general method of plotting them, we can pass
over to solving more complicated problems.
Let us see how to determine Q and M when the beam is acted upon
by a continuous non-uniformly distributed load whose intensity
changes along the beam length with x (Fig. 141). In other words, q
is a function of x or q—q(x). The bending moment and shearing force
will also be some functions of x:
M = M{x) and Q = Q (x)
The curve adceb representing the variation of q(x) is called the load
curve, and the area bounded by this curve is called the toad area.
Let us calculate Q and M in an arbitrary section at a distance xt
from the free end. Considering the shearing force as the sum of ele­
mentary forces q(x) dx acting on the left cutoff portion of the beam,
and replacing the summation by integration, we find
A, X,

Q (x j = — ^q (x) dx *s — £ d<a= — ©(xj (10.12)

0 0
Here ©(xi) represents the part of load area located to the left of sec­
tion c-C. Thus, the shearing force Q (x,) equal to resultant R* of the
continuous load over the length AC=xt may be calculated as the load
area ©(x,), lying to one side of the section.
The bending moment in the same section is equal to the sum of
moments of elementary forces q (x) dx, acting on the cutoff portion
of the beam, about point C, and may be calculated as the moment
Ch. 10] Internal Forces. Q- and M ‘diagrams 215

of resultant R Qt i.e.
M{ x l) = — R<lxr = — <o(x1)x r (10.13)
In other words, the bending moment of a continuous non-uniformly
distributed load is equal to the product of the load area lying to one
side of the section and the distance of the centre of gravity of this
area from the section under consideration (arm of the resultant).
Let us study how to plot the bending-moment and shearing-force
diagrams for a beam which is acted upon by a distributed force that
varies along its length as shown in Fig. 142.

Fig. 141 Fig. 142

Loads of this kind are applied to beams that support water and
earth pressure, for example, dam supports and columns for strengthen­
ing walls of water storage reservoirs. Connecting rods of steam ar.d
internal combustion engines are subjected to similar loading by forces
of inertia.
The load is characterized by the ordinate q0, the maximum intensity
of the load (in kgf/m). The reaction HA= 0; we have to determine
A and B.
For determining A we write down the'equation of moments about
point B. The moment of the load is equal to the moment of its resul­
tant, i.e. moment of the load area ^m u ltip lied by the distance of
its centre of gravity from point B. The resultant is shown in Fig. 142
by the dotted line; this emphasizes that the concentrated force equal
to the load area (3>—-jq d does not actually act on the beam, but we
216 Bending. Strength of Beams [Part IV

make use of it while taking the moment of the total load for determin­
ing the- support reactions.
The equations of moments may be written as
2 ^ = 0, A l— Jtol = 0, =^

2 ^ 1 = 0. — Bl + t4 = 0,

Thus, support A takes two-thirds of the total load <o= whereas

support B takes only one-third.
To plot the diagrams let us take a section at a distance x from the
right end of the beam. The ordinate q{x) of the load in this section is
determined from similarity of triangles:
Q M x x
~ T • 9 (*) = <7<>t
While determining Q and M >ve shall consider the right-hand por­
tion because it is acted upon by the concentrated force and triangular
load, whereas the left-hand portion is acted upon by the concentrated
force and trapezoidal load, which complicates the computations.
Shearing force Q will be the sum of the projections on the vertical
or reaction B and the hatched load (a {x )= ^ q (x )x = ^ q (iJj , i.e.

In this case the shearing-force diagram is represented by a quadratic

curve, and
at * = 0 ,
at x = l. <}=&=. + A

at x ,= = ~ 2 * «— ¥ 0 - f f 24
The shearing-force diagram is given in Fig. 142. It is clear from the
diagram that the maximum shearing force (in absolute value) occurs
in section A (at the support):
Q o u —+ = + ^

The shearing force passes through zero at x0 which may be deter­

mined by the following equation:

0.577/
Ch. JO) . Internal Forces. Q- and M-diagrams 217

We shall use this value of xa for determining the maximum value

of At. The shearing force achieves its analytical minimum in point Bt
where the intensity of the continuous load is zero. As is evident from
Eq. (10.3), the tangent to the shearing-force diagram in this section
is parallel to the *-axis.
Let us pass over to plotting the bending-moment diagram for which
we again consider the right cutoff portion of the beam. The moment
of the resultant of the hatched triangular load (Fig. 142) about point 0
is equal to its load area multiplied by the arm - j:

The bending moment in this section is

go*8
6/
This expression for M holds good for the whole length of the beam.
The bending-moment diagram is represented by a cubic curve. To
plot the cubical parabola we must calculate a few ordinates:
at x — 0, M= 0
at M
ai 'w=*2li
- T T Vf 1i - 1
at x = l, M =0
The bending moment is maximum in the section where (2=0, i.e.
at ; it is equal to

M m ax goP go/8 (10.14)

9/*3 15.58

The diagram is shown in Fig. 142. It is evident from formula (10.14)

that the maximum bending moment differs slightly from the moment
ai 2
at the middle of the span, which is equal to-j^-. In actual design
calculations for a beam loaded by a triangular force, the maximum
bending moment Afm#x may always be replaced by the moment in
the middle of the span equal to ; the error will not be more than
2 . 6 %.
Let us analyze the plotting of shearing-force and bending-moment
diagrams for a simply supported hinged beam (Fig. 143) loaded by
a continuous force whose intensity varies according to the following
parabolic law:
218 Bending. Strength of Beams [Part IV

Due to symmetry, the support reactions are

Here <o is the load area which is determined from condition ( 10. 12):

L.0 o J
J2 l r i i_i l l - 2 /, / (10.15)
l I 2 3 J 3

Hence the support reactions are

Let us now write down the expression for Q(x) and M (jc) In section
/•/ at a distance xt from the left-hand support A . Denoting the load
area of length xx by <o(x), we get
(a) The shearing force Q(jc)=
—A—to (jc>
d a The load area to the left of
the section is
x,
<a(x)=j<?(x)dx
________ I *

4 LJI J U 1 7

Fig. 143

Putting the values of A and <o(*) in the expression for Q(x), we get

- ^ < 3 * —2xt) (10.16)

(b) We shall calculate the bending moment with the help of for­
mula (10.4'):
Af(*) = $ Q( x ) dx +M( 0)

The constant of integration Af (0)=0 because there is no concentra­

ted moment at this end of the portion. Putting the expression for
Ch. 101 Internal Forces. Q- and M-diagrams 219
Q(at) under the integral sign, we obtain

MW= | (3/—2x)j dx

2<?oX% , q^xj
3 *l 31 3p“
OP

(10.17)

It is evident from Eqs. (10.16) and (10.17) that the shearing force
varies according to a cubical parabola, whereas the bending moment
varies according to a fourth-degree parabola.
We shall take a few values of the variable x to find points for plott­
ing these curves:
*=0, C (* )= ^ , M (* ) = 0

*■=4 , J W (* )= is 2 -
"24? 2 “ 5 5 ^ *
X = .l , « ( * ) - # - ■ f * . l ------- - Y - o

The corresponding shearing-force and bending-moment diagrams

are shown in Fig. 143. The maximum values of Q and M are respec­
tively equal to
Q »„=^ and (10.18)

We shall now discuss the order of plotting Q- and M-diagrams for

a two-span beam with an intermediate hinge; such beams are often
employed in bridge design. The dimensions of the beam and the forces
acting on it are shown in Fig. 144(a).
We must determine the reactions of the beam before plotting the
diagrams. It is clear from Fig. 144(a) that Uie arrangement may have
four support reactions: A, HA, B, and D. However, wecanwrite only
three equations of equilibrium for the whole beam. The fourth equa­
tion is determined from the condition that hinge C (on account of
its construction) cannot transmit bending moment, because it permits
relative rotation of one part of the beam (j4C) about the other (CD).
The last condition requires that the sum of moments of all forces
acting either to the left or right of the hinge about point C should
be zero. In other words, to maintain equilibrium the bending moment
229 Bending. Strength of Beams [Part IV

in the hinge must be zero. This additional requirement makes the

beam AD statically determinate.
First we shall determine HA. By equating to zero the sum of pro­
jections of all the forces on the beam’s axis we find that HA=0.

Fig. 144

Next we may write three equations for the moments as follows:

(1) by equating to zero the sum of moments of all the forces about
point A\
(2) by equating to zero the sum of moments of all the forces about
point B or D;
(3) by equating to zero the sum of moments of all the forces either
to the left or right of hinge C, about point C.
By solving these three equations we can determine all the three
unknown reactions A, B and D. However, the reactions can be deter­
mined more easily by breaking the beam arrangement AD into simple
beams. The suspended beam CD is supported by a hinge C at the end
Ch. W] Internal Forces. Q- and M-diagrams 221

of cantilever BC and by a movable hinge at point D. Therefore, we

may consider the whole beam (Fig. 144(6)) as a combination of two
beams. The suspended beam experiences reaction C through the hinge
at the end of the cantilever and in its turn presses this end with the
same force C.
By first analyzing the equilibrium of the suspended beam we find
its reactions D and C, then by taking into account the already known
force C acting at the end of the cantilever, we determine reactions
A and B. Jn this example
C - £ > = - j= :6 tf , 6 - |= 1 4 .5 tf

B« +c + 6-- = 23.5 tf
Check: 2 Y ^ A + B + D -q lt—P= 14.5+ 23.5+6-4X8 -1 2 = 0 .
Having determined the reactions we again assume the beam to be
a single unit with all forces and reactions and determine the moments
and shearing forces as in the general case. We shall check the values
by equating to zero the sum of moments about point C. It should be
borne in mind that hinge C does not represent the separation point
of sections of the diagrams if it is not acted upon by an external force.
The bending-moment and shearing-force diagrams are shown in
Fig. 144(q).
After determining the reactions it is more convenient to plot Q-
and M-diagrams separately for each suspended beam and the main
beam, laying the values of Q and M from a common x-axis.

§ 60. The Check of Proper Plotting of Q-

and M-diagrams
The differential relations between the bending moment, shearing
force and intensity of continuous load determine the relation between
shearing-force and bending-moment diagrams for any load. This
relation is of great practical importance in checking the correctness
of the plotted curves. We give below some concluding remarks which
may be helpful in plotting Q- and Af-diagrams.
1. It has already been slated 57) that the ordinate of the shearing-
force diagram Q = - ^ geometrically represents the slope of the tan­
gent to the bending-moment diagram at the corresponding point.
Identical geometrical relations exist between q and Q (Fig. 145).
2. If in a certain section
(a) Q>0, i.e. tan a > 0 , the moment increases;
(b) Q<0, i.e. tan a < 0 , the moment decreases;
 222 Bending. Strength o f Beams [Part IV

(c) Q passes through zero, changing its sign from plus to minus,
^max»
(d) Q=0, i.e. tan a = 0 , M=const.
3. If <7=0, i.e. ^ = 0, Q=const. Hence in portions free of contin­
uous load, the shearing-force diagram is bounded by straight lines
parallel to the x-axis; the bending-moment diagram*is made up of
inclined straight lines provided Q #0 (see item 2(d)). If ^< 0, i.e.
tan P<0, the shearing force decreases.

4. Over portions of the beam loaded by uniformly distributed force,

the bending-moment diagram is a parabola, whereas the shearing-
force diagram is an inclined straight line. If the load is distributed
non-unlformly, then the Q- and At-diagrams are represented by curves
whose shape depend upon the type of loading.
5. In sections under concentrated force the shearing-force diagram
undergoes a jump (equal to the force), and the bending-moment diag­
ram experiences a sharp change in the angle between the adjacent
regions (see, for example, section C in Fig. 137).
6. If the continuous load is directed downwards, i.e. =<7< 0 ,
or, in other words, if the second derivative characterizing the curva­
ture of the M-curve is negative, then the diagram is convex upwards.
On the contrary, if q> 0 (the load is directed upwards), the bending-
moment diagram in the corresponding portion is convex downwards
(Fig. 146).
7. In a hinged support the shearing force is equal to the reaction
of the support, and if there is no external moment acting on it, the
bending moment in the hinge is zero.
8. The bending moment on the free end of a cantilever is zero if
the end is not acted upon by a concentrated force couple. In the ab-
Ch. 10) Internal Forces, Q- and M ’diagrams 223

sence of a concentrated force on the free end the shearing force Q is

also zero.
9. At the fixed end, Q and Af are equal to the reaction and moment
of the support, respectively.
10. In sections where a force couple is acting, the bending-moment
diagram undergoes a jump equal to the moment of this force couple.
The shearing-force diagram however, remains unaffected.

The differential relations explained in § 57 and the remarks given

above help not only in checking the correctness of the diagrams but
will be used in future in plotting the diagrams too (Chapters 15, 16,
etc.).

§ 61. Application of the Principle of Superposition

of Forces in Plotting Shearing-force and
Bending-moment Diagrams
Analyzing the expressions for Q and M obtained in the previous
examples, we see that the external forces enter these expressions to
the first power; M and Q are linearly dependent upon the load.
Analyzing, for example, equation for M (x) in the section of a can­
tilever (Fig. 140),
Af (*) = P x— q -y-

we find that the ordinates of the bending moments in sections of this

portion consist of two components, Px and — the first component
representing the bending moment in the particular section due to
force P, whereas the second due to uniformly distributed load q.
We could have plotted the ben ding-moment diagrams for forces
P and q separately, and then added their ordinates algebraically.
This would be the application of the method of superposition of forces.
We shall illustrate with an example how to plot the total bending-
moment diagram. For the beam shown in Fig. 147 we have already
224 Bending. Strength of Beams [Part IV

plotted separately bending-moment diagrams under uniformly distri­

buted load q(M J and concentrated force P(Mp). The absolute maxi­
mum value of tne bending moments at the rigidly fixed end are
—^ and Mp = —Pl

To add the ordinates of two diagrams of similar sign, we place one

above the other as shown in Fig. 148(a). The bending moment in an
arbitrary section is the sum of moments:

and
Mp = — Px
The sign of M P changes if force P is directed upwards. To add (wo
diagrams having different signs, it is sufficient to superimpose one

Fig. 147 Fig. 148

of them over the other (Fig. 148(b)).

Suppose that in absolute value min Mv>m ax M P, i.e.

t |>IwI
Upon superposition of diagrams their ordinates get deducted automat­
ically, and in this example we shall get a negative ordinate at the
fixed end, the ordinates being positive over the span at a certain dis­
tance.
Obviously, in graphical summation both the diagrams must be
drawn to the same scale. In an identical manner we can plot the shear-
Ch. Ill Normal Stresses in Bending. Strength of Beams 225

ing-force diagram. The method of summation of diagrams is parti*

cularly useful in analyzing statically indeterminate solid beams
(Chapter 19).
To obtain the diagram in the conventional form, we may lay off
the summed ordinates from the horizontal axis (Fig. 148(a) and {b)).

C H A P T E R 11

Determination of Normal Stresses in Bending

and Strength of Beams
§ 62. Experimental Investigation of the Working
of Materials in Pure Bending
The bending*moment and shearing-force diagrams enable us to
determine the internal forces in an arbitrary section of the beam;
these forces are made up of normal and shearing stresses in the section
as a result of bending. We shall discuss how to determine these stres­
ses. It was earlier shown that shearing force in a section is the resul­
tant of elementary shearing forces, and the bending moment, of the
normal stresses which form force couples. If no shearing force Q acts
over a certain length of the beam, i.e. the shearing stresses in sections
within this length are absent, then these sections are acted upon only
by normal stresses which are easier to compute in this case.

A -P
(bj
Fig. 149

The type of bending in which shearing force is zero in sections nor­

mal to the beam’s axis is known as pure bending. Pure bending can
be achieved in practice if the system or external forces acting on some
portion of the beam can be reduced to force couples (see, for example,
Fig. 130). Actually, however, pure bending is possible only in those
cases when the dead weight of the beam is sufficiently small as compa­
red to the external forces acting on it and may therefore be neglected.
As an example we shall consider the bending of a wagon axle (sec
§55). The external forces acting on the axle (its weight is neglected)
are depicted in Fig. 149. Keeping in mind that due to symmetry both
226 Bending. Strength of Beams IP a rt JV

the support reactions are equal (A = B —P), for an arbitrary section

between points C and D we obtain
Q(x) = A — P = 0, M (x) = A x— P (x ~ a ) = P a ~ const
Thus, there is no shearing force in the middle portion CD of the
axle, and Af=con$t ^recall that ; the beam experiences pure
bending over the length CD.
Let us now return to finding normal stresses for this case. Let us
take a section at a distance x from the left support A and consider the
equilibrium conditions of the left cutoff portion (Fig. 149). This por-

Fig. 160

tion is acted upon by a force couple with moment M = Pa and normal

stresses in the section, which form force couples with a resulting mo­
ment M (x). Our task is to find the magnitude of these stresses at every
point of the cross section and determine their maximum value. How­
ever, the conditions of equilibrium between the external and internal
forces expressed by the relationship M (x)=Af are not sufficient for
determining normal stresses a because we know neither the magnitude
of these stresses nor their distribution over the section. The problem
is statically indeterminate, and for its solution we must study the
elastic deformation of the beam on the basis of experimental investi­
gations. Let us consider the results of experiments obtained from pure
bending of a beam by moment M acting in its plane of symmetry
(Fig. 150).
Lines 1-1 and 2-2 drawn on the beam surface perpendicular to its
axis are traces of two adjacent cross sections located at a distance Ax
from each other, whereas lines ab and cd joining them and parallel
to the beam’s axis represent longitudinal fibres of length Ax prior
deformation (Fig. 150(a)).
Experiments reveal" that after deformation (Fig. 150(6)):
(I) Lines 1-1 and 2-2 remain straight but turn with respect to one
another through an angle Aa. This leads us to the idea that the cor-
ch. n \ Normal Stresses in Bending. Strength of Beams 227

responding cross sections also remain planes but turn with respect
to one another through angle Aa.
(2) Lines ab and cd change their length: line ab gets shorter, whereas
cd elongates leading to the conclusion that the upper fibres are sub­
jected to compression and lower to tension.
(3) As shown in Fig. 150(c), the cross-sectional dimensions also
change: in the upper part the width of the beam increases, which cor­
responds to axial compression, whereas in the lower part (stretched
zone) it decreases.
As the deformation of the longitudinal fibres varies continuously
over the height of the beam, there must be a layer at a certain height
which does not change its length at all; this layer is called the neutral
layer and serves as the interface between the compressed and stretched
zones. In Fig. 150(6) the neutral layer is shown by dotted line; seg­
ment OiOa retains its initial length Ax.
The neutral layer is perpendicular to the plane of symmetry of
the beam in which the external forces act and intersects each cross
section of the beam along a straight line which is also perpendicular
to the piane of action of the external forces. The line of intersection
of the neutral layer with the plane of a cross section is known as the
neutral axis of the section. The neutral layer is an aggregate of the
neutral lines.
As the section is symmetrical with respect to the plane of applica­
tion of the external forces, both halves of the beam width must deform
symmetrically about this piane; this enables us to consider that lon-
'tudinal deformation of the fibres of an arbitrary layer parallel to
S e neutral one is independent of the location of the fibres along
the beam width.
It has been experimentally established that deformation in the la­
teral direction is related to the deformation of longitudinal fibres by
Poisson's ratio. This gives sufficient ground to presume that the lon­
gitudinal fibres do not press each other, and under pure bending ex­
perience only simple compression on the concave side and simple
tension on the convex, i.e. on the other side of the neutral layer.
At the same time, lateral deformation is instrumental in somewhat
distorting the beam section and making the neutral axis curved
(Fig. 150(c)), which leads to additional deformation of the neutral
layer making it doubly curved. However, as the elastic deformations
are small these distortions are ignored: in each cross section of the
beam the neutral axis is considered a straight line and the neutral
layer, a cylindrical surface.
Since the section is symmetrical w.r.t. the plane of application of
external forces, the beam axis also curves in the same plane in bending.
Such bending in which after deformation the beam axis remains in
the plane of application of external forces is known as uni-planar
bending.
228 Bending. Strength of Beams [Part tV

Experimental study of the bending of beams helps us to make, a

number of assumptions which have been used in deriving new con­
clusions:
1. In pure bending the cross sections which were planes prior defor­
mation remain planes during deformation too (the hypothesis of plane
sections).
2. Longitudinal fibres of the beam do not press on each other,
and therefore due to normal stresses experience simple uniaxial ten­
sion or compression.
3. The deformation of fibres does not depend upon their position
along the width of the section. Therefore, the normal stresses, though
changing along the height of the section, remain constant along its
width.
In addition to the assumptions made above, we shall introduce
three limiting conditions:
1. The beam has at least one plane of symmetry, and all the exter­
nal forces lie in this plane.
2. The beam material obeys Hooke’s law*, the modulus of elasticity
being the same under tension as well as compression.
3. The relation between the beam’s dimensions ensures that it
works under pure bending without warping or twisting.
It is known from experience that beams with a small width easily
loose their stability as far as the shape of the section is concerned
(they warp). If in a beam of rectangular section the ratio of height
to span is y , it works not as a beam but as a plate and it must
be analyzed in a different manner. *
In general, assumptions made above are only approximately true.
However, the theoretical error is so small (except in special cases)
that it can be ignored.

§ 63. Determination of Normal Stresses in Bending. 1

Hooke’s Law and Potential Energy of Bending

A. Lei us consider a beam subjected to pure bending by a moment M

(Fig. 151). Let us cut the beam in two parts by section /-/, and using
the method of sections consider the equilibrium of one of the portions,
say, the left, shown below In Fig. 151. For simplicity we consider a
beam of rectangular cross section. As the curvature of the beam is
practically negligible in comparison with its dimensions, the cutoff
portion may be drawn in undeformed shape.
The line of intersection of the plane of symmetry of the beam with
the plane of the section is taken as the z-axis (positive direction down-
wards); the neutral axis of the section has been taken as the y-axis,
its location along the height of the beam being not yet known. The
Ch. 11] Normal Stresses in Bending. Strength of Beams 229

x-axis has been taken along the neutral layer perpendicular to the
y~ .and 2-axis.
Every point in the cross section is acted upon by a normal stress o.
Let us isolate an elementary area dA about an arbitrary point having
coordinates y and z, and denote the force acting on it by dN=adA.
The cutoff portion of the beam maintains its equilibrium under the
action of external forces constituting a couple of moment M and

1 2

tl
1 1
w ------ X — dx

J# adk

Fig. 151

the norma! force dN which represents the influence of ihe removed

portion of the beam. The beam will remain in equilibrium only if
this system of forces satisfies the six static equations. Let us first
write down the equations of projections on the three coordinate axes
of x, y and z.
As the projection of moment M on any axis is zero, these equations
give us the condition that the sum of the projections of normal force
dN on the beam’s axis is zero. Replacing the summation over the
whole area by integration, we get
2X =0. J odA = 0 (U .l)
A

Expressions 2 ^ = 0 and 2 ^ = 0 give identities of the type 0=0,

because the force dN=<Jd/Tprojects on these axes into a point.
Let us now write down tne equations of moments about the axes
Ox, Oy and Oz. Let us note that moment At Iies in the plane xOz and
therefore does not give any moment about axes Ox and Oz.
Expression 2 M*—1Ogives an identity, because the forcedN=adA
is parallel to Tne x-axis:
2 ^ = 0, M — 2<W * = 0 or M — ^azdA = 0
A
230 Bending. Strength of Beams IPart IV

wherefrom
\<szdA = M (11.2)
A

or Jo*/d,4e=0 (11.3)
A

Thus, out of the six static equations we can use only three:
2X «0 or Jo d A ^ O (11.1')
A

2 ^ = 0 or J ozdA = M (11.2')
• A

2 A fi =s0 or J aydA = 0 (11.3')

A

However, the three static equations obtained above are not suf­
ficient to determine the normal stresses, because a varies with the dis­
tance z of area dA from the neutral axis according to a law which we
yet do not know. The distance z is also
unknown, because we do not know the lo­
cation of the neutral axis Oy.
B. Let us isolate an element of length dx
of the beam by two infinitely close sec­
tions I-J and 2 - 2 to study its deformation.
The shape of the element before and after
deformation is shown in Fig. 152.
For greater clarity the deformation of
the element is shown in a highly mag­
nified form. Both cross sections continue
to remain planes but turn about their
neutral axes (points 0, and Oz in the
front view) to form an angle da. The neut­
ral layer has been shown by a dotted line.
Line OiOa of the neutral layer retains
its initial length dx after deformation.
All fibres above the neutral layer shorten,
whereas those below it elongate.
We shall try to find the elongation of
an arbitrary fibre AB at a distance z from
the neutral layer and stretched by stress a. The initial length of this
layer is d x = _ 0 |0 8= p da. After deformation its length along the arc
AB becomes w AB=(p-f-a) da. The absolute elongation of the fibre
is A /=(p+ 2) da—p d a = 2 da. Relative elongation is equal to
Ch. 11\ Normal Stresses in Bending. Strength of Beams 231

i.e. the elongation of fibres is directly proportional to their distance

from the neutral layer.
Here p is the radius of curvature of the neutral layer, which may be
considered constant for the isolated (infinitely small) element. As*
suming that under bending the fibres do not press on each other and
that each fibre experiences simple (uniaxial) tension or compression,
we may make use of Hooke’s law in determining the tensile stresses:

o = £e or o= ~ (11.4)

Equation (11.4) shows that the normal stresses in bending vary in

direct proportion to distance z of the point of the section under con­
sideration from the neutral layer. This means that stresses vary along
the height of the beam linearly.
On the neutral axis 2=0 and cr=0. If we move into the zone of com­
pression (above the neutral axis), a along with z changes its sign to
minus (compression) and continues to increase in absolute value as

We move away from the neutral axis. Hence the maximum stress
occurs at the uppermost and lowermost layers of the section when
z=zma3t. The distribution of stresses along the height is shown in
Fig. 153.
Equation (11.4) only gives an idea about the nature of distribution
of normal stresses over the section; it cannot be. used for calculating
the magnitude of the stresses because both p as well as z are not known
since we do not know the location of the neutral layer in the height
of the section.
C. To determine a as a function of the bending moment, we shall
simultaneously solve Eq. (11.4) obtained from deformation con­
siderations and the static equations (11.1), (11.2), and (11.3).
Substituting the value of o from expression (11.4) in Eq. (11.1),
we get
2 * = 0 or z d A —Q
232 Bending. Strength of Beams [Part tV

Since — —consMO,
\z d A = Q (11.5)
A

This integral represents the static moment of the area of section about
the neutral axisOy, which becomes zero only about the central axis.
Therefore the neutral axis must pass through the centre of gravity
of the section. As the centre of gravity also lies on the axis of symmetry
Oz, the point of intersection of these two axes 0 represents the centre
of gravity of the section, and Ox represents the axis of the beam.
Thus, we have completely determined the location of the neutral
layer and neutral axis. The centres of gravity of all sections of the
beam are located on the neutral layer.
Now let us put the same expression (11.4) into Eq. (11.3):
= -zt/dA — 0, or -~-^zydA*=0

wherefrom it ensues that

\z y d A = 0 (11,6)
A

The above integral, which is the sum of the products of elementary

areas by their distances from the coordinate axes Oy and Oz is called
the product of inertia of the section with respect to the axes Oy and Oz.
The product of inertia of the section may be positive or negative;
consequently it may also vanish, because the coordinates Of elemen­
tary areas may have different signs.
According to expression (11.6), the product of inertia of the section,
which is generally denoted by
J t0*= jzydA

should in this case be zero.

As the section is symmetrical about axis Oz, for each elementary
area dA with coordinates (z, y) to the left of the z-axis we can find
a similar, symmetrically located elementary area to the right of
the z-axis. The 2-coordinates of these areas wi11 be the same by the mag­
nitude and sign, while the^-coordinates will be equal in magnitude but
will have opposite signs. Therefore the integral
\z y d A
A
will consist of two integrals equal in magnitude but of opposite signs.
Thus, for symmetrical sections this integral is always zero, and Eq.
Ch. Ill Normal Stresses in Bending. Strength of Beams 233

(11.6) changes into an identity. In our case the condition J (I(= 0

is satisfied. Bending will be uni-planar only under the condition that
the product of inertia of the section is zero about axes one of which
is In the plane of application of the external forces; then all sub­
sequent conclusions will be valid.
Finally, let us study Eq. (11.2); substituting expression (11.4)
into it, we get
2^ = 0, f — z*dA = M or
a p a
Let us introduce the notation
J .^ iA (11.7)

This integral, which is the sum of the products of elementary areas

by the square of their distance from the axis, is called the axial or
equatorial moment of inertia of the area about the */•axis and is denoted
by J y As the y-axis is the neutral axis, /„ is the moment of
inertia of the area of the section about the neutral axis *. From
the Iransformed expression of equation (11.2), we get
or (»-8)
P P J '
P
Putting this value of -r in Eq. (11.4), we get
Mz
<7 (11.9)
J

Hence, the normal stresses in any point of the section are directly
proportional to the bending moment and its distance from the neutral
axis, and inversely proportional to the moment of inertia of the sec­
tion about the neutral axis.
The neutral axis passes through the centre of gravity of the section
and is perpendicular to the plane of action of the external forces.
it is obvious from formula (11.7) that the moment of inertia is
measured in units of length to the fourth power and depends upon
the shape and size of the section. Methods of determining the moment
of inertia for various sections will be given below.
Let us modify formula (11.8) to understand the physical meaning
of this quantity:
I M
( 1110)
p*" EJ
* In future, when denoting the moment or Inertia about the neutral y*axls, we
shall often drop the Index y and denote it in short by J instead of / v-
234 Bending. Strength of Beams {Part IV

It is clear from this formula that the greater the moment of inertia
J of the section for a given bending moment, the greater will be the
radius of curvature of the neutral layer and, consequently, of the
beam’s axis, i.e. the less will be the bending of the beam.
The value of the moment of inertia characterizes the ability of
beam to resist bending depending upon the shape and dimensions of
its cross-sectional area. The modulus of elasticity E also characterizes
the ability of the beam to resist bending depending upon the material
of the beam. The product EJ is called the rigidity of die beam under
bending. The greater the rigidity, the less will be the bending of the
beam with a given bending moment.
D. The relative rotation of the sections is connected with the bending
of the beam’s axis. As is clear from the drawing (Fig. 152), the length
of segment OxOi=dx is equal to pda. Herefrom the angle of rotation
between two adjacent sections may be written as

1 M
Replacing — by its value we get
M dx
da ~e T (11.11)

i.e. the deformation and displacements in bending—the angle of

turningda and the curvature of the beam — are directly proportional
to the bending moment and inversely proportional to the rigidity
of the beam.
Repeating the reasoning employed in § 52, we can easily calculate
the potential energy accumulated by the beam during bending. If we
consider the bending of an infinitely small segment of the beam of
length dx, we can calculate the work done by the bending moment
over the da as follows:

d JT -I^ d a

Putting the value of da from Eq. (11.11), we get

d y = d r= 4 ^

Integrating over the whole length of the beam, we get

j, C M 'd x
( 11. 12)
u -\-w r
Ch. II) Normal Stresses in Bending. Strength of Beams 235

In pure bending of a beam (M=const) with a constant cross-section­

al area over the whole length (£J=const), the potential energy may
be expressed as
(1 U 2 ')
If bending moment along the beam’s length is expressed in terms of
different functions of x (for different portions), then integral (11.12)
breaks up into a sura of integrals (each within the limits of the cor­
responding portion), and the expression for potential energy of bend­
ing becomes
y=Z f w (1U2' )
The potential energy of deformation of the beam on account of shear
{§ 36) caused by shearing forces Q is usually neglected as it is rela­
tively small (for details, see Chapter 18).

§ 64. Application of the Results Derived Above

in Checking the Strength of Beams
Formula (11.9) solves the question about the magnitude and distri­
bution of normal stresses over the section. It has been derived for pure
bending, when the sections remain planes.
Experiments show that when Q is not zero, the sections not only
turn but also slightly warp under the action of shearing stresses.
This warping, however, does not alter the distribution of stresses in
fibres enclosed between the two adjacent sections. Therefore, for­
mula (11.9) may be used even when Q is not zero.
It should be noted here that as yet we can use this formula only if
the sections of the beam have an axis of symmetry, and the external
forces act in the symmetry plane.
The neutral axis of each section, from which z is measured, passes
through its centre of gravity perpendicular to the axis of symmetry.
Figure 154 shows examples of the distribution of stresses for beams
of various sections—rectangular, T-shaped, triangular. The normal
stresses are the same in all points located at equal distances from the
neutral axis. We get compressive stresses to one side of the neutral
axis, and tensile stresses to the other. The maximum stresses occur
in points which are farther from the neutral axis. For the accepted
convention of signs of M and z, formula (11.9) automatically gives
the proper sign of a, plus for tensile stresses and minus for compres­
sive stresses.
If the bending moment is positive, the beam bends with its convex­
ity downwards, the upper fibres are compressed (z<0), whereas
the lower fibres are stretched. The reverse picture occurs if the bend'
 236 Balding. Strength of Beams [Part IV

ing moment is negative. Therefore, in selecting the sign of normal

stresses while solving practical problems, we may follow the follow­
ing rules: if the point of the section under consideration is located
in the zone of stretching, a should be taken with a plus sign: if the
point is located in the zone of compression, a should be taken with
a minus sign. Obviously, in this case the absolute values of M and
z should be used in formula (11.9).

Fig. 154

To check the strength of a material w.r.t. normal stresses, it is

essential to find the maximally stretched or compressed areas. This
can be achieved by applying formula (11.9) to the critical section,
i.e. using instead of M and instead of z put z ^ , the distance
of the farthest point from the neutral axis. Then we get the following
formula for the maximum normal stress:
„ Mjnaxzm3X
“max j

Usually, this formula is transformed by dividing both the numera­

tor and the denominator by amax:
et __ Mmax
wmax= j
2m»x

Q uantity//2^JXis called the axial section modulus and is denoted

by the letter W. As / is measured in units of length to the fourth
power, W is measured in units of length to the third power, e.g. cm3.
Hence
A!max (11.13)
tfmasc
where
J
W (11.14)
zma x

If the section is symmetrical about the neutral axis, for example,

a rectangular section, the outer stretched and compressed fibres are
Ch. //] form al Stresses in Bending. Strength of Beams 237

located at equal distance from the neutral axis, and such a section
has a single definite value of the section modulus about the y-axis.
Thus, if we consider a rectangular section of height h (Fig. 155(a)), then

and W = j-
2

If the section is not symmetrical about the neutral axis, for example,
a T-shaped section, we get two values of the section modulus: one
for layer A (Fig. 155(6)), 1 ^ = ^ -, and the other for layer B,

■if tr *
t

zf z ’k t
<'b)
Fig. 155

Now in formula (11.13) we should introduce: Wi when calculating

stresses in point A and W* when calculating stresses in point B.
Let us write down the strength condition for tensile or compressive
stresses. The condition reflects the idea that the maximum stress
should not exceed the permissible:

From this condition we find that

W (11.16)

i.e. the section modulus determined from strength considerations

should be greater than or equal to the maximum bending moment
divided by the permissible stress.
Since W depends upon the shape and size of the beam section, by
selecting a particular shape (rectangular, T-shape, 1-shape) we can
find the dimensions of the beam such that its section modulus equals
the one obtained from formula (11.1G). We shall show below how this
238 Bending. Strength of Beams [Part IV

can be done in practice. The values of W and J for rolled sections are
given in specification tables (see Appendix).
We must differentiate between the following two cases when using
formulas (11.15) and (11.16).
The first case is more common in bending, when the material shows
equal resistance to tension and compression; in this case the permis­
sible stress is the same for both deformations:
KI= K l= M

In the case of a symmetrical section it becomes irrelevant whether

we check the strength of the stretched or compressed fibres, because
for both of them the section modulus W and the maximum actual
stress have the same value. In the case of an unsymmetrical section,
in formulas (11.15) and (11.16) W should be replaced either by Wi
or by Wt, whichever is less; it should correspond to the farthest fibre.
The second case deals with beams whose material has different
resistance to tension and compression. In this case we must write two
strength conditions instead of one—one for the stretched fibres and
the other for compressed:

« ( - + T rT < !* .]. » ,------< [ o J (11.17)

Depending upon whether the material has better resistance under

tension or compression, i.e. which of [o j or |crc] is greater, we have
to design the section by selecting its shape and size such that Wt
and W9 satisfy the strength condition.
The physical nature of section modulus is clear from formula (11.13):
the greater the section modulus W, the greater is the bending moment
to which the beam can be subjected without danger of failure. Thus,
section modulus characterizes the efFeclof shape and size of the select­
ed section on the strength of the beam when the stresses do not exceed
the limit of proportionality.
Formulas (11.13) and (11.17) cease to be valid for stresses exceeding
the limit of proportionality of the material.
Formulas (11.15) and (11.17) enable us to check the strength of
a given section (when the value of section modulus W is known).
If the beam's material has been selected and its permissible stress is
known, then with the help of formula (11.16) we can compute the nec­
essary value of section modulus, provided the maximum bending
moment Mmx is preliminarily calculated. Then, depending upon
the beam's profile, i.e. the shape of the section, the required cross-
sectional dimensions can be determined.
It was shown earlier that for this we must find the relation between
the cross-sectional dimensions and the value of the section modulus.
In § 73 we shall elaborate on this.
Ch. 12\ Moments of Inertia of Plane Figures 239

C H A PTER 12

Determination of Moments of Inertia

of Plane Figures
§ 65. Determination of Moments of Inertia and Section
Moduli for Simple Sections
While deriving the expression for normal stresses (§ 63) we had
obtained expression (11.7) of the type
J u= \z * d A
A

where z is the distance of any elementary area from the central tj-
axis. This integral, which covers the whole area of the cross section,
was called the moment of inertia of the area about the neutral axis.
The ability of a beam section to resist deformation in bending depends
upon the value of moment of inertia (11.10).
Apart from this, the strength condition in bending (§ 64) includes
the expression foY section modulus (11.14):

I t ensues from the above that we must learn to calculate the mo­
ment of inertia and section modulus for cross sections of any shape
to ensure strength and rigidity of the beam. Let us.start with the sim­
plest beam section, a rectangle of width b and height h (Fig. 156).
Draw axes of symmetry Oz and Oy through its centre of gravity 0.
If the external forces acting on the beam lie in plane xOz, then Oy
is the neutral axis (axis Ox is directed along the beam). Let us first
find the moment of inertia about this axis, and the section modulus
of the rectangle.
Elementary areas dA into which the whole area of the section should
be divided will be taken as narrow rectangles of width b and height
dz (Fig. 156). Then
dA = bdz
and integral J y

If we take the integral over the total area of the rectangular section,
It ft
z varies from — j to + -j*. Therefore

( 12- 1)
240 Bending. Strength of Beams |Pari IV

We get the section modulus about the neutral axis Oy by dividing

Jv by zn*x= 2" •
J v __M 3/ I 2 _ AM
( 12. 2)
?n»x 2 ~

If we have to calculate the moment of inertia and section modulus

of the. rectangular section about the axis Oz, then all that is required
is to interchange b and h in the above formulas:

and
hb»
W* = ~ T (12.3)

Let us note that the sum of the products z2dA does not change if
we displace all the strips dA =b dz (Fig. 156) parallel to themselves
in such a way that they lie within the parallelogram ABCD (Fig. 157).
a
r* - b ------ *1
~7
1
r
h . — L [.

/ft

+ y

7
Fig. 156

Hence the moment of inertia of the parallelogram ABCD about

the y-axis is equal to the moment of inertia of an equivalent rectangle
ABGE:

0 2 . 1)

As the moment of inertia of an area is an integral of the type

Jy—\ z z dA, we can immediately determine the moment of inertia of a
n
rectangular box section (Fig. 158) with the help of formula (12.1):
ch. m Momenta of Inertia of Plane Figures 241

The section modulus is

Jv BH* — bh9 _ BIP—bh3
12///2 = 577 ( 12.2')

Note that the section modulus cannot be calculated in the form of

difference W—Wt—W9 or —Ma/6, because this runs coun­
ter to the very concept of section modulus as the ratio J h!zmax.

Fig. 158

While determining the moment of inertia of a circle of radius r

(Fig. 159) we similarly divide its total area into elementary strips of
thickness dz along the axis Oz\ the width of the strips &=&(*) also
varies along the height of the section. The elementary area is
dA = b (?) dz
The moment of inertia is
J — J z*b(z)dz
A

As the upper and lower halves of the section are identical, it is

sufficient to calculate the moment of inertia for one half and double
the result. The limits in which z varies are from 0 to r:
r
J = 2[z*b{z)dz

We introduce now a new variable of integration, angle a (Fig. 159):

z=*rcosy, d2=» ——rsinyda, 6(2)=*2rsin*^
9 —3310
242 Bending. Strength of Beams [Part IV

The limits of integration are a=Jt at z = 0 an d a= 0 at z=r, therefore

J ~ — 2^ 2r‘ cos* ^ s i n * - y y d a = - ^ Jsin * a d a ‘^ - (12.4)

and
(12.5)

For a circle any axis passing through the centre of gravity is the
axis of symmetry. Therefore formulas (12.4) and (12.5) are valid for
ail such axes.

Fig. 161

Substituting r—j , we shall now express J and W through the circle’s

diameter:
ltd*
J 04 (12.4')

w = n32 « O.ld* (12.5')

The moment of inertia of a triangle (Fig. 160) about AB is:

z*bxdz, bz = b^-j^- — b ^ 1—
A
h

Later (§§ 66-68) we shall explain how to calculate the moment of

inertia of a section of any complex shape about an arbitrary axis.
The symmetrica! sections which we generally come across in prac­
tice are: for wood—rectangular and circular, for metals—I-shaped
and T-shaped (Fig. 161). For rolled sections we may use the GOST
ch. m Moments of Inertia of Plane Figures 243

tables * (specifications), which contain the dimensions and the values

of J and W for the sections manufactured at the rolling plants. These
tables are given in Appendix.
Generally, metal beams have complex cross sections, because a more
economic exploitation of the metal is possible in these sections as
compared to, say, rectangular or circular sections.
b

Fig. 162 Fig. 163

We saw in § 48 that the shafts are made hollow to remove the por­
tion of material that works under lower loading. In bending beams
the material near the neutral axis experiences very small normal
stresses (formula (11.9)) and is consequently not utilized fully. It is
therefore more expedient to modify the rectangular section by re­
moving the metal near the neutral axis and utilizing a part of this
metal in the upper and lower zones of the beam, which work under
more severe conditions, and saving the rest of it. Thus, from a rectan­
gular section we obtain an I-shaped section (Fig. 162), which has the
same strength but is lighter. The 1-sections should preferably be used
for materials which have equal resistance to tension and compression
(in the majority of the metals).
The T-shaped sections are used in the cases when this is dictated by
design considerations and when the materials, for example, cast iron
and concrete, greatly differ in resistance to tension and compression.
The latter condition requires that the stresses should be different in
the outer fibres.
It ensues from the above discussion that the most economic design
of the section should endeavour to obtain the maximum moment of
inertia and section modulus for the fixed area A. In this design the
greater part of the material will be located farther from the neutral
axis.
However, in some sections the section modulus may be increased
not by adding, but, on the contrary, by cutting off a part of the sec-

* GOST stands lor All-Union State Standard (in the USSR].

9*
244 Bending. Strength of Beams [Part IV

tion which is farthest from the neutral axis. If we cut off the hatched
segments of the circular section (Fig. 163), its section modulus some­
what increases, because the decrease in the moment of inertia is less
than that of distance znK from the outer fibres.
The most effective section in bending will be that for which the
ratio W/A of section modulus to the cross-sectional area is maximum.
It is more convenient to assess the effectiveness of section by the di­
mensionless coefficient a =W>\Ah), where h is the height of the sec­
tion. Table 11 contains the values of coefficient a for a few sections.
We see from the table that a is maximum for an 1-seclion.
Table 11
Coefficients of Profile Effectiveness

Shape of sectio n C oefficient ~ of

Shape
1_
s e c tio n
C o n fid e n t
a

I-sect! o n (depending upon 0.31-0.34 Rectangle 0.107

the profile No) Circle 0.125
Channel section (depen- 0.29-0.3J Triangle 0.083
ding upon the profile Hollow circular section 0.226
No) (when r/f?=i0.9)
T-soctlon 0.085

§ 66. General Method of Calculating the Moments

of Inertia of Complex Sections
While checking the strength of elements of structures we often come
across sections of complex shape for which the simple method used for
calculating the moment of inertia of sections like a rectangle or circle,
discussed in §65, does not hold.

W (b) (c)
Fig 164

Such a section may be, for example, a T-shaped section (Fig. 164(a)),
a pipe section working under bending load (in aviation design)
(Fig. I64(/>», a ring cross section of the neck of a shaft or a more com-
Ch. I2\ Moments of Inertia of Plane Figures 245

plex section (Fig. 164(c)). AH these sections may be divided into

simple shapes such as rectangles, triangles, circles. It can be shown
that the moment of inertia of a complex section is the sum of the mo­
ments of inertia of the parts into
which it is divided.
Let us take (Fig. 165) an arbit­
rary figure representing the cross
section of abeam; they»axls is drawn
in the plane of the section. The
moment of inertia of this figure
about the y-axis is (11.7)

A
where z is the distance of elementa­
ry areas dA from the y-axis.
Let us divide the area of the
figure into four parts: A 1t A t, Aa,
and A 4. Now when calculating the
moment of inertia according to
formula (11.7), the terms under the integral sign should be grouped in
a way such that we can carry out integration of the elementary areas
separately for each portion and then add the results. The value of the
integral will remain unchanged after this operation.
The integral will break into four integrals each of which covers
one of the areas A t, At, Az or A 4:
= \ z*dA — ^ zt dA + J z* dA -f 5 z*dA 4- J a* dA
j\ >1] /1| a4

Each of these integrals represents the moment of inertia of the cor­

responding portion about axis Oy; therefore
A, = (12.6)

where Jj, is the moment of inertia of area At about the (/-axis, J ll

is the moment of inertia of area A t about the same axis, and so on.
The result obtained above ttiay be formulated in the following man­
ner: the moment of inertia of a complex figure is equal to the sum of
the moments of inertia of parts comprising it. Therefore, to calculate,
for example, the moment of inertia of the section shown in Fig. 164(c)
about axis Oy, we must calculate the moments of inertia of appro­
priate triangles and rectangles about the same axis and add the re­
sults. We must know how to calculate the moment of inerlia of an
arbitrary figure about an arbitrary axis lying in its plane.
The solution of this problem forms the contents of this chapter.
246 Bending. Strength of B a m s [Part IV

§ 67. Relation Between Moments of Inertia About

Two Parallel Axes One of Which Is the Central Axis
The problem of obtaining the simplest possible formulas for com­
puting the moment of inertia* of any figure about an arbitrary axis
can be solved in a number of ways. If we take a number of axes pa­
rallel to one another, then the mo­
ment of inertia of the figure about
any of these axes can be calculated
if we know the moment of inertia
of the figure about the axis passing
through its centre of gravity and
parallel to the selected axes.
We shall call the axes passing
through the centre of gravity the
central axes. Let us take (Fig. 166)
an arbitrary figure. Draw the cent­
ral axis Oy and denote by J v the mo­
ment of inertia of the figure about
this axis. In the plane of the figure
draw axis 0 i{/i parallel to the
y-axis and located at a distance a
from it. We shall try to establish
the relation between J y and J'y, the moment of inertia about the
y,*axis. For this we shall have to write the expressions for J v and Jy.
Break the figure into elementary areas dA, and denote by z and Zt
the distances of points lying on the elementary area from axes Oy and
Od/u respectively. We find that
Jy — ^z^dA and Jy —^ z \d A

But it is evident from the drawing that

Zi = z + a
Therefore
J'y = J (z + af dA = ^ (2a+ 2az+fl8)<L4
A A

= J z*dA + 2 a $ zdA -f a* $ dA
A A A

The first of the three integrals represents the moment of inertia

about the central axis Oy. The second integral represents the static
moment about the same axis. It is equal to zero, because the y-axis
passes through the centre of gravity of the figure. Finally, the third
Ch. 12] Moments of Inertia of Plane Figures 247

integral represents the area of the figure. Therefore

J ^ J ^ a 'A (12.7)
i.e. the moment of inertia about an arbitrary axis is equal to the mo­
ment of inertia about the central axis parallel to the arbitrary axis
plus the product of the area of the figure by the square of the distance
between the axes.
Hence, our problem reduces to determining the central moments
of inertia. Knowing them we can calculate the moment of inertia
about any other axis with the help of formula (12.7). It is evident
from formula (12.7) that the centred moment of inertia is the minimum
of the moments about parallel axes, and it may be expressed as
J v = J'y-a * A (12.7')
We can similarly determine the product of inertia J'ox of the sec­
tion about axes Otyi and 0,Zi parallel to the central axes, if Jyt=
= \#z dA is known (Fig. 166). From the definition
A

where y t—y+ b, zt=z-f-a; therefore

I {y + t>){z+a)dA
A

= ^ y z d A + a b [d A + a [y d A -{-b ^ zdA
A A A A

The last two integrals are equal to zero because they represent the
static moments of the area about the central axes, Oy and Oz. Therefore
J 'y z ^ J ^ + a b A ( 12. 8)

The product of inertia of a section about two mutually perpendi­

cular axes parallel to the central axes is equal to the product of iner­
tia of the section about the central axes plus the product of the area
of the figure by the coordinates of its centre of gravity w.r.t. the new
axes.

§ 68. Relation Between the Moments of Inertia

Under Rotation of Axes
We can draw any number of central axes. But the question is: Can
we express the moment of inertia about an arbitrary central axis
in terms of the moment of inertia about one or two definite axes?
248 Bending. Strength of Beams [Part IV

We shall see how the moment of inertia about two mutually perpen­
dicular axes changes when Ihe axes rotate through an angle a.
Let us take an arbitrary figure and draw two mutually perpendi­
cular axes Oy and Qz through its centre of gravity 0 (Fig. 167).
Suppose that the moments of inertia about these axes, J v and J 2t
and the product of inertia of the section, Jy2, are known. Let us draw
the second system of coordinate axes Oyx and 0 zx at an angle a to the
first. This angle will be considered
positive if the rotation of the axes
about point 0 is anticlockwise.
The origin of coordinates 0 is re­
tained. L$t us express the moments
about the second system of coordi­
nate axes, J'y and J'z, through the
known moments Jy and J x.
The expressions for the moments
of inertia about these axes are as
follows:
= 5 z* dA, J^W dA
A i-l
J'u = \ z \ d A , J z = \tM A
(12.9)
It is clear from the drawing that the coordinates of area dA in the
system of rotated axes Oyt and Ozx are
y t = O E+EC = OE - f B D = y cos a -f- z sin a ^
zx= A D — DC = AD — BE = z cosct—//since j (12. JO)

Putting these values of y x and zx in formula (12.9), we get

J y’ — J (zcosa—ysina)*dA
A
= $ (z1 cos*a + y * sin*a—2yzsinacosa)dA
or
y^=-cos*aJ a*/L4-f-sin*a J y*dA—sin2a J yzdA ( 12. 11)
A A A
Similarly
(ycosa + zsin a)3dA
A
OF

/;= » s in * a J zadA + cos4a $ y*dA -f-sin 2 a J yzdA ( 12. 12)

Ch. 12] Moments of Inertia of Plane Figures 249

The first two integrals in expressions (12.11) and (12.12) represent

the axial moments of inertia, J y and J z, whereas the third represents
the product of inertia of section about the two axes, J„z. Therefore
J'u = J u cos* a -j- J t sin* a —J vt sin 2a
(12.13)
yz = Jy sin* a + J gcos* a -f sin 2a 1
To determine the product of inertia of the section we may require
formulas for passing over from one system of coordinates to the other.
For the rotated axes (Fig. 167) we get
= l clA
A

where y x and Zi are calculated according to formula ( 12. 10). Conse­

quently
Jyg= £ (2 $ in a + 0 cosa) (eco sa—y sin a) dA
A
*s=sinacos a ^ z*dA—sin a cos a ^ yt dA
A A
+ cos* a ^ y z d A —sin* a J yz dA
After simplification we get

J W = T («V— J z) sin 2 a -j- J vt cos 2 a (12.14)

Thus, in order to determine the moment of inertia about an arbit­

rary central axis Oyx, we must know the moments of inertia J 0 and J z
about a system of two mutually perpendicular central axes Oy and Oz,
the product of inertia of the section, J az, about the same axes, and the
angle between axes Oyx and Oy.
To calculate J v, J z, and J,fZ we must select the axes Oy and Oz
and break the area of the figure into parts in such a way that the above
values may be computed for each composite part by using the rule of
parallel axes only. We shall show in the example below how to do
this in practice. It should be noted that complex figures should be
broken into elementary areas for which the central moments of iner­
tia about a system of two perpendicular axes are known.
Lei us note that the above procedure and the final results (12.13)
and (12.14) would have been the same if we had taken the centre of
coordinates in an arbitrary point 0 other than the centre of gravity
of the section. Hence, formulas (12.13) and (12.14) hold true when
we transfer from one system of mutually perpendicular axes to an­
other rotated through an angle a , irrespective of whether the axes
pass through the centre of gravity or not.
250 Bending. Strength of Beams [Part IV

From formula (12.13) we may obtain another relation between the

moments of inertia when the axes are rotated. By adding the expres­
sions for J v' and J t' (12.13), we get
+ J'z — Jy (cos* a + sin* «) + h (sin* a + cos* a) = /„ + J t (12.15)
i.e. the sum of the moments of inertia about any two mutually per­
pendicular axes Oy and Oz does not change when the axes rotate. Put­
ting the values of J v and J z from (12.9) in formula (12.15), we get
$ z*dA + \y*dA = \{z*+ y*)dA = [ p ‘ dA = J p (12.16)
A A A A

where p—V y*+z* is the distance of elementary area dA from point 0.

As we already know, the quantity P*dA is called the polar mo­
ment of inertia about point 0 (§48).
The polar moment of inertia of a section about a point is equal to
the sum of the axial moments of inertia about two mutually perpen­
dicular axes passing through this point. This explains why this sum
remains constant when the axes rotate. Expression (12.16) may be
utilized for simplifying the computation of the moment of inertia.
Thus, for a circle we already have (§ 48)
nr*
~ 2

Due to symmetry in a circle / v= / „ therefore

which is the same as obtained by integration (§ 65).

Similarly, on the basis of formula (9.15) we get the following ex­
pression for a thin-walled ring section:

§ 69. Principal Axes of Inertia and Principal

Moments of Inertia
Formulas (12.7) and (12.13) solve the problem set before us in § 66:
knowing the central moments of inertia J,, and J z, and for a par­
ticular figure we can calculate its moment of inertia about any other
axis.
As the basic system of axes we select a system which will help
simplify formulas (12.13). To be precise, we may select a system for
which the product of inertia of the section is zero. Indeed, moments
Ch. 12] Moments of Inertia of Piane figures 251

of inertia Jy and J t are always positive because they are the sum of
positive terms; the product of inertia of the section
= l zydA
A

may be positive or negative, because the terms zy dA may have differ­

ent signs depending upon the signs of y and z for particular elemen­
tary areas. This means that it may
also be zero (§ 63, item C).
The axes about which the product
of inertia of section is zero are
called the principal axes of inertia. If
the centre of this system of axes
lies at the centre of gravity of the
figure, then they are called the
principal central axes of inertia. We
shall denote these axes by Oy0 and
Ozo', for these axes
y***= 0
Let us determine the angle Oo
between the principal axes and the
central axes Oy and Oz (Fig. 168).
In formula (12.14) for the product
of inertia, where we pass over from axes yOz to y\Ozu angle a is re­
placed by a«; then axes Oyi and 0 zt coincide with the principal axes,
and the product of inertia of the section vanishes:
at a = a n J yz —J 0U«aZ
Z»*—0
or
J =• — 4 sin 2a0 (Jx~~ J u) + J yt cos 2a,,= 0

wherefrom
tan2a0= 7^2~ - (12.17)
Jz Jy

This equation is satisfied by two values of 2a 9 differing by 180°,

or two values of a 0 differing by 90°. Thus, equation (12.17) determines
the location of two axes at right angles to each other. These are the
principal central axes of inertia Oyt and Oz0 for which Jy^ 0 = 0.
Using formula (12.17) and knowing J y, J Zi and J yz we may°obtain
formulas for the principal moments of inertia Jya and JZa. For this
we shall again use formulas (12.13); they give us the values of JVt
262 Beading, Strength of Beams IPart IV

and JZt if we replace a by cc0:

J y, = J ucos*a„ + J , sin*a0—./ff2 sin 2a0 1
(12.18)
JZv = Jvsin1a d+ cos1a 0 -\-Jyzsin2afl j
These formulas along with formulas (12.17) may be used in solving
problems. We shall show in §70 that one of the principal moments
of inertia is Jmi. and the other is Jmla.
Formulas (12.18) can be modified into a form which does not con­
tain a 0. Expressing cos1^ and sin*a0 in terms of cos 2a0, putting their
values in the first formula in (12.18) and simultaneously substituting
the value of from formula (12.17), we get

” 2 ' 2 COS/<Xo + 2 cos2a,

, Jp—Jt 1
— 2 "» 2 cos 2a ,

From formula (12.17), replacing the fraction

SSZ b* ± / l + ta-2a0= ± / l + J £ ^ ,
we get
+ (12.18')
mhi •

We would have obtained the same result by a similar transformation

of the second formula in (12.18).
Instead of Otj and Oz we may lake the principal axes Oy0 and 0t„
as the basic system of the central axes of inertia from which we can
pass over to any other system. Then the product of inertia of the
section will not appear in formulas of the type (12.13) (Jyotf>= 0).
Let us denote by p the angle that axis Oyx makes with the principal
axis Oy0 (Fig. 169). In calculating J'y, J ’z , and Jyt, angle a in for­
mulas (12.13) and (12.14) should be replaced by |5 ana J„, J t, and
Ju. should be replaced by Jy<>t JSo, and We find that
Jy = cos1p-M*o sin1 p 1
(12.19)
J yz sin 2p

The above formulas are exactly the same as the formulas for normal
stresses <ra and shearing stresses x0 (6.5) and (6.6) acting in two mu­
tually perpendicular planes in an element subjected to tension in
two directions (§ 30). Therefore Mohr’s circle can be used in this case
also. The axial moment of inertia should be laid off along the hori-
Ch. 12\ Moments of Inertia of Plane Figures 253

zontal axis, and the product of inertia of the section along vertical
axis. It is proposed that the reader should himself plot and analyze
the Mohr’s circle in this case. We shall only give the formula which
enables us to select that value among the two values of a# (formula
(12.17)), which corresponds to deviation of the first principal axis
(giving maximum J) from the original position of the ^-axis:
^ t<i
la n a 0 = (12.17')

This formula is exactly similar to formula (6.11).

We can finally state the law by which the moment of inertia of a
complex figure about an arbitrary axis can be found in the simplest
possible manner. It is essential to draw axes Oy and Oz through the
centre of gravity of the figure so that to divide the figure in simple
parts for which J t and J VI can be easily calculated. Then we de­
termine <xo from formula (12.17) and calculate the principal central
moments of inertia J Vo and J ^ according to formulas (12.18).
We can calculate the moment of
inertia about an arbitrary central
axis Oyi (Fig. 169), inclined at an
angle P to Oy0> according to formu­
la (12.19):
Jy = J y, cos1p + sin1p
Knowing the central moment of
inertia J,/, we can calculate the
moment of inertia about any paral­
lel axis y located at a distance a
(Fig. 169) from the centre of gravity
by formula (12.7):
J ^ J v + a'A
In a number of cases the princi­
pal axes of a figure can be drawn
straightaway. If the figure has an axis of symmetry, then this axis
will be one of the principal axes. Actually, while deriving the for­
mula o = ~ , we came across the integral ^ y zd A , which represents
the product of inertia of the section about the axes Oy and Oz. It
was proved that if Oz is the axis of symmetry, then this integral be­
comes zero.
This implies that in the present case Oy and Oz represent the prin­
cipal central axes of inertia of the section. Hence, the axis of symmetry
is always a principal axis, and the second principal central axis passes
through the centre of gravity at right angles to the axis of symmetry.
 254 Bending. Strength of Beams [Part IV

§ 70. The Maximum and Minimum Values

of the Central Moments of Inertia
We already know that the central moments of inertia are the mi­
nimum of all moments about a number of parallel axes.
Let us now determine the extreme (maximum and minimum)
values of the central moments of inertia. If we start rotating axis
Ot/i, i.e. changing the value of a, there is the change in the value of
Jv = Ju cos*a + J* sin2a — Juz sin 2a
The maximum and minimum values of this moment of inertia cor­
respond to angle at for which dJf/da vanishes. This derivative is
dJ *
-3 * = — cosa sin a 4-2 sin a cosa—2Jyz cos 2a
Putting a = a , in the above equation and equating it to zero, we get
(Jg— Jy) sin 20*— 2 JyZcos 2at = 0
wherefrom
2j
tan2a1 = -?-:g - « t a n 2 a B (see (12.17))
Thus, the axes about which the central moments of inertia are maxi­
mum or minimum are the principal central axes of inertia. When
these axes are rotated, the sum of the corresponding moments of iner­
tia does not change; therefore

When one of the central moments of inertia is maximum, the other

must be minimum, i.e. if
«fy* = «/jnax> th en =
Thus, the principal central axes of inertia are mutually perpendicu­
lar axes passing through the centre of gravity of the section, about
which the product of inertia of the section is zero and the axial mo­
ments of inertia have the maximum and minimum values.
In future we shall denote the principal axes of inertia by Oy and Oz
and the principal moments of inertia by J v and J z. We shall continue
to denote the axis of the beam by the x-axis as before.
§ 71. Application of the Formula for Determining
Normal Stresses to Beams of Non-symmetrical
Sections
By equating to zero the product of inertia of section about the prin­
cipal axes, we can show that the formulas given in § 63 are valid under
certain conditions for non-symmetrical sections as well.
Ch. 12} Moments of Inertia of Plane Figures 255

While deriving the formula for normal stresses (§ 63) we introduced

the limitation that the beam should be symmetrical about the plane
of action of the external forces, xOz, with the primary aim of (1) es­
tablishing that the neutral axis Off is perpendicular to the plane zOx,
and (2) proving that the sum of the moments of elementary forces
dN about the axis Oz is zero:
Mt = 0, -^^xydA = > 0t J zydA —Q (11.6)
A A
However, the conditions that the z• and «/-axes be at right angles
and the integral \z y d A be equal to zero may also be fulfilled for
A
a non-symmetrical section. For this it is sufficient that the z-axis
lying in the plane of action of the external forces and the neutral
y-axis be the principal central axes of inertia of the beam’s cross sec­
tion. The perpendicularity is then satisfied and the integral J zy dA
representing the product of inertia of the section about the principal
axes is also equal to zero.
Hence, the condition that the plane of action of the external forces
should coincide with the plane of symmetry may be replaced by an­
other condition: the plane of action of the external forces should coin­
cide with one of the two planes containing the principal axes of iner­
tia of the cross section. In a beam
these two planes are called the
principal planes of inertia.
The second principal axis, which
is perpendicular to the plane of
action of the external forces, repre­
sents the neutral axis, and the con­
dition ^ zy d A = 0 is automatically
satisfied.
Since we can always find the
principal central axes of inertia for
a beam of any shape, formulas
(11.9) and (11.13)
un — ^ jL and a ,x — ^
ana om Fig. 170

may be used for beams of any cross section, provided the external
forces lie in one of the principal planes of inertia of the beam and J
and W are taken about the other principal axis, which is perpendi­
cular to the plane of action of the external forces and represents the
neutral axis.
256 Bending. Strength of Beams [Part /V

A Z-shaped beam (Fig. 170) with principal axes Oz and Oy may be

taken as an example. The formulas given above are applicable to this
beam only if the external forces lie in plane xOz or xOy. In the first
case the ^-axis will be neutral, in the second the 2 -axis. As in this
case too the neutral axes of the section are perpendicular to the plane
of action of the external forces, the axis of the beam remains in this
plane even after deformation. Thus, if the external forces lie in one
of the principal planes of inertia, this will be the general case of uni-
planar bending.
Beams of Z-section are often employed as purlins, which are laid
over rusters. Under vertical pressure of the roof weight and snow the
purlins bend in the plane of action of the externa! forces (for corres­
ponding roof slope), i.e. they experience uni-planar bending.
It should be noted that in some cases additional normal and shear­
ing stresses connected with extra twisting of beam are set up in beams
of non-symmetrical (about the axis lying in the plane of action of
external forces) sections.

§ 72. Radii of Inertia.

Concept of the Momental Ellipse
We shall now introduce one more geometrical characteristic of sec­
tion which correlates the moment of inertia of the section, J, with
its area A by the following formulas:
Jv = i3vA and Js = i\A(12.20)
Quantities i„ and iz are known as radii of inertia and are respec­
tively equal to

and (i2.2i)

If J„ and represent the principal moments of inertia, iy and /,

are known as the principal radii of inertia. For example, for a rectan­
gular section we find with the help of formulas (12.1) and (12.3)

^ T m lg~ l2Wi = y H (12.22)

For a circular section formula (12.4) yields

<l2-23>
The values of principal radii of inertia for rolled profiles are given
in standard normal profile tables (see Appendix).
Ch. 12] Moments of Inertia of Plane Figures 257

The ellipse plotted on the principal radii of inertia as its major and
minor axes is known as momental ellipse. To plot the momenta! el­
lipse we lay off from the centre of gravity of the section the radii of
inertia: iy is normal to the central
y-axis, i.e. along the z-axis, and iz
is normal to the z-axis (along the
y-axis). If J v= J max, the major axis
of the ellipse 2L will lie along the
z-axis (Fig. 171).
The momenta] ellipse has the fol­
lowing remarkable property: the
radius of inertia about an arbitrary
axis Ox drawn through the centre
of gravity of the section is equal to
the normal dropped from the centre
of ellipse to the tangent parallel to
the above axis. Hence, with the
help of the momental ellipse we
can graphically find the radius of
inertia ix for an arbitrary axis Ox
making an angle P with the princi­
pal axis Oy. For this it is sufficient to draw a tangent to the ellipse
parallel to the x-axis and measure distance ix between the axis and
tangent (Fig. 171). Knowing the measured radius of inertia ix, we
calculate the moment of inertia about the x-axis by formula
Jx =i$A (12.24)

Some sections like circle, square, etc. (Fig. 172), which are com­
monly used in engineering practice, have equal moments of inertia
about the two principal axes of inertia. Consequently, the principal

Fig. 172

radii of inertia are also equal (iu—i*), and the momental ellipse chan­
ges to the momental circle. For such sections every central axis rep­
resents a principal central axis of inertia; this is also evident from
formula (12.19) for the product of inertia of section, which vanishes
for every value of (5 if J v—Jz (see § 69).
258 Bending. Strength of Beams [Part tV

A bar of such a section displays equal resistance to bending in all

directions, which is particularly important in axial compression of
long bars (Chapter 27).

§ 73. Strength Check, Choice of Section and Deter­

mination of Permissible Load in Bending

Formulas (11.15), (11.16), and (11.17) derived in Chapter 11 for

expressing the strength condition in bending, combined with the abil­
ity to calculate the moments of inertia and section moduli (Chap­
ter 12), enable us to solve the fun­
damental problems of strength of
materials in bending (§4), name­
ly:
(a) check the beam strength when
the beam dimensions and the forces
acting on it are known;
h- 1 0 0 (b) determine the required cross-
sectional dimensions if the axial
dimensions of the beam and the
forces acting on it are known;
(c) determine the permissible load
which the beam can withstand if
its axial and cross-sectional dimen­
sions are known.
Fig. 173 It is assumed that the permissible
stresses are known in all the above
cases.
We shall illustrate with examples how to apply the strength con­
ditions for solving the above problems.
A. Suppose it is required to check the strength of a rectangular 60X
------steel
X 100 mm ■ bar weakened by two symmetrical holes o? diameter>fd
d=10 mm (Fig. 173), if the bending moment in the critical section
Mmax^l-S tf-m and the permissible stress [o]=1600 kgf/cm2.
Let us calculate the moment of inertia of the section about neutral
axis Oy:

J , - 2 ( M fl* + £ ) = ^ - 2 («X I X3» + ^ ) = 3 9 1 cm*

The section modulus is

78.2 cm3
Ch. 12] Moments of Inertia of Plane Figures 269

The maximum stress in critical section is

° » . X - % 2= l37| r~ = 1660 kgf/cm* > 1600 kgf/cm*

Overstressing of — X 100—3.75% is permissible.

B. Let us now solve a problem on selecting a section. Let a hinged
I-beam of span 1=4 m be loaded along its length by a uniformly
distributed force q = 2 tf/m and a concentrated force P = 6 tf at the
centre of the span. It is required to select a section for the beam if
the permissible stress kl=1600 kgf/cma.
The maximum bending moment at the middle of the span can be
calculated by applying the principle of superposition of forces (§ 61).
Bending moment due to the distributed load q according to for­
mula (10.11) is

Bending moment due to concentrated force P (10.100 is

The maximum total bending moment in the critical section is

Mm x==M4-f Af^=4 + 6 = 10tf-m>= 10 x lO'kgf-cm
According to strength condition (11.16) the required section modulus
=625 cm3

From standard tables (see Appendix) we find the profile No. which
satisfies this condition: I-section No. 33 having section modulus
1F=597 cm3 (overstressing of about 5% is permissible).
C. Let us now consider an example in which we have to determine
the permissible uniformly distributed load which may be safely ap­
plied to a hinged jib beam of span /= 10 m. The beam has 1-section
No. 60 strengthened by two 200 x 20 mm plates welded to it (Fig. 174).
Permissible stress loj=!400 kgf/cm8.
The moment of inertia of the section will be found as the sum of
moments of inertia of the I-section and the two plates about the neut­
ral axis (as per standards) and it will be calculated with the help
of formula for moment of inertia about parallel axes (12.7):

76 806 + 2 [ ■ ^ ^ ■ + 2 0 x 2 x ( 3 0 + 1)4] « 153700 cm4

200 Bending, Strength of Beams [Part IV

Section modulus
4820 cm’
2 ro»x

According to strength condition (11.16) the maximum bending mo­

ment yWm-x= must not exceed [a] IF, wherefrom
8X H00 X 4840-. ,r. __
1<7J ^ ir *■— i^ ]< loo4" " * 54 kg*/crn
or 5400 kgf/m.
If the dead weight of the beam is taken into account, then we should
subtract the distributed load due to the weight of two plates, 2 x

X20X 2X 100 x 0.007 85=63 kgf/m, and the weight of the I-beam
(see Appendix) 108 kgf/m (</#=63+108=171 kgf/m in all).
Hence, the beam may be loaded by a service load
<?= [<?]—ft “ 5400— 171 » 5230 kgf/m
D. Finally, we shall now discuss the analysis of a composite beam of
non-symmetrical section. Suppose it is required to determine the per­
missible bending moment for a beam fixed rigidly at one end in a
wall, if a force couple is applied at the other end in the principal
plane of inertia. The dimensions of the section are given in Fig. 175.
The span of the beam is 7=0.6 m. The permissible stress is [ol=
= 1600 kgf/cma.
First of all it is necessary to locate the centre of gravity of the sec­
tion. For this we select an arbitrary system of coordinate axes yiOzi.
Ch. 12\ Moments of Inertia of Plane Figures 261

It is convenient if the whole figure lies in the first quadrant. The dis­
tances of the centre of gravity from these axes may be determined by
the formulas
-f- and

where and S2| are static moments of the area about axes Oyt
and 0 zL, respectively.
To determine the static moments we divide the area into two rec­
tangles, vertical / and horizontal / / . The area of the figure is A =
= 1 x 1247x1= 1 9 .0 cm3. The static moments* are
Sty = ^ 101,1 4 - 1 2 x 0.54-7(1 + 3.5) = 37.5 cm3
Syi = AzZu t e= 12 x 6.0 + 7 x 11.5= 152.5cm8
The coordinates of the centre of gravity are
yc = 1.97 cm » 2.0 cm

2f= ! ^ « 8 . 0 0 c m

Let us now choose the coordinate system of the central axes of iner­
tia Oy0 and Oz<>. The simplest way is to direct these axes parallel to
the arms of the figure; this will be helpful in calculating the moments
of inertia of the section about these axes.
The moments of inertia of individual rectangles about axes Oy0
and Ozo can be calculated from the formulas of parallel axes 112.7)
and ( 12.8 ), and the moments of inertia of the rectangles about their
own axes from formula ( 12 . 1).
Table 12 (see Fig. 175) contains the plan of computations with the
help of which we can determine the angles between the principal
axes and axis Oy9:
tan!&.r-= 2x97 *<jn
lan z/Xq yo_ ju — 100—278 ” 1*oy
2a; = —47*40' and a ;= -2 3 ° 5 0 '

The minus sign shows that angle should be laid off In the clock­
wise direction:
sinaj = —0.404, cosa;=0.915, sin2a; = —0.74, cos2a; = 0.673

• In the indices of the coordinates the first subscript denotes the axes tty, or
(fej, and the second subscript denotes ttie area.
262 Bending. Strength of Beams [Pari IV

Table 12
Determination of the Moment of Inertia

CbArdinafaf * Moments of inertia of the areas (cm4)

of the areas
so. (cm)
I 4 Azl TT +A«\ V* *=Ayyt,
o J«9
w
OJ wO
.0 id bh> At* fcft» J°V*
E
9 e *# IT 4 IT Aul ±
2 <c

l 12 -1.5 -2.0 144 48 192 1.0 27 28 + 36

2 7 2.5 3.5 0.6 85.6 86 28.6 43.8 72 + 61

2 19 144.6 133.6 278 29.6 70.8 too + 97

The principal moments of inertia are

J y = Jy cos4a„ + J\ sin* (X*—7 ^ sin 2a 0
= 278 x 0.915*+ 100 x 0.404*+ 97 x 0.74 = 320 cm*
J x — 7I sin* oc0+ J%cos* a 0+ 7 ^ sin 2a0
= 278 x 0.404*+ 100 x 0.915*—97 x 0.74 = 58 cm*
Let us check whether the results obtained are correct:
1. 7„ + 7, = 320 + 58 = 378cm‘ = 7 2 + / “= 2 7 8 + 1 0 0 = 378cm‘
2. 7yr= y ( 7 “—7®)sin2a+722COs2a

= — *5*(278— 100) 0.74 + 97 x 0.673 = 0

It is clear from the calculations that 7 „= 7m,x, and 7*=7min.
Hence, it is advantageous to apply the bending couple in the plane
xOz so that axis Oy becomes the neutral axis (Fig. 176).
Let us now find the section modulus. For this it is necessary to
determine the distance of the farthest fibre from the neutral
axis Oy. This can easily be done by drawing the section to a certain
scale and marking the principal axes * on it. For the section under
consideration the measured distance was found to be zmiu=8.1 cm.
Therefore the section modulus about the i/-axis is
lFtf= A = f ? = 39.5cm»
Zfliax
Formulas (12.10) may be used for analytical determination of 2max or ym n '
ck. m Shearing and Principal Stresses 263

We determine the maximum permissible bending moment from the

strength condition *:

wv
wherefrom
max [M ]< [o ] Wy
M0= max [Af] < 1600 x 39.5 = 63 200 kgf *cm « 0.63 tf •m
If the bending moment is applied in plane xOy, then the distance
of the outer fibre from the axis Oz being yma,t=4.12 cm, the section

Fig. 176

J 58
modulus will be Wz= ==14-1 cn,3«and the magnitude of the
moment which can be applied safely will be
A f?=m ax[M ,]<1600 x 14.1 = 2 2 560kgf•cm « 0 .2 2 6 tf-m
which is three times less than the moment which can be applied in
plane xOz.
Let us note that if the moment is located in a plane other than the
principal plane, for example, parallel to the flange of the angle sec­
tion, then the bending of the beam will not be uni-planar, and the
strength condition will be different (§ 120).

C H A P T E R 13

Shearing and Principal Stresses in Beams

§ 74. Shearing Stresses in a Beam of Rectangular
Section
Let us try to determine, first of all, the shearing stresses in sections
perpendicular to the beam axis when the sections are rectangles
(Fig. 177).

* in this example we do not account for the additional normal stresses which
appear due to restrained torsion.
264 Bending. Strength of Beams \Part IV

Suppose a positive shearing force Q acts in a section when the beam

is subjected to bending. Let us make the following assumptions re­
garding shearing stresses x in this section:
(I) alt shearing stresses in the section act parallel to the shearing
force Q, which is the resultant of the former;

(2) shearing stresses acting in planes which are located at the same
distance z from the neutral axis are equal in magnitude.
Both these assumptions were put forward by D. I. Zhuravskii.
The theory of elasticity reveals that the assumptions are valid for
rectangular beams if the height of the beam is greater than its width.

We shall now try to calculate the shearing stresses and ascertain

the law of distribution of shearing stresses along the height of section.
Let us consider a beam loaded by a number of forces (Fig. 178).
Let us isolate a part of length dx cut out by sections /-/ and 2-2. It
will be assumed that section 2 - 2 on the right side of the cutout portion
experiences shearing stresses x, which give resultant shearing force Q
acting downwards, then on the other side section /-/ will experience
shearing stresses acting upwards, which also give a resultant shearing
Ch. 13J Shearing and Principal Stresses 2G5

force Q. It is quite natural that in the absence of distributed load on

the isolated part of the beam, the shearing forces should be equal in
magnitude. Sections /-/ and 2 - 2 will also experience normal stresses
which, however, are not shown here.
According to the law of complementary shearing stresses (Chap*
ter 6, formula (6.8)), similar shearing stresses should be expected to
act in planes parallel to the neutral layer. Therefore, if we take two
horizontal sections of the beam at dis­
tances z and z+dz front the neutral axis
and isolate an element of sides b, dx,
and dz (Fig. 179), then the vertical faces
of this element will experience shear­
ing stresses x, whereas the horizon­
tal faces will be acted upon by equal
but opposite shearing stresses t \
As the fibres parallel to the axis of
the beam do not press on each other
in the process of deformation, the sec­
tions of the beam parallel to the neut­
ral layer do not experience any normal
stresses. Therefore, instead of determin­
ing the shearing stresses x over the
beam cross section, we shall determine
equal stresses t ' acting in a plane pa­
rallel to the neutral layer (Fig. 180).
At first it seems strange that shear­
ing stresses appear in planes paral­
lel to the neutral layer. However, we
can explain this phenomenon with the
following example.
Let us suppose that the beam con­
sists of two identical rectangular rods
placed over one another (Fig. 181(a));
the friction between the rods may oe
ignored, it is assumed that the beam
bends under the action of at least one force P acting in the middle or
the span. The bent beam is shown in Fig. 181(6) in a highly magni­
fied scale. The lower fibres of the upper beam AiBy stretch, where­
as the upper fibres of the lower beam A 2 B2 shorten as compared
to their initial length AB.
If the beam were a single rod, it would have bent as shown in Fig.
181(c). Fibres AB would be in the neutral layer and would not have
changed their lengths. Therefore in bending of a solid beam, shearing
stresses r / preventing the upper and lower halves of the beam from
shear along the neutral layer are transmitted from the upper half to
the lower through the neutral layer, and vice versa (Fig. 181(d)).
266 Bending. Strength of Beam [Part IV

Figure 182 shows a part of the facade of a rectangular beam sub­

jected to uni-planar bending. Let us draw two very close sections
/-/ and 2 - 2 at a distance dx from each other. Let us also draw a hori­
zontal section at a distance z from the neutral layer.
Thus, we shall be able to isolate from the beam an element ABCD
having sides dx, h/2 —z, and b. An axonometric projection of the ele­
ment is shown in Fig. 182. Let M be the bending moment in section

Fig. 182

I ‘I, and M+dM in tf|e adjacent section 2-2. The side faces of the
element will be acted upon by normal stresses a which are lower to
the left and greater to the right. The horizontal section will experience
shearing stresses x'=x.
We have not shown in the diagram the shearing stresses x acting
in sections 1 - 1 and 2 - 2 because they do not enter the condition of
equilibrium of the isolated element, which is obtained by equating
to zero the sum of the projections of ail the forces on the axis of the
beam.
To obtain the condition of equilibrium of the isolated element, we
must calculate all those forces acting on it which are parallel to the
axis of the beam. The elementary shearing force dT on the elementary
area bdx is
dT = xbdx
The normal stresses acting on an infinitely small area dA of the side
face at a height Zi from the neutral axis are
Me.

The force dNt acting on this area is

d N ,= ^ - d A
Ch. 13] Shearing and Principal Stresses 267

The whole of the side face AD is acted upon by a force Nx (Fig. 183):

At Ai

Integral $ zxdA is the static moment about the neutral y-axis of

the part of section GFAD enclosed between the section at a height h
and the edge of the beam (Fig. 184). Let us denote it by S£. Thus,
a; msV /io n
Ni ~ - j f " 03.1)
Identically face BC of the element is acted upon by a force
(M +dM )S%
AV (13.2)

The difference of the normal forces

dM S«
Nt - N x

when projected on axis Ox (Fig. 183) is balanced by the shearing force

dT. Therefore

But therefore
(13.3)

implying thereby that this formula represents the shearing stress at

height z in a section perpendicular to the axis of the beam.

t
k ____ k h

* *
^ __ "d*
V C c n
Fig. 183 Fig. 184

Let us derive the formula for Sj} for a rectangular beam (Fig. 184)
of height h and width b. The static moment of area GFAD about axis
OiOt is equal to the area multiplied by distance zh of its centre of gra-
268 Bending. Strength of Beams IPart IV

vity from axis OiOi. The area of GFAD is equal to

and distance z* is

Hence
SS= K t - z) t ( t + j)= T 1( 1-|? -) 03.4>
While computing the static moment of the area of a section it is
immaterial whether we take the portion of section which is below the
level 2 or the bigger portion, because both the static moments are

Fig. 185

equal in magnitude. Generally, we take the static moment of the

portion which is easier to compute. Since for a rectangle *///= -jy»
formula (13.3) lakes the form
Q W i2 / 42* \3 Q / . 4 z» \
M F¥\ h* 2WT\ ~Kr ) *13‘5)
Hence, shearing stress t changes along the height of the rectangular
section according to a parabolic law. The shearing stress vanishes at
the lower and upper ends of the. section where z = ± ; this is in strict
conformity with the law of complementary shearing stresses. It at­
tains maximum value on the neutral axis (where the normal stress is
zero) where z = 0, and in the section where Q(A)=Qm, x:
r —- ( 1 3 .G )
,na* — 2 ~ W
at, m Shearing and Principal Stresses

Thus, the maximum shearing stress in a rectangular section is 1.5

times greater than its average value. Figure 185 shows the distribution
of shearing stresses when the shearing force is positive.
Shearing stresses somewhat distort the accepted picture of defor­
mation of a beam. We had assumed that under the action of bending
moments the cross sections of a beam turn w.r.t. each other, although
they continue to remain planes (Fig. 186(a)). Due to shearing stresses
the elements of the material enclosed between two sections warp.

nr n?

r
i

m * B< *
Fig. 186

In accordance with the variation in the value of the shearing stress,

the warping increases from the edges of the beam towards the neutral
axis. Therefore the sections are deformed (Fig. 186(6)). However,
warping has almost no effect on the deformation of the fibres along
the beam, therefore formula cr= ^ can be used even if a shearing
force is acting on the beam.
Thus, in addition to the strength check for maximum normal stres­
ses (11.15)
a*., = -> * ■ < l«J

we must check the strength of the material for maximum shearing

stresses
T „ „ = 2 5 ^ < [tJ (13.7)

We shall solve a numerical example to get an idea of the order of

the magnitude of x in rectangular beams.
Let us determine the maximum normal and shearing stresses for
a rectangular beam with the following data: the beam lies on two
supports and over its total length 1 = 4 m takes a uniform load of in­
tensity <7=1.2 If/in; Afmax—2.4 tf-m; <?max=2.4 tf; /i=27 cm; 6=
270 Bending. Strength of Beams [Part IV

—18 cm; [a]=110 kgf/cm*; lr]=22 kgf/cm*.

oOTtx=» —109-5 kgf/cm* < 110 kgf/cm*

W = ^ = s ! i f - T 3 = 7 . 5 k g f / c m ’ < 22kgf/cm>
We see that a rectangular beam designed to take the maximum nor­
mal stress equal to the permissible remains highly understressed as
far as the shearing stresses are concerned.
However, in practice we may come across just the reverse case; it
may occur when the shearing force is large whereas the bending mo­
ment is small. In such cases of loading, even in a rectangular section
the decisive part in determining the dimensions of the beam is played
by the shearing stresses.
The formula for shearing stresses in a rectangular section was first
derived by the Russian engineer D. I. Zhuravskii when he was de­
signing wooden bridges for the St. Petersburg-Moscow railway line
in 1885. Zhuravskii employed a slightly different and more compli­
cated method in obtaining this formula without using the relation
AM

§ 75. Shearing Stresses In I-beams

As the sections of I- and T-beams may be considered as consisting
of rectangles, then with a certain degree of approximation the for­
mulas derived for rectangular sections in § 74 may be applied to these
sections too. The shearing stresses in a point at a distance z from the
neutral axis may, for an I-section (Fig. 187), be expressed by the
same formula
x = ^ L n s si
Here Sf is the static moment of the area enclosed between level z
and the edge of the beam about the neutral y-axis. As for the quantity
b(z), the width of the section, it has been written as a function of z
to emphasize that in the denominator of formula (13.3) the width
at level z should be used. If we examine the derivation of formula
(13.3), we see that b is the multiplier in the term t b dx, i.e. it is the
lateral dimension of the area which is being acted upon by the stress x \
Thus b is the width of the beam at level z. Therefore, when applying
formula (13.3) to an I-section for calculating the shearing stresses in
web sections, instead of b(z) web thickness bw should be used. Static
moment may be computed as the sum of static moments of the two
rectangles hatched in Fig. 187(a). Upon computation we get
*w
Ch, 13] Shearing and Principal Stresses 271

It is evident from the formula that along the web height the shear­
ing stresses vary by a parabolic law (Fig. 187(6)) and become maxi­
mum on the neutral axis of the section.
Formula (13.3) cannot be used for calculating shearing stresses in
the portions lying in the flanges of the I-section, because these stresses
are far from being equal along the flange width. In the area around
the z-axis they may be assumed to vary approximately as shown by
dotted lines in Fig. 187(6). However, in the remaining area of the

Fig. 187 Fig. 188

flange, i.e. along almost the whole of flange width, they vary as shown
in Fig. 188 and do not achieve large magnitudes due to the conditions
on the flange surface and the law of complementary shearing stresses.
Knowing now the laws of distribution of normal and shearing
stresses along the height of I-section, we can draw the following con­
clusion about the working of an I-section.
The flanges of an I-section, being located at a considerable distance
from the neutral axis, experience over their whole area normal stresses
that are maximum or close to maximum. Shearing stresses in the
flanges of an I-section are negligible.
As we move towards the neutral axis, the normal stresses in the
web of the I-section tend to zero. Within web limits the static moment
S£ does not change much for various values of z. Therefore shearing
stresses along the web height are sufficiently large (see the curve in
Fig. 187(6)). In short, it may be summarized that the flanges of an
I-section bear normal stresses, and the web bears shearing stresses.
Let us check the shear strength of a beam acted upon by a shearing
fpree Q=2A If, assuming the permissible shearing stress 1x1=1000
kgf/cma. The section is shown in Fig. 189. From Table I of Appendix
we find 7=1290 cm4. The static moment of half of the section is
81*4 cm3= S 3. For calculating the stresses at point 2 the static
moment can be found by subtracting the static moment of half of
the web from Smax:
$2 = 81.4—0.5x0.51 x8.19* = 81.4— 17.1 = 64.3 cm3
272 Bending. Strength of B eam [Purt IV

The shearing stresses are

tj = 0, t 8= =3 236 kgf/cma

t, « ^ » 297 kgf/cm*

The diagram of distribution of shearing stresses along the I-section

height is shown in Fig. 189. It can be seen from the diagram that the
maximum shearing stress is con­
siderably less than the permis­
sible which may be attributed to
the large thickness of the web in
the rolled profile. Much belter
utilization of metal can be
achieved in composite beams
(see § 80), riveted and welded.
Let us determine that fraction
of the shearing force which is
taken up by the web. For this we
multiply theordinatesof theshear-
ing-stress diagram by the area of
Fig. 189 the web of I-section: 236x0.51 X
X 16.38 -f (297 — 236) X 0.51 X
X l6 .3 8 x y =2312 kgf, which comprises 96% of the total shearing
force.
The method of determining shearing stresses in an I-beam which
has been explained here may also be used for other sections made of
rectangles: hollow rectangular section, T-section, etc.

§ 76. Shearing Stresses in Beams of Circular

and Ring Sections
Let us consider a beam of circular section. In this beam the shear­
ing stresses can no more be parallel to the shearing force. If there
are no forces acting on the side surface of the beam, the shearing
stresses on elementary areas t and 2 in the vicinity of section contour
must act along the tangent to the section contour (Fig. 190(a)). These
tangents will intersect the line of action of the shearing force at
point C. Since shearing force <? is the resultant of shearing stresses
(Fig. 190), the shearing stresses on arbitrary elementary areas 3 and 4
at the same distance z from the horizontal diameter act along the line
passing through the same point C. Each of these shearing stresses x
may be broken into two components: vertical x* and horizontal xt.
The horizontal components in the left and right halves of the section
balance each other, whereas the vertical components add up into
Ck. 131 Shearing and Principal Stresses 273

shearing force Q. Hence, in round beams vertical stress components

play the same role as the total stresses t in rectangular beams.
We can thus apply formula (13.3) to round sections too, but it will

Fig 190

give us only the vertical component of shearing stress at an arbitrary

point. In subsequent discussion we shall write x instead of

Here, as in the previous case, S® is the static moment of the area

between the edge of the section and level z and is expressed by the
formula
Sy ~ $ z\ dA = J zxb fo) dzx
A A

It is more convenient to introduce a new variable, angle q>,j, in

computing the static moment; if r is the radius of the section, then
z — r sin <p„ a, = r sin <pzl, b fo) = 2r cos
dzx= r cos<pzl d<pzl, b (a) = 2rcos<p*
We shall limit ourselves to determining x^,*:

T“a* (13.7)
n/2
^rnax* $ 2r cos q>zlr sin cpzlr cos <pzl dyti
o
2r*
T (13.9)
tO—3310
274 Bending. Strength of Beams [Part IV

Since J= and bz=0 = d = 2 r, we get

Q X 2ra X 4 4Q
3 X 2rnr* = 3 n rI
Thus, for a circular section
- .1 Q (13.10)
max 3 nr*

i.e. Tmax 's *-33 times greater than the mean value of r.
Even in rectangular sections, where Tmax is i.5 times greater than
the mean value, check for shear strength is often not required and this
is all the more so for circular sections. It should, however, be noted
that shearing stresses may be of a considerably higher magnitude in
pipe-section beams.
Example. Find the maximum shearing stress in an iron pipe of
external diameter d=*10 cm and wall thickness f = l cm; Qn>ax= 2 tf.
Maximum shearing stress occurs in points of the neutral layer and
is expressed by the formula
0 .max*-1max
"^aiax Jyb (13.7)

here J y is the moment of inertia of the pipe section; Sma!{ is the static
moment of the semicircular ring, b—'2 t is the double thickness of the
pipe wall.

'» = “ " V ( ' + - £ r ) « " 'l l (1S.U)

where r9 is the mean pipe radius.
The static moment of a semicircular ring is equal to the difference
of the static moments about the diameter of the inner and outer semi­
circles; the static moment of a semicircle is expressed by the formula
S (r)= ^~ (13.9)

The required static moment of the semicircular ring is

[ (r . + 4 - ) * - ( r , — f )*] = 2 rV [l + ^ > ] « 2 r * ( (13.12)

Therefore
Qx2r\t Q 20 _ 2 X 2000
nr ft n X9x 1
= 141.4 kgf/cm9
2/ x Jirit
The maximum shearing stress in a semicircular ring is twice the
mean stress. Let us recapitulate that this ratio is 1.5 for a rectangular
section and 1.33 for a solid circular section.
ch. m Shearing and Principal Stresses 275

§ 77. Strength Check for Principal Stresses

In the previous discussion we worked out two criteria for checking
the strength of beams under bending under normal stresses (11.15)
and shearing stresses (13.7):
(11.15)

(13.7)

We shall consider the elements of beams whose strength may be

checked by these formulas.
Figure 191 shows a part of the front view of the beam being analyzed
in the sections of maximum bending moment and maximum shearing
force. The diagram shows the elements whose strength is checked by
&mox

Fig. 191 Fig. 192

conditions (11.15) and (13.7). The first formula is used for elements
located near the top and bottom edges of the section with Afmax. These
elements are subjected to simple tension or compression. The second
condition, (13.7), applies to an element located near the neutral axis
with Qmax; this element experiences pure shear.
Thus, when checking the strength of the beam under normal and
shearing stresses according to the universally accepted method of
stress analysis, we actually check the strength of material in three
elements shown in Fig. 191.
Generally speaking, it cannot be said with certainty that these
three elements are the maximally loaded. Therefore, we must learn
how to check the strength of every element of the beam taken in an
arbitrary section at a distance z from the neutral axis. Only then can
we be sure of defining the maximally loaded element and eheck its
strength.
10*
276 Bending. Strength of Beams [Part IV

Let us take an element of the material (Fig. 192) in an arbitrary

section at a distance z from the neutral layer. The faces of this element
perpendicular to the axis of the beam will be acted upon by normal
stresses a, whereas shearing stresses x will act on all the four side
faces. The front faces of the element will be free of stresses.
Stresses a and t may be expressed by the following formulas:
a-HHi
J ’ UT

where M is the bending moment, and Q the shearing force in the isolat­
ed element.
Let us consider the case when both a and t are positive. We shall
have to take recourse to the theories of strength to check the strength
of the element because it is in a compound stressed state; the computa­
tions must be started by calculating the principal stresses.

As the front face A BCD of the element (Fig. 192) and faces parallel
to it do not experience shearing stresses, they must lie in one of the
principal planes. The principal stress acting in this plane is zero,
because the plane is free of normal stresses. Thus, we are to study a
problem of plane stressed state.
Our aim now is to determine the remaining two principal stresses
knowing the normal and shearing stresses in two mutually perpendi­
cular planes, one of which is parallel and the other perpendicular to
the axis of the beam (Fig. 191). We solved an identical problem in
§32 by plotting the stress circle. There the method was applied to
the more general case of a stressed state, where two mutually perpen­
dicular planes with normals a and p are acted upon by stresses ca ,
anc^ Tp==—T<*- this Pr°klern we shall attribute index a
to the face of the element perpendicular to the axis of the beam, and
index f) to the face parallel to the axis (Fig. 193).
Let u$ lay off from point 0 the segment OKat representing c ^ -rr,
in the positive direction and another segment KaDa equal to x on
Ck. 13] Shearing and Principal Stresses 277

the perpendicular to the o-axis at point Point Da on the stress

circle corresponds to the plane perpendicular to Ihe axis of the beam.
In a plane parallel to the axis of the beam Op=0; this means that
point Kf, coincides with point 0. Segment KpDfi laid off downwards
represents the shearing stress xp=—x and gives the second point on
the circle, Dp. Joining the two points we get the centre of the circle,
point C, and the radii CDa and CDp. After plotting the circle we get
segments OA and OB representing the principal stresses, which re­
main to be determined. It is evident from the drawing that these
stresses have different signs. Therefore, the numbering of principal
stresses may be done as follows:
a l = ' 0 A > 0, cr2 — 0, a 3 — O B < 0
Making use of formula (6.13) given in §32, we get
V o‘ + 4t* = 4-1° + y ^ + T F I
a, —f V 4- 4t* ■=i I(-o - V o* + 4t*|
a8= 0
The formulas for <r, and <r3 may be written in an integrated form as

5 } = 4 |0 ± V o * + ‘l f | (13.13)
We have plotted the stress circle and computed the stresses on the
assumption that both o and x are positive. If any of Ihe stresses is
negative, then the corresponding sign in formula (13.13) should be
changed. A similar change would also have been essential in graphic
determination of <r, and or3 by plotting the stress circle.
Knowing all the three principal stresses, we can write down the
conditions of analysis for all the theories of strength.
According to the first theory, the theory of maximum normal
stresses,
® ,< [o]. or | [ a + | / o-- + 4 i< |< [o ] (13.14)

According to the second theory, the theory of maximum strain,

[c1-i» (fft + (ji)K [ff]
Putting the values of <rx, or*, and a, we gel
[ y (<r + ]/ o2+ 4x2) —y p (<r—V o* + 4x4)] < [o]

Assuming )i=0.3, wre find

[0.35o + 0.65 y V + 4 t» ] ^ [a] (13.15)
278 Bending. Strength of B eam [Part IV

According to the third theory, the theory of maximum shearing

stresses,
[oi—o3] < [ a ]
or
4 |o + V o* + 4t« _ o + (/<!* + 4t*J < [o]
which yields
/< j24-4x*<[a] (13.16)
Finally, according to the fourth theory, the theory of maximum po­
tential energy’ of distortion, we have
[<«i—og* + (of,—o,)* + (<J3—o,)*J < 2 [a]1
wherefrom
V a * + 4 t* ) ’ -f(o —V ^ + W ) ' 4- (2 1 /o * + 4t*)*] < 2 [<j]‘

After simplification we get

[o*+ 3t*K {> J4, |/<r2+ 3Ta < f c ) (13 17)
Now we shall try to find the points of the beam in which its strength
for principal stresses should be checked.
As the reduced stress depends both upon a and t , the strength check
should be carried out for those elements of the beam which simulta­

neously experience maximum cr and t. This is possible if the following

two conditions are fulfilled for the element:
(1) Bending moment and shearing force are maximum in the same
section.
(2) Beam width changes sharply near the edges of the section (for
example in an I- or a box section). The bending-moment and shearing-
force diagrams for such a section (Fig. 194) reveal that the shearing
and normal stresses near the region where the flange becomes the web
are close to maximum (points a and b).
Ch. 13] Shearing and Principal Stresses 279

Thus, the two above conditions determine whether an additional

strength check is necessary and also determine the element where
this check should be carried out. In the cases where these conditions
are not satisfied, we limit ourselves to selecting a few points where
the maximum reduced stresses can occur. As for selecting lhe proper
formula for analysis, the best, of course, is the one based on the theory
of maximum distortion energy (13.17).

Fig. 195 Fig. 196

In practice, however, the theory of maximum normal stresses

(13.14) is still used in the analysis of beams, because it often gives
smaller dimensions of the section.
Example. A simply supported beam AB (Fig. 195) is loaded by
symmetrically acting forces P = 6.4 tf located at distances n=50 cm
from the supports; the permissible stress is lol=1400 kgf/cma. Select
an I-section and check its strength in the region of transition from
flange to web.
The maximum values of Af and Q occur in the same section under
the load:
Mmax = P a ~ 0 .5x6.4=»3.2 tf-m
QW*X= P = 6.4 tf
The required section modulus is
r _ A W - 320 000 220
[oj 1400 m

We should take an I-beam No. 22 having 1F=232 cm3; /=2550cm 4.

The dimensions of the c f o s s section have been schematically shown
in Fig. 196. For the selected section
320000
232 = 1380 kgf/cm*
280 Betiding. Strength of Beams [Pari IV

An additional strength check should be carried out for z—10.13 cm;

at this height
320000x10.13
<r 2850 = 1271 kgfcm*

The static moment of the flange is

S ' - 11 x 0.87 X 10.565= 101 cm*
The shearing stress is
0400 x 101 k cm2
0.54x 2530 W KgI
The strength condition according to the first theory of strength (13.14)
is
i f 1271 -f-F 1271*-f- 4x469* | = 1426 kgf/cm* > 1400 kgf/cm*
The strength condition according to the fourth theory of strength
(13.17) is
Y 12719-|- 3 X469s •= 1510 kgf/cm* > 1400 kgf.cm2
As the reduced stress according to the fourth theory is 8ao greater
than the permissible stress, the dimensions of the I-scction should
be increased by taking an l-beam No. 22a. After computations we
get for this section om,ix=1260 kgfcm®, and for 2 = 10.11 cm, a —
= 1158 kgf cm* and t=442 kgf cm*. The reduced stress according to
the first theory is 1329 kgf cm®, and according to the fourth theory of
strength, 1423 kgf/cm-.

§ 78. Directions of the Principal Stresses

In the preceding section we determined only the magnitude of the
principal stresses for an arbitrarily selected element without con­
cerning ourselves with their direction. The results obtained were good

Fig. 197
Ch. 13} Shearing and Principal Stresses 281

enough for materials which have equal resistance to tension and

compression. For materials like reinforced concrete, however, it is
extremely important to know the direction of tensile stresses in every
point so that we can place the reinforcement rods in this direction.

The direction of the principal stresses may be determined with

the hetp of the stress circle (Fig. 197). Suppose a« and t*. acting in
a plane perpendicular to the axis of the beam, are positive;
m Bending. Strength of Beams t Part IV

and

After plotting the stress circle we see that the relative position of
the lines of action of stress <ra and the maximum (algebraically) prin­
cipal stress Oj is the same as the relative position of line BDa and
the x-axis; the latter two make an angle a in the stress circle (Fig. 197).
To mark the direction of Oi on the drawing we must lay off angle a
from the direction of <ra clockwise.
The principal stresses change their direction within the limits of
the section. Near the edges of the beam one of the principal stresses

Fig. 199

is zero, whereas the other is directed parallel to the axis of the beam;
at the neutral layer the principal stresses make an angle of 45° with
the axis of the beam.
Figure 198 shows the stress circles and directions of the principal
stresses in various points of the section. It is assumed that the bending
moment and shearing force in the section are positive.
Having obtained the directions of the principal stresses in an ar­
bitrary point of the given section, we continue one of the lines till
it intersects the adjacent section. We determine the direction of the
principal stress in this new point and continue the line till it inter­
sects the next section. We thus obtain a broken line which in the limit
changes into a curve the tangent to which coincides with the direc­
tion of the principal stress in the point under consideration. This
curve is known as the trajectory of the principal stress. The directions
of the trajectories of principal stresses depend upon the type of load­
ing and the working conditions of the beam. We can draw two tra­
jectories of principal stresses through every point of the beam—one
for the tensile stresses and the other for compressive stresses. The tra-
Ch. 14] Sltear Centre. Composite Beams 283

jectories for compressive stresses are shown by dotted lines and those
for the tensile stresses by solid lines (Fig. 199, the middle drawing).
The reinforcement in reinforced concrete beams should be placed
in such a way that it is located approximately in the direction of
the trajectory of the principal tensile stresses (Fig. 199, the lower
drawing).
Theoretical investigations on principal stresses in bending that give
the present-day design formulas were first carried out by N. A. Bele-
lyubskii in connection with the design of bridge beams (his results
were published in 1870-76). In his works principal stresses were called
“oblique stresses".

C H A PTER 14

Shear Centre. Composite Beams

§ 79. Shearing Stresses Parallel to the Neutral Axis.
Concept of Shear Centre
A. Beams of thin-walled sections experience shearing stresses parallel
to the //-axis in addition to shearing stresses parallel to shearing force
Q, i.e. perpendicular to the neutral axis (y) that were discussed in
§§74-76. The validity of this statement can be easily confirmed by
considering the parallelepiped having sides A H —y, A B = tf and
BC=dx (Fig. 200(a) and (6)) which is isolated from, say, the flange
of an I-sectioii by sections 1-1 and 2-2 and plane ABCD parallel to
plane xz.
Let us assume that bending moment M X-=M in section 1-1 is less
than the bending moment M2=M+dAf in section 2-2. The resultant
Ni of internal normal forces acting on the front face (ABGH) of the
parallelepiped will be less than the resultant /V2of the normal forces
on the rear face (Fig. 200(c)). The difference between N 2 and A'i (see
formulas (13.1) and (13.2)) is calculated by the formula
dN — N.l— N l (14.1)

where 5^ is the static moment about the neutral axis of area ABGH
of the front face or a similar rear face where the internal normal
stresses are summed up. The difference between Nt and Af, can be ba­
lanced only by internal shearing stresses acting on face ABCD because
the top, bottom and side faces of the parallelepiped, being external
surfaces, are free from forces and there is no possibility of any addition­
al forces appearing on the front and rear faces which could counter­
balance the difference (Fig. 200(6) and (c)).
284 Bending. Strength of Beams [Part tV

Hence, on face ABCD of the parallelepiped we have shearing stres­

ses and in accordance with the law of complementary shearing
stresses similar stresses x, appear on face ABGH, j,e. in the cross
section of the beam (Fig. 200(6) and (c)). On account of the fact that
flange thickness tf and length dx of the isolated element are small
quantities, these shearing stresses can be considered to be uniformly

distributed over the area of face ABCD. Consequently, the sura of

elementary internal shearing forces acting here will be
dT —Xfljdx
The equilibrium condition of the isolated parallelepiped can be
written as follows:
= N t + d T — N t = d T — <W«=0
or

wherefrom

Thus, Zhuravskii’s formula (13.3) can also b§ employed for shear­

ing stresses parallel to the neutral axis in thin-wallcd sections if
quantity b in the denominator is taken as the width of the layer in
which shearing stress is calculated, irrespective of whether the thin-
wailed section is assumed to be cut parallel or perpendicular to the
neutral axis.
Ch. 14\ Shear Centre. Composite Beams 285

In our case (with the assumption that N£>Ni) shearing stresses x t

in the left half of the top flange act in the cross section from left to
right. It can be easily seen that in the left half of the lower flange,
where the normal stresses are compressive and as before la 2|> |O i |,
shearing stresses xf act in the opposite direction (Fig. 200(d)); in the
right half of the top flange they act from right to left (Fig. 200(e)),
whereas in the right half of the lower flange from left to right.
rf -diagram

tei (W
Fig. 201 Fig. 202

The shearing stresses in the flanges and web ol ihe thin-walled sec­
tion form the so-called shearing stress “streaml,lies'', the streamlines
for an 1-section are depicted in Fig. 201,
Let us write the expression for shearing stresses x t. One of the qu­
antities in formula (14.2) is the static moment of the flange area
hatched in Fig. 201:
h—~tf
= 'V o , = ytj
Therefore
Q Sl <?(*—/ / ) *
(14.3)

i.e. shearing stress x, varies linearly along the flange length (in for­
mula (14.3) the ^-coordinate is to the first power). This stress becomes
maximum when y —bi:
Q(h— t f)bi
Vfflax — 2 J„ (14.4)

When &!<</<&!+/«:, the whole web of the I-section lies in the vertical
section. The shearing stress is not distributed uniformly along the web
height, therefore Zhuravskii’s formula cannot be employed for its
calculation. The shearing-stress (xr ) diagram for I-section is shown
in Fig. 201. The diagrams of shearing-stress distribution in the flan­
ges and web of a channel section are depicted in Fig. 202(a); for a
286 Bending. Strength of Beams [Part IV

C-section, in Fig. 202(b). The shearing-stress streamlines are shown

in the cross section for each of these profiles.
When the shearing stresses have to be determined in the flange
of a closed thin-walled profile symmetrical about the axis of loading
(a-axis), for instance, at point K of the flange of a box section (Fig. 203),
then one imaginary section must pass through point K and the
other through a symmetrically lo­
cated (with respect to the axis of
loading) point /(,. In the numera­
tor of formula (14.2) we introduce
the static moment of the area of
flange bounded by these two sections
(the area is hatched in Fig. 203),
and in the denominator the double
thickness of the web (due to two
sections). We obtain a formula for
determining t , which is similar to
formula (14.3). Figure 203 shows the
shearing-stress diagrams in the flan­
ge and the web and also the shear­
ing-stress streamlines in the profile.
If the web or flange of the thin-
walled section is inclined to the
plane of loading at an angle a, then this circumstance must be taken
into account while computing shearing stress by formulas (13.3)
and/or (14.2) by introducing a factor cos a in the denominator of these
formulas. Let us assume that an equal leg angle section beam is loa­
ded in the plane of symmetry zOx (Fig. 204). The sum of projections
on the 2-axis of internal shearing forces, replaced in Fig. 204(a) by
forces T, will be equal to 2T cos a. As this sum of projections of the
internal forces is equal to the shearing force Q,

Hence, shearing stress xf , which may be considered uniformly dis­

tributed over the flange thickness, may be determined at some point
K of the angle flange by the formula

T QSj _
f J y t f cos a J y i j c o sa Jy

where SJ is the static moment of the hatched area of the flange. Shear­
ing stress xf is maximum at point N on the #-axis where u= umM=*
Ch. 14\ Shear Centre. Composite Beams 287

/>
= b -- 2 a — - f :
(H.5)

B. If we consider Figs. 201*203, we note that when the I- and box beams
are loaded in the plane coinciding with the principal central plane
of inertia xOy for xOz), which is also the plane of symmetry of the
beam, the internal shearing forces give a resultant equal to shearing
force Q and directed along the axis of symmetry of the section (the
shearing-stress streamlines are, so to say, in equilibrium).

Fig. 204

However, if we consider channel, C- (Fig. 202(a) and (b))t T- (Fig.

207), equal leg and unequal leg angle (Figs. 208 and 209) sections also
loaded in the plane coinciding with the principal central plane of
inertia, xOzt but which is not the plane of symmetry of the beam,
the internal shearing forces in the sections give the aforementioned
resultant and a force couple about the x-axis of the beam. This implies
that the resultant of internal shearing forces of the section equal to
the shearing force Q passes not through the centre of gravity C along
the principal central axis of inertia Oz, but parallel to this axis
through some other point in the section. The beam consequently
experiences torsion in addition to uni-planar bending.
The point through which the resultant of all internal shearing for­
ces of the section passes (the moment of all internal shearing forces
of the section about this point is zero) is known as shear centre or
flexural centre, and the line parallel to the x-axis and joining the shear
centres of all sections of the beam is called the shear-centre line. Ob­
viously, for the beam to experience only uni-planar bending without
torsion of the thin-walled section, the plane of application of exter­
nal forces must pass through the shear-centre line parallel to one of
the principal central planes of inertia of the beam. This ensures ful­
filment of the condition of equilibrium according to which the product
288 Bending. Strength of Beams [Part tV

of inertia of the section about the line of loading and a perpendicular

neutral line must be zero, i.e. the beam experiences uni-planar bend­
ing. At the same time, the moment of external forces as well as the
moment of internal shearing forces about the shear centre will be zero,
i.e. the beam will not be subjected
to torsion.
Let us take the channel section
1 (Fig. 205 and 206) as an example and
explain how to determine the shear
Say centre, point A. Neglecting the
shearing stresses parallel to the axis
f a if L y. H ,r in the flanges, we assume that in­
ternal shearing forces in the walls
of a channel section give a resul­
JsM* tant approximately equal to shear­
ing force Q and directed along the
middle line of the wall. The resul­
tants of internal shearing forces in
the flanges, acting parallel to the
F ig. 205 neutral line of the section, will be
denoted by T and assumed to be ap­
plied at the middle of flange thick­
ness. Keeping in mind that shearing stress T/ in the flange varies li­
nearly, with a maximum value according to formula (14.4) equal to
Q (h -if)bx
Tfttax —‘ 2/„

we may write the following expression for resultant Tz

T/ m a x |-0 Q<h-t/ibV/
Mi 4T„

The condition according to which the moment of all internal shear­

ing forces in the channel section about the shear centre is equal to
zero may be written as follows:
Q e - T { h — lf) = Q
wherefrom
T{h—tf) (h -tfp b*t,
3 “ 4 y„
(14.6)

In the more complex cases the shear centre location can be deter­
mined by special methods which are discussed in the theory of bend-
in* and torsional deformations of thin-walled bars.
Ch. 14) Shear Centre. Composite Beams 289

Let us note that the shear centre coincides with the centre of grav­
ity of the section if the latter has two axes of symmetry (Figs. 201,
203). If the section has one axis of symmetry, the shear centre lies
on this axis (Figs. 202, 207, 208). If the section consists of rectangles

whose middle lines intersect at one point, the shear centre lies at
this point (Figs. 207, 208, 209). In these figures the shear centre is
denoted by A, while / and U show the directions along which the
loading leads to uni-planar bending of the beam without torsion.

§ 80. Riveted and Welded Beams

In the examples of selection of cross-sectional dimensions of beams

which were discussed in preceding sections the required values of
section moduli of I-beams were such that we were able to select
rolled profiles in all the cases. The biggest rolled profile manufactured
in the Soviet Union, the 1 section
No. 60, has a section modulus of
about 2560 cm’.
In practice, however, we often re­ 3
quire profiles of considerably bigger
size. In such cases we use composite
beam sections by riveting plates and
angles or by welding plates.
A riveted beam (Fig, 210) consists
of a vertical plate /, a number
of pairs of horizontal plates 2 and
angles 3. The angles and plates C 1

are joined by rivets. A welded

beam (Fig. 211) consists of vertical
and horizontal plates joined by
w elds. Fig. 210 Fig. 2U
290 Bending. Strength of Beams [Part IV

The design of welded and riveted beams is treated in the courses

on metal structures. There it is pointed out how to determine beam
dimensions if the maximum bending moment is known.* Given
below is an example on checking the strength of a welded beam.
A schematic diagram of the beam, the forces acting on it, and the
bending-moment and shearing-force diagrams are shown in Fig. 212.

r o o t , VM IP-IOtf
Htmnmmnnimmimmi

Fig. 212

The cross-sectional dimensions of the beam are given in Fig. 213.

We have to check the strength of the beam as a whole and of the welded
joints.
Let us calculate the moment of inertia of the whole section, working
as a rigid one, about the principal x-axis:
I V1943
/* = t -1- 2 x 35 x 2 x 633= 159 000 + 555 600=714 600 cm*
The section modulus
714600
64
—11 160 cm8
The maximum normal stress in the beam at the middle of its span
- Mmax 177.6x10*
u inax
i i 160
—1592 kgf/cm*
is less than the permissible stress which is 1600 kgf/cma.
The shearing stresses at the upper or lower ends of the web
• 94xl08x 35 x2 x 63
-

714600x1
= 580 kgf/cm*
* The design of riveied and welded beams has been treated in detail in the pre­
vious editions of this book. See N. M. Belyaev, Strength of Materials, Nauka, Edi­
tions 7-14 (in Russian).
Ch. 14| Shear Centre. Conposite Beams 291

These shearing stresses will be taken up by a pair of welded seams

(one on each side of the web) along planes /-/ of dimension m each
(Fig. 214). Therefore^ while calculating shearing stresses in the seams,
thickness 2m of the two seams must be substituted in the denominator
of the formula instead of bw . The minimum design thickness of the
seams is taken equal to 0.4 cm. For this value of m the shearing
stresses in the seams are
94xl0:,x35x2x63
714600 x 2 X 0.4 = 725 kgf/cm*

or, what is the same

t, = 58o| | = 580 ^ = 725 kgf/cm*

These stresses do not exceed the permissible shearing stress for

welded joints.
The joint may also be made by intermittent seams (Fig. 215).
The shearing force acts over length a and is taken up by seams of

length c. Therefore, everything else remaining the same, the stresse

in intermittent seams (welded keys) will be times greater than
stresses in continuous seams of the same size. Now automatic welding
of parts with continuous seams is generally used. Therefore, joints
made with the help of intermittent seams are gradually becoming
obsolete. The intermittent seams have the additional shortcoming
that the beginning and end of each seam are pockets of local stress
concentration, which is not taken into account by the design formulas.
The strength of the web should be checked against principal stresses
at the base of the weld seam. This area will experience normal stresses
of a considerable magnitude (from Af«= 174.2 t f m) and shearing
stresses which are just slightly less than 580 kgf/cm*. The combina­
tion of these two stresses may considerably raise the principal and
reduced stresses at this level.
 PART V
Deformation of Beams
Due to Bending

CH A PTER IS

Analytical Method
of Determining Deformations
§ 81. Deflection and Rotation of Beam Sections
When external forces act in one of the principal planes of inertia
of a beam, its axis is observed to bend in the same plane and uni-
planar bending occurs.
In Fig. 216 the deformation of a beam rigidly fixed at one end and
loaded at the other by a concentrated force is shown in an enlarged
scale. The centre of gravity 0 of a section having abscissa x moves
to Oi.
Displacement 0 0 { of the centre of gravity of a section in a direc­
tion perpendicular to the axis of the beam is called the deflection of
beam in the particular section or the deflection of the particular section
of the beam. We shall denote deflection by y.
Strictly speaking, since the beam axis lies in the neutral layer it
does not change its length and the displaced point 0, must be slightly

to a side from the perpendicular to the beam axis. However, deflection

y is usually small as compared to the length of the beam and the dis­
placement of the perpendicular to a side represents a small quantity
in comparison with deflections; it is therefore neglected.
During deformation sections of the beam remain plane and turn
w.r.t. their original position. In Fig. 217 sections O-Oi and B-Bi
are shown before and after deformation.
Ch. 15] Analytical Method of Determining Deformations

Angle 0 by which each section turns w.r.t. its original position

is called the angle of rotation of Ute section. We must learn to calculate
the deflection and angle of rotation in each section for practical ap-
pi ication.
Maximum deflection can serve as a measure of the degree of distor­
tion of the structure when it is acted upon by external forces. Gene­
rally, to prevent the beam joints from loosening and to reduce vib­
rations under a dynamic load the value of maximum deflection is
restricted for a loaded beam. Thus, in steel beams, depending upon
their designation, the maximum deflection should not exceed 1/1000-
1/250 of the span.
Besides, we also require the value of deformation when solving
statically indeterminate problems, in which the number of reactions
is more than the number of equations of statics. The additional equa­
tions can be written only by studying the deformation of the struc­
ture. We must know how to calculate deflection y and angle of rota­
tion fl for every section in order to be able to determine the deforma­
tion completely. Both y and 0 are functions of x— Ihe distance of
the section from the centre of coordinates; there is a definite relation
between y and 0 in each section.
Let us decide upon a coordinate system, which we shall use in fu­
ture. The centre of coordinates will be a point on the original posi­
tion of the beam axis, which we shall always select as the A'-axis,
and the y-axis shall be directed upwards, perpendicular to the beam
axis before deformation. Under these conditions the equation
y = f( x ) (15.1)

represents the equation of a curve along which the beam bends when
it is loaded; it is the equation of ihe deflected axis of the beam.
The tangent to the deflected axis of the beam (Fig. 217) at point Oi
makes an angle 0 with the x-axis, i.e. an angle equal to the angle of
rotation of the section about its original position. On the other hand
we know that Ihe tangent of the angle between the tangent to the
curve y= f(x) and the x-axis is
(15.2)

Since in actual practice the deflection of a beam is generally small

as compared to its span, angle 0 is also very small and generally does
not exceed 1°. For such a small value of the angle we may consider
that the tangent of the angle is equal to the angle expressed in radians.
It ensues that
(153)
294 Deformation Due to Bending [Part V

i.e. the angle of rotation of a section Is equal to the first derivative

of the deflection in this section w.r.t. x.
Thus, the problem of studying the deformation of a beam narrows
down to obtaining the equation of the deflected axis y~f(x); knowing
the equation, we can calculate the angle of rotation in any section
by differentiation.

§ 82. Differential Equation of the Deflected Axis

In order to obtain y as a function of x, we must establish a relation

between the deformation of a beam due to external forces and its
size and material. We had obtained such a relation before in § 63.
Let us make use of formula (11.10), which we had obtained while
studying pure bending. Extending the formula over the general case
of bending, i.e. neglecting the effect of the shearing force on defor­
mation, we get the relation
I M ix)
<>1*5 ej

where p(x) is the radius of curvature of the deflected axis between two
adjacent sections at a distance x from the centre of coordinates, M (x)
is the bending moment in the same section, and EJ is the rigidity of
the beam. Generally, the effect of Q(x) on the deformation of beam
is not large; the method of taking into account its effect is given
in § 108.
Figure 218 depicts the change in the radii of curvature as the bend­
ing moment is increased. In order to obtain the equation of the de­
flected axis we shall employ the mathematical relation between the
radius of curvature of the axis and its coordinates x and y:

i
(15.4)
pt*>

Putting this value of curvature in formula (11.10), we get

a differential equation which relates y, x, M(x) and EJ:

12 M (* )
3 EJ
(15.5)
i - i “» y
Y

This is known as the differential equation of the deflected axis or

quite often differential equation of the elastic curve.
Ch. IS1 Analytical Method, of Determining Deformations 295

In a vast majority of practical cases we find that , represent­

ing the angle of rotation of a section of the beam, is a very small
quantity. Therefore, its square may be neglected in comparison to
unity; consequently, equation (15.5) may be written in a simpler
way:
or 05.6)

This relation is known as the approximate differential equation

of the deflected axis.
The convention for the sign of bending moments is decided irres­
pective of the direction of the coordinate axes; it is known that the
second differential is positive if the concave side of the curve faces

Fig. 218 Fig. 219

the positive direction of the 0 -axis and negative if the convex side
faces it (Fig. 219). Hence, the sign of the bending moment does not
depend upon the location of the coordinate axes but the sign of the
second differential does.
If the 0 -axis is directed upwards, then the positive sign should be
used in equation (15.6); the negative sign should be used if the 0 -axis
is directed downwards.
In the future we shall always direct the 0 -axis upwards, and the
differential equation (15.6) may be written as:
£/g=M (x) (15.7)

The sign of the bending moment shall be selected according to the

above convention.
296 Deformation Due to Bending [Part V

The deflection may be obtained from the differential equation of

the deflected axis by integrating equation (15.7). Bending moment
M (x) is a function of x\ therefore, upon integration we get
£ 7 jj* = J Af (x)dx-\-C

Integrating once again,

EJy = J dx J At ix) d x ± C x ± D

Thus we get the following equation for the angle of rotation:

e “ C = Z 7 [ f M W ‘f c + C] (15.8)
and the following equation for deflection:
y = -gj \d x JiVf (x )d x + C x ± D \ (15-9)

These equations have two constants of integration C and D. The

method of calculating these constants will be shown in examples
below.
Before we take up practical problems, we deem it necessary once
again to emphasize that equation (15.7) is approximate; the error
that we allow by neglecting the quantity ( ^ j in comparison to
unity is small only in those cases when the deformation of the beam is
small in comparison with its size. If this condition is not satisfied,
then the angles of rotation are found to be large enough so that their
square cannot be ignored anymore; in such cases it becomes essential
to integrate the whole of equation (15.5).
Examples of such cases are the deformation of thin springs and
thin veneer and, generally, bending of flexible beams.

§ 83. Integration of the Differential Equation

of the Deflected Axis of a Beam Fixed at
One End

Consider a beam fixed at end A and loaded by a concentrated force

P at the other end and a uniformly distributed force q along the whole
length of the beam (Fig. 220); let I be the span of the beam. We shall
designate point A as the centre of coordinates, direct the i/-axis up­
wards and the x-axis towards the right. The differential equation
of the deflected axis is:
E Jy”= M (x)
Ch. I5\ Analytical Method of Determining Deformations 297

The bending moment in an arbitrary section at a distance x from

the centre of coordinates is
(JC)----- P ( l - x ) - q i = f £ - (15.10)

£ ;/= -P (< -* )-< ? (15.11)

We integrate this equation twice:
£ / » ' - - P ( / * - £ ) —| ( f * * - « , + y ) + C (15.12)

W - - p { $ - r ) W { ¥ - J T + !t t ) + Cx + D <I3 I3 >
To determine C and D we must locate such sections of the beam
where the deflection as well as the angle of rotation are known. One
of these sections lies over support A\ in this section at x=0, 0
and y = 0. Putting these values first
in equation (15.12) and then in
(15.13), we get C =0 and D = 0. It
is evident from the expressions of
deflection and angle of rotation that
constants C and £>, when divided
by rigidity EJ of the beam, give
the corresponding values of angle
of rotation and deflection in a sec­
tion which lies at the origin of coor­
dinates A. The constants C and D
have the following dimensions:
[C] = force x (length)4 and
|D] = force X (length)3
The fact that the constants of
integration are found to be zero is
a direct outcome of selecting the fixed end of the beam as the origin
of coordinates. In plotting bending moment and shearing force diag­
rams we measured abscissa x from the loaded end of the beam; here it
is more expedient to measure x from the fixed end to reduce the
amount of calculations required to determine C and D\ this somewhat
complicates the expression for M (x) but simplifies the determination
of deformations. Having determined C and D, we can now transform
the expressions for y and 0 in such a manner so that the brackets con­
tain only dimensionless numbers, which is helpful in calculating the
deflection and angle of rotation:
(15.14)
(15.15)
298 Deformation Due to Bending 1Part V

With the help of these expressions we can determine the maximum

values of y and 6. From the designer’s point of view the maximum
absolute value of deflection y is of greater interest; therefore, besides
the analytical maximum of function y at 0=-^j--=O, we must also
And its value at the ends of the span. In the example under considera­
tion the maximum deflection y occurs at point B , where 0 is not zero.
The analytical maximum of function y is equal to zero at point A .
We shall denote the deflections in various sections by letter f with
an index showing the section in which it occurs. Thus at x = l

h — m ~ & <l5-l6>
The minus sign shows that the deflection is in the downward direc­
tion. Obviously, the maximum angle of rotation wilt occur in the
same section; it will be
Pi* ql*
(15.17)
2EJ 6 EJ

The minus sign indicates that the section turns in the clockwise di­
rection.
Both expressions (15.16) and (15.17) show the separate effect of P
and q on the deflection and angle of rotation, respectively. When one
of the forces is absent, the corresponding part of expression becomes
zero.
To have an idea about the magnitude of deformation let us take
P = 2 tf, <7=0.5 tf/m, 1=2 m, £= 2 x 1 0 ° kgf/cm2, permissible stress
[at—1400 kgf/cm2, and select an 1-beam from the specification table.
The strength condition for the beam may be written as:
10®
T (2x2+0-5¥ ) — = 357 cm8
W [cr| MOO
From the standard table for I-beams (see Appendix) we And I-beam
No. 27 having W = V \ cm8, 7=5010 cm4. The angle of rotation and
deflection may be calculated as
/2x2* 23’ 107
------- V 2 0.5 T 2 x l0 4X 5010 215 radian
/ 2 x 2s 10* ——0.G3 cm
/* = - I 3 2xl0*x5010

The maximum deflection constitutes (0.63/200)= ^ of the beam

span, while the square of the maximum angle of rotation
= I '46 000, i.e. it is negligibly small as compared to unity in for­
mula (15.5).
Ch. 15\ Analytical Method of Determining Deformations 299

§ 84. Integrating the Differential Equation of the

Deflected Axis of a Simply Supported Beam
Let us find the deformation of a simply supported beam, loaded uni­
formly by a continuous force q (Fig. 221). The origin of coordinates
lies at the left support and the x-axis is directed towards the right.
A distinguishing feature of this problem as compared to the previous

Fig. 221

one is that the support reactions must be determined in order to find

an expression for M (x).
From symmetry A = B = a n d HA= 0. We calculate In the fol­
lowing order:
B J % = M{x), M(x) = + ^ x - - 2 |L = + |( / x - x > )

We know the following values of deflection: at support A , i.e. at

x=0, deflection y = 0 , and at support £, i.e. at x= l, deflection y= 0.
Applying formula (15.19) to section A first, we get:
D= 0
Then applying it to section B, we get the following equation:

wherefrom
300 Deformation Due to Bending [Part V

The formulas for y and may now be written as follows:

(*5.20)

In order to find the maximum deflection we must determine the

section in which 0=0; from symmetry this must be the middle sec­
tion. By p u ttin g y = y in formula (15.20) we find that 0 = -^ becomes
zero; under these conditions:
r Sql*
/m a x - 384£y

The maximum values of 0 occur at x= 0 and x=l:

0°max = =n j£ _
f 24CJ

In this example also we find that is the deflection of the beam

Q
at the origin of coordinates and -gj is the angle of rotation of the sec­
tion at support A, which coincides with the origin of coordinates.
In all the above examples, if we direct the //-axis upwards and the
X-axis towards the right, then a negative value of U corresponds to

F ig . 2 2 2

clockwise and a positive value corresponds to anticlockwise rotation

of the section.
Let us determine the deformation in one more case of a simply sup­
ported beam. Assume that the beam is acted upon by a moment Af
at the right-hand support (Fig. 222). Reactions A and B give a mo­
ment M and are equal to
, „ .Vf
Ch. 15\ Analytical Method of Determining Deformations 301

We assume the left-hand support to be the origin of coordinates;

therefore
EJ ^£5 = M (x) — -f- Ax — + y x
E Jy^^+ C x + D

The constants of integration are determined from the condition

that the deflection at supports A and B is zero: al * = 0 deflection
Ml
y= 0 , wherefrom D—0; at x= l deflection 0, wherefrom C = — g -.
Therefore
•“ £ — <15-22>
M ix ( , *2\
GEJ \ l 10

The maximum deflection occurs in the section where ^ *=0, therefore

l_ 3 -j| = 0

The abscissa of this section is

*„ = y i = 0.577I (15.23)

Maximum deflection is
f- Mlxl / i \ ______ Ml* Mt*
' g KT E J\ 3f* ) 9 Y z EJ ~ ~ M.6EJ

and deflection at the middle of the span is

£ Ml* r . /a l Ail*
" /* ~ “ T2e 7 L , “ 4PJ5“ “ TCE7
The deviation from the maximum deflection is of the order of 2.5%;
thus even for such a highly unsymmetric loading as this we can with
sufficient accuracy assume that'the maximum deflection in a simply
supported beam occurs at the middle of the span.

§ 85. Method of Equating the Constants o! Integration

of Differentia! Equations When the Beam Has a
Number of Differently Loaded Portions
In the examples discussed above the beams were identically loaded
along the whole length and there were two constants of integration,
C and D. Every portion of new loading adds two constants of integra-
302 Deformation Due to Bending [Part V

tion and complicates solution of the problem if we do not follow Ihe

rules which reduce the number of constants of integration to two, ir­
respective of the number of differently loaded portions.
Let us consider a beam with three differently loaded portions
(Fig. 223). Let us decide to have a common origin of coordinates for
,
A-
. i (
ilium f
iiiiim
r
mimrrntmTr
B\
__ >
-ajj ^ ^ I 1 ft 7r/RTe
H \
1
1
* 111
--------------- S i ------------ —^
*-----------------— / ------- ----------- H
Fig. 223

all the port ions—at the left or right end of the beam—and while writ­
ing the bending moment expression consider that portion which in­
cludes the origin of coordinates.
Let F be the origin of coordinates. We write the equation of the
deflected beam axis in the first portion and integrate it twice:
EJy\ = — Pxl
EJy[ = ~ Z f + C i (15.24)

EJyi = “ ^ + Cxx t + Di (15.25)

The bending moment expression for the second portion should be
written in such a way so that summands EJy", E Jy' and EJy coin­
cide with identical quantities of the equations of the first portion in
the boundary section (over support A). This will take place if (x—a),
which represents the arm of the force that is absent in the first portion,
is integrated with respect to d(x—a) or, in other words, without
opening the brackets. Let us point out that x is the abscissa of an ar­
bitrary section of the portion under consideration; a is the abscissa
of the starting point of this portion.
Let us now write three equations for the second portion:
EJij, = — Px, + A U , - a , ) _ ! I S j p E

= + (15.26)

EJy, — + C ,* , + P , (15 27)

In the section at support A the angles of rotation computed from
equations (15.24) and (15.26) must be equal, i.e. the beam axis must
Ch. 15] Analytical Method, of Determining Deformations 303

bend smoothly over support A. The deflections at the support, ca!*

culated from equations (15.25) and (15.27), must also be equal. In
other words, y\=y\ and y x= y 2 when Xi=x 3 =at. From these conditions
we find C1 —C2 —C and Dl= D a=D.
Let us now pass over to the third portion. There is no distributed
load on this portion. In order to retain the bending moment expres­
sions due to distributed load in their previous form, it is necessary
to extend the distributed load of the second portion to the end of
the beam and for compensating this extra load apply an identical
load of opposite sign. These transformations will not disturb the equi­
librium of the beam nor will they change the support reactions.
To ensure that the new load in the form of a concentrated moment
M does not change the structure of the three equations of the third
portion as compared to the second, moment M should be multiplied
by (x~~a) to the power zero; this affects neither the units of the force,
nor the conditions of equilibrium.
In the light of the above, let us now write the equation of the de­
flected beam axis and integrate it twice:
E jy ;= — P x ,+ A + " ‘V **+ M (x .-a ,)"
*?»..»_ . A(Xy-at)* q{xa—axY
E Jy* = ----- 2— I------- 2--------------6—
+ T ( o - « J , + A ) f e _ a |)+ c > (15.28)

ct P xl | .d(X»—fll)3 <?(xa- g |)4

CJ yt — g 1 G 24
+ M O r ,- - ,) ’ + c A + Da (15.29)

At the boundary section (where M is applied) we have the following

conditions for equating the constants of integration: y3 *=y* and y%—y 3
when Xa=xa= a 4. Substituting the first of these conditions in equations
(15.26) and (15.28), we find that Cs=Ca=C. Substituting the second
condition in equations (15.27) and (15.29) we find Dt—D*=D.
The constants of integration are reduced to two: C and D. For
determining these constants we employ the following conditions:
deflection of beam at supports A and B is equal to zero, i.e. at xx=au
i/i=0 and at x 3 —l, y*=0. After substituting these conditions in equa­
tions (15.25) and (15.29) we get the following two equations:
+ 4 -0 = 0 (15.30)
Pi 8 , A V - a i) * 4 f/-o i)«

<
1 (1 - at)* .
f C / + D = 0 (15.31)
24 r 2
304 Deformation Due to Bending [Part V

The values of C and D are obtained by simultaneous solution of equa­

tions (15.30) and (15.31).
We considered all the three portions and wrote three equal ions for
each of them to show how to reduce the number of constants of integ­
ration to two: C and D. While solving other problems it is not at all
necessary to again write all equations for each portion, it is sufficient
to write three equations only for the portion which is farthest from
the origin of coordinates. All summands of the right-hand side of
the equations will pertain to this portion. At th:s stage it <s desirable
to mark the summands which pertain to the previous portions. One
of the ways of marking is shown in the next lines:
EJy'
2
Til.
, £*)" M ix —na)| -f-C (15.32)
f> i ■ 13
i4(x—Q|P q j x — g,)4!
E Jy
G T

f ~ I + C x+ D (15.33)
_____ la

Here C and D pertain to all the portions. Sometimes C, Cx and D

are written in the beginning of the right-hand side of equations (15.32)
and (15.33).
The method of equating the constants of integration was first pro­
posed by R.F.A. Ciebsch.

§ 86. Method of Initial Parameters for Determining

Displacements in Beams
If we take a careful look at the equations for angle of rotation (15.32)
and deflection (15.33), obtained by Ciebsch’s method and discussed
in the preceding section, we can note that load in the form of concent­
rated moment M was respectively reflected in these equations as
Af (x—«) and M ~~c-a

The significance of the parentheses was earlier explained in §85.

Let us recall that x is the abscissa of an arbitnry section in the por­
tion of beam under consideration and a is the abscissa of the starting
point of this portion.
Concentrated force P and support reaction A were reflected fn the
same equations as
P(A---g)3 and P ( x — o)-»
2 6
Ch. 15] Analytical Method of Determining Deformations 305

Uniformly distributed Force <7 entered the equations as

Q(*-<*)* an>i P jx—a)*
6 ana 24

It was mentioned in § S3 that the constants of integration C and D

are angle of rotation (6 0) and deflection (y0) at the origin of coordi­
nates, respectively, multiplied by EJ. We may, therefore, write
C—EJQ0 and D —EJyn.
Keeping in mind that in a beam having a number of differently
loaded portions there may be a number of concentrated moments M,
a number of concentrated Torces P and the .uniformly distributed force
q may be acting on a number of portions, the equations for angle of
rotation (15.32) and deflection (15.33) may be written in the following
general form:
£./y' = £ y e , + £ M ( * — + + (15.34)

(15.35)
This method of writing the displacement equations is known as
the method of initial parameters, while the equations are called the
general equations of the method of initial parameters.
This method was first mentioned in the works of Prof. N. P. Puzy-
revskii and Academician A. N. Krvlov.
The application of this method will be illustrated in an example in
the next section.

§ 87. Simply Supported Beam Unsymmetrically

Loaded by a Force

Let us write the displacement equations (15.34) and (15.35) for the
beam shown in Fig. 224 by the method of initial parameters:
E Jy' - £70, + A P (15.36)

E Jy = E Jy 0 + E J% x+ A (15.37)

According to the fhrst condition (at *=0, i/A=0) equation (15.37)

changes into an identity: 0 - 0 . The second condition (at x= l, </$=0)
applied to equation (15.37) yields the following:
E j e Ai + A ^ - P ^ =0

t l —3310
306 Deformation Due to Bending [Part V

After substituting A = ~ and (/—a)s=b, we obtain the following

expression for the initial parameter, the angle of rotation at support A:
Pb
QEJt ('*-*>*)

Knowing now the angle of rotation and deflection at the origin of

coordinates and keeping in mind the above two substitutions, we re­
write equations (15.36) and (15.37) in the final form:

W — f - I P - t ’l + x - - (15.38)
EJy = b') - f t (15.39)

As it has been assumed in Fig. 224 that a>b, the maximum deflec­
tion will occur in the first portion between the middle of beam and
point of application of force/*. Weshall therefore not include in further

calculations the last summands having the factor (x—a) as they per­
tain to the second portion.
The angle of rotation is zero in the section of maximum deflection
(at x —xc); therefore

E J y '— t £ - < ( '- 6’) + - ^ = 0

wherefrom
(15.40)

Let us now calculate the maximum deflection in this section;

max
6/ 6 1 3

f Pb yj Pbl* ^ Y
27E7l — 27EJ (15.41)
Ch. I5\ Analytical Method of Determining Deformations 307

If force P is shifted to the centre of the beam, i.e. if we take a = 6 =

=*0-5/, the deflection in the section of application of force P becomes
, Pi* /ic o n
------- h l x s -------!-T S B 7 <I5*42)
If, on the other hand, force P is shifted towards the right support
so that in the limit b tends to zero, then for b-+Q
Xr — —4 s r = 0.577/
c VT
Thus, when force P is shifted from the middle of the beam to sup­
port B, the point of maximum deflection changes its abscissa merely
from 0.5/ to 0.577/ (Fig. 225) (see also formula (15.23)).
If force P acts as shown in Fig. 224, the deflection at the middle
of span is
h it— m r <15-43>

By substituting the numerical values of ail the quantities in for­

mulas (15.41) and (15.43) we can confirm that the difference between
the magnitudes of the two deflections is very small, which makes it

*1
I
I

Fig. 225

possible to calculate the deflection for practical purposes at the middle

of span without determining the location and magnitude of maximum
deflection. This is valid in all those cases in which the bending mo­
ment diagram is unique.

§ 88. Integrating the Differential Equation

for a Hinged Beam
In the preceding examples the portions into which the beam was
divided for writing the equation of the deflected axis corresponded
to similar portions of the bending moment diagram. The continuity of
the beam axis is broken by the hinge. Therefore, while integrating
the equation of the deflected axis, the portion containing the hinge
should be divided into two, although the bending moment equation
is the same on both sides of the hinge. Only the deflections of the de­
li*
 Deformation p u t to Bending {Part V

ments joined by a hinge are equal at the joint; the angles of rotation
of the sections are different. Therefore the equation of the deflected
axis is different for parts of the beam which are joined by a hinge.
Let us consider the beam shown in Fig. 226; there is a hinge in sec­
tion C. To keep the calculations simple we shall load the beam only
by a moment M acting in section B. The reaction B can be found

Fig. 226

easily by equating to zero the sum of the moments of all forces (i.e B
and M) to the right of the hinge with respect to point C. We get B= ~ .
Reaction A may be determined by taking the projection on the ver­
tical axis of all the forces acting on beam AC& (i.e. forces A and B).
We get A = ~ . The reactive moment M A is equal to the sum of mo­
ments of forces M and B about point A:

Let us select point A as the origin of coordinates. The bending mo­

ment in any section of the beam between A and B can be expressed
by the formula:

To obtain the equation of the deflected axis we must consider two

portions, AC and CB. The differential equations and their integrals
may be written as follows:
First Portion Second Portion

+ C lX + D. -{-CjX-j-Dj
Ch. 15\ Analytical Method of Determining Deformations 309

We have the following four conditions from which to determine the

constants of integration:
in section A: ^ - = 0 and at **=0
in section C: y t = y 3 at x=~a
in section B : y 2 = 0 at x = a + l
From the first two conditions we gel:
C, *=i0, O 1= 0

From the last two conditions we get:

iL <r>_\
i ( * - 2 ) - I VO ) + Cia +
M r (rt-MF
/ l-ir - a (d I) “t~Eit = ®

wherefrom
^ M (a + / ) * ( / - 2a) „ M( a +f ) Ht - 2 a ) o
0/2 • u z=‘

We shall explain the outlines of the solution for determining the

deflection of the beam shown in Fig. 227.
The beam has six portions, therefore we get 12 constants of integra­
tion when w’e write down equations for determining the deformations.

Fig. 227

It is evident that sections separating the different loaded portions,

the supports and the hinges will give us the required 12 equations.
The fixed end A gives two equations: the deflection equal to zero and
angle of rotation equal to zero. The hinged end Ogives oneequation:
deflection equal to zero. Hinges B and D give one equation each:
the portions to the left and right of the hinge have equal deflection
at the hinge.
Sections separating the differently loaded portions (section C,
where the distributed load finishes, and section F, where the concent­
rated force P is applied) give two equations each: the deflections of
portions to the left and right of the section are equal and the angles
of rotation of these portions are also equal. The hinged intermediate
support E gives three equations: equality of the deflections, equality
310 Deformation Due to Bending [Part V

of the angles of rotation and that both the deflections are zero in this
section.
In Fig. 227 the number of equations that each section gives has been
circumscribed.

§ 89. Superposition of Forces

Hooke’s law is true not only for the beam materia! but for the beam
as a whole; the deflections and angles of rotation are directly propor­
tional to the external forces. This is a direct outcome of the linear
relation between the bending moment and load, and the curvature
and the bending moment. For a beam fixed rigidly at one end and
loaded with a distributed force q and a concentrated force P acting
at the free end the bending moment in a section at a distance x from
the fixed end can be written as a function of force according to the
following formula:
jM(x) = — P (t— x )— q {± = ££ (15.10)

The relationship between curvature and bending moment is also li­

near:
E J $ £ = M ( x ) ~ - [ P ( l - X) + 2 l ! - (15.11)

Therefore upon integration w.r.t. x we get an expression for y as a li­

near function of external forces:
_ P^r x l ql*x* r 4* v*l
QEJ / J~~24£7 T + 7r J

In cases of compound loading this result enables us to obtain the

equation of the deflected axis by adding the ordinates of curves cor­
responding to individual forces. This simplifies the computation of
maximum deflection in some cases.
Let us study the application of the method of superposition of for­
ces in determining the deformation of a cantilever’s end A of a single­
span beam ABC (Fig. 228). By replacing the effect of the distributed
force q of cantilever AB on portion BC by a moment Af0—— ~ t
we can calculate the angle of rotation of the beam in section B by
using formula (15.22) given in § 84:
ft _ Mpl qazl
B ~ 3£7 ~ U£7
C h . 15] Analytical Method, of Determining Deformations 311

When section B rotates, the straight axis of cantilever AB also bends

by an angle 0fl and the deflection of the cantilever’s end A will be
M pla_ gto8
f'A = AA, = —QBa =
3EJ ~~ OCJ

Under the action of distributed force q the cantilever does not re­
main straight; it bends and acquires position AtJB (without, however,
changing the angle of rotation 0B in section £), and the deflection at

Ai
h
Fig. 228 Fig. 229

end A of the cantilever can be expressed by the same formula which

is employed for deflection of cantilever beams under bending (see
§83, formula (15.16))

the total deflection of cantilever end A will be:

r _ t' j f " Qlai <
1 °* _
lA lA -r /A -ggy 8 fy 24EJ

The displacements of hinged beams can also be determined by

using the method of superposition of forces. For this the beam should
be divided into the number of beams comprising it, each of these
beams should be studied separately and then the individual dis­
placements should be added up.
Thus, for example, the schematic diagram of the beam discussed
In § 88 (Fig. 229(a)) may be replaced by the diagram shown in Fig.
229(b). In this diagram the “suspended” beam CB is supported at its
left end C by the right end C of the main beam >1C. The effect of the
hinge may be replaced by forces C (Ffg. 229(a) and (d)).
312 Deformation Due to Bending [Part V

Force C can be determined by studying the equilibrium of beam

CB; for this beam, force C is a passive force as it is the reaction of
beam AC. An active force C of the same magnitude will act on beam
j4C—this force is the pressure of beam CB on beam AC.
The deflection of beam CBy shown separately in Fig. 229(c), can be
determined at any point. The deflection of beam AC may be deter­
mined as shown in Fig. 229(d). Both these cases were discussed in
§§ 83 and 84.
The deformation of beam ACB is shown in Fig. 229(c). Portion AC
of beam ACB experiences the same deflection over its whole length,
as beam A'C separately. The deflection of portion CB of beam ACB
consists of two deflections: deflection f Xl which is a component of
deflection fc (directly proportional to the distance from point £),
and deflection f a, calculated for beam CB
m in im according to the schematic diagram in
Fig. 229(c).

UHLJ (11|W § 90. Differential Relations in Bending

In §§ 57 and 82 we obtained differen­
tial relations for continuous load q{x),
Mm shearing force Q(x), bending moment
M (x), angle of rotation 0 and deflec­
i/itiif.. Ilk tion y:

dx
dy = 0
£2
dx
After certain transformations these
......... .... relations can be written in the follow-
ing -sequence:
Fig. 230
*37
dx (EJy) = EJ&

- i f r ( £ ' e> = ! M t o = Q to

= « W = 4 to

From the equations it is evident that knowing force q (£) and the
types of supports, we can obtain Q(x), M (x), EJQ and E Jy by succes­
sive integration: conversely, knowing the equation of the deflected
Ch. /5 | Graph-analytic Method 313

axis, by successive differentiation of EJy w.r.t. x we can obtain

EJ.Q, M (a-), Q(x) and <?(*). In graphic representation of these relations
we shajl lay off the positive values of the above quantities upwards
and the negative values downwards; the positive direction of thex-axis
will be towards the right, rotation; of the section in the clockwise di­
rection will be considered negative and in the anticlockwise direction
positive. Figure 230 contains diagrams depicting the taw of variation
of all quantities, which characterize the bending of a hinged beam
loaded with a non-uniform distributed force q (x) (the load is negative
as it is acting downwards).

C H A PTER 16

Graph-analytic Method of Calculating

Displacement in Bending
§ 91. Graph-analytic Method
The method, of integrating the differential equation of the deflected
axis gives us equations of deflections and angles of rotation, with the
help of which we can calculate the deflection and angle of rotation
in any section of the beam..
In a number of problems (statically indeterminate beams, determi­
nation of maximum deflection) it is sufficient to determine the de­
flection and angle of rotation for a
few definite sections. In such cases
it is more appropriate to use the
graph-analytic method. This method
is based on the resemblance of differ­
ential relations between deflection,
bending moment and intensity of m r f i a k J
continuous load.
Imagine a beam with an arbitra­
t 1 i
rily loading (Fig. 231). The differ­ (,• T’T 'T vJ
ential equation of the deflected 0 »
axis of this beam may be written
($82) as:
(16.1)
Fig. 231
Below the beam we draw another
beam of the same length loaded
by an, as yet unknown, continuous force qj, the positive direction
of which is taken upwards; we shall refrain from specifying the type
of supports also and shall only point out that the support reactions
must balance the external force qr. The second beam will be hereafter
314 Deformation Due to Bending [Part V

mentioned as the fictitious beam; all quantities relating to this beam

will be denoted with a symbol f. For this fictitious beam we shall de­
termine the bending moment M f in each section by integration, using
the differential equation that correlates the bending moment with the
intensity of the continuous force (§§ 57 and 90):

s r = < t/ (16.2)

Let us compare equations (16.1) and (16.2). If we assume that

qf = M(x)
i.e. if we load the fictitious beam with a fictitious force, which changes
according to the bending moment of the real beam, then
d3 (EJy) (PM f
dx* = dxi

If in integration we can achieve equality of the constants of inte­

gration on the left- and right-hand sides of the equation, i.e. Ct=Cr
and £>*=£>„ we shall obtain
d(E Jy) dMf
dx 1=1 dx ' EJy — My
dMf
Considering that and solving these equations for y
and 6, we get the following formulas:

Sf = | f (16.3)
8= | f (16.4)
Thus, deflection in the section of the real beam (due to the given
load) is equal to the bending moment in the same sect ion of the ficti­
tious beam (from the fictitious load), divided by the rigidity of the real
beam. Similarly, the angle of rotation of the real beam (due to the
given load) is equal to the shearing force in the same section of the
fictitious beam (from the fictitious load), divided by the rigidity of
the real beam.
In the analytical method of determining deformations, the constants
of integration were found from boundary conditions, i.e. by equating
to zero the deflections at the supports and equating the deformations in
sections common to two adjoining portions of the beam.
In the method under discussion the equality of constants of integra­
tion, while integrating equations (16.1) and (16.2), can be achieved
by fixing the ends (or intermediate sections) of the fictitious beam in
Ch. 16\ Graph-analytic Method 315

such a way so as to satisfy the following conditions, which directly

ensue from expressions (16.3) and (16.4):
(1) if deflection f of the real beam is zero, then the bending moment
in the corresponding section of the fictitious beam must be zero;
(2) if the angle of rotation 0 of the real beam is zero, then the shearing
force in the corresponding section of the fictitious beam must be zero;
(3) if the deflection and angle of rotation of the real beam are not
equal to zero, then the corresponding bending moment M f and shear­
ing force Qf must also not be zero.
Table 13 contains conditions for all types of supports of the real beam
and gives the constraints in corresponding sections of the fictitious
beam, which satisfy the conditions of constraint of the real beam.
In Fig. 232 are depicted the widely prevalent combinations of real
and fictitious beams for statically determinate structures. In each
pair any beam may be taken as the real, then the second automatically
becomes fictitious; this can be easily checked with the help of Table 13.
Tabic 13
Conditions for Obtaining the Proper Fictitious Beam

Real beam F ic titio u s beam

Type of support Conditions Required Constraints of the ficti­

for y and 6 conditions tious beam satisfying
for and these conditions
o,
Hinged support (no deflec­ #=o M ,= 0 Hinged support (no mo­
tion; rotation of section 0*0 Q /f 0 ment; support reaction
is possible) is possible)

Fixed end of the beam p=0 A f,= 0 Free end of the beam (no
(no deflection and no 0=0 Q/=0 moment and no concen­
rotation) trated force)

Free end of the beam y -f 0 Af; # 0 Fixed end of the beam

(both deflection and ro­ 8 *0 Q/j^O (both support reaction
tation are possible) and reactive moment
occur)

Intermediate support (no 0=0 M ,= 0 Intermediate hinge (no

deflection; rotation of 0*0 Qj £ 0 moment; hinge trans­
section is possible) mits force)

Intermediate hinge (both 0 9^ 0 Af ^ * 0 Intermediate support (both

deflection and rotation 0*0 Q/ 5*0 moment and support
of section is possible) reaction are possible)
.'JIG Deformation Due to Bending IPart V

In multispan beams with intermediate hinges the fictitious beam may

bo selected according to the method explained for the beam in Fig. 233.
It must be noted that the fictitious beam corresponding to a stati-
. caliy determinate real beam must also be
•m* statically determinate.
J Thus, in order to determine deflection y
X and angle of rotation 0 in a section of the
given (real) beam, we must follow the proce­
dure explained below:
\ (a) draw the given beam alongwith the
§ ) forces;
(b) draw the bending moment diagram
M(x)\
(c) assume the zero axis of the bending
} moment diagram as the axis of the fictitious
beam and the bending moment diagram M (x)
as the fictitious load q,; if the bending mo­
ment is positive the ordinates of force qj
1 must be directed upwards, if it is negative
s j qt must be directed downwards;
" (d) draw the supports of the fictitious
Fig. 232 beam in accordance with the conditions given
in Table 13 and Figs. 232 or 233:
(e) calculate the read ions of the fictitious beam due to the fictitious
load (i.e. the fictitious support reactions); for cantilever beams this
step may be bypassed;
(f) calculate bending moment M } in that section of the fictitious
beam which has the same abscissa as the section of the real beam in
which deflection f is required to be determined;

ft
S
Fig 233

fg) calculate shearing force Q/ in that section of Ihe fictitious beam

which has the same, abscissa as the section of the real beam in which
the angle of rotation 0 is required to be determined;
(h) calculate / and 0 according to formulas (J6.3) and (16.4).
The graph-analytic method of determining deformations relieves
us from calculating the constants of integration in each particular
case and with the help of data given in Table 13 and Figs. 232 o r233
offers a direct solution, which is in agreement with the given initial
conditions.
a t. i6) Gruph-mxatytic Method 317

Fictitious moments have the dimensions of force X (length)*,

fictitious shearing forces have the dimensions of force x (length)1,
and intensity of the fictitious load is measured in units of force X length.

§ 92. Examples of Determining Deformations by the

Graph-analytic Method
Let us find the deflection at point B of beam AB shown in Fig. 234.
The bending moment diagram for the above beam is a triangle with
the maximum ordinate in section A equal to PI. We shall take the axis
of the bending-moment diagram as the axis of the fictitious beam and

Fig. 234 Fig. 235

consider the bending-moment diagram as the fictitious load (this can

be done by putting downward arrows at the ends of the ordinates as
the ordinates are negative).
Following the instructions given in Table 13 we take point B as
the rigidly fixed end of the fictitious beam and point A as its free end.
Now we have to calculate the bending moment in section B of the
fictitious beam. The moment of the triangular load about point B
will be equal to the product of the area of the load, to, with the dis­
tance of its centre of gravity from section B:

Dividing this expression by EJ we get the deflection at point B\

The formula is exactly similar to the result obtained in § 83.

The shearing force in section J3 of the fictitious beam is numerically
equal to the area of the triangle:
$\6 Deformation t)ue to Bending [Part V
Therefore the angle of rotation in section B of the real beam is
Pit
0 * = Sejl 2 EJ

Let us find the deflection at the point of application of the force for a
simply supported beam loaded in the middle of the span by force P
(Fig. 235).
The bending-moment diagram of the real beam is a triangle with the
ordinate at the point of application of the force equal to + Con­
sider the bending-moment diagram as the fictitious load, with the
arrows pointing upwards, as the ordinates of the diagram are positive
in this example.
The supports of the fictitious beam can be determined according to
Table 13 so as to satisfy the conditions of constraint of the real beam.
From symmetry the reactions of the fictitious beam must be equal,
and each must be equal to half of the total load:

Af tf/ ~ 2 x 2 x 4 x * ' nr
Bending moment in section C is equal to the sum of the moment of
reaction (with a minus sign) and the moment of half of the triangular
lead (with arm //6):

M/ c = — A / X y + y Xy x X x lx i
_ />/* l p i* t pp
16 x 7 + ic x 6 48
wherefrom
pp
fc 48EJ

The angle of rotation at the left support is

a _Q/A Af PP
•4 EJ “ EJ “ 16EJ
because the shearing force at the support is equal to the support reac­
tion (in this example with a minus sign because A f is directed down­
wards). At support B

It is evident from the above examples that for the convention of

signs of fictitious load, bending moment and shearing force decided
earlier, the minus sign in the formula for deflection corresponds, as
before, to deflection downwards and in the formula for the angle of
Ch. 16) Graph-analytic Method. 319

rotation to rotation in a clockwise direction; positive sign corresponds

to the reverse directions.
Let us find by the graph-analytic method the deflection in the mid­
dle of the span and at the ends of cantilevers for the beam shown in
Fig. 236.
The bending-moment diagram is a trapezium with maximum ordi­
nates M = —Pa. Let us change the bending-moment diagram to a ficti­
tious load acting downwards. The fictitious beam consists of two small

i
Pa

Fig. 23G

cantilevers supporting the suspended beam AB. Deflection in the mid­

dle of the span (point F) is equal to the fictitious moment at this point
due to the distributed load divided by the rigidity of the beam:
_ P afi
' F ~~ 8EJ — 8EJ

The deflection at point C can be determined by first calculating the

fictitious bending moment in the section; the deflection is caused by
the triangular force acting on beam CA and the reaction of the suspen­
ded beam equal to A f— ^ ~ (Fig. 236):
P axa 2 Pa9 1 Pa*
Mj r - Aja 2 3
a=
~T 3

wherefrom the deflection in section C is

“ <557-(3* + 2a)
ic -~ B T
320 Deformation Due to Bending [Part V

In the last example we shall determine deflection in section. Dof

the beam shown in Fig. 237.
Let us first draw the bending moment diagram. The moment is
+Af in section C and 2eF0 in section B>. The moment changes linearly
over the whole length of the beam. We change the bending moment
diagram to the fictitious load and draw the fictitious beam according
to the conditions given in Table 1'3. Considering the cantilever AB,
we determine the fictitious bending moment at point B:

The corresponding deflection is

££ AH*
JYIl
'* 1E7

§ 93. The Graph-analytic Method Applied

to Curvilinear Bending-moment Diagrams
The bending-moment diagram of a uniformly distributed load is a
parabola.

>
Fig. 238 Fig. 239

The convex parabola (Fig. 238) is characteristic of simply supported

beams. The ordinates of this parabola are given by the expression
M (x) = -Q x— j X 3

The area of this parabola can be found from the expression

Ch, 16\ Graph-analytic Method 321

This.area may also be found* as two-thirds of the area of the circum­

scribed rectangle:

Abscissa jc0 of the centre of gravity of half of the parabola area

{Fig. 239) is calculated as follows:
//2

. 1 “ A «" 9.
* ~ ^ = l6

The distance between the centre of gravity of half of the parabola

area and the centre of the whole parabola is

The concave parabola represents the bending moment diagram of a

beam rigidly fixed at one end (Fig. 240). The ordinate of any point
on this parabola is found from
the expression Af (x)——qx% i2.
The area of the parabola r i* ’' i r ■i ' ■ 1
found as iS *1 >'
U ---------------/ — !-------------►

= ^ M (x) = ^ dx ~
q o
This area is also equal to one-
third of the circumscribed rec­
tangle

1 3 2 1 G
Ordinate x 0 of the centre of gravity of the parabola is

•qlx
*^0~~
i " 1'
The distance between the centre of gravity of the parabola and the
section of maximum bending moment (rigidly fixed end) is equal to
one-fourth of the beam span.
Let us solve the following examples using the relations derived here,
322 Deformation Due to Bending I Part V

Find the deflection at the middle of the beam shown in Fig. 238. The
area of the bending-moment diagram is taken as the fictitious load.
The diagram is positive, therefore the fictitious load is directed up­
wards. The fictitious support reactions are
© -as qP
—
Af 2 24

These reactions are directed downwards (Fig. 238).

The fictitious bending moment in the middle of the span is equal
to the sum of static moments of the fictitious forces located on one side
of the middle section, say, on the left-hand side. The forces located
to the left of the section are A f and the left half of the parabola. The
arm of force A } is equal to half of the span; arm of the half parabola is
3/
•g*. Therefore, the fictitious bending moment in the middle of the
span is
M J = — A / x 4 - + - y X -j| /

x T + 1 T x T§/ = -3 S 4 ?/4

and the deflection in the middle of the span is

Consider a beam rigidly fixed at one end and loaded by a uniformly

distributed force q (Fig. 240(a)). Let us find the deflection of the free
end. The bending-moment diagram of the real and the fictitious beams
is shown in Fig. 240(6).
The bending moment at the fixed end A of the fictitious beam is
equal to the product of the area «o of the complete diagram with the dis­
tance between A and its centre of gravity, i.e.
qP
8

and the deflection in section A is

t
' A~ WT

Consider beam ABC with one cantilever as shown in Fig. 241(a).

Using the method of breaking the diagrams and the method of super­
position of forces, find the deflection and angle of rotation in section C.
The beam is loaded all along its length by a uniformly distributed force.
Ch. 16} Oraph-anatytic Method 323

The possible bending moment diagram is in Fig. 241(6). Replace this

diagram by its components: from force q over length I (Fig. 241(c))
and from force q on the cantilever of length a (Fig. 241(d)). The ordi­
nates of the last two diagrams can be taken from the problems solved
earlier. The important ordinates have been written in Fig. 241(c)
and (d). The fictitious beam is g
shown in Fig. 241(c).
Let us isolate the fictitious
beam BC: it is acted upon by
pressure B f from the suspended
beam AB and the parabolic
force of maximum ordinate ~ -
(Fig. 241(f)).
In the fictitious beam AB,
taking the sum of moments of
all fictitious forces about sup­
port A, we find (Fig. 241(g))
that
,_ H ga* , I qP
t ~ 3 2 T * 1 2 12
qP
24"

Returning to Fig. 241(/) we

calculate

Q /c----- 4
qan
6

M/ c ----- B,a—

Fig. 241

The required values of deflection and angle of rotation in section C

are:

8C_ Q/c
~e T

fc = ~ w f = — t ’ + S a 1)
324 Deformation Due to Bending \Part V

CH A PTER 17

Non-uniform Beams
§ 94. Selecting the Section in Beams
of Uniform Strength
All preceding discussions were on beams of uniform section. In prac­
tice, however, we often have to deal with beams in which the cross-
sectional dimensions change either gradually or sharply.
We give below a few examples on selecting the dimensions of the
cross section and determining the deformation of non-uniform beams.
We know that bending moment usually varies along the length of
the beam; therefore, by determining the cross-sectional dimensions
from the condition of maximum bending moment we provide an extra
margin in all sections of the beam except the one which corresponds
to M -a„. Beams of uniform strength are used to achieve greater economy
of inelal and in some cases also to increase flexibility. Under this term
come beams in which the maximum normal stress is the same in all
sections and equal to the permissible stress (or less than it).
The dimensions of such a beam are calculated for the following con­
dition:

and
= ^ (17.2)

Here Mix) and ^ (x ) are the bending moment and section modulus
in any arbitrary section of the beam; in each section WP(x) must vary
in direct proportion to the bending moment.
Conditions (17.1) and (17.2) are true also for the section with the
maximum bending moment; if we denote the section modulus in the
section of maximum bending moment Afmax by fl?d, then
Afflux Mix ) (17.3)
= l*|

We shall explain the order of computations with Ihe help of the

following example. Consider a beam of span / rigidly fixed at end A
and loaded at the other end by a force P (Figs. 242 and 243). Assume
the beam to be of rectangular section. The problem of obtaining a
varying section modulus can be solved by changing either the height
or the width of the beam or both simultaneously.
Suppose the height of the beam is fixed, h=lu, and the width varies,
b—b{x). The section modulus at a distance x from the free end will
be BP(*)= , and the bending moment will b e —Px\ the section
Ch. 17\ Non-uniform Beams 325

modulus in the support section is W0~ M l , and the maximum

bending moment at the support is A i^ ^ -fP /j. Only the absolute va­
lues of M (.v) and A1WiS are required for computations. From formula
(17.3) we get
PI x G Px x 6
bch2 ** b {x) ft*
wherefrom
b { x )^ b n± (17.4)

i.e. width varies linearly as a function of x. At x= l the width is be.

The front view and plan of the beam are shown in Fig. 242. This
shape is obtained if we consider the strength of the beam only w.r.t.
the normal stresses; in section B the width of the beam is zero.

However, we must ensure sufficient strength of the beam under

shearing stresses also. The minimum width of the beam according
to this condition is determined from the following equation
X — 3<?n,ax . . »
wix ^mi«» 11
and, since Qmnx**P,
u _ 3P
ro,n ~ 2/i It]
The corrected shape of the beam is shown in Fig. 243.

§ 95. Practical Examples of Beams of Uniform

Strength
The example discussed above finds practical application in design
of springs. If we ignore its small curvature, a spring may be looked upon
as a simply supported beam (Fig. 244(a)) loaded with a force P in
the middle of its span and having reactions -y at its ends.
We design such a bar by the same principles as a beam of uniform
326 Deformation Due to Bending IPart V

strength of constant height /i0 and variable width b(x)-, as the loading
is symmetric it is sufficient to study just one half of the span.
The section moduli W (*) and W0 can be expressed by the same for*
mules as in the preceding example.
The maximum bending moment in
the middle of the span is:
Pi
4

Bending moment in any arbitrary

section is:

Solving, as in the preceding exam­

ple, we get:
»(*)“ &. •r (17.5)
The maximum width required to
successfully resist the shearing force
y can be determined from the fol­
lowing formula:
b
"min “- 34 A(|p[tj

Fig. 244 The front and top views of the

spring are shown in Fig. 244(6) and
(c). However, such a shape of the spring is highly inconvenient from
the practical point of view; therefore the shape is slightly modified
without affecting the performance of the spring. Imagine that the spring
is divided into thin strips when seen front the top, as shown in
Fig. 244(d). If we place these strips not adjacent to each other but
one over the other and neglect the friction between them, then without
affecting its working the spring may be given a shape the top and front
views of which are shown in Fig. 244(e) and (/), respectively.
Obviously, in actual practice each spring plate, the 1st, 2nd, etc.,
is manufactured in one piece and not in two halves.
Non-uniform beams are often used in mechanical engineering For
example, shafts are often designed as beams of uniform strength.

§ 96. Displacements in Non-uniform Beams

When determining the deflection and angle of rotation of a non-
uniform beam, it should be borne in mind that the rigidity of such a
beam is a function of x. Therefore, the differential equation of the de-
C h. 17) Non-unifortn Beams 327

fleeted axis may be written as

where J (x) is the variable moment of inertia in different beam sec­

tions.
Before integrating this equation we must express J (x) in terms of
/ , i.e. the moment of inertia of the section in which the maximum bend­
ing moment acts. Having done this, we can carry out the computa­
tions in the same manner as for a beam of uniform section (§ 82).
Let us show this through the example discussed earlier. We shall
determine the deflection in a beam of uniform strength (Fig. 242), which
is fixed at one end, loaded at the other by a force P and has a fixed
height. Let the free end of the beam be the origin of coordinates. Then
M {x )~ -P x , J( = (17.6)

The differential equation may be written as

p j &y _ P xl _
dx3 ~ x ~
PI (17.7)
Integrating twice,
EJ = — P lx + C , E Jy= —
We have the following conditions for determining the constants of
integration: at point A (x=l) deflection y —0 and angle of rotation
4^
dx
= 0 . Therefore
0 = — P ia-f C and 0= + C /+ D
wherefrom
C = PP and D= —
The expressions for y and 0 may be written as follows:
dy
dx
Pi
EJ
.. . Pi*
EJ
P t1
EJ 0- t )
PI PP PP
y= 2 EJ EJ 2EJ ~ - - S H '- * 4 + *)
Maximum deflection at the free end is obtained by putting x=0:
PP
/ m ax 2EJ
328 Deformation Due to Bending [Part V

If we had a beam of uniform section with a moment of inertia «/,

then the maximum deflection would be

3L7
or two-thirds greater.
Hence, non-uniform beams are more flexible than beams of uniform
section of the same strength. It is because of this property and not
due to saving of metal that non-uniform strength beams are used in the
manufacture of elements such as springs.
Equation <17.7) indicates that in this example the curvature of the
beam is constant, i.e. the beam axis deflects along a circle. Bui upon
integration the equation obtained was that of a parabola. It is sug­
gested that the Feader should explain, the reason for this.
When the graph-analytic method is used for determining the de­
formation qf non-uniform beams, it does not present any difficulties.
Instead of dividing the bending moment and shearing force in the
fictitious beam by EJ to compute / and A, we obtain Ihe fictitious load
by dividing the ordinates of the bending moment diagram of the real
beam by rigidity EJ. Then
an<J [ -M i, e = Qi

When applying this method to non-uniform beams, we assume that

Mix)
4i= FJ (x)
Then we load the fictitious beam by this force and obtain the required
deflection and angle of rotation as the bending moment and shearing
force in sections of the ficlitious beam.
Pxl PI
In Ihe example discussed above i.e. the
fictitious beam should be loaded not by a triangular force but by a
uniformly distributed force (Fig. 245). The deflection of section B,
which is equal to the bending moment in the fixed end of the fictitious
beam, can be expressed by the formula
pp
f-M } 2£7
We could have obtained the same result by assuming that the beam
has constant rigidity EJ and its bending moment diagram is obtained
by multiplying each ordinate by the ratio-J/ (Jtj ; Ihe ordinates of the
bending moment diagram thus obtained are
J n Jt
Ch. 17] Non-uniform Beams 329

(Fig. 246). Thai according to tihe general principle of the graph-ana*

lytic method
pp
Mj — — P l x l - j — ■— and y= 2 EJ

Thus, deformation of non-uniform beams can be calculated by the

same method as that for beams of uniform rigidity. The only differ­
ence is that the bending moment diagram used in this case is obtained
by multiplying with the ratio -J-..

4^ ' — I,
r
1 i
A
[J 4 H
It
l l l l l i i l l H I la 1
M- I -------- PI

Fig. 245 Fig. 246

Let us determine by thegraph^analytic method the deflection under

force P for a simply supported beam loaded at the middle of the span
by the above force P (Fig. 247(c))k The moment of inertia of the sec­

tion is J in the left half and 0.5J in the right half. Let us obtain the
new bending moment diagram by multiplying the ordinates of the
right half of the real bending moment diagram (Fig. 247(d)) with the
ratio y/0.5y=2; the fictitious beam with the new loading is shown
330 Deformation Due to Bending [P a rt V

in Fig. 247(c). The reaction of the left-hand fictitious support is:

* _ i pi i 2 , i pi i l pi*
At 1 4 2 3+ 2 2 T 3 13 12

The fictitious bending moment in section C and the deflection of point

C are:
M
m/ c ~----- TA ~ ±i ~4 T
- 1 1 — L
T~T
pn . Pi* PP
24 + 96 “ 33
and
pp
/c = 32PJ

Let us determine the deflection of a beam rigidly fixed at one end

and loaded at the other by a concentrated force (Fig. 248(a)). The cross-
sectional area of one half of the beam is greater, and J£> h. In order
to transform the bending moment diagram (Fig. 248(6)) into the ficti­
tious load, we must multiply the ordinates of the left-hand portion
of the diagram by ~ (Fig. 248(c)).
Deflection under force P may be calculated as follows (for J^= 2 J^\
M, ' ~ Tl Pp Tl Tl T2 Tl " ~ WT
D, 2 i l P I*
T
3P P
w j;
 PART VI
Potential Energy.
Statically Indeterminate Beams

C H A P T E R 18

Application of the Concept of Potential

Energy in Determining Displacements

§ 97. Statement of the Problem

Besides the methods of determining deflection and angle of rotation
discussed above, there is a more general method, which can be used
for determining deformation of any elastic structure. It is based on
the law of conservation of energy.
When a static tensile or compressive force is applied to an elastic
bar, transformation of potential energy from one form to another
takes place; a part of the potential energy of the force acting on the bar
changes into potential energy of strain. If we load the bar by successive
addition of small loads dP at its end (Fig. 249), then each addition will
be accompanied by a decrease in the level of the load and the potential
energy of strain will correspondingly increase.
This phenomenon is true for all types of deformation of an elastic
structure provided the loading is static. Such a construction may be
looked upon as a machine which converts one type of potential energy
into another.
We have agreed (§ 2) that a load will be called static if it increases
gradually, so that acceleration in the elements may be ignored; trans­
mission of pressure (force) from one part of the structure to another
does not affect the motion of these parts, i.e. their velocity remains
constant and acceleration is zero.
Under these conditions deformation of the structure is not accompa­
nied by any change in kinetic energy of the system; only conversion
of one form of potential energy into another takes place. In making
this statement we neglect the magnetic, electric and thermal effects,
which do not alter the deformation considerably.
As the motion of the elements of the structure does not change with
time, at each instant every part of the structure will be in equilibrium
under the action of external forces and forces of reaction, and each
element of a part will be in equilibrium under the external forces
332 Potential Energy. Statically Indeterminate Beams [Part VI

and stresses acting on it. Deformation of the structure, stresses in the

various parts, and reactions transferred from one part to another, ail
follow the increase in load. r
Thus, we may say that total conversion of one form of potential
energy into another takes place if deformation occurs without violating
the equilibrium of the system. Work done by the forces acting on the
structure serves as a measure of the energy transformed into another
form.
Let us denote the accumulated potential energy of strain by U and
the decrease in potential energy of the external forces by UP. The quan­
tity Up is determined as the positive work \VP done by Ihese forces;
on the other hand, the accumulated potential energy of strain U is
equal to the negative work W done by the internal intermolecular for­
ces (negative because the direction
of displacement of points of the
body due to deformation is opposite
to the internal forces).
The law of conservation of energy
for elastic systems may be expressed
as follows:
UP = U (18.1)
In this formula, replacing Up and
U by the corresponding values of
work Wp and W, we get a modified
iP form of the same law:
Wp = — W or lF/, + F = 0 (18.2)
Fig. 249 This formulation of the law of con­
servation of energy coincides with
the principle of virtual work as applied to elastic systems: equation
(18.2) states that the sum of work of all forces acting on a body is
zero if deformation of the body occurs without violating the equilib­
rium of the system.
Thus, the principle of virtual work as applied to elastic systems is
a corollary of the law of conservation of energy.
It ensues from formula (18.1) that the potential energy of strain U
is numerically equal to work Wp done by the external forces in causing
this strain:
U ^W p (18.3)

The following interpretation of this equation sometimes given in

books on structurat mechanics is erroneous: "Work done by the exter­
nal forces in deforming a body changes into potential energy of strain.”
Actually, only a different form of energy can change into potential
ch. m Potential Energy in Determining Displacements 333

energy of strain. As a rule, this is the potential energy of the external

forces. Work done by the external forces during this conversion is
only a numerical measure of the converted energy.

§ 98. Potential Energy in the Simplest Cases

of Loading
We have already derived the expressions for computing potential
energy in tension and compression (§ 10), shear (§ 36), torsion (§ 52),
and also in pure bending (§ 63(D)).
Let us write all the above-mentioned formulas in Table 14.
Table 14
Potential Energy ot Strain in Simplest Cases

T ype o» drform al Ion P o te n tia l energy of deform ation

1 aKI_ P * t -V*£ A
Tension or compression
T p s l 2Wa 5T
Shear
X v 20 A 2a
I API %*GJy
Torsion
— Mi<f‘~ 2 a T = ^ i r
1 . ,ii 0*EJ
Ppre bending
t * — a-

Let us have a look at the contents of the right half of the table.
The potential energy of strain is equal to half of the product of force
or moment of force couple with the displacement of the section in
which the force or force couple is applied. Let us use the term general­
ized force for every load that causes displacement, i.e. it may be a con­
centrated force or the moment of a force couple. The displacement
corresponding to the generalized force will be known as generalized
displacement. The word “corresponding” implies that we are talking
of displacement of the section in which the force under consideration
is acting. Elaborating further, we are talking of displacement which
when multiplied by the force gives us the work done. For a concentrat­
ed force this displacement will be linear in the direction of the force
(deflection, or elongation). For the moment of a force couple it will be
the angle of rotation of the section in the direction of the moment.
The formulas in the first column may be stated in a general manner as
follows: the potential energy of strain is numerically equal to half
of the product of the generalized force with the generalized displace­
ment.
334 Potential Energy. Statically Indeterminate Beams [Part VI

The second column in these formulas shows that the potential energy
of strain is a second order function of the independent external forces.
Potential energy is always positive.
The third column shows that the potential energy of strain is a sec­
ond order function of the finite values of generalized displacements—
elongations, angles of rotation, deflections—and is completely deter­
mined by the latter.
Consequently, although these formulas have been derived on the
assumption that the load increases statically without violating the equi­
librium of the structure during the process of loading, they are valid
for all types of forces provided the force and displacement are linearly
related and are considered at an instant when the structure has at­
tained equilibrium.

§ 99. Potential Energy for the Case

of Several Forces
Imagine a beam acted upon by several forces: Pu Ps, P*, . . . .
Let 6], §i, fi3, • . . denote the displacements of the beam in the sec­
tions of application of the forces and in the direction of their action.
In Fig. 2o0 the solid line shows the straight axis of the beam while
the dotted line shows it after deflection. We will assume that the fol­
lowing conditions are satisfied: (a) all forces are applied statically (their
magnitude increases gradually from zero to a finite value P(); (b)
all deformations are within the elastic limit and are linearly related
to the external forces; and (c) a decrease in the potential energy of the
applied force is accompanied by an increase in the potential energy
of strain of the beam.
Any of the forces Pt shown in Fig. 250 can be considered a general­
ized force. Here, the generalized force P, will not be just the active
force but a balanced force system (including support reactions) which
produces displacement 6 { at the point of application of the force in
the direction of its action.
All the forces and displacements are related to each other by the fol­
lowing expressions:

Here, a denotes constants and the subscripts must be interpreted as

follows: the first is the serial number of the displacement, or the “point
of displacement’’ (for instance, number 1 as “the first subscript denotes
the displacement in the section of application of force Pi)\ the second
is the serial number of the force causing the displacement, or the "cause
 Ch. 18] Potential Energy in Determining Displacements 335

of displacement” {number 2 as the second subscript denotes lhat the

displacement has been caused by force P*).
The system of equations (18.4) is known as ihe generalized Hooke's
law for a deformable body. The basic idea behind each line is that any
displacement represents the sum of displacements of the given point
due to each of forces Pt.
The generalized Hooke’s law (18.4) may also be called the law of
cumulative action of forces, or the law (principle) of superposition of

Fig. 250 Fig. 251

forces. We have used these formulas on more than one occasion for
deriving design equations (for example, equations (6.18) in § 33).
When a number of forces are acting, the potential energy should
be calculated by Clapeyron's theorem:

y = r = | 1 p , s a ... (i8.5)

the notations here are the same as in formula (18.4).

Clapeyron’s theorem may be stated as follows: the strain energy of
an elastic system due to a number of generalized forces is equal to
one-half of the sum of the products of the generalized forces and gener­
alized displacements caused by the simultaneous action of the former.
In conclusion, it should be pointed out lhat in principle anv group
of acting force factors that can be defined by one parameter can be
taken as the generalized force. However, from the practical point of
view, it is convenient to partition the complicated load acting on
the structure into simple generalized forces.
Let us consider an example. A beam that is rigidly fixed at one end
is loaded at the free end by a concentrated force P and a force couple
of moment M (Fig, 251). We shall calculate the potential energy of
strain of the beam.
Clapeyron’s theorem in this case can be written thus:

t/ = r = ! ( / > / „ + m „ ) (18.6)
386 Potential Energy. Statically Indeterminate Beams [Pari VI

The displacements may be taken from examples solved earlier or

from a handbook:
PI* Ml* A Ml PI*
/ » ~ 3EJ 2EJ ’ ** EJ T eT
(18.7)

The minus sign shows that the direction of displacement does not
coincide with that of the corresponding force. Let tis substitute the
displacements into dapeyron's theorem
u ~ ww
u 2 M 3IV
*2 EJ
£ . U ±2 -VI ^ EJ 2 EJ )
P*l* , M*l PM P
(18.8)
~ 6£7 + 2 EJ 2EJ

The leader’s attention is drawn to the fact that while calculating

potential energy due to a number of forces it is wrong to calculate the
potential energy due to each force separately and then sum them up.

§ 100. Calculating Bending Energy Using

Internal Forces
In general, the bending moment M (x) is a variable quantity. It
has a corresponding shearing force Qf*) in every section. Therefore,
it is expedient to consider the equilibrium of a small element of length
A,
r i * : rdd ■r /
r /IS j
Bat Mffl titx) / ! V B(x)
&___ & !__ L j
Qo) g)
¥(X C /
Of 4
\
4w
L \i
. dx -*■
i i
Fig. 252 Fig. 253 Fig. 254

dx instead of the whole beam (Fig. 252). Due to the bending action of
forces, the sections of the clement turn and make an angle 40 with
one another (Fig. 253). The shearing forces tend to shear (Fig. 254)
the element; thus the displacement due to the normal stresses is per­
pendicular to the direction of the shearing stresses, and vice versa. This
enables us to calculate independently the work done by the normal and
shearing forces.
Usually the work done by the shearing forces is small in comparison
to the work done by the normal forces; therefore, we shall not take it
into consideration for the time being. Elementary work done by the
Ch. J8\ Potential Energy In Determining Displacements 337

normal forces (as in simple bending) is:

clWf ^ d U = j M ( x ) d ^ ^ ^ M ( , x ) - ^ § ~ (18.9)

or dU - M>-$ r X (18.10)
Total potential energy of bending can be obtained by integrating
this expression over the whole length of the beam:
u = ^ M ^ L= ^ M H x)dx (18. n )

The limit of integration indicates that integration should cover the

whole length of the beam; if there are a number of zones for M (x),
then integral (18.11) must be divided into a sum of integrals.
We ena this section by calculating the potential energy of a simply
supported beam loaded by force P
(Fig. 266).. The bending moment di­
agram has two zones; therefore
(* Mldx , C M\dx (18.12)
J 2EJ 1 A 2EJ p kd
Dh V
M x — + A xl ---- h - I *1* A M \ “ 2L
p /l
h - f A '
M t — -f B x 2 = -f Pa
I x u --------- - / — —

§ 101. Casttgliano’s Theorem

Let us now explain the method of determining displacements via the
potential energy of strain. We shall determine the displacement of the
points of an elastic system in the direction of the forces acting on it.
We shall solve this problem in a number of stages, starting with the
simple case (Fig. 256) when concentrated forces Pu P a, P*. . . . act
in sections 1 ,2 ,3 , . . . of the beam. Due to these forces the beam bends
into curve / and there retains its equilibrium.
Let us denote by y u ya, y9, . . . the deflection of sections 1, 2 ,3 , . . .
in which forces Pi, P 2, P3, . . . are acting. We shall calculate one of
these deflections, say, yi (the deflection of the section in which force
Pi is acting).
12-3310
338 Potential Energy. Statically Indeterminate Beams [Part VI

Let us shift the beam from position I into an adjacent portion //,
shown in Fig. 256 by a dotted line, without disturbing its equilibrium.
This may be achieved by various methods: by adding a new force, by
increasing the existing forces, etc.
Let us assume that an infinitesimal increment dPi {Fig. 256) is
applied, in addition to force Pi, to shift the beam from position /
to deformed state //. In order to retain the equilibrium of the beam

Fig. 258 Fig. 257

during this sliift we assume that the increment is applied statically,

i.e. increases from zero to the final value slowly and gradually.
As the beam shifts from position I to position II all forces fall in
level, meaning thereby that the potential energy decreases. Since Ihe
equilibrium of the beam remains undisturbed, the decrease in potential
energy d(Jp of the forces may be considered to be completely transformed
into the potential energy of strain dU of the beam; dUp is measured
by the work of the external forces in shifting the beam from position
I to position II:
dU = dW r (18.13)
The change in the potential energy of strain, the energy being a
function of Torces Pu P2, P9, . . occurred due to an infinitesimal
increment in one of the independently applied forces, Pi. Therefore,
the differential of this composite function will be

Quantity dWp, in its turn, represents the difference in the work done
by all the forces in position II and in position I:
dWP= W a— Wi
If all the forces increase simultaneously and gradually, then work
Wi can be calculated as follows:
= y P ilh 4- y + y P*ys + •••
While calculating W2 we must consider that it depends entirely upon
the final shape of the deformed beam (§ 100) and not upon the order
in which the forces are applied.
Ch. 18) Potential Energy in Determining Displacements 339

Suppose we first load the beam by force dPi, the beam bends slightly
(Fig. 257, position III) and its deflections in sections 1, 2, 3 are dylt
dye, dys, respectively. The work accomplished by the static force
dPi is - j dPidyt. We now start loading the beam gradually and simul­
taneously by increasing forces Pi, P«, P«.
Deflections yu y9, y 3 will be added to the original deflections dyu
dy2t and dy9 (Fig. 257). In this stage of loading, forces P u Pa* Pa will ac­
complish work ^ P iy A - y P ^ r l - y ^ 8= ^ 1- In addition, force
dPJt which is already acting on the beam, will also accomplish work
(it traverses a distance y t\ since it remains constant during the second
stage of loading, the work done is dPxyi). The beam occupies position
II shown in Fig. 257 by a dotted line.
Hence, the total work done by the external forces in shifting the
beam from the undeformed state into position / / i s (Fig. 257):
ir,= i.d P id !,1+ r 1+ d p 1x !,1

Now we can calculate

dU = dWP« Wt - W t = dPxdyt + dPt x yt

Neglecting the second-order term, we get

dWP = dPiLX y i (18.15)
Putting the values of dU (18.14) and dWp (18.15) in equation (18.13),
we get

or

Hence, in this example the deflection at the point of application of

the concentrated force Pi is equal to the partial derivative of the poten­
tial energy of strain with respect to this force.
The result obtained above can be generalized. Suppose moments M
act in various sections of the beam besides the concentrated forces
(Fig. 258). We may repeat the preceding discussion for the case when
the beam is shifted from position I to position II due to the addition
of an infinitesimally small moment dMk to the original moment Mi.
The reasoning remains unchanged. However, when calculating the
work done by the moments, the Tatter should be multiplied not by the
deflections but by the angles of rotation Oj, 0a, . . etc. of the sections.
340 Potential Energy. Statically indeterminate Beams [Part VI

where the above moments are applied. ThendU will be equal t o ^ - x

xdM t, dWp will be equal to dMi and formula (18.16) will take
the form
(18.17)

As tfi is the displacement corresponding to force Pi and 0, the dis­

placement corresponding to moment Afj, the conclusions arrived at can

Fig. 25a

be formulated more broadly as follows: the derivative of strain energy

with respect to a generalized force is equal to the generalized dis­
placement.
This result is known as Castigliano's theorem. It was published in
1875.
We note that if the beam were acted upon by a distributed force, the
preceding derivations would still remain valid because every distrib­
uted force can be considered as consisting of a large number of con­
centrated forces.
The above discussion pertains to a beam, but it should be absolutely
clear that it can be repeated for any structure in which deformation
follows Hooke’s law.
For bending we obtained a formula which correlates the potential
energy with the bending moment:
y = (18.H)

Let us calculate the partial derivative of U w.r.t. one of the external

forces, for example, Pt:

We have to deal in this case with differentiation of a definite integral

w.r.t. parameter, as M (*) is a function of both Pi and jr; we integrate
w.r.t. x and differentiate w.r.t. P t-. We know also that if the limits of
integration are constant, then it is sufficient to differentiate the func­
tion under the sign of integration.
Ch. 18] Potential Energy in Determining Displacements 341

Thus, deflection at the point of application of concentrated force

Pi will be:
dU (* M(x)dx dM (x) /ioio\
^ = 'Sp7 = 3 — e7-------- W T (l818)
l
and the angle of rotation in the section under moment M i will be:
n dU P M (x)dx a.M(x)
=W ------ omT (l8,19)

The limit of integration, /, shows that integration is over the whole

length of the beam.

§ 102. Examples of Application of

Castigliano's Theorem
Let us calculate (Fig. 259) the deflection of the free end B of a beam
which is rigidly fixed at Its other end A. The beam is loaded by a con­
centrated force acting at point B. In this case we can directly apply
—H
B\ ^injrriTrfnTiiiiiiiiii B
I — , ------ 1 ,8

Fig. 269 Fig. 260

Castigliano’s theorem, because we are required to find the deflection

of the section where concentrated force P is applied:
dU p M ( x ) d x dM(x) /,Q io \
y* dP J El dP (18.18)

The origin for abscissa x may be selected arbitrarily, the only consid­
eration to be kept in mind being that the formula for M (x) should
be as simple as possible. Measuring x from point B, we gel the follow­
ing expression for the bending moment in an arbitrary section:
Af (jf) = — Px and = —x

Substituting these values in the formula for y B and integrating

over the whole length of the beam from 0 to I, we obtain:
„ P (-P x)dx, v P f .. .p i *
Vh = 3 El f EJ j x d x — + 3£y
o o
342 Potential Energy. Statically Indeterminate Beams (Pari V t

We have obtained the same formula as before, with the only differ­
ence that y B is positive. We have determined the displacement corre­
sponding to the force with respect to which the equation was differen­
tiated. By the term “corresponding” we mean that the product of the
force and corresponding displacement gives us the work done by the
above force. If the displacement is positive, the work will also be po­
sitive, which implies that the displacement is in the direction of the
force. If, however, the deflection or angle of rotation is negative, then
displacement occurs in a direction opposite to that of the force. Thus,
in this problem point B deflects downwards.
Let us consider an example in which it is essential to calculate the
reactions prior to calculating the bending moment M(x).
Let us calculate the angle of rotation at support B of a simply sup­
ported beam of span I (Fig. 260) loaded with a moment M acting at the.
above support and a uniformly distributed force<7over its whole length.
The required angle of rotation is:
q ... dU C M(x)dx OM (x)
i ~ E i -------m ~
Bending moment (Fig. 260) is expressed by the equation
M (x )-+ A x — 2 f

When we calculate the derivative of M (x) w.r.t. M the expression

for M (x) must contain only the independent external forces, which
are considered in Castigliano’s theorem. Therefore, reaction A must
be expressed through Af and q; if this is not done there is a chance
of making a mistake during differentiation by overlooking the fact
that A is a function of M and q. Reaction A is:
A ~ l L 4 .lL
A 2 + I
Therefore
M (x )^ x -!£ + 1 £

and the derivative is:

dM (x) , X
~ s r = + T
The limits of integration are determined from the condition that
the formula for the bending moment must be valid for the total length
of the beam. The required angle of rotation can be calculated as follows:
1
OU C \ f ql qx3 , /Wx \ x . qP . M l
0» = T j r = ) - E r ( - § - * - - r - + — ) 7 dA= 2 f e r + w
Ch. J8] Potential Energy in Determining Displacements 343

If the bending moment is expressed by different functions of x

in different portions, then the integral should be divided inio separate
integrals for each portion. The total displacement will then be equal
to the sum of all the integrals, which will be equal in number to the
number of different portions in the beam. When solving such a problem
it is extremely important to select the proper limits of integration.
I P

II
4
Fig. 261

Let us consider a beam of span I rigidly fixed at one end (Fig. 261),
loaded by moment M acting at point C at a distance a from the support,
and a force P acting at the free end. We have to determine the angle
of rotation in section C.
The point of application of moment Af divides the beam into two
portions; BC and AC. Therefore the angle of rotation of section C is:
a _ 9U _ p Mtdx dM^ , r AM* 3A*S
c dM J EJ dM M~

where Afi and are bending moments in the sections of the first
and second portions, respectively. The limits of integration can be
written only after we decide the point of reference from which to mea­
sure abscissa x for each section in the two portions.
Let us consider an arbitrary section in the first portion at a distance
x from the free end B. The bending moment in this section is:
Mi — -—Px and ~ ^

the limits of integration in this portion being 0 and I—a.

When calculating the bending moment in sections of the second por­
tion, we shall continue to measure x from the free end B\ then
A f , - —P x + M and ^ .= + 1

the limits of integration being I — a and L However, it is better to

measure x for the second portion in such a way that the lower limit
becomes zero (this simplifies calculations). Obviously, point C—the
initial point of the second portion—should be taken as the origin.
In this case we get:
Mt = - P ( x + l - a ) + M and -^ -= + 1

the limits of integration being 0 and a.

344 Potential Energy. Statically Indeterminate Beams [Pari VI

Considering the second version, we get:

A , f ° M t dx dMr , c M i d x 0Afa
] iJ ' 'M + ) - i r S M

The first integral is zero; therefore

Oc~-gj- j [ - P ( x + l - a ) + M ]d x = ----f(2 2^ ° ) ° + ~ E r

The required angle of rotation is the sum of two terms: one due to
force P in the clockwise direction (against the direction of M) and the
other due to moment M in the anticlockwise direction.

§ 103. Method of Introducing an External Force

Let us consider a beam of span I fixed rigidly at end A and loaded

at the free end B by a force P. Our aim is to determine the angle of
rotation of section B.
Direct application of Casligliano’s theorem is not possible, because
in this case the force does not correspond to the nature of deformation.

Fig. 262 Fig. 263

We have a concentrated force acting in section B instead of a moment.

To solve the problem we apply an additional moment at point B
(Fig. 262) in an arbitrary direction, say, for example, in the anticlock­
wise direction. For a beam loaded in this manner the angle of rotation
of section B can now be found by applying Cast igliano’s theorem.
The angle of rotation can be expressed by a formula consisting of two
terms: one depending on P ahd the other on M a. This formula is true
for all numerical values of P and M a including Afft=0. Therefore, by
assuming that Afa*=0 in the final expression, we obtain the expression
for angle of rotation due only to force P. The calculations are as follows:
a OU c M d x dM
B ~ d M a = , J EJ ~5M7
dM
M — -f Ma— Px and TMZ — -f-1
Ch. t8\ Potential Energy in Determining Displacements 345

The limits of integration are 0 and I; therefore

(18.20)

We may put Afa=*0 after integrating the above expression. But the
result will be the same if we put Afa= 0 in equation (18.20) and then
integrate. We require the additional force only to calculate, the partial
derivative of the bending moment w.r.t. this additional force. Having
found the partial derivative* we can safely equate the additional force
to zero.
Hence, the angle of rotation of section B due to force P is:

e = T T ^ ~ ' Px^dx==- ‘w r
o
The minus sign indicates that rotation occurs against the direction of
moment M<,, i.e. in the clockwise direction.
If it is required to calculate the deflection in a section of the beam
where no concentrated force is acting, we must similarly apply an addi­
tional force Pa in the above section and after obtaining the expression
for deflection equate the force to zero.
Let us determine the deflection of free end B of the cantilever shown
in Fig. 263. The beam is loaded by a uniformly distributed force. We
apply an additional force Pa in section B in order to calculate its de­
flection. The beam has two distinct portions: BC and CA. The de­
flection of B will be a sum of two integrals:
.. _ <>V _ f Mi dx dMi , r M-idx dMa
Jb dPa \ EJ dPa - r j ~ E J WZ

The reactions at the supports will be

n O. a) V(/*-<**)
I ~r 2 1
21

B P g-H . c)a
a l 21

The additional force should in no case be ignored while calculating

the reactions. We solve the problem by considering the additional
force as one of the active forces.
The way x coordinates are measured is shown in Fig. 263 for both
the portions.
In the first portion:
dMt
Af, —— Pax <l** 2 ’ OPa

the limits of integration being x= 0 and x —a.

346 Potential Energy. Statically Indeterminate Beams [Part VI

In the second portion:

M ,= + A X~ l f ----- p j s + S l S f U z - J *
OMt __ a
S E T ------7 x
the limits of integration being x= 0 and *■=/. Hence
a
~-p **— *£-) (—*) &
o

+ - e r j [ ~ p «T -t + x~ i£ ] ( — t x ) i x

Assuming Ptt= 0, we get

The first factor represents the deflection due to the load on the canti­
lever and the second the deflection due to the load between the supports.

§ 104. Theorem of Reciprocity of Works

With the help of the concept of potential energy we may derive the
following relation between deformations in various sections of a beam.
If we apply a static force P%in section 2 of a beam already loaded
by force Pu then to deflection yn of the point of application of force
P^ due to this force will be added a deflection yu due to force
Pt (Fig. 264). The first number in the subscript of y indicates the point
the deflection of which is required to be determined; the second num­
ber indicates the force causing this deflection.
The total work done by the external forces will consist of three terms:
work done by force Pi in causing deflection ylu i.e. - j Piyu \ work
done by force P 2 in deflecting the point of its application by y 22i i.e.
F*yi2; and, finally, work done by force P, over deflection yn caused
by the force P8, i.e. P\yi2.
Therefore, the total accumulated energy due to the two forces is:
^ ==y F 1yu - |- y P ^ „ - l- P 1y13 (18.21)
C h . /S ) Potential Energy in Determining Displacements 341

(the potential energy of strain depends only upon the final values of
forces and deflections and not upon the order of applying the external
forces).
Now, if we apply force Px to a beam already loaded by force Pt,
then reasoning in the same way we obtain
U = T p & M + 'jp iyu + p # n ( 18-22)
Comparing the two expressions for U, we get
(18.23)
i.e. the work done by force P, (or the first group of forces) over dis­
placements caused by force Ps (the second group of forces) is equal to
the work done by force Pa over displacements caused by force Pi.

i
Fig. 264

This is known as the theorem of reciprocity of works. It can be slated

in another way: work done by the first force (Pi) under the action of
the second force (P9) is equal to the work done by the second force un­
der the action of the first.
By taking a particular case when P t= P t, we obtain the theorem of
reciprocity of displacements: yii= un, i.e. deflection of point 1 due to
the force acting at point 2 , is equal to the deflection of point 2 due to the
force acting at point 1 .

§ 105. The Theorem of Maxwell and Mohr

Deflection of a beam at the point of application of concentrated force
P is:
.. CM{x)dxdM(x)
y jj EJ dP (1 8.18)

a similar expression can be obtained for the angle of rotation by re­

placing with Let us elucidate the physical meaning of
these derivatives.
If a beam is acted upon by an arbitrary number of concentrated
forces Pi, P i , . . ., moments M i, M t......... and distributed forces qt.
348 Potential Energy. Statically Indeterminate Beams (Part VI

qt, . . then moment M (x) in any section of the beam is a linear

function of all these factors:
M(x) = uJPl + a iPi + . . . . . . +£,<7i + c 2<72 . . .
(1&.24)
Coefficients au as.........bx, bit . . ., cx, c*t . . . are functions of the
beam span, the distances of the points of application of the various
forces and moments from the supports, and the abscissa x of the section
in which the bending moment is required to be calculated. Suppose

(a)
I
(b)
A

Pig. 265

we have to find the deflection of the point of application of force Pi.

dM
Then =au because in this differentiation P2, P3, . . Mr, M t,
.. <71, <?., . . .. a,, a2, . . ., bu Cu c........... are all constant
quantities. However, at may be taken as the numerical value of moment
M in an arbitrary section due to a unit force, i.e. P ,= l: it is evident
that by putting P i= l and equating all other forces to zero in equatiort
(18.24) we get M —ax.
For example, for the beam shown in Fig. 265(a) the bending moment
is:
M ( x ) ~ — Px — ?y

The derivative = —x, which is also the expression of the bend­

ing moment for the beam, if we load it by a unit force acting at point
B—the point of application of force P (Fig. 265(b))—in the same direc­
tion.
Similarly, the derivative of M (x) w.r.t. force couple Aft is numeri­
cally equal to the bending moment due to a unit force couple acting
in the same section as M x.
Hence the calculation of derivatives of a bending moment may be
replaced by the calculation of the bending moment due to a unit force.
We shall denote such moments by M°.
Thus, to determine displacement 6 (deflection or angle of rotation)
of an arbitrary section, irrespective of whether the corresponding force
acts in this section or not, we must write down the expressions Tor the
bending moment M (x) due to the given load (we shall denote it simply
Ch JB\ Potential Energy in Determining Displacements 849

by M) and for M 9 due to a corresponding unit force acting in the sec­

tion in which displacement 6 is required to be found. Then this dis­
placement will be given by the formula

&=l * T T dx (18.25)
U
This formula was first derived by J. C. Maxwell in 1864 and applied
in design practice by 0. Mohr in 1874.
If in formula (18.25) we want to define deflection as 6, then moment
M ° should be calculated for a unit concentrated force applied in the
section where the deflection is required. If, however, we want to calcu­
late the angle of rotation, then a unit moment should be applied.
For the example considered in Fig. 265 we have:
— Px — %
y (Fig.265(a))
Af*= - l x x = - x (Fig.265(b))

o
The plus sign indicates that the direction of displacement coincides
with the direction of the unit force; a minus sign would indicate the
opposite direction.
If the beam has to be divided into a number of portions to calculate
the bending moment in a section, then the integral in formula (18.25)
will also break into a sum of the respective integrals.

§106. Vereshchagin’s Method

Pirlet and A. N. Vereshchagin and before them H.F.B. Miilier-Bres-
lau proposed a simplification in calculations according to formula
(18.25). As the unit load is usually either a concentrated force or a
force couple (moment), Ihe Af#-diagram is bounded by straight lines.
In such cases, for any shape of the bending moment diagram, integral
^ MM°dx can be calculated as follows. Suppose the bending moment
diagram (Fig. 266) is represented by a curve, whereas the diagram for
M 9 is a straight line. The product Af dx may be considered as the area
element dto, which is shaded on the bending moment diagram.
As the ordinate M °=x tan a, the product M dx Af°=d© x tan a.
Hence integral J MM 9 d*=tan a ^ x d ® represents the static moment
of the area of the bending moment diagram about point A multiplied
by tan a. However, the static moment is equal to the total area co
350 Potential Energy. Statically Indeterminate Beams [Part VI

of the bending moment diagram multiplied by the d ista n c e ^ of its

centre of gravity from point A. Therefore
^ MM* dx = (oxc ta n a

But xc tan a is the ordinate of the Af0-diagram under the centre

of gravity of the bending moment diagram. Therefore
J MM* dx = (oMh

and the required displacement is

6 = -^ (18 26)

Hence, in order to determine displacement 6, we must calculate area

© of the bending moment diagram, multiply it by ordinate Af?; of
the unit bending moment diagram under the centre of gravity of area
co, and divide it by the rigidity of the beam, E J.

(ai

(d

Fig. 267

Let us determine by this method the angle of rotation of section D

of the beam shown in Fig. 267(a). The beam is loaded by a moment M
acting at the end B of the cantilever AB. The bending moment dia­
gram is shown in Fig. 267(6). Let us apply a unit moment in section D
in an arbitrary direction (Fig. 267(c)). The bending moment diagram
due to the unit load is shown in Fig. 267(d). As M is zero in portions
DC and CB, we are left with only one integral for portion AB.
Area © is equal to -FM/, and the ordinate of the M°-diagram under
the centre of gravity of area © is equal to -f ^ . Therefore the required
Ch. 18\ Potential Energy in Determining Displacements 351

angle of rotation 0o is:

= e j ( + MI) ( + 2 ?) * + lS c j
The plus sign indicates that rotation is in the direction of the unit
moment, i.e. in the clockwise direction.

§ 107. Displacements in Frames

Let us calculate the angle of rotation 0 of section C and horizontal
displacement A of point D of the frame shown in Fig. 268(a) with the
help of Mohr’s theorem.

Let us calculate the reactions and bending moments for all the three
states in Fig. 268(a), (b), and (c), respectively;
(a) due to the given load:
W= 0, D= £ = A , M,= + g x „ M .-+ M , M, = 0
(b) due to unit force:
H— D = \ = A, A?J= -fx,, A +x3
(c) due to unit moment:
= 0, D=^= A , « } = '+ g , /WS=0, MS=0
The deformations are:
(a) and (b)

MtM$dx+
a
^ 4 d x + \ M { a + x t)dx
352 Potential Energy. Statically Indeterminate Beams [Part VI

(a) and (c)

2Ma
.JL f *Sr £idx
EJ ) 2 a X i2aa x ZEJ

In solving the same problem by Vereshchagin’s method we must plot

bending moment diagrams for all the three loading arrangements in

Fig. 268. The diagrams, which are shown in Fig. 269(a), (6), and (c),
enable us to determine the following quantities:
©i = Ma, o)a = Ma, <o5= 0 (a)
M ^± a, Mc* = Y a (b)

.% = |. Met = 0. Met = 0 (c)

If there are three zones of loading, formula (18.26) can be written as

6 = ^7 K Mc i + + co3M%z)

The required displacements are:

A -r7 (* » T ‘ + * 4 « ) - ? £
2 2 Ma
* = 4 jM a 3 3 EJ
Ch. 18] Potential Energy in Determining Displacements 353

§ 108. Deflection of Beams Due to Shearing Force

In calculating deformations we have considered only the bending
moment. However, shearing forces also cause deflection. The Russian
scientist Prof. I.G. Bubnov was the first to determine the deformation
of a beam by considering the shearing forces.
Let us consider a beam rigidly fixed at one end and loaded at the
olher by a force P. Due to shearing stresses two adjacent sections
a,b\ and a*ba (Fig- 270(a)) separated by a distance dx will warp.
Maximum distortion will occur near the neutral axis; elements locat­

ed at the top and bottom surfaces of the beam will not warp.
The planes will occupy certain intermediate positions (dotted lines
c M i and c2 0 2 dt) making an angle y 0 with the original (Fig. 270(b)).
As in this case the shearing stresses are the same in all sections, they
will all turn by the same angley0and due to exclusive effect of shearing
stresses occupy the position shown in Fig. 270(b); end B will lower
w.r.t. support A. The deformation due io bending moment, which is
in the form of rotation of adjacent sections, is not shown in the figure.
Absolute deflection of the second section w.r.t. the first will be equal
to segment 0*0-1, i.e.
\dtjQ\ = 0 , 0 , (18.27)
In the general case, when the shearing force Q (x) is not constant but
varies along the length of the beam, angle y, will also vary. However,
the overall picture of deformation will remain unaffected; only dyQ
will be different for different elements of length dx.
Absolute deflection of the second section w.r.t. the first, \dyQ\.
may be calculated from the condition that potential energy' of strain
accumulated in the element of length dx during shearing is equal to
the work of external forces acting on the element: dUQ—dWp.
For the beam under consideration the external forces will be the
shearing stresses (the shearing force Q(x)). If the increase in load and
deformation is gradual, then the work done by these forces over a rela-
354 Potential Energy. Statically Indeterminate Beams [Part VI

tive displacement [dyQ | is

3 W |d s ,,| (18.28)

As the shearing stresses are not uniformly distributed over the sec­
tion, we have to take recourse to the method of differentiation in order
to determine the potential energy accumulated by the beam due to
these stresses.
Let us cut a small element of dimensions dx, dz, b(z) at a distance
x from the origin of coordinates and at a distance z from the neutral
axis (Fig. 270(a)) out of a rectangular beam (or a beam made of rec­
tangular beams), in addition to the normal stresses, the sides of tills
element will also be subjected to shearing stresses
_ Q{ x) S( z)
Jb\z )

For this element the potential energy of shear will be expressed by

the formula

Energy in the element of length dx and height A will be

Integration is carried out w.r.t. z, and the limits of integration are

selected so as to cover the whole section.
The above expression may be modified by multiplying and dividing
it with the cross-sectional area A:

where k is a dimensionless number which depends only upon the shape

and size of the beam and is:
(18.30)

Equating the values of dllQ and dWp, we get

wherefrom
ck. m Potential Energy in Determining Displacements 35S

The sign of deflection may be determined as follows. If the shearing

force Q (x) is positive and the y-axis is positive upwards, then the re­
lative deflection dyQ will be negative if we move from the left-hand
section towards the right (Fig. 270(b)). Consequently,
i y Q------(18.31)
Total deflection of any section having abscissa x is obtained by in­
tegrating expression (18.31):
y q =*—J k (18.32)
The constant of integration CQ depends upon the type of constraints.
Since Q(x)=*dM {x)!dxt we have
(18.33)

i.e. the deflection of the beam due to the shearing force is directly pro­
portional to the ordinate of the bending moment diagram with the
opposite sign; the ordinates are measured from a definite axis of abscis­
sas.
Constant k may be calculated for all types of sections. For a rectan­
gular section
'- f i . * w -* .
Therefore
/i/2 h/2
u A c S * (z)d z_ 9 r < V
} -? w — a j v ~ I T r z s
-h/2 0
Let us use the above result in determining the deflection of a beam
of span I, fixed at its left end A and loaded at the free end B by a con­
centrated force P. Assuming point A as the origin of coordinates, we
get:
M (x )------;> (/_ * ) and ( ,,= + * q ^ + C « = S £ ^ + C <,
6pi
At x = 0 the deflection y A=0; therefore CQ= — . D eflection^
may be written as
6 Px
#Q ~ 5GA

Maximum deflection occurs at point B, i.e. at the end of the beam

(where *= /):
' , 6 PI
IQ = 8GA
356 Potential Energy. Statically Indeterminate Beams \Part V!

Total deflection of point B is:

f Pi 3 6 Pi p p / , i 18EJ \
•** 3F7 S CA** 317V 1
1 /l*
As =* f°r a rectangular section, we get
/ ____ PP / i . 3 A* E \

a E 5
Assuming -y equal to y for metals and 20 for wood, we obtain:

f“ -J R ( 1 +7 7*) (for meta,s)

/ “» (1 + 6 y ) (for wood)

Thus we see that additional deflection due to the shearing force depends
upon . Therefore in comparatively short beams, especially in
wooden beams, it may acquire a high value. For example, if y « = y ,
then for a wooden beam 1+6^- = 1.375, i.e. deflection due to the
shearing force is 37.5% of the deflection due to the bending moment.
It should be noted that in a number of courses k is taken as 1 .5 and
not as 1.2 (for a rectangular section). This result is obtained by as­
suming that the deflection of the beam due to a shearing force depends
upon the shearing strain at the neutral surface, but this assumption
is erroneous.
It should be further noted that the displacements described above
Will not occur over some length near the fixed end (Fig. 270), but this
reduces the calculated deflection of the beam by a very small amount.

CHAPTER 19
Statically Indeterminate Beams
§ 109. Fundamental Concepts
Until now we have been considering only statically determinate
beams, in which the three support reactions can be determined from
equations of equilibrium. Very often the conditions in which the struc­
ture works require that the number of supports be increased; the beams
in these cases become statically indeterminate.
For example, to decrease the span of a simply supported beam
(Fig. 271(n)), we may put an additional support at the middle
cl m Statically Indeterminate Beams 357

(Fig. 271(6))', to reduce the deflection of the beam rigidly fixed at one
end (Fig. 272(a)), we may prop its free end (Fig. 272(6)).
The cross-sectional dimensions of these beams, as of tne beams dis­
cussed earlier, are obtained by plotting the shearing force and bending
moment diagrams (obviously after determining the support reactions).
A 0 B
fa
W
A c 8
fa i
(b)
F ig. 271 Fig. 272

In all such cases the number of possible support reactions exceeds

the number of static equations.
If the number of support reactions exceeds the number of static equa­
tions by one, the beam is known as single-degree statically indeterminate.
If the difference is greater, the beam becomes statically indeterminate by
two degrees, three degrees and so on. In this book we have considered
mostly single-degree statically indeterminate beams and also multiple-
degree statically indeterminate continuous beams.
The basic method employed for removing the static indeterminacy
of beams was proposed by C.-L.-M. Navier in 1826 and is based
upon integration of the differential equation of the deflected beam axis.
This method will be discussed in the next section.

§ 110. Removing Static Indeterminacy Via

the Differential Equation of the Deflected Beam Axis

If one hinged support is added to a statically determinate beam, it

makes the bpam single-degree statically indeterminate and simulta­
neously creates one new condition for determining the unknowns:
the deflection of the beam at the support is equal to zero. Therefore,
when the differential equation of the deflected beam axis is integrated
twice, the overall number of equations and unknowns is found to be
equal.
Lei us consider the beam shown in Fig. 273. The static equations for
the beam are: HA—0 (1), A + B —ql (2); and Bl—q ^ 0(3).
One reaction is immediately known. We are left with two equations,
(2) and (3), and three unknown support reactions: A, B, and MA.
The beam is thus single-degree statically indeterminate.
 358 Potential Energy. Statically Indeterminate Beams [Part VI

Let us now write the differential equation of the deflected beam axis
and integrate it twice:
E J f-B x -q Z (19.1)
E J y 'B a B y — q ^ - + C <19.2)
E J y -ir -jf+ c x + D (19.3)

On account of C and D the number of unknowns has increased to

five, but now the two static equations are supplemented by three con­
straint conditions: (1) y = 0 at x= 0 , (2) ^ = 0 at x= l, and (3) «=>0 at
x= l.

Fig. 273

It ensues from ( 1) that 0 = 0 . Conditions (2) and (3) when substituted

into equations (19.2) and (19.3), respectively, give

T T -£ + c = 0 (19-4)

T - U + C, = ° <19-5)
Dividing (19.5) by / and subtracting (19.5) from (19.4)
BP qP BP . qP A
T “ 6— T + 2?,= 0
From this equation we find
‘ (19.6)

From theequation of statics (2) we find A —^ q l. Next we determine

from equation (3) the moment in the rigidly fixed end,
The fact that the support reactions are positive indicates that their
directions shown in Fig. 273 are correct.
Substituting B in expression (19.4), we obtain
C ----- _______________
8 v* 2 T v § £48
Ch. 19] Statically Indeterminate Beams 359

Now, substituting B and C in equations (19.2) and (19.3) we get the

final equations for angles of rotation and deflections:
(19.7)

< 1 9 '8 >

Having determined support reactions 5 , A, and MA, we can now

plot the bending moment and shearing force diagrams by the usual
method.
With the help of equations (19.7) and (19.8) we can determine the
angle of rotation and vertical displacement of an arbitrary section of
the beam in the same manner as for statically determinate beams.
If the beam has a number of differently loaded zones, the static inde­
terminacy may be removed either by using the method of equating
the integration constants (Clebsch’s method, § 85) or by the general
equations of the method of initial parameters (§ 86).

§111. Concepts of Redundant Unknown and Base Beam

After considering the beam shown in Fig. 273 we established that
the number of equations of statics was one less than the number of the
unknown support reactions. One of the reactions is a superfluous or,
as it is sometimes called, a “redundant” unknown. This term has taken
deep roots in technical literature although it can be applied only with
certain reservations. Obviously, the extra reaction and the correspond­
ing support constraint are redundant only from the point of view
of their necessity in the equilibrium of the beam as one rigid body.
From the engineer’s point of view in a number of cases the extra sup­
port is not redundant but is actually a helpful tool in designing
structures.
In a number of methods employed for removing static indeterminacy
of beams, we write down conditions expressing the compatibility of
displacements in that section, where the “redundant” reaction is acting.
These conditions along with the usual equations of statics enable us
to determine all the unknown support reactions.
In § 110 for the beam shown in Fig. 273 we had two equations of
statics for determining three unknown support reactions A, B, and MA.
Any of the three can be taken as the redundant reaction. Let us choose
the reaction of support B as redundant. In this case we can argue that
the given beam is obtained from the statically determinate beam AB
with end A rigidly fixed (Fig. 274); end B is later propped up by an
additional support.
The statically determinate beam obtained from the statically inde­
terminate beam by removing the “redundant” constraint is known as the
360 Potential Energy. Statically Indeterminate Beams [Part V/

base beam. By selecting one of the reactions as redundant we at the

same time select the base beam.
Let us now try to transform the base beam (Fig. 274) into a beam
which is completely indentical to the given statically indeterminate
beam (Fig. 273). For this we load the base beam with the distributed
force q and apply a ‘'redundant” reaction B at Us end B (Fig. 275).

J - ■B gl i m ’ll ii m uItLU
tii-i 11
i i
!•«- I
Fig. 274 Fig. 276

However, this is not sufficient. In the beam shown in Fig. 275, point
B may move vertically under the action of forces q and B, whereas
in the actual statically indeterminate beam (Fig. 273) point B does not
have this freedom: it must remain attached to the supporting hinge.
Therefore, to make Figs. 273 and 275 Identical, we must add the con­
dition that the deflection of point B due to forces <7 and B must be zero:
=0 (19.9)
This is the additional equation which enables us to determine re­
action B. It represents the condition of joint deformation as applied
to this case: end B does not detach from the support. This additional
equation can be solved by a number of methods.

§ 112. Method of Comparison of Displacements

Equation (19.9) ffl= 0 , which was obtained in § 111 and which
expresses the condition of joint deformation, may be solved as follows.
The total deflection of point B of the base beam due to forces q and
B is made up of two deflections: due to force q and f sa due to
force B. Therefore
0 (19.10)

We have to calculate these deflections. First load the base beam only
by force q (Fig. 276(a)). The deflection of point B will be
f = ^
YEJ

Let us load the base beam by “redundant” reaction B (Fig. 276(6)).

The deflection of point B in this case will be
an
C/t. 191 Statically Indeterminate Beams 361

Substituting these values in equation (19.10) we get

— $ n + § £ = 0

wherefrbin B = 2$L t I.e. the same as obtained earlier in § 110 (19.0).

In this method we first allow the base beam to deform under force
q, and then select a force B which returns point B to its original posi­
tion. Thus we select the unknown reaction B such that the deflections
a_________ £_______
iiiiiiimmiiimmiiiifumi

Fig. 276 Fig. 277

due to q and B neutralize each other. This method is known as the

method of comparison of displacements.
The remaining reactions are (see § 110)

5t - m

The bending moment expression is obtained by considering the right-

hand side of the beam (Fig. 275) and substituting the value of B cal­
culated above (19.6):

Shearing force Q is expressed by the formula

Q = — B + qx=*— q ( f — x)

The bending moment and shearing force diagrams are shown in

Fig. 277. The section of maximum bending moment corresponds to
362 Potential Energy. Statically Indeterminate Beams [Pari VI

abscissa x«, which may be obtained from the following relation:

dM 3ql_
T 0| i.e. 8 ‘ •<7.vo= 0
3/
wherefrom -g-. The corresponding ordinate of the bending mo­
ment diagram is:

Mmax - T 8 2 64

§ 113. Application of the Theorems of Castigliano

and Mohr and Vereshchagin’s Method
The indeterminacy of the beam discussed in §§ 110-112 can also
be removed by Castigliano’s theorem (§ 101).
The “redundant” reaction B (Fig. 278(a)) is replaced by a redundant
unknown force B, which acts on the statically determinate beam AB
(Fig. 278(b)) along with the given
force q.
By differentiating the expression
for potential energy w.r.t. B and
equating the deflection f B thus ob­
(1 tained to zero we may write equa­
* 1 00 tion (19.9) as follows:
i
q i
f 4 r^ = ° c 9-1')
ci iin n iiiin H in iiiH iiiim 0
We now have to calculate M and
(b) t g^-and integrate within the appro­
priate limits
Fig. 278
M = + B x —^ , = x (19.12)
We assume that the beam has a uniform section all along its length;
after dividing by EJ equation (19.11) may be written as
t
J ( a : - 2 r ) j(‘i* = 0 09-13)
wherefrom
B (19.6)
After this the solution is the same as in the method of displacement com­
parison. . .
Ch. 19\ Statically Indeterminate Beams 363

After the indeterminacy of the beam has been removed, displacements

in statically indeterminate beams are determined in a manner exactly
similar to that used for statically determinate beams. If an additional
force has to be applied for determining displacements (§ 103), the force
should be assumed to act on the base beam. Under these circumstances
the additional force only affects the main reactions and the redundant
reaction must lie regarded as an active force, as before.

Fig. 279 Fig. 280

If the same problem (Fig. 273) is solved by Mohr’s method, then in

addition to the first state when it is loaded by the given forces and the
redundant unknown force (Fig. 279(a)) we must show the beam in the
second state to be loaded by force P6 (Fig. 279(b)). Using the notations
of Fig. 279, we obtain
M = Bx— q £ , (19.14)

i.e. the same as obtained by applying Castigliano’s theorem.

When solving the same problem by Vereshchagin's method, in addi­
tion to the two loading diagrams used above (Fig. 280(a) and (b))
we must also plot bending moment diagrams due to forceq (rig. 280(c)),
force B (Fig. 280(d)) and force F °= l (Fig. 280(c)).
364 Potential Energy. Statically Indeterminate Beams [Pert VI

The areas of the bending moment diagrams are:

for force q: <*>„- — 7 7 / “ — j

for force B\ co* = y B / x / = y

The corresponding ordinates of the bending moment diagram of unit
force arc:
3/
multiplication factor for w,: =
<>/
multiplication factor for (aB: Af£ =
Deflection of point B is:
f I (8 1 s 21 qP 3/ \ n
t* - T r\ T T “ ttJ = °
wherefrom
B -T ll
After this the solution is the same as explained in §§ 110 and 112.

§ 114. Solution of a Simple Statically

indeterminate Frame
Plot the bending moment diagram for the given frame (Fig. 281).
The elements of the frame have uniform rigidity, which is constant
*
p
G
I
e

7)7. ’’7.
Fig. 281 Fig. 282

along their length. Denoting the reactions by A , H, MAt and C, we

write the following equations of statics:
H = P , A + C = 0, P a -C a — MA = Q
There is one redundant unknown; let this be reaction C. The base
beam loaded with the force P and the redundant unknown is shown
in Fig. 282.
Ch. 19) Statically Indeterminate Beams 365

Let us solve the problem by applying Castigliano’s theorem. In

the equation of joint deformation
dU

potential energy U is the sum of the energies of the first portion, CB,
and the second portion, BA. Con­
sequently, equation f c= 0 can be h*-------- a-------- h
written as
a

h l M* i i r d x = 0

The moments and their derivatives

are:
Mi t-Cxly

= + C a —Pxa, + Fig. 283

Substituting these values in the equation of joint deformation, we get

<z a
$Cx*dA*-{-^ (Ca— PxJ a d x—0

After integtating we get

C
—m3+ Ca*----Pfl*
r = 0a, a
C = 3j Pn
3
The bending moments are: in the first portion M i=~^Pxx and in
3
the second portion Afa = ■§■Pa—Pxt. The bending moment diagram
is shown in Fig. 283.
In solving the above problem by Vereshchagin’s method we depict
two states of loading of the beam: with the given forces and reaction
C (Fig. 284(a)) acting on the beam, and with a unit force acting in
the direction of reaction C (Fig. 284(d)). Next we plot the bending
moment diagrams M and M°. Areas of the bending moment diagrams
(Fig. 284 (fr) and (c)) for the given load are:
I Cn^
e>1= y C a x a = -f - -2 -

(o'= -\-Caxa = Cai and <«£= — y P a x a = —

vOO Potential Energy. Statically Indeterminate Beams [Part VI

W (b) (c) (0 (e)

Fig. 284

The ordinates of the unit bending moment diagrams corresponding

to the centres of gravity of the bending moment diagrams for the given
loads are (Fig. 284 (6) and (c)):
M'c- + f a , A**c,= + a, M%=+a

The condition of joint deformation (after factoring out EJ) is:

2 . paz
y y a+Ca*a—-~^-a = 0

wherefrom C = ^ P , which is the result obtained earlier by applying

Castigliano's theorem.

§ US. Analysis of Continuous Beams

From a practical point of view a very important category of indeter­

minate beams are the continuous beams, which lie on a number of in­
termediate supports to which they are hinged. The ends of such beams
may be either hinged or rigidly fixed. Let us first discuss a case when

Fig. 285

the beam has hinged supports. In continuous beams one of the end sup­
ports is usually fixed whereas all others are capable of moving. The
numbering of supports and spans will be from left to right, the extreme
left support will be denoted by 0 and the extreme left span by 1 .
Lengths of the spans will be denoted by letter / with the number of the
corresponding span as a subscript. We shall assume that the beam is
of uniform section and consequently its rigidity EJ is constant. Fig­
ure 285 shows a continuous beam with appropriate notations, and
also the support reactions. It can be easily seen that the number of
CU. I9\ Statically Indeterminate Beams 367

redundant support reactions is equal to the number of intermediate

supports.
If we were to follow the method discussed above, we would take the
reactions of the intermediate supports as redundant unknowns and a
beam simply supported at points 0 and n+2 as the base beam. Addi­
tional equations would be obtained by equating to zero the deflections
of the points of intermediate supports of the base beam. However, there
is a simpler and more popular method which makes use of a different
type of base beam and redundant unknowns; in this method there are
not more than three unknowns in each equation.
Selection of the redundant unknovws and base beam is closely in-
terlinked. The statically determi­
nate base beam is obtained from the
statically indeterminate beam by t
removing constraints that are the
redundant unknowns.
The problem can also be approa­
ched in a different manner. Convert
by some method the statically in­
determinate beam into a statically
determinate beam and study which
of the reactions and constraints
must be removed to achieve this.
These reactions will constitute the
redundant unknowns in the stati­
cally indeterminate beam.
Thus, in the two-span continuous Fig. 286
beam (Fig. 286(a)), the reaction
of intermediate support B may be taken as the redundant unknown.
Then the base beam will be a beam simply supported at points A
and C; the beam can, however, be made statically determinate by in­
troduction of a hinge at point D (Fig. 286(6)). The base beam system
will consist of cantilever CBD and suspended beam AD. By introduc­
ing a hinge we impose the condition that the bending moment and
hence the normal stresses in section D should be zero. Thus, when we
consider the base beam system we actually equate to zero the normal
stresses in section D acting from the left portion on the right and vice
versa. These stresses give resultant moments equal in magnitude to
the bending moment in section D. These moments reapplied to the
base beam are shown in Fig. 286(c).
While transforming our statically indeterminate beam into stati­
cally determinate by introducing hinge D, we- select the bending mo­
ment in section D as the redundant unknown instead of one of the sup­
port reactions.
Point D may be selected arbitrarily. However, the computations
are simplified considerably if we select point D in the section of the
368 Potential Energy. Statically Indeterminate Beams [Part VI

beam just above the intermediate support, point B, i.e. if we consider

the moment at support B as the redundant unknown. Now the base beam
system will consist of two simply supported beams hinged at points
A, B, and'C and having a common support at point B.
This precisely is how the base beam system is selected in continuous
beams. The bending moments M„ M .......... M a„lt M n, Afn+, at
the intermediate supports are taken as the redundant unknowns.
Such a selection of the redundant unknowns simplifies the equations
from which the former are calculated. The equations may be written
in general form with the help of the theorem of three moments.

§116. The Theorem of Three Moments

To derive the theorem of three moments let us consider a continuous
beam having a number of spans of different lengths, l», etc., and
loaded by vertical forces acting arbitrarily (Fig. 287(a)). Let us first
show all the reactions which may occur in this case. From the equilib­
rium of the beam it is evident that the horizontal reaction He—0.

firm ft nm
ll vffl?. 1 p
to

The base beam (Fig. 287 (b)) is obtained by introducing hinges at all
the intermediate supports. Then the redundant unknowns are the bend­
ing moments Mu Aftt, etc., acting at the intermediate supports. Mo­
ments at the end supports must be zero. Let us load the base beam by
the external forces and the moments acting at the supports (Fig. 287 (c)).
As the direction of support moments is not known, we consider them
positive. After the solution is completed the sign of the result will
show whether the assumed direction is correct or not.
The next step is to write down the condition which imposes the same
restrictions on the deformation of the base beam as are present in the
continuous beam. In the base beam the spans on both sides of the nth
hinge which separates them may rotate (Fig. 288) due to the external
load independent of each other. Let us denote by the angle of ro­
tation of the span to the left of the nth hinge, and by the angle of
rotation of the span to the right of the nth hinge. These possible angles
of rotation of two adjacent spans are shown in rig. 288. In a continuous
Ch. 191 Statically Indeterminate Beams 3C8

beam both sections coincide and simply represent two sides of the ssme
section. Therefore, the condition of joint deformation may be written
as
o ;-e ;= o (19.15)

This is the condition which must be satisfied by adjacent spans at

support n of the base beam loaded by the external forces and support
moments. Such a condition may be written for all intermediate sup­
ports and, consequently, the number of additional equations that we
obtain is equal to the number of re­
dundant unknowns.
Let us take an example to eluci­
date how condition (19.15) can be
expressed mathematically. Consider
a two-span continuous beam (Fig.
289(a)) loaded by dilferent distri­
buted forces and qit acting on
the two spans.
The base beam loaded only by
the external forces is shown in Fig.
289(b). For clarity the two adjacent
spans have been shown slightly se­
parated at support /; actually, hin­
ges V and I" coincide.
Both sides of section 1 of the sup­ Fig. 288
port will turn as shown in the diag­
ram. The deformations must be the same in a continuous beam; this
can be achieved by loading the base beam by a negative support mo­
ment Mi (Fig. 289(c)) of such magnitude that the deformations become
equal. It follows from the above discussion that the deformations
will be equal only when the following condition is satisfied:
(19.15')

Returning to the analysis of statically indeterminate beams by

strain comparison and considering Fig. 287, we must expand equation
(19.15) by calculating Ihe deformations involved in it.
In the base beam the angles of rotation at support rt depend only upon
the deformation of two adjacent spans In and ln+i. Let us isolate these
two spans together with the forces acting on them (Fig. 290). Span
ln is acted upon by the external forces applied to it as well as support
■moments Afn_* and Af„, and span J„+i is acted upon by support mo­
ments M n and M n+i in addition to the external forces applied to it.
For clarity the two adjacent spans have been shown slightly separated
at support n; actually, hinges n ' and n* coincide.
13—3310
£70 Potential Energy, Statically Indeterminate Beams [Part VI

We shall determine % and 0„ by the graph-analytic method. The

fictitious beams, shown below their respective spans, are also hinged.
The fictitious load of the left span is made up of:
(a) the bending moment diagram of the external forces, obtained
by multiplying load area <on with the distance an of its centre of gra-

&) 1 Miiiiiauiiiuuimiiumumiiiiiimiiuti
2

Fig. 289

vity from the left support (as the bending moment diagram is positive,
the load ordinates are drawn with arrows pointing vertically upwards;
if the ordinates of the bending moment diagram are negative, area
o>R is used in calculations with a minus sign);

Fig. 290

(b) the triangular bending moment diagram from positive support

moment
(c) the triangular bending moment diagram from positive support
moment M n.
The right-hand fictitious beam corresponding to span /rt+1 is acted
upon by the following forces:
c/t. m Statically Indeterminate Beams 371

(a) the bending moment diagram of the external forces, obtained

by multiplying load area <an+, with the distance bn +t of its centre of
gravity from tne right support;
(b) the triangular bending moment diagram from positive support
moment M„;
(e) the tpiangular bending moment diagram from positive support
moment Mn+X.
The angle of rotation of section n is equal to the shearing force of
the fictitious beam in this section divided by the rigidity of the beam:

The shearing force at the support is equal to the support reaction R'n
of the fictitious beam.
Let us calculate this reaction. Load area is distributed between
the supports of the fictitious beam as in a lever arrangement, exerting
a force of 2" at support ri. The triangular load with the maximum
ordinate M n gives a reaction at the support which is two-thirds of its
total value, whereas the triangular load with the ordinate Mn^x
gives a reaction which is one-third of its total value. Hence

The fictitious shearing force (& is equal to this reaction taken with a
positive sign:
Qk = R'„
The angle of rotation % is:

( K f j + 2 M A + AJ..-A)
In a similar manner we obtain reaction Ri, for the right span:

The shearing force in this case is equal to the support reaction taken
with a negative sign:

The angle of rotation 0£ is:

e« ----- g ^ ( 6t0
13*
372 Potential Energy. Statically Indeterminate Beams [Part VI

Substituting the values of and 0« in equation (19.15) and can*

celling out 6EJ, we obtain:

or
M„,ll„ + 2 M ,(l.+ ‘. . t ) + M nilt ' . i ----- 6 ( ^ L + ^ ± j h ± l ) (19.16)
\ ln *«+i /
which is the equation of three moments.
We can write as many equations of this type as the number of inter­
mediate supports, i.e. as the number of unknown support moments.
Once the support moments are known the problem becomes one of ana­
lyzing a number of simply supported beams loaded by the external
forces and known support moments.
The brackets on the right-hand side of equation (19.16) contain the
sum of fictitious reactions at the middle support due to the given load
acting on the adjoining spans. Consequently, the theorem of three mo­
ments (19.16) may be formulated in short as follows:
iM.-l/.+ 2 A lll(/. + / „ 1) + « . +i / . H - - 6 / ? t (19.17)
Here Rft represents the fictitious reaction of support (n) due to ben­
ding moment diagrams M (*) of the given load on the two adjoining
spans.

117. An Example on Application of the Theorem

of Three Moments
Let us consider a three-span continuous beam of uniform section
which is loaded as shown in Fig. 291(a). Start numbering the supports
from left to right. The equation of three moments should be written
twice: for supports I and 2 .

We shall need the areas of the bending moment diagrams of externa!

forces acting on the base beam. These diagrams are shown in Fig. 291(b).
Let us write the equation of three moments (19.17) for support J. Assum-
Ch. 19) Statically Indeterminate Beams 373

ing n= 1, we obtain
*W „>1 = .M 0 = Ot g>„ = w, = 0

®r«+i=«>i= + + 1T> *«+i —

The equation can be written
2A*i (/, + / , ) -f-Ay2= — -|p /5 (ig.18)

Let us pass over to support 2 now. Assuming n= 2, we get

M n + 1 ~ Mt —0,

0),'/i +j © — -i- Ha 2^1 as 4.

0)3 - + 7 T “ + T2
. I*
?n+l T
The second equation of three moments is:

+ = (19.19)
The redundant unknowns Afi and Af# can be calculated by sol vine
equations (19.18) and (19.19).
If we consider a particular case and assume and al=P
(Fig. 292 (a)), we obtain

M‘= - m p' and M>— m pl

Knowing the support moments, we can easily plot the bending mo­
ment diagram of the continuous beam without any additional calcula­
tions. To do this we first draw the bending moment diagrams of the
base beam system due to the given load (Fig. 292(d)). The bending
moment diagrams due to support moments Afi and M , are shown in
Fig. 292(c). The resultant diagram with the characteristic ordinates
is shown in Fig. 292 (d). The bending moment diagram may also be plot­
ted without moving the sections apart at the support; in the given
problem this was done for the sake of clarity.
The support reactions may be calculated for each span separately.
The two reactions determined separately for each intermediate sup­
port may then be algebraically summed up.
The support reactions can be computed in another way. The sum
of the moments of all forces to the left of support J about the point of
support is equated to the support moment Afi:
374 Potential Energy. Statically Indeterminate Beams [Part VI

which implies that

(the minus sign means that reaction A acts vertically downwards).

We now consider the two left spans. The sum of the moments of all

Fig. 292

forces about point 2 is equated to the support moment Afa:

A21 + fl/ — P x 0.5f = Mt = — ^ PI

After substituting the value of A and certain computations we obtain

57
p (directed upwards). Next we consider the extreme left
span:
W - 0 .B M - M ,-----

And, finally, we determine C after considering the two right spans:

IH U -’ p i + C l - i P t - M , — T ^p.
Ch. 19] Statically Indeterminate Beams 375

Let us finally check whether the calculations were correct:

We see that they are. With all the reactions calculated the shearing
force diagrams for all the spans can be plotted without any difficulty
(Fig. 292(d)).

§ 118. Continuous Beams with Cantilevers.

Beams with Rigidly Fixed Ends
The theorem of three moments can be easily applied to situations
when the beam has cantilevers or when the ends (one or both) of the
beam are rigidly fixed.
Let us consider a two-span beam with a cantilever (Fig. 293(a)).
which works under the following conditions:
/x= 6m, /2= 5m, c= 2m , q = 4tf/m
The moment Mc may be considered as known and equal to the bend­
ing moment in section C due to the load acting on the cantilever.

Fig. 293

Thus, we have the following data to write the equation of three mo­
ments:

The equation is:

2Af, (/, + /a) - 2 £ ! i%» - 6 & y (19.20)
376 Poientiat Energy. Statically Indeterminate Beams [Part VI

or

wherefrom
M jss—3.86tf-m
When determining the support reactions, it is recommended to
study beam AB and the beam with cantilever BCD separately. The
bending moment and shearing force
diagrams are shown in Fig. 293 (b)
and (c).
In order to explain how to solve
the problem when one end of the
beam is rigidly fixed, we must first
study the design of the constraint
(Fig. 294).
The fixed end may be considered
as propped from below at point A
and above at point B or vice versa.
Such a construction cannot be con­
sidered absolutely rigid, because the
portion of length l\ between points
A and B is capable of undergoing
Fig. 294 deformation, and the beam section
which coincides with the front face
of the watl can turn as a conse-
quence. The smaller the length h and the greater the moment of iner­
tia of the portion and the lower the pliability of the wall, the more ri­
gid is the constraint. We shall get an absolutely rigid fixation by as­
suming that in the limit /j= 0 (or Ji=oo). While analyzing continuous
■beams with fixed ends, we must replace the fixation by an additional
span, write the equation of three moments and then obtain the condi­
tions for the actual beam by substituting
li — 0 or soo
Let us consider a beam rigidly fixed at both ends and loaded by a
force P acting at distances a and b from the left and right supports
respectively (Fig. 295(a)). We assume that supports A and B do not
impede longitudinal deformation of the beam. We remove the const­
raints and add a span on each side thus reducing our problem to the
analysis of a three-span continuous beam (Fig. 295(b)).
We have the following data for writing the equation of three mo­
ments at support 1 \
Ch. /$! Statically Indeterminate Beams 377

The equation is:

2Mt ((, + y + AV, - 6 ^ (1 + £ ) (19-21)

The data for writing the equation of three

moments at support 2 (n=2) is as follows:
= m„ £ = £ 2 5 (i + £ )
«,1+i = “ j = 0
Therefore
Mll, + lM ,(l, + l , ) ---6ir(l + £)
(19.22)
Now we substitute /t—/3= 0 and U—l in
the above equations of three moments (19.21)
and (19.22). We obtain the following set gf
equations:

M,l + U \J * =Pab (1+ f )

Solving the equations, we get
Pab‘l Pa*b
Ml= I* * Mt = I*
The moment in the section under force P is:
Mtb . Mta 2 PaW
Mp cf*P y •+ " T ’ + T “ 73
The bending moment diagram plotted for this data is shown in
Fig. 295(a).
 PART VII
Resistance
Under Compound Loading

CHAPTER 20

Unsymmetric Bending

§119. Fundamental Concepts

Until now we have studied problems in which the elements of a

structure are subjected to only one of the fundamental deformations:
simple tension or compression, torsion, or planar bending. In actual
practice a majority of the elements of structures and machines are
acted upon by forces which give rise to two or more types of deforma­
tions simultaneously.
Shafts in machines are subjected to torsion as well as bending. Be­
sides tension or compression bars of trusses (rafters, bridges and
cranes) also experience bending, because of welded and riveted joints
at corners instead of hinges for which the trusses are actually designed.
All such cases in which we have a combination of fundamental defor­
mations are cases of compound loading.
Analysis of compound loading is usually based on the principle of
superposition of forces, i.e. it is assumed that the effect of deformation
caused by one of the forces on the deformation caused by the rest of
the forces is negligible. Experiments confirm that this principle can be
applied when deformations are small (exceptional situations when it is
not applicable at all will be discussed later). Hence the principle
of superposition of forces may be applied to determine total stresses
and deformations in an elastic system subjected to compound loading
of an arbitrary nature, i.e. stresses and strains corresponding to the
various types of fundamental deformations may be added geometri­
cally.
Let us first study the particular cases of compound loading and then
the case when the elastic system is subjected to the most general com­
pound loading.
Ch. 20] Vnsymmeiric Bending 379

§120. Unsymmetric Bending.

Determination of Stresses
Till now we have been using the formula o = ~ for calculating
the normal stresses in bending. However, normal stresses in a section
of the beam can be completely determined by this formula only in
case of uni-planar bending*, when the beam bends in the plane of
action of the forces and the neutral axis is perpendicular to the plane
of loading and represents the principal axis of inertia.

Fig. 296

In actual practice we often come across cases when the plane of ap­
plication of the forces does not coincide with any of the two principal
axes of inertia of the section. Experiments show that under such load­
ing the axis of the bent beam does not lie in the plane of application
of the forces; this is known as unsymmetric bending.

Roof beams are usually acted upon by forces the plane of applica­
tion of which makes a considerably large angle with the principal axes

• Speaking more accurately, this will occur when all the forces lie in one of the
principal plsnes of inertia of the section passing through the bending centre; in a
number of cases the bending centre coincides with the centre of gravity of the cross
section (§79).
380 Resistance Under Compound Loading [Part VII

(Fig. 296). We also often come across cases when the plane of applies-
tion of forces is only slightly inclined to the principal axes of inertia.
We shall explain the method of checking the strength and calculat­
ing the deformation in case of unsymraetric bending with the help of
the following example.
Consider a beam rigidly fixed at one end and loaded at the other by
a force P which acts on the face of the beam and makes an angle <p
with the principal axis Bz (Fig. 297). The second principal axis By
is perpendicular to the first; let us select the direction of the coordinate
axes such that force P lies in the first
quadrant.
For checking the strength of the
beam we must first find the point
which experiences the maximum nor­
mal stress. Let us derive an expres­
sion for the normal stress at any poi nt
of an arbitrary section at a distance
x from the free end of the beam.
Let us divide force P into compo­
nents, P z and Py, which are directed
along the principal axes Bz and By.
The values of these components may
be calculated by the following for­
mulas:
Fig. 298
P£ = P cos tp and P]t, = Psin<p
Thus, we have reduced unsym-
metric bending to a combination of
two planar bendings caused by forces Pt and Py which act in the prin­
cipal planes of inertia of the beam. Adding the stresses and deforma­
tions for each of these bendings, we find their total values in unsymmet-
ric bending.
The bending moments due to forces P z and Pu in the section having
abscissa x are:
I I = Ptx = Pxcos cp= M cos ip |
J Me (= PyX Px sin <p= Af sin tp j t •)
The subscripts y and a of Af denote the principal axes about which
the moments have been calculated; Af denotes the bending moment in
the plane of application of force P, and its value in the given section is
Px. If we depict the moments in vector form, we notice that we can
obtain Af„ and Af, by directly resolving the total bending moment Af
along the principal axes (Fig. 298).
To determine the signs of the bending moments in a three-dimen­
sional problem like this, it is necessary to find additional conditions (we
 CA. 20\ Uasymmetric Bending 381

shall explain this point below). Let us restrict ourselves to finding the
magnitude of the bending moments only; the effect of the direction of
bending moments on the sign of stresses will be taken into account
when the latter are calculated.
We determine the stresses at point C (having coordinafes y and z)
lying in the first quadrant (Fig. 297), We can separately calculate the
normal stresses at this point caused by moments M tJ and M z which
bend the beam in principal planes xz and xy respectively. The formu­
las derived for planar bending are valid in this case too.
The normal stress at point C due to bending moment M y is compres­
sive (negative) and may be expressed by the formula
M„z
■1< •] -----Aj—
t*
COS (f>
Jv
where J v is the moment of inertia of the section about the t/-axis which
is also the neutral axis for bending due to moment M v. Moment M t
will also give rise to compressive stresses at point C equal to

where J t is the moment of inertia of the section about thez-axis. Total

stress at point C is obtained as the algebraic sum of the stresses calcu­
lated above:
( 20. 2)
Jy Jl \ Jy JZ }
The above formula may be used for calculating the stresses at any
point in any section of the beam. As the formula has been derived for
a point with positive coordinates y and z, we shall always get the stres­
ses with their proper signs if we substitute y and z with proper signs in
formula (20.2).
Thus, at point D (Fig. 297) y is positive but z is negative. Conse­
quently. the first term in formula (20.2) will become positive whereas
the second will remain negative as before.
Although formula (20.2) has been obtained by considering a particu­
lar case of a beam rigidly fixed at one end and loaded at the other, it
is not difficult to notice that it is a general formula for calculating
stresses in unsymmetric bending. Only the rules for finding the pro­
per sign of the stresses will be different for beams which are loaded or
constrained in a different manner. If the positive direction of the
principal axes of inertia passing through the centroid is always se­
lected in such a way that the plane of application of forces always pas­
ses through the first quadrant, then the sign before the right-hand side
of formula (20.2) should be in accordance with the nature of deforma­
tion which takes place due to the bending moment (or its components)
at any point in the first quadrant (a positive sign in case of tension and
382 Resistance Under Compound Loading [Part VII

a negative sign in case of compression). Now it would suffice to use

the proper signs of y and z to obtain the proper signs of the stresses
within the elastic limits at any point from formula (20.2).
In order to determine the maximum normal stresses we must first
locate the critical section of the beam and then the maximum stressed
point of this critical section. It is evident from formula (20.2) that the
critical section is the section in which the bending moment M is maxi­
mal.
While finding the maximum stressed point we must bear in mind
that in uni-planar bending the deformation due to normal stresses is ro­
tation of the sections about their respective neutral axes. In unsymmet­
ric bending, which is a combination of two uni-planar bendings, there
is simultaneous rotation of the sections about two axes which intersect
at the centre of gravity of the section.
We know from kinematics that rotation of a body about two axes
may be replaced by rotation about an axis passing through the point
of intersection of the two axes. Hence, in unsymmetric bending also
in every section we have a line which passes through its centre of gra­
vity and about which the section rotates during deformation of the
beam. This axis will be the neutral axis: the fibres of tiie beam mate­
rial lying in its plane will neither elongate nor shorten and the normal
stresses at points on the neutral axis will be zero. In relative rotation
of two sections the maximum deformation (tension or compression)
occurs in the fibres which are farthest from the neutral axis.
Hence, the problem of determining the maximum stressed points in
unsymmetric bending is reduced to locating the neutral axis and the
points farthest from it.
Equation of the neutral axis can be written from the condition that
normal stresses are zero at points lying on the neutral axis. Let us de­
note the coordinates of a point on the neutral axis by y 0 and z«; substi­
tuting these values for y and z in formula (20.2), we get the value of
a equal to zero:
q ^ __Af cos <pX20 M sin q>Xy0

Dividing by —M, we get

cosyxz, + sinyxgq (20.3)
Jv Jg
This is the equation of the neutral axis. It represents a straight line
passing through the centre of gravity of the section (at y o = 0 and zo=0).
Figure 299 shows two beam sections; the y • and the z-axis are the
principal axes of inertia. Assuming that the beams are loaded as in
Fig. 297, the projection of force P has been shown in both the sections
and the proper signs of the normal stresses have been given for each
quadrant; the signs above and below the section are for stresses due
Ch. 20\ Unsymmetric Bending 383

to moment M Vi whereas the signs to the right and left of the section
are for moment M z. For a beam which is loaded and constrained in a
different way (Fig. 300), the signs of the stresses will also change ac­
cordingly.

Approximate location of the neutral axis is shown in Fig. 299 As

the neutral axis passes through the centre of gravity, it is sufficient
to know angle a which it makes with
the //-axis in order to locate it fully.
It is evident from Fig. 299 that the tan­
gent of this angle is equal to the abso­
lute value of the ratio of 20 to #0:
ta n a = Is.

From equation (20.3) we obtain

tan ct = I— tan <p-r' (20.4)
Hence, the location of the neutral
axis does not depend upon the magni­
tude of force P, but only upon the
angle which the plane of application
of external forces makes with the 2-axis
and upon the shape of the section.
After calculating angle a from for­
mula (20.4), we plot the neutral axis Fi6- 300
on the diagram, and by drawing tan­
gents to the section parallel to the neutral axis we find the maximum
stressed points, which are the points farthest from the neutral axis
(poults / and 2 in Fig. 299).
384 Resistance Under Compound Loading [Part VIJ

Substituting the coordinates of these points {yu it, or yt, z t) with

their proper signs in formula (20.2), we calculate the maximum tensile
or compressive stresses. The strength condition for the beam may be
written as
K .X I = Mm , ( ^ '. + < fo) (20.5)

where y{ and Z\ (or u, and are the coordinates of the point (in the
coordinate system of principal axes passing through the centroid) far­
thest from the neutral axis.

For section with corners in which both the principal axes of inertia
are the axes of symmetry (rectangle, I-beam), i.e.
KIHi/a| = Iynml and K IH 2«IH*«nJ
formula (20.5) may be simplified and the expression for o<1>a) may
be written as follow's:

( 20. 6)

The strength condition for such sections is as follows:

!^fa«| = % ^ (c o s « p -l-^ s in « p )< [o j (20.7)
w
While selecting the section we set the value of »and knowing to],
M*,-, and angle <pwe find by trial and error the values of Wy and Wz
which satisfy the strength condition (20.7). In unsymmetric sections
without comers, i.e. when we use strength condition (20.5), the loca-
Ch. 20J Uasymmetric Bending 385

tion of neutral axis and the coordinates of the farthest point (yu z*)
must be determined everv time beforehand. For a rectangular section,
w h ' h
= Therefore assuming the ratio -y known we can easily find
Wu and the dimensions of the section from condition (20.7).
The diagrams showing the distribution of stresses in a rectangular
section are given in Fig. 301.
It is clear from equation (20.4) that angles a and <p are not equal,
i.e. the neutral axis is not perpendicular to the plane of application of
external forces as was the case in uni-planar bending. The perpendicu­
larity can be achieved only if
J u = Jz (20.8)
but then all axes become the principal axes, and unsvmmetric bending
becomes impossible; irrespective of the plane of loading we shall have
uni-planar bending. This w-ill be true for square, circular and all
other sections which satisfy equation (20.8).
The shearing stresses may also be calculated by a method similar to
the one adopted for determining the normal stresses; the total shearing
stress will be equal to the geometric sum of the stresses due to bending
in each of the principal planes. Usually the value of the shearing
stresses has no practical importance.

§ 121. Determining Displacements in Unsymmetric Bending

We shall again apply the principle of superposiiion of forces lode-
termine the deflection in various sections of a beam subjected to unsym-
metric bending. Considering the same example discussed in the preced­
ing section, we shall first find the deflection of point B (free end of the
beam) only due to force F,; the deflection is in the direction of the
z-axis and is

where i is the span of the beam. Similarly, the deflection of point B

due to a single force Pu is in the direction of the y -axis and may be ex­
pressed
t pvp PP*i*9

Total deflection f of the free end of the beam is equal to the geomet­
ric sum of the two deflections:
/ - K / J + fl (20.9)
Also
[ u __ 5 in <fJy
( 20. 10)
fs COS <p T f
38$ Resistance Under Compound Loading IPari VII

and
fy h
1 sin a cosa
It follows from this relation that the angle between total deflection
/ and the z-axis is equal to os, i.e. deflection / is perpendicular to the
neutral axis. The beam bends not in the plane of application of forces,
but in a plane perpendicular to the neutral axis (Fig. 302).
We shall consider the s/-axis the principal axis with the maximum
moment of inertia, then plane xOz will be the plane of maximum rigi­
dity, because the deflection of the beam is minimal in this plane. If,
as in the examples discussed above,
and hence a>q>, the plane
of bending deflects from the plane of
maximum rigidity more than the
plane of application of external for-
j
ces. The greater the ratio the
*Z
greater the difference. Hence, in
narrow and high sections in which
the ratio of the principal moments
of inertia may be quite great, even
a small deviation of the plane of ap­
plication of forces from the plane of
maximum rigidity will give rise to
considerable deviation of the plane
Fig. 302 of bending of the beam.
As long as the externa! forces act­
ing on the beam of such a section
lie in the plane of maximum rigidity xOz, the beam deflects in the
same plane and the magnitude of deflections is small because of the
moment of inertia J u being large. But as soon as the plane of applica­
tion of forces deviates from axis Oz by a small angle <p, there is a large
increase in the deflections in the direction of y-axis (the designer very
often overlooks this factor). The deflections in the direction of the
2-axis are, however, almost unaffected. Let us take a numerical exam­
ple to study this phenomenon. Consider a timber beam (Fig. 297)
A=20cm high and b—6 cm wide. Then

^ = ^ ~ ^ « 4°0°cmS / a= £ ^ « 3 6 0 c m «

Ratio of the moments of inertia is:

Ch. 20] Un$ymmtric Bending 387

When the plane of application of forces deflects by 5° from the z-axis,

we obtain
tana= tan< p 7^=0.0875 x 11 «* 0.963, a«44°
Ji
Deflections in the direction of the ^-axis will almost be equal to the
deflections in the direction of the z-axis:
fv = ft tan a —0.963/*
Moreover, the deviation of the plane of application of forces from
the plane of maximum rigidity is accompanied by a considerable in­
crease in the normal stresses. In the example discussed above the ma­
ximum normal stresses (as compared to uni-planar bending when (p=0)
increase in the ratio (see formula (20.6))

Wy ( c°3<p^ sm<t>) „ ^j + JLtan9) cosc p « ( l+ ^ 0.0875) 1= 1.29

Fig. 303 shows the relative location of the neutral surface, the plane
of bending and the plane of loading.
Beams in which the principal moments of inertia of sections differ
considerably from one another, work satisfactorily if bending occurs

in the plane of maximum rigidity (high rectangular sections, I-beams,

channel bars). They, however, fail under unsymmetric bending. There­
fore in situations where the designer is not very sure of a sufficiently
accurate coincidence of the plane of loading with the principal plane,
he should avoid using such sections or make additional provisions
(by putting constraints) to prevent lateral deformation, which might
occur during unsymmetric bending.
However, careless reinforcement of the existing structures may be
extremely harmful. We know a case when a beam of channel section
388 Resistance Under Compound Loading {Part Vli

consisting of a plate and two angles (Fig. 304(a)), working und^r a load
acting in plane xOz was reinforced by welding to it an extra angle
(Fig. 304(A)). This resulted in deviation of the principal axes from the
plane of loading and gave rise to deformation in the lateral direction,
which was completely unforeseen.
Example- Select the section for a wooden lath of height h and width
b and determine the. deflection of its middle point. Assume that y —2;
the length of the beam (distance between the two supporting trusses)
is /= 4 m and the roof is inclined at 23° to the horizontal; the load due
to the lath’s weight and the weight of snow on the roof may be consid­
ered as uniformly distributed and having the intensity q—A00 kgf/m.
The lath is simply supported. Permissible stress is 100 kgf/cm3, and
the modulus of elasticity Is £=10* kgf/cm3.
Maximum bending moment will occur at the middle of the span;
it will be
- 800 kgt • m

As angle <p is equal to the angle of slant of the rod, i.e. 25°, it fol­
lows from formula (20.7) and condition y B=2 that
m bh2 A3 Afmax / , A . \
" 12 ^ “W " ( C0S‘P+ 7 3,0 ^ J

™ n ^ ° - 906 + 2 x 0 -423) = 1402 Cm*

wherefrom IC ^ ]/12 x 1402=25.6«*26 cm and b—\Z cm.

Maximum deflection of the beam occurs at the middle of the span.
Moments of inertia of the section are:
T _bh* _ 13X26* _ . m 139X26
Jy —~J2 —fs— — lUuoUcm , = jj- =4760cm 4

The angle of inclination, a, of the neutral axis can be determined as

follows:
ta n a = tan<p;jr = tan250i ^ j - = 1.865

wherefrom a=>61°50', and the angle made by the plane of bending

with the plane of loading is:
a —<p= 61°50'—25°= 36°50'
The deflection in the plane of maximum rigidity is:
hql* cos <p 5x4X 44x0.906xl0®
u : m J vE ‘ 384X19059X104
= 0.64 cm
Cfc. 21] Combined Bending and Tendon 289
Total deflection is:
0.64
/=
'
Js-
cos a 0.472 1.35 cm

Deflection in the direction of they -axis (parallel to the arm of width 6)

is:
f v — f t tan a = 0.64 x 1.865 * 1.19 cm
Hence, in this example the deflection in the direction of axis Oy
is much greater than in the direction of Oz and is almost equal to the
total deflection.

cm apter 24

Combined Bending and Tension or Compression

§ 122. Deflection of a Beam Subjected to Axial

and Lateral Forces

In engineering practice we often come across; cases when a beam is

subjected to combined bending and tension or compression. Deforma­
tion of this type may occur either by the simultaneous action of axial
and lateral forces or by the action of axial forces only.
AL 9
ilHIIIIIIIIIIIIIIIIIIIIIII

-------------
Fig. 305

The first of these cases is shown in Fig. 305. Beam AS is acted upon
by a uniformly distributed force q and an axial compressive force P.
If we assume that the deflection of the beam is negligible as compared
to its cross-sectional dimensions, we can also assume with sufficient
accuracy that after deformation force P will give rise to axial compres­
sion only-.
Applying the method of superposition of forces, we can find the nor­
mal stresses at any point of an arbitrary section as the algebraic sum
of stresses caused by forces P and q.
Compressive stress aP due to force P is uniformly distributed over
the cross-sectional area, 5, and is equal in all sections:
P
380 Resistance Under Compound Loading {Part V/J

In a section with abscissa x the normal stresses due to bending in the

vertical plane are given by the formula
_ M( X) Z
Jy
where x is measured, say, from the left end of the beam.
Hence, at a point of this section having coordinate z (measured from
the neutral axis) the total stress is:
P i M( x ) z
a = o >+0* s
Figure 306 shows the diagrams of stress distribution in the given sec­
tion due to forces P and q and also the resultant diagram. Maximum
stress occurs in the uppermost fibres, where both deformations are

[ •" " y

compressive; the fibres below the neutral axis experience either ten­
sion or compression depending upon the numerical values of ap and
crff. In order to write the strength condition let us determine the maxi­
mum normal stress.
As the stresses due to forces P are equal in all sections and uniform­
ly distributed, the critical fibres are those which experience the maxi­
mum bending stresses. These fibres are the outer fibres of the section
in which the maximum bending moment occurs; for them
n ___ l. M nm
u </ max — -3- jp

Thus, stresses in the outer fibres 1 and 2 (Fig. 306) of the middle
section of the beam may be expressed

Z;}— (2i.i)
and the design stress
l«»»l = |o,l = | 4 + % 5| ,(21.2)
C k. 2 1 1 Combined Bending and Tension 391

If forces/1were tensile, the sign of the first factor would be reversed

and the lowermost fibres would be the critically loaded ones.
By denoting the compressive or tensile forces by JV we can write
the general expression for checking the strength of such beams:

< % » ,- ± ( - jT + n p ) < [ « J (21.3)

In writing equation (21.3) we have assumed that the section is symmet­

ric about the neutral axis and the beam material has equal resistance
to tension and compression.
The above method can also be applied to beams subjected to in­
clined loading (Fig. 307). The inclined force can be decomposed into

a normal component, which bends the beam, and an axial component,

which stretches or compresses it.
Example. An inclined beam (Fig. 308) is loaded at the middle of its
span by a force P —2.5 tf. Find the maximum compressive stress in
the beam.
The upper half of the beam only bends; the lower half is bent as
well as compressed. Bending is caused by the force P cos 30e, whereas
compression by the force P sin 30°. Maximum bending moment

Pi 2.5X3
^mtx 4 “ 4 1.875 tf-m

The section modulus and the cross-sectional area are, respectively,

2400cm3, S = 16x30 = 480 cm8

The maximum compressive stress (in the uppermost fibre of the beam
in a section to the left of the applied force) is:
___ P Sin 30° AJjnax_____ 2500 137 500
" 5 W =* 2X 480 2400
= —2.6—78.1 * —80.7 kgf/cma
m Resistance Under Compound Loading [Part V II

§ 123. Eccentric Tension or Compression

A second situation in which the bending and axial deformations are
added up is eccentric tension or compression, which results from axial
forces only. This type of deformation occurs when the bar is acted upon
by two equal and opposite forces P which act along AA parallel to its
axis (Fig. 309). The distance of point A from the centre of gravity of
section 0 is OA—e and is known as the eccentricity.

Fig 309 Fig. 310 Fig. 3)1 Fig. 312

Let us first consider eccentric, compression, because it is of greater

practical importance. Our task is to calculate the maximum stresses
in the material of the bar and check its strength. To solve the problem
let us apply two equal and opposite forces P at points 0 (Fig. 310).
This will not violate the equilibrium of the bar as a whole and will
not affect the stresses acting in its sections.
Forces P crossed by a single stroke will cause axial compression,
whereas the pairs of forces P crossed by two strokes will give rise to
pure bending of moments The design scheme of the bar is
given in Fig. 311. Since plane OA of the bending moment may not co­
incide with any of the principal planes of inertia of the bar. the defor­
mation, in general, will be a combination of axial compression and
pure unsymmetric bending.
As the stresses are equal in all sections in axial compression and pure
bending, we may check the strength of the bar in any section, say sec­
tion C-C (Fig. 311).
Let us remove the upper portion and consider the equilibrium of the
lower portion (Fig. 312). Assume Oy and Oz to be the principal axes of
inertia of the section. Let y P and zP be the coordinates of point A , (he
point of intersection of the line of action of force P with the cross sec­
tion. We shall select the positive direction of axes Oy and Oz in such
a way that point A always remains in the first quadrant. Then both
y P and zP will be positive.
Ch. 21] Combined Bending and Tension 993

In order to find the maximum stressed point of the section, we write

the expression for normal stress a at an arbitrary point B having coor­
dinates y and z. The stresses in section C-C are made up of axial com­
pressive stress due to force P and bending stress due to pure unsymmet-
ric bending by moment Pe, where e—OA. The compressive stress due
to fo rce P is eq u a l to ^ at every point, where S is the cross-sectional
area of the bar; unsyrametric bending may be replaced by bending mo­
ments in (he principal planes. Moment Pyp bends the bar in plane
xOg about the neutral axis Qz and gives rise to normal compressive
stress at point B. Similarly, the normal stress at point B due to
bending in plane xOz caused by moment Pzp is also compressive and
is equali to
* PzpZ
-j— .
Summing the stresses due to axial compression and bending in two
plants and. considering the compressive stresses to be negative, we get
the following formula for the stress at point B:

— <2 I <>

This formula is valid for calculating the stresses at any point of an

arbitrary section, only the coordinates of the point in the system of
principal axes should be substituted for y and z with proper signs.
In the case of eccentric tension, the signs of all the terms in the ex^
pression for the normal stress at point B will be reversed. Therefore,
in Order to obtain the stress with the proper sign from formula (21.4),
regardless of whether it is eccentric tension or compression, we must
consider the sign of force P in addition to the signs of coordinates y and
z\ in eccentric tension there should be a positive sign before the expres­
sion

'( t + E + 3 ?)
and in eccentric compression a negative sign.
The above formula may be modified somewhat. Let us factor out -g;
we obtain
T ( I + f + £f ) <2I5>
Here it and iv are the radii of gyration of the section about the princi­
pal axes (recall that JZ~%S and
To find the maximum stressed point we must select the y- and the
z-axis in such a way that a attains the maximum value. The varying
terms in formulas (21.4) and (21.5) are the last two, which reflect the
394 Resistance Under Compound Loading [Part VII

influence of bending. Also, since the maximum stresses in bending oc­

cur at points which are farthest from the neutral axis, it is essential,
as in unsymmetric bending, to locate the neutral axis first.
Let us denote the coordinates of points on the neutral axis by yt
and z%. As the normal stresses are zero at points on this axis, after sub­
stituting y 0 and z0 in formula (21.5) we get

1+ = 0 ( 21. 6)
2 ly
This is the equation of the neutral axis; it is the equation of a straight
line not passing through the centre of gravity of the section.
The simplest way of plotting this line is to calculate the segments it
cuts on the coordinate axes. Let us denote these segments by a9 and az.
In order to find segment ay cut on the #-axis, we put in equation (21.6)
*o = <L 00 = ^
and obtain
1 = hence a = — JL (21.7)
v yp '
Similarly, assuming that
& = <>.
we obtain
(21 .8)

If yp and zP are positive, segments av and at will be negative, i.e.

the neutral axis will be located on the other side of the centre of gra­
vity than point A (Fig. 312).
The neutral axis divides the section into two parts, compressed and
stretched. In Fig. 312 the stretched part has been shaded. Drawing
tangents to the contour of the section, tangents parallel to the neutral
axis, we obtain two points Dt and Da which are subjected to the maxi­
mum compressive and tensile stresses.
Measuring the coordinates y and z of these points and substituting
them in formula (21.4), we calculate the maximum stresses at points
Di and Z>* by the formula

<*<!,*-- P ( , T + - 7 r ~ + “' 7 7 7 (2L9)

#If ihe bar’s material has equal resistance to tension and compres­
sion, then the strength condition may be written as
(21.10)
Ch. 21) Combined Bending and Tension 395

For sections with comers in which both principal axes of inertia are
also the axes of symmetry (rectangle* I-beam, etc.), y t~ y max and zt=
=zmax. Therefore formula (21.10) may be simplified and written as
follows:

If the material of the bar has unequal resistance to tension and com­
pression, then its strength must be checked in the stretched as well as
compressed zone.
However, in some cases one check may suffice for these materials
also. It is evident from formulas (21.7) and (21.8) that the location of
point A of application of force and that of the neutral axis are inter­
related; the nearer point A is to the
centre of gravity the smaller the
coordinates y P and zP and the great­
er the segments au and az. Thus,
as point A approaches the centre of
gravity of the section, the neutral
axis moves away from it, and vice
versa. Therefore, in certain positions
of point A the neutral axis will
pass outside the section and the
whole section will experience either
tensile or compressive stress. Ob­
viously, in such cases it is always
sufficient to check the strength of
the material at point only.
Let us analyze a case of practical Fig. 313
importance, when a bar of rectan­
gular section (Fig. 313) is eccentrically loaded by force P at point A
on the principal axis Oy. The eccentricity OA is equal to e, and the
dimensions of the section are b and d. Applying the formulas obtained
above, we get
yP— +<?. zP= 0
The stress at point B is
( 21. 12)

because

The stresses are eoual at all points on a line parallel to axiz Oz. Loca­
tion of the neutral axis is determined by the segments
$ b*
at = oo (21.13)
av ~~ V2 e *
396 Resistance Under Compound Loading {Part V I!

The neutral axis is parallel to the xr-axis; the points which experience
maximum tensile and compressive stresses are located on the sides /•/
and 3’3.
The values of or*,* and <rmin may be obtained by substituting y —
in formula (21.12):

(21.14)

§ 124. Core of Section

In designing elements from materials which have poor strength un-

der tension (concrete, stone), it is highly desirable that the whole
section should work under compression, this can be achieved by li­
miting the value of eccentricity, i.e. not shifting the point of applica­
tion of force P loo far from the centre of gravity.
It is desirable that the designer should know beforehand the value
of eccentricity which may be permitted for a particular section with­
out the risk of stresses of two types occurring in it. Here we require
the concept of core of section. The term core of section defines an area
about the centre of gravity within which force P may be applied at
any point without giving rise to stresses of different types.
As long as point A remains within the core, the neutral axis does
not intersect the contour of the section, the complete section lies to
one side of the neutral axis and hence works only in compression. As
we move point A away from the centre of gravity, the neutral axis ap­
proaches the contour. The core boundary is determined from the condi­
tion that when point A lies on the boundary, the neutral axis passes
close to the contour just touching it.
Thus, if we move point A in such a way that the neutral axis rolls
along the contour without intersecting it (Fig. 314), then the locus of
points A will form the core of section. If there are “depressions” in the
contour, then the neutral axis should roll along the envelope of the
contour.
To plot the core of section we must draw the neutral axis in a number
of positions touching the contour, determine the segments oy and at
and calculate yP and zP—the coordinates of the point of application
of force—with the help of relations (21.7) and (21.8):
(21.15)
They represent yt, and zc—(he coordinates of points on the core’s
boundary.
If the contour of the section is a polygon (Fig. 315), we can find the
coordinates y e and zc of points on the core’s boundary by successively
Ch. 21] Combined Bending and Tension 397

drawing the neutral axis coinciding with the sides of the polygon and
calculating the segments ay and a 2 for the corresponding sides.
In the transition from one side of the contour to another the neutral
axis will rotate about the apex between the two sides; the point of
application of force will move on the core’s boundary between the

Fig. 314 Fig. 315

points already obtained. Let us establish how the point of application

of force P should move so that the neutral axis always pass through one
and the same point B(y„t 2 n) and rotate about it <Fig. 316). Substitut­
ing the coordinates of this point of the neutral axis in equation (21.6)

we see that coordinates y P and 2 P of point A, the point of application

of force P, are related to each other linearly. Thus, if the neutral axis
rotates about a fixed point B, the point of application of force moves
along a straight line. Conversely, the motion of force P along a straight
line is consistent with the rotation of the neutral axis about a fixed
point.
Figure 316 shows three positions of the point of application of force
on this line and three corresponding positions of the neutral axis.
Hence, if the contour of the section is a polygon, the core’s boundary
between points corresponding to the sides of the polygon consists of
straight-line segments.
If the contour of the section is made up of curved lines either parti­
ally or fully, the core’s boundary will be drawn with the help of points
(formula (21.15)). Let us study a few simple examples on plotting the
cores of sections.
For plotting the core of a rectangular section (Figs. 313 and 317)
we shall employ the formulas derived at the end of the preceding sec­
tion.
398 Resistance Under Compound Loading [Part VII

For plotting the core’s boundary when point A moves along axis Oy
we must find that value of eccentricity e=e 0 corresponding to which
the neutral axis occupies the position H%Oi (Fig. 317). From formula
(21.13) we find that

wherefrom
e„ = i (21.16)

Hence, along axis Oy the core's boundary will lie at a distance of

~ from the centre of gravity (Fig. 317, points / and 3). Along axis Oz
the core’s boundary will be determined by a distance of -jj- (points 2

and 4).

To draw the core’s boundary completely, let us plot the neutral axis
in positions HtOt and //*0S corresponding to the extreme points /
and 2 .
As the force moves from point / to point 2 along the core’s boundary,
the neutral axis must pass from position H\0 %to f f tOs all the time
touching the section, i.e. rotating about point D. For this to occur, the
force must move along the straight line 1 -2 . It can be similarly proved
that lines 2-3, 3-4, and 4-J will form the other boundary lines of the
core.
Thus, for a rectangular section the core is a rhombus with diagonals
equal to one-third of the corresponding sides of the section. Hence, if
the force is applied on the principal axes, the whole section experi­
ences stresses of a particular sign, provided the point of application of
the force does not lie beyond one-third of the distance of the corres­
ponding side from the centre of gravity.
Ch. 21J Combined Bending and Tension 399

Figure 318 shows the distribution of normal stresses along the height
of a rectangular section when the eccentricity is equal to zero, less
thon, equal to and more than one-sixth of the width of the section. It

Pig. 318

should be noted that for all locations of force P the stress at the centre
of gravity of the section (point 0 ) is the same and equal to and force
P does not have any eccentricity along the second principal axis.

Fig. 319 Fig. 320 Fig. 321

From symmetry considerations the core of a circular section of radi­

us r will also be a circle of radius r0. Let us consider an arbitrary posi­
tion of the neutral axis tangent to the contour. We shall direct axis Oy
perpendicular to this tangent. Then
af = <x>, 2e — 0
nr4
4nrx r
—=- >s»—- ------= — —
ay r 4

Hence, the core is a circle of radius four times less than the radius
of the section.
400 Resistance Under Compound Loading IPart VII
In an I-beam the neutral axis will not intersect the section while
moving round ii, if it remains tangential to rectangle ABCD described
around the I-beam (Fig. 319). Therefore, in an I-beam the core is a
rhombus, as in a rectangular section, but only with different dimen­
sions.
Jn a channel bar, as in an I-beam, points /, 2, 3, 4 of the core's
boundary (Fig. 320) correspond to positions of the neutral axis when
it coincides with the sides of rectangle ABCD. The distances may be
found from formulas (21.15).
Example I. A cutout element of a chain (Fig. 321) is made of steel
wire of diameter d=50 mm; o*=60 mm. Find the maximum permis­
sible value of force P if the permissible tensile stress in section A is
lcrl=l200 kgf/cm*.
In the given section the wire material is subjected to eccentric ten­
sion. Eccentricity e is equal to o + y . Selecting axis Oy in the plane
passing through force P and the axis of the straight part of the wire as
the other axis, we get

Stress at an arbitrary point of the section is given by the expression

or= + (q+ 4)

Substituting the limiting values y = ± . ^ t we find the maximum and

minimum stresses:
'» « ! 4 P , 64 Pde 4P r. . AP
. ff- . j “ ^ r ± T5ar“ 'sar L1:tTj ' juP
The strength condition may be written as

AP r *(.+*)i 4P
nd*
wherefrom
» ?uP[(T l 3 . 1 4 x 5 aX J2 O 0 _

Example 2. A long bulkhead h=3 m high and 6=2 m wide

(Fig. 322) supports an earth mound. Earth pressure per metre length
of the bulkhead is H= 3 tf and acts at a height of -|- from the founda­
tion. Specific weight of brickwork is 7 = 2 tf/ma. Find the limiting va­
lues of stresses in a section taken through the foundation.
Ch. 22] Combined Bending and Torsion 401

If we isolate a portion one metre long from the wall (Fig. 322), we
can consider it as a bar fixed at one end and subjected to. bending due to
earth pressure and compression due to its own weight. As the maximum
stresses occur at the fixed end, it is sufficient to check the strength of
the cutoff portion in this section only. The problem of determining
stresses in this section is equivalent to analyzing a bar subjected to si­
multaneous compression and bending.
The forces transmitted through this sec­
tion are the weight of the cutoff portion
N = \ X2X 3X 2= 12 tf and earth pres­
1 M W
i
sure H—3 tf. The bending moment in
this section due to force H is equal to I
M = H ^ —3 If* m. We shall employ
n r -
formula (21.3) for calculating the max­
imum normal stresses at the edge of
the foundation. The section which is
being checked is a rectangle with di­
i
V/,W V //////
t
VA > 1
1 2
mensions b — 2 m and d = l m; there­ |i
fore the maximum compressive stress
on side 1 - 1 of this section is

< w —
H
5
M 12
2X 1
6X3
1x2*
iH l N
»
d~tn

'
= — 6 —4.5 Fig. 322

—— 10.5 tf/ms —— 1.05 kgf/cm*

The compressive stress at points on side 2-2 of the section is

^mln -------- 5 ’+ ^ = ~ ‘ 6 + 4 *5
= - 1 . 5 tf/m4= - 0.15 kgf/cm4

CHAPTER 22

Combined Bending and Torsion

§ 125. Determination of Twisting and Bending Moments
In Chapter 9 we discussed a problem on checking the strength of a
shaft under torsion. However, machine parts such as shafts rarely work
under pure torsion. Even a straight shaft working under torsion bends
due to its own weight, weight of the pulleys and the pressure exerted
by the belts. Hence, a majority of machine parts working under tor­
sion are actually subjected to combined bending and torsion. Crank­
shafts belong to this group.
1 4 -3 3 1 0
402 Resistance Under Compound Loading [Part VII

In the analysis of elements subjected to combined bending and tor­

sion the first thing to do is to find the design values of the bending
moment Mb and twisting moment M t.
Let us consider a straight circular shaft with a pulley and crank.
The loading diagram of the shaft is shown in Figs. 323 and 324. A pul­
ley of weight Q is mounted at its left end, the belt pressures on the
tight and slack sides of the pulley are T and t, respectively

and a horizontal force P acts at the crank pin at the right end of the
shaft. Let us consider the instant when the crank is vertical. All di*
mensions are given on the diagrams.
We have to calculate the bending and twisting moments for shaft
AD. Forces T and t (pull of the belt) acting on the pulley may be re­
placed by a forceT-M acting at the centre of the pulley and a moment
{T—QRt, where R 9 is the radius of the pulley. Force T-\-t along with
the weight of pulley, Q, bends the shaft; moment (T— £)R9, which
twists the shaft, is balanced by the moment applied to its right end.
Let us replace force P acting at the crank pin by an identical force
P acting on the shaft at point N of its extension and a moment Pha.
Thus, the ends of the shaft are acted upon by moments Ph0 and
( T — f)R 9. In equilibrium, when the machine runs uniformly, these
two moments are equal to the twisting moment Af t=Pht= ( f —t)R0.
If we know the number of revolutions of the shaft per unit time, n,
and the transmitted power, N, then the twisting moment can be found
from formula (9.3) of § 46:
M , - ' 716.2$
n , hence p _Af*
ho
i >

where
II
3

As far as bending is concerned, the shaft is acted upon by vertical

force Q and horizontal forces T + t and P. Therefore, we must plot the
bending-moment diagrams for the horizontal as well as the vertical
forces (Fig. 325 (a) and (5)), considering the shaft to be simply support-
Ch. 22] Combined Bending and Torsion 403

ed at the bearings B and Cc(one of the bearings permits horizontal dis­

placement). ■: •• , .
Having plotted the bending-moment diagrams for the vertical and
horizontal forces, we can find the total bending moment Mb in a sec­
tion as the geometric sum of the two/ The geometric addition of the
vector^ representing the bending moments in section B is shown in
Fig. 326. The total bending moment in section B may be written as

Each section will have its own plane of bendipg. However, since
the section moduli of a circular body is the same about all axes passing
through its'-centre of gravity, we can superpose the bending-moment

planes of all sections in the plane of the diagram and then plot the re­
sultant bending-moment diagram, without irt any way affecting the
final results. This precisely has been done in Fig. 325(c). We wish to
point out without a formal proof that between sections B and C the
resultant bending-moment diagram does not have a maximum.
It is evident from the shape of the diagram that critical loading
occurs either in bearing B or in bearing C, depending on the numeri­
cal values.
14*
404 Resistance Under Compound Loading [Part V II

§ 120. Determination of Stresses and Strength

Check in Combined Bending and Torsion
Having calculated the maximum bending moment M b and the ma­
ximum twisting moment M t, we can now find the maximum stresses
in tiie shaft material and write the strength condition. Let us assume
the shaft to be cut in the critical section C {Fig. 327) and analyze it
with the help of the principle of superposition of forces. We shall cal­
culate the stresses in the cross section due to the bending moment and
add to them the stresses due to torsion.

P ig . 3 2 7 F ig . 3 2 8
The bending moment acts in the horizontal plane, the neutral axis
will be vertical, and the maximum normal stresses oh will occur at
points c, and c2 at the endpoints of the horizontal diameter. Torsion
will give rise to shearing stresses only, which will be maximal at
points on the contour.
Thus, pointsC\ and c 2 in the sectional plane, will experience maximum
normal as well as maximum shearing stresses. At points c3 and cx on
the vertical diameter the shearing stresses due to torsion will be sub­
stantiated by shearing stresses due to bending. However, these stres­
ses are small in magnitude and experimental studies show that points
ci and cz are the critically loaded points. Let us isolate cubic elements
of the shaft’s material around these points (Fig. 327). Four faces of
these elements will be subjected to shearing stresses t* (two of these
four faces will experience normal stresses too) and the other two faces
of the cube will be completely free of stresses (Fig. 328). Hence, the
element is in a two-dimensional stressed state. It is known (§§ 39 and
77) that in order to check the strength of a material in two-dimensional
stressed state, we must find the principal stresses o, and <r3 and sub­
stitute their values in the strength condition written on the basis of
one of the theories of failure.
An element of a bent beam cut at a distance z from the neutral axis
is also in a similar two-dimensional state, of stress. We discussed how
to check the slrengih of such an el’ement (§ 77) while studying the
Ch. 22] Combined Bending and Torsion 40?
strength of beams that experience both a bending moment and a shear­
ing force. The only difference was that both the normal stress a and the
shearing stress t were caused by bending in that case.
For checking the strength of an element cut from the shaft, we can
directly apply the formulas derived in § 77 by substituting or, and xt
in place of a and t , respectively. Then we obtain the fol lowing strength
conditions according to four different theories of failure:
the theory of maximum normal stresses:

Y + V<ti>+ 4xjr) ^ [<*]

the theory of maximum strain:
(0.350,-1-0.65 /o f + T c ? ) < [o] (22.1)
the theory of maximum shearing stresses:

the theory of distortion energy:

V o i + 3xi < [a]
We must calculate <jh and t, to correlate the strength check with
the numerical value of moments Mb and M t and the dimensions of the
shaft. Stress o,„ which is the maximum normal stress due to bending
moment Mb, is

For a circular shaft W = -j-, where r is the radius of the circular sec­
tion. On the other hand, the maximum stress x t due to torsion is
Mt Mt Mt Mt
T< Wfj nr^ n w*
2 1 4

Substituting these values in the first expression of formula (22.1), we

obtain

We can similarly obtain design formulas for the other theories of

failure. It is evident that all these formulas can be represented by a
single general formula of the type
( 22. 2)
406 Resistance Under Compound Loading [Part V ll

where Ma is the design moment whose value depends not only upon
the moments Mb and M i but also upon the theory of failure applied.
According to
the theory of maximum normal stresses: '

the theory of maximum strain:

Mdi ** 0.35Af*+ 0.65 V Aff+ M\ .(22.3)
the theory of maximum shearing stresses:
Mda = VM% + M\
the theory of distortion energy:
Af«4« V M%+ 0-.75Af?
Formula (22.2) is similar to the formula by which we check the
strength under normal stresses due to bending by a moment Md.
Therefore, the strength check of a round shaft under combined bending
and torsion may be replaced by a check due to bending only by the ben­
ding moment Md.
In some constructions the shafts are subjected to tension or compres­
sion due tb an axial load N in addition to bending and torsion. The
effect of the axial forces on the strength of the shaft may be taken into
account by the addition of stresses a 0 caused by them to the maximum
bending stresses ab: oa= j , where A is the cross-sectional area of the
shaft.
From formula (22.2) we obtain

4 ^ foj
wherefrom the radius of the shaft

'> V IW ’ d = 2r (22.4)
For using this formula all we have to do is to establish which of the
theories of failure should be used and, consequently, which of the ex­
pression in formulas (22.3) should be employed for calculating the de­
sign moment.
We can straightaway reject the theory of maximum normal stresses
(see § 39), because shafts are usually made from steel and from ductile
materials in general. Until recent times, shafts in machine tool indust­
ry were designed by the formula based on the second theory (the theory
of maximum strain). The formula is sometimes also referred to as
Ch. 22) Combined Bending and Torsion 407

Saint-Venant's formula:
^ (0.35Atb+ 0.65 V M \+ M?) < [a]

It was used despite the fact that the hypothesis underlying it is definite­
ly not true for ductile materials. For some ti me now shafts are designed
by formulas which are based either on the third theory (theory of
maximum shearing stresses) or on the fourth theory (distortion energy
theory):
-/Af}+Af?<[c] and ^KM|+0.75yWf<[o]
Table 15 compares the values of shaft diameters for different ratios
of Mb and M h using the same permissible stress in all the theories
of failure. The diameter obtained by the theory of maximum strain
(Saint-Venant’s formula) is taken as unity.
Table 15
Comparison of the Shaft Diameters
S h a ft d ia m eter according to

II th eory III theory IV th eory

M bB&0 1 1.15 1.10

Mb= - L M t 1 1.07 1.03
Mb = Mt I 1.03 1.01

It is evident from the table that, firstly, the difference in the dimen­
sions of the shaft is not large regardless of the theory used and, second­
ly, the diameter by Saint-Venant’s formula is in all cases less than the
diameter obtained by the other two theories. This helps to explain
the fact that Saint-Venant’s formula is still used sometimes in engi­
neering practice, although it is based on a hypothesis which has been
proved inapplicable to ductile materials.
A designer must remember that transition to the new design formu­
las based on more accurate theories would have been practically im­
possible if the old values of permissible stresses had been retained.
This would have led to the use of shafts of bigger diameters, where
old shafts of smaller size designed according to Saint-Venant's for­
mula were working satisfactorily.
The idea is that when we change over to a new formula we cannot re­
tain the previous factor of safety and the previous permissible stress.
408 Resistance Under Compound Loading IPart V II

More accurate design and deeper knowledge about the working of

materials must, as a rule, be accompanied by a reduction in the factor
of safety and consequently an increase in the permissible stress, Icrl.
Therefore, when we are calculating the design moment by the new
formulas, we must increase the permissible stress la] by such a value
so that we can justify the dimensions of shafts already working satis­
factorily on the basis of the new theories and reliable experimental in­
vestigations.

CH A PTER 23

General Compound Loading

§ 127. Stresses in a Bar Section Subjected
to General Compound Loading
The methods of finding stresses and deformations used in solving
particular problems of compound loading may also be employed in
situations of more complex loading of the body. Limiting our discus­
sion to prismatic bars, in which the centre of bending coincides with
the centre of gravity of the section, we assume that such a bar
(Fig. 329(a)) is in equilibrium under the forces acting on it; the ori­
entation of these forces in space is arbitrary. To simplify the diagram,
only concentrated forces have been shown in Fig. 329(a). However,
distributed loads and moments may also be applied. This will not af­
fect our discussion.
For finding the stresses at an arbitrary p o in ts of the bar, let us draw
a section mn perpendicular to the bar’s axis; the section cuts the bar
into two parts (7 and //). Let us remove one part, say, the right one,
and transfer all forces acting on it to the centre of gravity of section
mn, to point C. In the future discussion we shall use a right-hand rec­
tangular coordinate system with the x-axis passing through the centre
of gravity of section mn and normal to it and the other two axes coin­
ciding with the principal axes of inertia of the section passing through
the centroid.
When a force Pk is transferred to point C (Fig. 329(b)), we obtain a
force Ph. acting at the centre of gravity in general and not coinciding
with any of the coordinate axes and a moment M k=P)fik acting in
genera] in a plane which is inclined to all the coordinate planes. Pro­
jecting force Ph on the x-, y- and z-axes, we find the components Phx,
Phy and Phi', similarly, projectingMh on the x -, y -, and z-axes, we ob­
tain the components Mhv, Mkl/ and M hz (Fig. 329(c) shows the resolu­
tion of vector Lh representing moment Afft into the components Lux,
Lh 0 and /„*,). By doing the same operation with all the forces Pk acting
on the right part of the bar, we can replace the force system Ph by a
statically equivalent system consisting of six components: three forces
Ch. 23] General Compound Loading 409

along the coordinate axes acting al the centre of gravity of the section,

£ v Q ,= £ P *
410 Resistance Under Compound Loading IPart V It

and three moments about these axes,

M ,= y Mt,
*»1 k■1 A?=1
(Fjg. 329(d)). Forces N, Qy, and Q? will be considered positive if they
coincide with the positive directions of the coordinate axes, while
moments Mx, My, and Mt will be considered positive if they act in
the anticlockwise direction about the corresponding axes (all the force
and moment components* shown in Fig. 329(d) are positive).
From the earlier discussion we know the simple forms of deformation
which result from the action of each of these components. It should be
borne in mind that these forces transferred from the right cutout part
to the left reflect the action of the right part on the left and therefore
in section mn are manifested as stresses. Thus, N is the sum of normal
stresses distributed over the section, M x is the sum of moments about
the x-axis of all shearing stresses acting in section mn} and so on.
It is evident that N gives rise to tension or compression, Qy and Qz to
shear In the direction of the y• and the 2-axis, respectively, M x to tor­
sion, and My and Mz to pure uni-planar bending about the y- and the
2-axis, respectively. Thus, in the most general case of loading of the
bar, the latter experiences four simple deformations: tension or com­
pression (N), torsion (Mx) and uni-planar bending about two axes,
(Af„ and Qz) and (Mz and QM). Three force factors, N, M v, and Afz,
give rise to normal stresses in section mn while the remaining three,
Qv* Qz* an£l to shearing stresses (Fig. 330(a) and (c)).
Let us first study the case when only normal stresses appear in the
bar section. It can be easily seen that this is a particular case of com­
pound loading—tension or compression with pure bending in two
principal inertia planes passing through the centroid.
§ 128. Determination of Normal Stresses
Let us assume that the forces acting on the removed part of the bar
can be reduced in section mn to three components: the normal force N
and two bending moments M.y and M 2; we shall assume all the compo­
nents to be positive (Fig. 330 (a)). Let us derive a formula for determin­
ing normal stress at a point A located in the first quadrant of section
mn and having coordinates y and 2 .
The positive normal force N gives rise to the uniformly distributed
tensile stress, o '—±A(IS, the positive bending moment My gives in the
first quadrant tensile stress a’,—+Myz/Jfn while the positive bending
moment M t gives compressive stress o*'~—M ty/Jz (see Fig.330 (6)).
Summing up these components of normal stress we get the following
expression for calculating the total normal stress at point A:
o«=o#-f a" + (23.1)
Ch. 23] General Compound Loading 411

For calculating the total normal stress at any other point of the
bar’s cross section, it is sufficient to substitute in formula (23.1) the
values of N , M„, and M z and coordinates y and z with the proper
signs; this gives us the total normal stress with the proper sign.
It is obvious from formula (23.1) that the normal stresses are linear
functions of coordinates y and z; they must attain maximum at those
points of the section which are farthest from the neutral axis (at the

neutral axis the normal stresses are zero). Figure 331 (fl) depicts a sec­
tion of the bar; in all the quadrants the signs of normal stresses, fl) a ,
(2) a" and (3) o'", are shown in the assumption that N ,M y and M t are
positive. It is obvious that the neutral axis will intersect the quad­
rants with normal stresses of different sighs and in the given case will
not pass through the centre of gravity and the top left quadrant.
Assuming that in formula (23.1) stress <j is equal to zero and denoting
the coordinates of a point on the neutral axis by y„ and z„, we get the
following equation of the neutral axis:
N , Mv
0
T+77
Equating to zero first zn and then yn>we find the intercepts cut by
the neutral axis on the axes of y and z, respectively (Fig. 331 (b)):
N Jt , NJy
(23.2)
SM t and 0 .— -3EJ
As the presence or absence of factor W S in.formula (23.1) does not
affect the inclination of the neutral axis with respect to the coordinate
axes, the inclination may be determined from the equation
4J2 Resistance Under Compound Loading [Part VIJ

wherefrom it ensues that

ta n a is .J L s .^ 1 (23.3)
Vn Jt Mg

By geometrically summing the moments M„ and M , acting in sec­

tion mn in planesxz and xy, we obtain the resultant bending moment
Mb^ V 'M * + M i (23.4)
Angle <p between the plane in which Mb acts and the vertical plane
xz may be found from the expression

ta"'<’ = lT7 (23.5)

This expression enables us to write formula (23.3) in the following

form:
t a n a = - p ta n <p (23 6)

Angles a and <pwill be considered positive if they are laid in the anti­
clockwise direction from the corresponding axes (a from the tr-axis and
<p from the a-axis).
It is clear from (23.3) that in general the neutral axis in the section
is not perpendicular to the trail of the resultant bending moment

(Figs. 331 arid 332) acting in the same section. The neutral axis will be
perpendicular onlv when angles a and <p are equal. This, in turn, is
Possible only under the following conditions: (1) (p—0, i.e. Af*=0;
(2) i-e. 0; and (3) In the first two cases the bar
Ch. 23} General Compound Loading 413

experiences uni-planar bending in one of the principal planes of iner­

tia irrespective of the magnitude of the principal moments of inertia;
in the third case all the axes of inertia are principal central axes of
inertia (circle, square, etc.) and
therefore bending is uni-planar in
all directions. The whole discussion
leads to the following genera) con­
clusion: bending is uni-planar and
the neutral axis is perpendicular to
the projection of the plane of action
of the resultant bending moment if
this plane intersects the section per*
pendicular to one of the principal
central axes of inertia alright angles.
In general the neutral axis divides
the cross section into two zones:
a stretched zone and a compressed
zone. Drawing lines parallel to the
neutral axis and tangent to the con­ Fig. 332
tour of the cross section, we find the
points Ot and 0 * of maximum tensile and compressive stresses which
lie farthest from the neutral axis (Fig. 3310)) for both zones. Substi­
tuting tiie coordinates of these points {y0x and z0l, or y0t and Zq)
with their proper signs in formula (23.1), we find the maximum tensile
and compressive stresses:

?max "J' r j^ j ^ lf9 (23.7)

While solving practical problems it is sometimes more convenient

to replace general formula (23.7) by the following formula:
\N \ , \M yz<t\ t lM >y,l
'm a x = ± (23.8)

in which the absolute values of N, Af„, M z, y 9 and z%are substituted,

and the signs of the terms are ascertained in each particular case from
the actual direction of force factors and location of the points in the
section.

§129. Determination of Shearing Stresses

Shearing stresses in a bar’s section occur due to torsion of the bar

about the x-axis, Af*, and shear in planes xy and xz (Qy and Qz); see
Fig. 330(c). For a bar of circular or ring section the shearing stresses
xt due to twisting moment M x can be calculated by the well-known
414 Resistance Under Compound Loading [Part VU

formula
(23.9)

For a bar of any other cross section the maximum shearing stresses
may be determined by the formula
m ax tt = ^ (23.10)

using the data for Wt given in the section on twisting of bars of non­
circular section. In ail the cases the maximum shearing stresses (tor­
sion) occur at the contour of the section and act along the tangent to it.
The shearing stresses due to forces Q„ and Qt are, as a rule, of se­
condary importance; they are determined by Zhuravskii’s formula
Q*$l
and (23.11)
T'~"V> (*)
In rectangular and round sections these shearing stresses attain their
maximum on the corresponding principal axes of inertia: xv on the
z-axis and T,on the#-axis. At those points of the contour where the di­
rection of maximum shearing stresses due to shearing forces (max t„
or max xz) coincides with the direction of maximum shearing stress
due to torsion, the two are arithmetically summed up and the maximum
total stress is used for strength analysis:
Tmax = max t , + max or Tm#S( = max x, -f max xx
Since the normal stresses due to bending and the total shearing stres­
ses due to shear and torsion are both maximum at the contour of. the
section, it is logical to search for the maximally stressed points and
also to check the strength of the bar’s material on the contour. The
points experiencing maximum shearing stresses do not always coincide
with the points subjected to maximum normal stresses. In such cases
the strength of the bar’s material should be checked at those points on
the contour where the combined effect of normal and shearing stresses is
most unfavourable.

§ 130. Determination of Displacements

If we recall that in the general case of compound loading the bar ex­
periences, besides other types' 6f elementary1deformations, two planar
bendings in the principal planes of inertia, it becomes clear that in
general the deflected axis of the bar must be represented by a curve in
space. The curvature of the axis in plane xy is
(23.12)
Ctu 23\ General Compound Loading 415

and in plane xz it is

<23,I3)
If the curvature vectors, x* and x„, are laid on the corresponding
Coordinate axes, vector x of the total curvature of the deflected axis,
which represents the geometrical sum
x = V X + x* (23.14)
makes an angle v with the a-axis, and the tangent of this angle (see
Fig. 331(c)) is found from the formula
tan y ft* (23.15)
fiAg Jt
A comparison of formulas 123.15) and (23.3) assures us that angles
a and y are equal, i.e. the total curvature vector is parallel to the neut­
ral axis, and if there is no .normal force .the two coincide. Hence,
the resultant curvature plane, which is perpendicular to the total cur­
vature vector and. tangent to the deflected axis of the bar in the given
section, is always perpendicular to the neutral axis. The centre of
gravity of the given section gets displaced perpendicular to the neutral
axis only when the bar is subjected to.bending in one plane (when tp=
const and a=cdrtst along the whole length of the bar, for instance, in
uni-planar and unsyinmetric. bending).
If as in the example! depicted in Fig. 331, then according to
(23.3) tan a > ta n <p and cCxp, i.e. the centre of gravity gets displaced
in a direction which is inclined to the plane of action of bending mo­
ment Mb and tends towards the y-axis. It can be easily noticed that
the centre of gravity always deflects from the plane of the resultant
bending moment towards the axis about which the moment of inertia
is maximum.
It follows from the above that the deflected axis of the bar can be
represented by a curve in a plane only if the total curvature vector
makes a constant angle y = a with the tf-axis along the whole length of
/ Af °
the bar; i.e. (see (23.15)) if the product is independent from
the x-coordinate. The last may occur, few* instance, when a bar of uni­
form section is loaded by forces that act in a single plane.
Applying the principle of superposition of forces, we can use the dif­
ferential equations obtained from (23.12) and. (23.13) for finding the
total displacement of the centre of gravity of an arbitrary section.
After integrating and finding the constants of integration from the
boundary conditions and then determining for the given, section two
displacement components f y and f T in the direction of the principal
axes of inertia y and a, we can determine the total displacement as
416 Resistance Under Compound Loading I Pari VII

the geometric sum:

f~ V r,+ n (23.16)
Besides the analytical method, the displacement can also be found
by the graph-analytic method and Castigliano’s theorem, which is
particularly useful when dealing with crank rods (see below). When
Castigliano’s theorem is employed for determining displacements un­
der compound loading, the potential energy of deformation, U, must
be expressed as a function of all the six force components: N, Q„, Qz,
M x, M y, and M z. Neglecting the energy of shearing stress due to shear­
ing forces, we may write
U = U(N) + U (At,)+ U ( M ,) + U (Mz)
Assuming that in general normal force N and twisting moment Mx
do not remain constant over the whole length of the bar, we can write
the following expressions for the energy stored in an clement of length
dx:

and the following expressions for the energy stored in a segment of

length I of the bar:

and

We have the following expressions for the energy due to normal stres­
ses in uni-planar bending (see § 100):

and

Keeping the above expressions in mind, we can write the formula for
U as follows:
(23.17)

where subscript I shows that the expression is integrated over a length

/ of the bar for which the functions of the ^-coordinate, i.e. N, Mx,
M ft, and M z, are continuous. If the bar contains a number of such seg­
ments, then separate integrals should be calculated for each of them
and then summed up.
Applying Castigliano’s theorem, we can find the displacement in the
direction of any of the forces P from the following expression:
Ck. 23] General Compound Loading 417

By P and 6 we denote here a general force and the displacement cor­

responding to it. The formulas used in the methods of Mohr and Vere­
shchagin can be derived in a similar manner.

§ 131. Design of a Simple Crank Rod

Crank rods are often used in engineering practice as parts of crank

gear and other mechanisms, crankshafts, etc. The design of crank rods
is a little more difficult than that of straight rods. As an example, we
shall explain how to design the crank rod shown in Fig. 333. The rod
consists of two parts: a vertical part (of rectangular section) and a hori­
zontal part (of circular section), rigidly connected at right angles to
each otner. The following loads are applied to the crank rod. In section
A: Pi—1200 kgf, P 4=1000 kgf,
n Ann A AA

this section about the longitudinal

axis of the first part. Fig- 333
The analysis starts by plotting
diagrams that show the variation of
all force factors on each part of the crank rod. Each may be assigned
its own system of rectangular coordinates, choosing the axes in such a
way that force N is always a normal force, M x a twisting moment, and
M y and Af* bending moments. In Fig. 334 such diagrams for the first
part arc shown to the right of the crank rod (Fig. 334(a)), and for the
second part below the crank rod. On these diagrams the values of
force factors at the beginning and end of each part are written in arith­
metic form. After plotting the diagrams for all the parts we are in a
position to locate the critical sections, in which the combined effect of
all the force factors is most unfavourable. In our example the critical
sections are: horizontal section 1 - 1 in the first part and vertical section
2-2 in the second part, both in the vicinity of section B (see Fig. 334(a)).
These sections and the forces acting on them are shown in Fig. 334(d)
and (c).
418 Resistance Under Compound Loading [P a ri VII

Let us check the strength of the rod in section /•/. The forces acting
on this section may be reduced to the force N = P i—1200 kgf, twisting
moment Mx= M t = —80000 kgf -cm, and two bending moments:
He »a

y Neutral axis >

(lip flea I tmanh B) (RtgM-hand view aim/} BC)

Fig. 334

/V i= —1000X120=— 120000 kgf-cm and M t= P ah * = m x

X 120=48 000 kgf -cm. Normal stresses at any point of this section
may be calculated by the formula (see (23.1))
N Mvz Mxy 1200 120000 48000 160
a “ T + 7 7 ‘^ ~ t~ l 2 0 2250Z* §40“ ^ ~ ^ 3~Z _

since S=W i=8x 15=120 cm*, J //=Wi,/12= 8x 15s/'12±=2250 cm4, and

J 2= / i6*/12=15x 83/12=640 cm*. The neutral axis culs the following
Ch. 2<JJ General Compound Loading 419

intercepts on the y- and the z-axis:

N J, 1200 x 640
av ~ S M Z = 120 X 48 000
= 0.133 cm
N JV 1200 x 2250 3
az — 5m „“ I& x 120 555 —T5 = 0.1875 cm

Maximum normal stresses occur at point 1 which is farthest from

the neutral axis and has coordinates ^ = —4 cm and 2i = —7.5 cm
(this is a case of uniaxial stressed state):
cr(1) = 10 + ijp x 7.5 + 75 X 4 = 710 kgf/cm* < [o]
= 800 kgf/cm*
Hence, the rod's material at point 1 is sufficiently strong. We must now
check the strength of the rod’s material at points 2 and 3, which expe­
rience torsional shearing stresses in addition to the normal stresses.
The normal stresses at these points are

*ia) = 4 - 7 f ^ = 10+ 75 X 4 = 310kgf/cm*

o(s>= - j + 7 “ = 10 + x * 7-5 = 410 kgf/cm*

To find the shearing stresses we determine J t=>ab* and

From the ratio n = h / b = ^ = 1.875 we find from Table 9 (§ 54) by linear
interpolation the coefficients a=0.416, p=0.406, and y=0.808.
Hence /<=0.416x84=2506 era4 and IF(=0.406x83=233 cm3. At
point 2
- S a r - 343 ks f' cra'

and at point 3
T(a) = Y^max = 0-^08 x 343 = 277 kgf/cm*
We will check the strength of the rod at these points by the third
strength theory:
aX U = V 0 * w +4tS1>= y3IO* + 4 x 343*
= 753 kgf/cm* < 800 kgf/cm*
Odll(3, = V<rSi>+ 4 T ? „= V 4 l0 * + 4 x 277*
= 689 kgf/cm* < 800 kgf/cm*
Hence, at these points too the rod is sufficiently strong.
420 Resistance Under Compound Loading \Pari VII

Let us now undertake the analysis of the second part. In critical

section 2 - 2 the force factors are
W= P 8-f = 400 -f 6000=6400 kgf
Mx ^ - /> ,/, = — 1000 x 120=* — 120000kgf-cm
M;/ = / y i = 400 x 120 = 48 000 kgf •cm
Mt = — Af, = —80000kgf -cm
As the section is a circle and has equal moments of inertia for bending
about the two principal axes of inertia, i.e.
J J z ~ Jb ~ “ p and J t = J j>~ ~ j - a=2.Jb

bending moments M y and M t can be geometrically summed up into

the resultant bending moment
Mb= = l/4 8 000*-^0000* = 93 280 kgf -cm
The projection of the plane of action of the resultant bending mo­
ment passes through the centre of gravity making an angle tp with the
z-axis such that
Mg 80 000 5 . CAoo/
M,x ~ 4 8 ^ 0 0 1=1 3 a n t ^ V - 5 9 2

The critical point of the section is point O, which lies at the inter­
section of the projection of the Af*,-plane with the contour. At this
point the normal stress is
N . 4Mb 6400 , 4 X 93280 2 0 3 7 .1 1 8 770
= oT7 ' .*ty* 5 .1 4 3 + 3. 14ry 33 r* ' 7?

and the shearing stress is

- 2/Mjf 2 x 80 000 76300

To determine the radius of the section for this part of the crank rod
we shall again use the third strength theory:
tfies (O) = i f 0 (0 ) *1“ 4t(0>

<[<r] = 800 kgf/cm*

Taking the square of both sides of this equation and multiplying
the result by r•, we obtain
2037V2+ 2 x 2037x 118 770r +118 770a +152 600s < 64 X 10
C}i. 23] General Compound Loading 421

or
(!>= r«_6.485ra—756.1 r —58 516 > 0 (*)
While solving equation 0 = 0 by trial and error we neglect the fac­
tors containing r2 and r in the first approximation (i.e. we ncgLct ihe
relatively small normal stress NIS). This gives
r> { /5 8 5 !6 = 6 .2 4 c m
If we substitute a slightly larger value of r=6.3 cm in the equation
0 = 0 , we get
0 = 6.3«—6.485x6.3*—756.1x6.3—58 516
= — 1013 kgf -cm*
We thus see that the selected value of r is not sufficient for satisfying
the inequality (*); let us try r=6.4 cm. In this case O —+5099 kgfx
Xcm". By interpolation we find that r=6.32 cm. This corresponds to
5 = nr* = 3.14 x 6,32* =. 125.5 cm*
I ^ 3 3.14
'' ' Xy 6.32* — |OK 5 /.m 4
Izoocm*
J ( = j p = 2Jb = 2 x 1253 = 2506 cm*
We shall apply Castigliano’s theorem to calculate the displacements.
In Table 16 force factors N, M x, M y, and Mt and their derivatives
Pi, P^ and Afo are expressed for each of the crank rod parts as func-
Table 16
Force Factors and Their Derivatives
First pari ( 0 < x K lt)

Force factors N M x M j, Mz
Formulas for them Pi -M 0 - P t x Pax
dJ6Pt 1
dJdPt — X
did P s x
did la* —1

Second part (0 «. x < /,)

Formulas for force ■- J V I P h -P l* P»X-\-PoX —

factors -M o
d/dP x — X
did P i - ti X
d/dP a I /l
did M o —1
422 Resistance Under Compound Loading [Part VII

tions of x. Using the data of the table, we easily find that

,120 A 80
_ 1 ( f 1260 . , p 1200*a—48000* j \
= ? n w (J w " + j — es—
42 060 a a.
‘J>TIo5 ~ 0 -2 I cm

/ uu MXj dMgt •
>= aP7 &*■£?**+j E Jn ’ d P V ^

.120 80 80
_ 1 ( p I000sa dx , p I0 0 0 x l2 0 x l2 0 d x , p (130Qxa—80000*)dx\
2 X 10^ J 6250 l" j 0.4x2506 hJ 125$--------- I
1 278 n COA
= —5—=0.639 cm

c P M t i OAtgi j , r i A^j d N% j , p ^ys

f’ -nF; li\r tfls p r ix+]n essf7d*+}a »
,n
¥->+?t ,dx+‘§
i
fxTi
'f
,120

.)
^0
640+r
0
80

b
125.5
0
80
400x* dx , p6400<i* , pi

5
1553"
V
0.9167
= 0.408 cm

• -T C -f& ^ + fx fc T J ? *

- t ( | « g fc * + | & * & } -- * » *-1>^ )

Ch. 24} Curved Bars 423

( 120
P 80000dx
80
, (*(1300*—80 000) ( - 1 )
y 1555
15 870 j 0.4X1704 '
=0.00794 = 0.456°
The total displacement is
/ » |/ /=+/»+/* = j/0 .2 1»-f 0.639* + 0.408*
e=0.787 cm

CHAPTER 24

Curved Bars
§ 132. General Concepts

Besides straight-axis bars in structures we often come across bars

in which the axis, i.e. the line passing through the centres of gravity of
successive cross sections, is a curved line. Chain links, lugs, hooks,
arches, vaults, hoisting crane frames, etc., all belong to this group of
dements (Fig. 335). Strictly speaking, no bar has an absolutely
straight axis. All bars, which we design as straight bars, have a slight
curvature. Therefore, a study of the effect of curvature of the bar's axis
on the distribution of stresses will, on the one hand, enable us to check
the strength of bars having appre­
ciable curvature and, on the other
hand, judge the influence of a slight
deviation of the axis from a straight
line on the strength of a straight bar.
We introduce the following res­
trictions for checking the strength of
curved bars:
(a) sections of the bar have an
axis of symmetry;
(b) the axis of the bar is a flat
curve which lies in the plane of sym­
metry;
(c) all external forces also lie in
the same plane.
On account of symmetry* defor­ Fig. 335
mation of the bar will also occur in
the same plane, the bar's axis will remain a flat curve lying in the plane
of external forces and the picture will be identical to that of uni-pla­
nar bending of beams.
424 Resistance Under Compound Loading [Part VII

By writing the above restrictions we cover almost all situations of

the working of curved bars. Our task is to find the maximum stresses,
check the strength and determine the deformation of curved bars.
The solution will be similar to the case of bending of straight beams.

§ 133. Determination of Bending Moments and Normal

and Shearing Forces

Imagine a curved bar (Fig. 336) loaded by external forces Px, P», P3,
and Pt acting in the plane of symmetry of the cross sections. The sup­
port reactions, not shown in the figure, lie in the same plane.
To determine the stresses in sections perpendicular to the bar’s
axis we draw a section ma which divides the bar into two parts, I and
II. Let us remove part I and con­
sider the equilibrium of part II
(Fig. 337(o)). Part II is acted upon
by force Pu the reactions at the
fixed end (not shown in the figure)
and stresses in section mn which
appear because of the action of the
removed part on the part under con­
sideration. What are the stresses
in section mn?
The section will experience nor­
mal as well as shearing Stresses (not
shown in Fig. 337(a)). With the res­
trictions of § 132 the normal stres­
ses will give the following result­
F ig . 3 3 6 ants: the bending moment M and the
normal force N. The shearing stres­
ses in the section will yield a re­
sultant shearing force Q. These three forces are shown in Fig. 337(a).
Let us now consider part I of the curved bar (Fig. 337(b)). All ex­
ternal forces acting on this part of the curved bar may be reduced in
general to a resultant force R and a moment M v. Resultant R mav be
resolved into two components, Rx and R t. These three resultants’ are
depicted in Fig. 337(c). They also represent the action of part I of the
curved bar on part II.
A comparison of Figs. 337(a) and 337(c) immediately reveals that
bending moment M in section mn will be equal to My, normal force
N will be equal to Rx and shearing force Q to R z.
The internal forces in curved bars—the bending moment, normal
force and shearing force-can be calculated as in the bending of
straight bars through external forces acting on one side of the cross
section. Their compulation amounts to solving the equations of statics.
Ch. 24) Curved liars 425

The bending moment is equal to the algebraic sum of the moments

of all the forces located on one side of the section about the centre of
gravity of the section.
The norma! force is equal to the algebraic sum of the projections of
all the forces located on one side of the section on the tangent to the
bar’s axis drawn through the given section.

The shearing force is equal to the algebraic sum of the projections

of all forces located to one side of the section on the vertical axis of
the section.
The bending moment will be considered positive if it increases the
curvature of the bar. The normal force will be considered positive if
it tends to detach the portion under consideration from the removed

portion. The shearing force will be considered positive if it is obtained

by rotating the positive direction of the normal force through 90°
clockwise (Fig. 338).
As in a beam, while determining Af, N, and Q we may consider the
equilibrium of either the left or the right portion of the bar, into which
it is divided by the particular section; the selection is arbitrary and
depends upon the convenience of computations.
426 Resistance Under Compound Loading [Part V I/

The sign conventions decided above for the bending moment, nor­
ma! force and shearing force are. independent of whether the left or the
right portion is considered.
Let us study an example for determining M, jV, and Q. Consider a
bar representing one quadrant of a circle of radius R0, fixed rigidly at
one end and loaded at the other by a force P (Fig. 339), Draw an arbi­
trary plane with the centre of gravity at 0 . Location of the plane is de­

termined by angle q>which it makes with the vertical. We consider the

right portion of the bar to determine. M, N, and Q. This spares us the
trouble of determining the reactions in section C.
The bending moment is equal to the moment of force P about
point 0:
M — + P xOD = + P R 0 sin ip (24.1)
Projecting force P on the normal to the section and on the plane of
the section itself, we obtain
N - — P sirup, Q= +Pcosqp (24.2)
Hence the maximum bending moment and normal force occur at
<P—90°, i.e. at the fixed end. Figure 340 shows the A4-, JV-, and Q-
diagrams. The axis of the bar has been taken as the zero line. Ordi­
nates have been cut along the radii of curvature of the bar.

§ 134. Determination of Stresses Due to Normal

and Shearing Forces
The shearing stresses acting in a section of the curved bar add up
to form the shearing force, Q. We can derive precise formulas for cal­
culating shearing stresses in curved bars using the same approach that
was employed in calculating shearing stresses in straight beams. How­
ever, theoretical investigations reveal that the distribution of shearing
stresses in curved bars closely resembles their distribution in straight
beams. It is therefore permissible to calculate shearing stresses in
Ch. 24} Curved Bars 427

curved bars ;by Zhuravskii’s formula, which was derived for straight
beams:

The strength condition for shearing stresses in curved bars may con­
sequently be written as
(13.7)

Let us now determine the normal stresses due to the two resultant
internal forces: bending moment M and normal force N. Let us first
consider the normal force. _
Considering an..element of length ds of the curved bar which is acted
upon by forces N (Fig- 341), we notice that these forces acting at the

Fig. 34!

centres of gravity of the sections result In simple axial tension or com­

pression of the element under consideration. Therefore, the corre­
sponding stresses must be normal to the sections and uniformly distri­
buted over the cross-sectional area, A :
(24.3)

The sign of the stress is determined by the sign of force N.

§ 135. Determination of Stresses Due to Bending

Moment
The task of finding the law of distribution of normal stresses due to
the bending moment over the section and deriving appropriate formu­
las for computing them is statically indeterminate and, as in straight
beams, requires that besides-writing and solving the static equations
we must consider the corresponding deformations and write down ad­
ditional equations. While determining stresses due to Q and N we
could manage without similar computations, because we made use of
known solutions; to determine the normal stresses due to the bending
moment M we propose to carry out all the computations, which, inci­
dentally* we have already given while deriving the formula for normal
stresses in straight beams.
428 Resistance Under Compound Loading [Part V II

Under ihe action of external moment M the curved bar A B (Fig. 342)
experiences pure bending over its whole length.
Figure 343 depicts the part DC of the bar which is acted upon by in­
ternal forces transmitted from the removed parts AD and CB. Bending
moment M is shown on the left and the elementary normal force <j dA
on the right.
The location of the neutral layer along the height of the section is
not known and has to be determined; we shall assume that it does not
pass through the centres of gravity of the sections. Let the origin of

coordinates be located at point C, which lies on the neutral axis, y>

but does not coincide with the centre of gravity 0; moreover, distance
OC is as yet to be determined. The e-axis is the axis of symmetry, and
the x-axis is perpendicular to the plane of the section. Bending moment
M lies in the plane of symmetry xCz, and each elementary area dA
with coordinates y and z is acted upon by a force o dA. We’can write
six equations of equilibrium for the portion, which retains its equilib­
rium under the influence of M and c d A .
The projection of the external forces on the x-axis is zero; the sum of
projections of forces odA may be represented by an integral over the
whole cross-sectional area:
l< JdA~0 (24.4)
A

Equations that represent the projections of all the forces on the y

and the z-axis,
J j Y = 0 and 2 Zb=0
become identities, because the o’s are perpendicular to the y - and the
z-axis. We similarly get an identity from the equation of moments
about the x-axis:

because neither force o dA, which is parallel to the x-axis, nor bending
moment M, which lies in plane xCz, give a moment about the x-axis.-
Ch. 24) Curved Bars 429

By similar logic the moment M about the 2-axis must also be zero;
as for elementary forces o d A , their moment about the above axis is
given by the integral \ a dA y.
A
The fifth equation of equilibrium will therefore be
2A *r = 0, \a d A y = 0 (24.5)
A
The last integral is zero on account of the symmetry of the section
about the z-axis.
Now we equate to zero the sum of moments of all the external forces
about the #*axis. The equation may be written thus:
2 ^ = 0, M — \c d A z = * 0 (24.6)
A

Hence we get the following two equations by considering the static

equilibrium of the portion:
5 vdA = 0 (24.4)
A
M — $odAz = 0 (24.6)
A

We still do not know the law of distribution of normal stresses over

the height of the section. For this reason let us first study the deforma­
tion of the bar.
As in uni-planar bending of straight bars, we shall make use of the
hypothesis of plane sections, which has been experimentally found

applicable for curved bars also. We shall assume that, under the in­
fluence of bending moment, sections perpendicular to the axis of the
bar remain flat and simply turn w.r.t. one another (Fig. 344). The
fibres of the neutral layer CiCt-CjC^ retain their original length, and
fibres located at equal distances from the neutral axis elongate and
430 Resistance Under Compound Loading [Part VII

shorten by an equal amount and: hence experience equal stresses oyer

the width of the section. Let us establish a relation between the rela­
tive angle of rotation and deformation of fibres for two adjacent sec­
tions. Let us cut from the curved1bar
which is being acted upon by only
a bending moment (Fig. 345) an ele­
ment bv two close sections making
an angle dq> with one another. This
element is depicted in Fig. 346:
OrOa .js the axis of the bar,; and
/> ; ^ C^Cj is its neutral layer.
L*"“ The normal stresses acting in the
cut planes form couples;.: due to
these force couples the angle be­
tween adjacent Sections /-/ and 2-2
changes by 6 d<p*on account of rela­
tive rotation of these sections.
Let us determine the normal stres­
Fig. 346 ses in these sections at points A\
and At which lie at a distances from
their respective neutral axes. We
select the positive direction of thez-axis towards the outer fibres. Fibre
A\A% elongates by AjDt; the corresponding stress is
cr = e £

where e is the relative elongation of fibre A xA t . It is equal to the ratio

of absolute elongation AiDz to the initial length of the fibre AiAt:

At
Denoting the radius of curvature of fibre AtAa by p, we obtain
A aD9 = z 6 d<f, A 1 Ai = pdq>
„ z 6d<p 2 0dq> - ,n .
f f = P*p £ <2 4 J>
Formula (24.7) gives the distribution of normal stresses due to bending
moment M over the height of the section.
As ^ and E are constants for each section, or depends only upon
the z-coordinate and the radius of curvature of fibre A iA t (p=r+z,
where r is the radius of curvature of the neutral layer).
For a straight beam we had obtained a linear law of distribution of
normal stresses; in a curved bar o varies according to a hyperbolic law
(Fig. 347). It is also evident from formula (24.7) that in fibres which
are on the outside w.r.t. the neutral layer the increase of stresses
Ch. 24] Curved Bars 431

is slower than that of z, whereas in fibres which are on the inside w.r.t.
the neutral layer stresses increase faster than z, because z changes sign
from positive to negative.
Hence in a curved bar the normal stresses at the “inside” outer fibre
are greater, and at the “outside” outer fibre are less than the stresses
for the same fibres of a straight bar. This is quite understandable since
the initial length of the inside fibre
of a curved bar is much less than
that of the outside fibre; in a straight
bar these lengths are equal. This
explains the difference in relative
deformation and hence the differ­
ence in stress for these fibres.
Let us proceed with the solution
of equations of statics. (24.4) and
(24.6) with the help of relation
(24.7) obtained by considering the
deformation of the bar. Substitute
expression (24.7) in equation (24.4):

A
f«"-I*T?f"-°
A Fig. 347

Factoring out the constant quantities, we obtain

If " - 0
(24.8)

This equation enables us to determine the location of the neutral axis.

Equation (24.8) implies that in the case of a curved bar it is not the
static moment about the neutral axis, j z dA, that is zero. This clearly
shows that in bending of curved bars, the neutral axis really does not
pass through the centre of gravity of the section. Substituting in equa­
tion (24.8) z=>p—r (Fig. 346), we find
dA
I * ? " - A! " - ' !A
A

wherefrom it ensues that

A
'TZ a (24.9)
432 Resistance Under Compound Loading [Part. VII

The method for computing rwill bedifferen.t.for each particular sec­

tion. Substituting now expression (24.7) in equation (24.6), we get

<24-10>

where M is the bending moment; integration is over the whole cross-

sectional area. The integral in the above equation may be modified
as follows:

1 T d
A A
A = fA A
J M

On the basis of equation (24.8) the second of the last two integrals is
equal to zero, whereas the first is equal to the static moment of the
cross-sectional area about the neutral axis. This integral may be com­
puted as the product of the cross-sectional area by the distance of
its centre of gravity from the neutral axis, z0 (Fig. 347):
S = Azt (24.11)
Hence, equation (24.10) may be written
m - e ^ s ~ o (24.12)
wherefrom
6dq> M
H i 15 (24.13)

and the formula for normal stresses, (24.7), becomes

M 2
S p
(24.14)

Equation (24.12) confirms that the static moment 5 of the cross-

sectional area about the neutral axis is not zero, i.e. in bending of
curved bars the neutral axis does not pass through the centre of gravity
but is slightly (by z0) displaced. In Fig. 347 we depicted this displace­
ment towards the centre of curvature of the bar. After actually deter­
mining r from equation (24.9) for a number of sections we find that the
neutral axis really gets displaced towards the centre of curvature.
This displacement occurs on account of the equality of the total com­
pressive and tensile forces acting in the section. Since the stresses
due to bending moment are less at the “outside” fibre and greater at
the "inside" fibre, as compared to stresses in the corresponding fibres
of an identical section of a straight bar (Fig. 347), the neutral axis
must get displaced towards the inside fibres in order to maintain an
equality of the total tensile and compressive forces.
Ch 24] Curved Bars 433

Adding to Ihese stresses the stresses due to the normal force obtained
in the preceding section, we get the following formula for total normal
stresses in a curved bar:
N , m z
A ' S J (24.15)
The maximum tensile and compressive stresses will occur at the outer
fibres / and 2 (Fig. 347).

§ 136. Computation of the Radius of Curvature

of the Neutral Layer in a Rectangular Section
Equation (24.9) is the expression for determining r:

Let us solve this equation for a bar of rectangular cross section. Let
h be the height and b the width of the section, Ra the radius of curva­
ture of the bar, Rt the radius of curvature of the outer fibres, R a the

Fig. 348 Fig. 349

radius of curvature of the inner fibres and r the radius of curvature of
the neutral layer (Fig. 348). If we divide the section into elementary
strips of area d A - b d p , then equation (24.9) may be written
„_ bh h h
' Ch dp Tfl = . R. (24.16)

wherefrom
7 r-, fi> —!i__ (24.17)

15-3810
434 Resistance Under Compound Loading JPart VII

Formulas (24.16) and (24.17) enable us to determine r and z0 and,

hence, S for a rectangular section.
Location of the neutral layer in sections consisting of a number of
rectangles is determined by the same method as in the rectangular
section of a curved bar; only formula (24.16) becomes more compli­
cated.
Let as consider an 1-beam having flanges of different sizes (Fig. 349).
The denominator in equation (24.9) is calculated as follows:

The radius of curvature of the neutraZ layer is determined from the

expression
j ._
(24.18)
in Aj Aft *n Aft

§ 137. Determination of the Radius of Curvature

of the Neutral Layer for Circle and Trapezoid
To determine the radius of curvature of the neutral layer for a circu­
lar section of diameter d, we cut the disc into elementary strips of area
dA by lines drawn parallel to the neutral axis (Fig. 350).

Fig. 350 Fig. 351

Let us express dA and p as functions of angle <p subtended at the

centre. It is clear from the diagram that
p = /?,+■—sin cp, dA = b 14p
Ch. 24] Curved Bars 435

But
bg = dco$tp and dJp=ycos<pdq>
(P
which implies that d A = y C o s 8<pd<p. The denominator in equation
(24.9) may be written
+ 3 1 /2
f* d A _ f* d1cos8 <prf<p
J p ” J 2/?0+ d sin ©
A -3 1 /2

After integration we get

+ 31/2

I 2R<,+ rf5inqp = :nC(2#o—

-n/2

Putting this value in equation (24.9) and substituting nd2/4 for A, we

obtain
(24.20)
4 ( 2/?6 —■ 4/?o— d*)

For a trapezoid (Fig. 351) we again use equation (24.9). The area of
the trapezoid is

The width of the trapezoid at a distance p from the centre of curvature

is
b (p) = &i+(&S- A ) » dA = b(p)dp

The integral ^ y has the following value (dropping the intermediate

A
operations):
) ^ $ - ( b > - 6,)

Now from equation (24.9) we get

(24.21)

When bi=bu i.e. when the trapezoid becomes a rectangle, the above
formula becomes identical to formula (24.16).
15*
436 Resistance Under Compound Loading [Part VII

When 61 - 0 . we obtain the formula for determining the neutral

axis of a triangular section:
hb
r— (24.22)
2/?ir ,n

§ 138. Determining the Location of Neutral

Layer from Tables

With the reasoning of §§ 136 and 137 for rectangular, circular and
trapezoidal sections, we can calculate r and z0 for an arbitrary section.
The results for a few shapes are given in Table 17. In this table the
values of z0 are given as fractions of the radius Ro depending upon the
ratio — , where c is the distance of the inner fibres from the centre of
C
gravity of the section. In the extreme left columns are given values of
Role. At the top of all the other columns is given the shape of the parti­
cular curved bar. The quantity Zo is obtained by multiplying the cor­
responding tabulated value, k, with Ro, i.e.
Zo = kR*
n
It is evident from this table that when the ratio ~ increases, the
ratio rapidly approaches zero, i.e. the neutral axis approaches
the centre of gravity. This means that the difference in the working of
material In a curved and a straight bar diminishes to the point of be­
coming immaterial. It follow's that in the limit the neutral axis passes
through the centre of gravity of the section. Hence, when & is large,
the location of the neutral axis and the stresses in the curved bar are
determined, with a small error, by the same formulas which are used
for straight bars.
For a ratio of — equal to ten, the quantity z0 may be considered
equal to zero for all practical purposes.

§ 139. Analysis of the Formula for Normal

Stresses in a Curved Bar

Substituting the coordinates of the farthest points of the section in

the formula for normal stresses (24.15),
for point /, 27 and R+ (outside fibres)
for point 2 , —z% and R%(inside fibres)
Ch. 24] Curved Bars 437

Table 17
Locating the Neutral Layer from Tables

K.

1.2 0.361 0.336 0.352 0.269

1.4 0.251 0.229 0.243 0.182
l.C 0.186 0.168 0.179 0.134
1.8 0.144 0.128 0.138 0.104
2.0 0.116 0.102 0.110 0.083
2.2 0.096 0.084 0.092 0.068
2-4 0.082 0.071 0.078 0.057
2 .6 0.070 0.061 0.067 0.049
2.8 0.060 0.053 0.058 0.043
3.0 0.052 0.046 0.050 0.038
3.5 0.038 0.033 0.037 0.028
4.0 0.029 0.024 0.028 0.020
6-0 0.013 0.011 0.012 0.0087
8.0 0.0060 0.0060 0.0060 0.0049
10.0 0.0039 0.0039 0.0039 0.0031

1.2 0.418 0.408 0.453 0.399

1.4 0.299 0.285 0.319 0.280
i.C 0.229 0.208 0.236 0.205
1.8 0.183 0.160 0.183 0.159
2 .0 * 0.149 0.127 0.147 0.127
2 .2 0.125 0.104 0.122 0.104
2.4 0.106 0.088 0.104 0.088
2.6 0.091 0.077 0.090 0.077
2.8 0.079 0.067 0.078 0.0G7
3.0 0.069 0.058 0.067 0.058
3.5 0.052 0.041 0.048 0.042
4.0 0.040 0.030 0.036 0.031
6 .0 0.018 0.013 0.016 0.014
8 .0 0.010 0.0076 0.0089 0.0076
10.0 0.0065 0.0048 0.0057 0.0048
438 Resistance Under Compound Loading [Part V II

we may write down the following strength condition of a curved bar:

- H i M Zx c i
a >“ T + — fiT < W 1
N M Zt ^ . r ’ (24.23)
0« - T — n j < W J
If the material has unequal strength under tension and compression,
then to] will have two different values. On account of the fact that two
factors, M and N, give rise to normal stresses, it is more complicated to
determine the critical section for a curved bar than for a straight one.
In some cases (see § 133) M and N attain maximum values in the same
section, which obviously is the critical section. If the situation is dif­
ferent, the strength of the material has to be checked in a number of
sections and the critical section can be determined only after appro­
priate calculations.
If the radius of curvature Ro of the bar is large as compared to the
height of the section h (precisely, when R£>5h), then the ratios
~ or become negligibly small and the normal stresses which
depend upon the bending moment differ only slightly from the normal
stresses calculated by using the formula for a straight bar. This state­
ment can be easily verified from the data given in §§ 135 and 136. Let
us take, for example, equations (24.10) and (24.7). Eliminating
and substituting r-fa for p, we obtain
a= (24.24)

If we neglect - , then formula (24.24) becomes the same as the formula

for calculating normal stresses in a straight bar:

Let us determine the error that is made if we determine the maximum

normal stresses due to the bending moment in a curved rectangular bar
from the formula for straight bars
R 0 = Sh
The radius of curvature of the neutral layer is
r = _______!1 ______ r - - r— h —r i o fi< m
5.5 0.20067
‘”5 ^ 0 3 5 lnt 5
Ch. 24\ Curved Bars 439

and consequently
2»= ^ o - r = 00167/i, or ?o= 0.00334/?o
i.e. the neutral axis passes from the centre of gravity at a distance
which is only ^ of the height of the section.
The normal stresses due to bending, when calculated from the for­
mulas for curved bars, are found to be
M 2, AfxO 5167/1 0.5167xAJx6
S R x ~WiXO OI67/iX5.5A= 0.55I1M’
_ M 2a Mx0.4833/i 0.4833xA !x6 ,M
1 S Rt Wtx0.0167A X 4.5/i------ 0 4509«t* = 1,U/1 W

l.e. the values of stresses differ by ±7% from those calculated by the
formula for straight bars.
This is the main reason why curved bars are divided into two groups
for purposes of strength check. Bars with a large curvature
fall in the first group. In such bars the normal stresses should be cal­
culated by the following formula:
N
° w “ T ± T 3 ? r ,< W (24.23)
Practical examples of this group of bars are machine parts like hooks,
chain links and rings. To the second group belong bars with a small
curvature, in which the racfius of curvature of the axis is large as com­
pared 1o the dimensions of the cross Section ( ~ > 5 j . In such bars
the normal stresses due to bending may be calculated according to
the formula for straight bars:
V M
a^ = JA ± W ^ ^ (24-25)

This group generally consists of curved bars used in various structures:

arches, domes, etc.

§ 140. Additional Remarks on the Formula for

Normal Stresses
While checking the strength of curved bars, we often obtain consi­
derably high stresses at the inside fibres. These stresses (Fig. 347) start
decreasing fairly sharply at a small distance from the edge of the sec­
tion. Thus, they bear strong resemblance to local stresses and their ef­
fect on the strength of a material must be taken into account according
to the recommendations given in § 15: ductile materials (mild steel) do
440 Resistance Under Compound Loading [Part V II

not face danger of failure because of these stresses exceeding the yield
stress, if the loading is static.
The fundamentals of the theory of analysis of curved bars, discussed
in § 135, were first put forward by the Russian Academician A. V. Ga-
dolin between 1856 and 1860. The accurate theory of bending of curved
rectangular bars was formulated in 1880 by Kh. S. Golovin; results ob­
tained by him prove that sections of curved rectangular bars remain
planes after bending. Experiments conducted for the verification of
this theory show satisfactory concurrence of the results with the theore­
tically computed values.
The hyperbolic law of distribution of stresses can be distinctly seen
by beaming monochromatic polarized light on a transparent model of
a deformed curved bar. We notice rows of dark and light strips in the
model; the sharper the change of stresses the narrower and more fre­
quent are these strips. Figure 352 shows the distribution of strips un-

S
Fig. 352

dor pure bending for a model which has a straight as well as a curved
portion. The strips are spaced uniformly in the straight portion be­
cause the stresses change linearly, i.e. uniformly. In the curved portion
we notice a concentration of the strips on the concave side and an op­
posite picture on the convex side, which corresponds to sharp and non-
uniform increase of stresses in the former zone and a considerably
slower change in the latter.
While studying the distribution of normal stresses in curved bars
we ignored the radial normal stresses which occur due to mutual com­
pression of fibres of the bar material. These stresses have much greater
importance for curved bars than for straight bars, as is seen from exper­
iments on gypsum (brittle) models. These stresses are particularly high
in sections in which the width changes suddenly (I-beams).
Ch, 241 Curved B a n 441

§ 141. An Example on Determining Stresses

in a Curved Bar
A bent frame is acted upon by two forces P of 800 kgf each. Find the
maximum stresses in section AB. The radius of the axis is R, —80 mm,
and the cross section is a rectangle 80 x 30 mm in size (Fig. 353).

As— < 5 , we must use the formula applicable to bars with large
curvature. Let us determine radius r of the neutral layer:
h
r= -
In
Ht
In our example /i=80 mm, #i= 120 mm, and Rt —40 mm; therefore
80 _ 80
f = . 120 1.099
'72.8 mm
, n 40
Now we calculate the quantities necessary for analysis:
2 %— Rq— r = 80— 72.8 = 7.2 mm=0.72 cm
$ = 4*0 = 8 x 3 x 0 .7 2 = 17.3 cm3
Z| =*2 ' “F *0= 4+ 0.72 = 4.72 cm

* ,= -y —z0= 4—0.72 = 3.28 cm

Bending moment about the centre of gravity of the section is
M. = —800 x 2 5 = —20000 kgf-cm
Normal force AT=+800 kgf. The cross*sectional area 4= 2 4 cm2. The
normal stresses at point 4 (<r2) and B (orj) are
a, = + f £ - X ± y = + 3 3 - 455 — 422 kgf, cm*
<’. = + | p + ^ x X 5r ‘= + 3 3 + 9 ‘>8 = +98* kgt+nr
 442 Resistance Under Compound Loading j Part VII

Had we ignored the curvature of the bar and computed the stresses
by the formula

we would have obtained

o 800
^ = + 3 3 T 625 = { -5 9 2 } kgf/cm«
0 24 +

The stresses in the inside fibres would have been

9 8 1 -6 5 8
sin 100 —33%

less, which is beyond the provided factor of safety. Hence, we may

conclude that considerable overstressing may occur if the cross section
of a curved bar is designed without taking into consideration its curva­
ture.

§ 142. Determination of Displacements in Curved Bars

Analysis and experiments show that though the curvature must be
accounted for while determining stresses in bars of large curvature, the
same may be ignored in majority of cases when deformation is being
determined. Let us study how to
calculate potential energy expended
in bending of a curved bar.
Let us cut from the bar an ele­
ment of length ds by two cross sec­
tions (Fig. 354). Both faces of the
element will be acted upon by shear­
ing stresses, which give a result­
ant force Q, and normal stresses,
which give a normal force N and a
bending moment M.
For determining the potential
energy accumulated in the element,
we must compute the work done by
all the forces acting on the element.
Fig. 354 While determining the potential
, , . energy of a beam we neglected the
work done by the shearing forces. This simplification is all the more
justified in case of curved bars because the effect of shearing forces is
still less.
Now all we have to do is to calculate N and M. If wc neglect the
curvature of the bar, then this is equivalent to assuming that the
C /t. 2 4 ) Curved Bars 443

deformation of the element under force couple M is identical to that of

a beam. The potential energy due to this deformation is equal to ;
the only difference when compared to the expression for potential
energy of a beam lies in a different notation of the length of the element,
ds instead of dx.
On account of the fact that we ignore the curvature of the bar, the
neutral axis must pass through the centre of gravity of the section.
Therefore, when the section rotates under the action of Af, the centres
of gravity 0i and 0 a do not move and N does not do any work. Conse­
quently, we may calculate the work done by N independent of Af and
then add to it the value obtained from the expression given above.
Forces N acting on the element produce simple tension or compres­
sion; the potential energy accumulated during tension or compression
is given by the expression The potential energy accumulated in
the element is
M*ds N* da
dU = 1ET
2EA

The potential energy accumulated in the whole of the bar is obtained

by integrating the above expression over its total length:

M w + I w <24-2«>
$ s

According to Castigliano's theorem, the first derivative of this ex­

pression w.r.t. concentrated force P gives us the linear displacement of
the centre of gravity of the section in which force P is applied. Similar­
ly, the first derivative of U w.r.t. Af0 gives us the angle of rotation of
the corresponding section:
rd U f AidsdM , (* N d s ON
' “ ■ a r - j E J dP 1 ,J ~Ta "5p (24.27)
’ M ds dM , f*N ds dN
CD

11
II

I EJ dM0 1 J EA dM9 (24.28)

6
Mohr’s method may also be used for determining displacements in a
curved bar. Formulas (24.27) and (24.28) are replaced by
(2 4 2 9 )
s s
Let us apply this formula for calculating the vertical displacement
oi end B of the curved bar whose axis is described by radius /?*. The
bar must be drawn in two states:
(a) when it is loaded by the given force, P (Fig. 355(a));
444 Resistance Under Compound Loading |Part VI!

(b) when it is loaded by a unit force P °=I acting on section B in

the direction of the required displacement (Fig; 355(6)).
Let us calculate Af (*), M°, N(x)t and N°:
M = -j-/>i?0s«n(p, Af° = /?,sin<p
W= — Psinip, N* = — sin <p, ds = R t dip

Substituting the above values in formula (24.29), we obtain:

where i is the radius of gyration of the section.

The first term in the parentheses show's the effect of the bending mo­
ment on deflection, the second term shows the effect of normal force.
Since in the majority of cases is a small quantity, the effect of
i e
normal force on the deformation of curved bars is in a number of cases
comparatively small.

Fig. 355

If we want to find the horizontal displacement of point B, we

should apply a unit horizontal force P9—l at the above point. We can
similarly find the angle of rotation of this section by applying Afu= l .
If it is required to break the bar into a number of portions for cal­
culating M and N, then each of the corresponding integrals in formula
(24.29) becomes a sum of integrals with appropriate limits.
Ck. 24\ Curved Bars 445

§ 143. Analysis of a Circular Ring

Let us find the stresses in the critical section of a circular ring
(Fig. 356) subjected to two tensile forces P. The radius of the ring is
Ro and its rigidity EJ. The problem is statically determinate as far as
the external forces are concerned. However, with respect to internal
forces it is statically indeterminate.
Let us cut the ring by a horizontal section into two parts; the upper
part is shown in Fig. 357. The sectioning plane will experience inter­

Fig. 356 Fig. 357

nal forces transmitted from the lower (removed) part: normal force
N=Q.5P and bending moment M A drawn arbitrarily as shown in the
figure (there is no shearing force in the horizontal sections). We have
exhausted all equations of statics in drawing these conclusions from
the symmetry of the ring, but moment M A still remains unknown. Let
us now consider a section making an angle q>with the sectioning plane
(see Figure). The following forces will act in this section:
M y = M^ + 0.5/, /?0( l — cos(p) (24.30)
Nv = 0.5 P coscp (24.31)
—O.SPsiny (24.32)

Since the section is symmetric, the angles of rotation of horizontal

sections to which moments M A are applied will be zero; therefore the
partial derivative of potential energy with respect to M A will also be
zero:

-£J ^ [Ma 4- 0.5P/?o (1 —cos <p)] dtp «*=0

446 Resistance Under Compound Loading [Part VII

The equation is solved as follows:

n/ 2
| (Ma + O.SPR9 — 0.5PR0 cos<t>)d<p= MA£+Q .5PR0%— 0.5PR 0 = 0

M A = ~ 0.5PR 0 (1 - 1 ) = - QA82PR,

Hence, moment M A acts in the opposite direction to the one shown in

the figure.
With the help of formulas (24.30)-(24.32) we can determine the in­
ternal forces in any section of the ring. The section that evokes maxi­
mum interest is section B. In this section at q>=n/2 we have
MB = — 0.182PR, + 0.5 PR 6 = 0.318PR<>
Q8 = 0.5P, Nb = Q
Thus, we see that for a ring section B , where force P is applied, is
critical although the normal force in this section is zero.
The reader is advised to plot the bending-moment, normal-force and
shearing-force diagrams for the ring section using formulas (24.30)
and (24.32).

CH A PTER 25

Thick-walled and Thin-walled Vessels

§ 144. Analysis of Thick-walled Cylinders
Wc were perfectly justified in considering the distribution of stres­
ses as uniform over the thickness of the wall in a thin-walled cylindri­
cal reservoir subjected to internal pressure {§ 29). This assumption has
minimal effect on the accuracy of design.
However, such an assumption in the case of cylinders having con­
siderable wall thickness as compared to their radius is sure to result
in large errors. The analysis of such cylinders was worked out by
G. Lame and A. V. Gadolin in 1852-4. The latter gained worldwide
fame thanks to his works on analysis of curved bars in application to
strength analysis of artillery guns. Figure 358 shows the cross section
of a thick-walled cylinder of external radius r, and internal radius r2;
the cylinder is subjected to external pressure pt and internal pres­
sure p2.
Let us consider a very thin ring of radius r in the cylinder wall. Let
dr be the thickness of the ring and let AB (Fig. 359) depict a small ele­
ment of this ring subtending an angle c/0 at the centre.
Suppose the element has unit thickness in a plane perpendicular to
the plane of the figure and suppose ar and or+dor are the stresses acting
Clu 25] Thick-walled and Thin-walled Vessels 447

at the inner and outer faces of element AB\ also suppose at is the stress
at its side faces. It is obvious from the symmetry of the section and
the load that element AB will not warp and that no shearing stresses
will act on its faces. Faces of the element which lie in the plane of the
<Sr+&Gr

figure will experience the third principal stress, at, caused by the pres­
sure on the cylinder base. This stress may be considered constant over
the cylinder’s cross section.
In the plane of the figure, element AB is acted upon by two forces
0 t drX l, making an angle dQ between themselves, and a radial force
(or 4- d<jr) (r+ dr) dOx 1—a,r d0 x 1
This force is directed towards the outer surface. In equilibrium the
three forces constitute a closed triangle abc. It is evident then that the
radial force represented by segment ab is connected with force o t dr
(segment ca) by the following relation:
ab — cadQ
or
[(ar -1- dar) (r 4 - dr)—<rrr] d0 = at dr dft
Neglecting the small quantities of higher order, we get
<sf dr-\-dofr = o t dr
wherefrom
a' - o- + - i3 r r = 0 <251>
The equilibrium conditions give us only one equation for determin­
ing two unknown stresses. The problem is statically indeterminate and
we must consider the deformation of the cylinder.
Deformation of the cylinder consists in its elongation and in radial
displacement of all points of its cross sections*" Let us denote the radial
448 Resistance Under Compound Loading | Part VII

displacement of points of the internal surface of the element by u

(Fig. 360). Points on the outer surface will get radially displaced by
u+du. Thus, thickness dr of the element will increase by du and the
relative elongation of the cylinder
materia! in the radial direction
will be V = ^ - .
In the direction of stresses at the
relative elongation et will be equal
to the relative elongation of arc ab
when it occupies position cd. As
the relative elongation of the arc is
the same as the relative elongation
of radius r, «<= ■-. From Hooke’s
law (formulas (6.21), §34),
e r = 4 - ( ° '—per,— = ~jL
Fig. 360
8#= -^ (o,—p<Tr—JW,) = 7
(25.2)
As both e, and er are functions of u, they must be compatible. Dif­
ferentiating e, w.r.t. r, we get
du
d*t dr r 11 I ( du u\ 1 , v
"37 — r* T i T F - T r T l * '- * ') (25.3)
This is the condition of joint deformation. Substituting in it the
values of e, and e( from (25.2), we obtain the second equation correlat­
ing <j, and o>:
7F | t — iw *)] = 7 *4 r - ( a , - * * )
or
d<t/
"37 <crr—crr) (25.4)
Substituting in this equation the value of or—a t from (25.1), we gel
doi dor ... do.
-ar-c-jr^-e+^-jf
or
4 r+ 4 r= ° (25.5)
For simultaneous solution of equations (25.1) and (25.5) we differen­
tiate the first w.r.t. r and substitute in it the value of ^ from the
Ch. 25] Thick-walled, and Thinwalled Vessels 449
second. This gives us
dar dot
dr dr f r
The differential equation may be rewritten as
3 dor
r dr (25.6)

The solution of this equation is

(25.7)

which can be checked. Constants A and B are calculated from the

boundary conditions at the internal and external surfaces of the cy­
linder:
(Or)r=r,----- Pit (Or)r=f, -------------------------- p 9 (25.8)
The negative, sign in the right-hand sides of these expressions signifies
that the positive direction of a, corresponds to tensile stresses (Fig. 359).
From the expression (25.8) we get
a s (Pt— P i)r\r\
A — P t ' i — P i 'i
n r-—r*'~
1 rt /'*—fi
1 ri
The values of the constants and equation (25.7) give us (he final
formulas for ar and o,:
(Pt-Pi)rirj
r 'i'f- r l)
(25.9)
. _ pA —Pin . iPt—Pi)Ar%
rl ri
It is obvious from these formulas that the sum crr+crf does not de­
pend upon r y i.e. the strain along the axis of the cylinder is the same
at all points of the section (as <rz is the same for all points), and the
section remains a plane.
A situation in which only internal pressure pt acts on the cylinder is
of considerable practical importance. Here

Figure .361 shows the distribution of stresses over the thickness of the
cylinder waifs when />(=0. As <t? is usually much less than or and in
450 Resistance Under Compound Loading [Part VII

magnitude, only the latter two are considered in checking the strength
of the cylinder. From the third theory of failure (theory of maximum
shearing stresses) we find that the maximum difference of principal
stresses,
(25.11)

occurs at points of the internal surface of the cylinder and is always

considerably greater in magnitude than the internal pressure.
Thus, permanent deformation begins at the internal surface of the
cylinder when (cr,— becomes equal to the yield stress of its ma­
terial; any attempt to curb the appearance of permanent deformation

Fig. 361 Fig. 362

by increasing the external radius n is accompanied by an increase of

the numerator as well as denominator in formula (25.11). Therefore,
although the difference of principal stresses (at—or)max becomes less,
the decrease is very slow. However, when permanent deformation be­
gins at the internal surface of the cylinder, this does not mean that the
maximum lifting capacity of the structure has been exhausted; we can
properly evaluate the strength of the cylinder only by analysis based
on the method cf permissible loads.
Lifting capacity of thick cylinders in the elastic range may be im­
proved by creating initial stresses. For this the cylinder must be made
of two cylinders, one fitted into the other. The external diameter of
the internal cylinder is madea little more than the internal diameter
of the outer cylinder. The outer cylinder is put on the internal one in
healed stale and upon cooling gives rise to reactions at the surface of
contact; the reactions compress the internal cylinder and stretch the
outer. The analysis given below wil 1 show that these initial stresses
improve the working of the composite cylinder which is subjected to
internal pressure.
C h. 28] Thick-walled and Thin-walled Vessels 451

Figure 362 shows a composite cylinder after it has cooled. Stresses

in a tangential direction will be: for the outer cylinder (tensile)
f_ Psf3 | P tf ?/•$
' rl ~ rsr i(r*~r$)
and for the inner cylinder (compressive)
% 22
__ Pzr *___________P%r*r%
* rl ~ rl
Figure 363 shows the distribution curves of these initial stresses for
the following numerical data:
r, = 11.50 cm, r 8 = 5.70 cm, r 3 = 8.25 cm, pu= 280 kgf/cm*
For the outer cylinder stresses at the externa! surface are
2r\
= + Ps -rY ^ ri °= + 613 kgf/cm*
rl r9

and stresses at the internal surface are

= + r\ _ rl - + 695 kgf/cm*
rl '9.

For the inner cylinder stresses at the internal surface are

* 2r2
ah = — A> = — 1080 kgf/cm*
'a 'a
and stresses at the external surface are

= — Pa = — 600 kgf/cm*
« 'a

Let us now assume that the cylinder is subjected to an internal pres­

sure of p4=3400 kgf/cm1. The distribution of ut without considering
the initial stresses p 3 will be given by formula (25.10):
__ Pan
I r*
The limiting values of these stresses are:
at the external surface 0^=4-2245 kgf/cm8
at the internal surface <^’= 4-5620 kgf/ctn2
The corresponding curve is’depicted in Fig. 363. When internal pres­
sure and initial stresses act simultaneously, the total stress may be
452 Resistance Under Compound Loading [Pari V II

taken as the sum of ordinates of curves crt+o't and cr,+cr?; the curve
of total stresses has the shape of a tooth, as is shown in Fig. 363.
The shape of the resultant curve shows that when initial stresses
p3 are acting, the stresses in the outer cylinder increase whereas the
stresses in the inner cyl inder decrease.
As a consequence, the material
works more uniformly—the maxi­
mum stress comes down to 5620—
—1080=+4540 kgf/cma and the mi­
nimum stress grows up to 2245+
Mini itnsses
+613=2858 kgw'cm*. Of course, this
y /z \ distribution holds well only when
the material is working within the
woo • ^ 2 elastic limit.
Cl Let us determine the difference of
m
Iadiii "’hj ch !*1» se" ‘ial
3000 /P to create the required initial stress
mo
5000\
A p 3 (here ra is the initial external
radius of the inner cylinder, and rl
is the initial internal radius of the
m ol flto jtmsa o u (er cylin(fer).
As the outer cylinder cools, these
Pig 353 radii tend to become equal due
to a decrease in r3 by Ar”9 and an
increase in r3 by A/y, the sum of ab­
solute values of these deformations must be Ar,:
|A r;|+ |A r ;| = Ar3
Relative tangential elongation of the material at the internal sur­
face of the outer cylinder is

= t ( ° ; ( - £ ^ l +n)

In this formula for r3 we substituted the radius r 8=r^—Ar^ common to

both cylinders; this is possible because Ar; is a small quantity and the
error committed by it is very small. Relative elongation of radius r'a
will also be e'(; therefore
Ar; = ejr, _ Pa'3 / ri + ri
£ V r f- r l
Relative tangential compression of the material at the external sur­
face of the inner cylinder is
Pa (
T K .- K .)
C/I. 2 5 ] Thick-mailed and Thin-walled Vessels 453

and the shortening of radius r\ is

Ar',= P9r 9

The sum of absolute values of Ar\ and ArZ will as before be

!> ,',('l+ 'i , / 'S + 1 \ 2P.1/-2 ra _ rJ
l 4
~ V ‘? R f + l v + " J " v 7F T ^ ~ ‘V s

Hence, in order to create the desired initial stress pa, we must pro*
vide a difference of diameters
r a _ ra
4ftr* ri rs
Ad,
E W -4 ) v t- 'i)
The minimum temperature 1 ° up to which the outer cylinder must
be heated before it is put on the in­
ner cylinder can be determined from G r*dfjr
the following relation:
a tr3i° = A r,
wherefrom
ft_r®
f> = 2pjr i ____________
_ri rt = 66°C
** (1 -D W -D
(We have assumed the following
numerical values: a —125x10"%
£ = 2 X 10* kgf/cm% Ad8=0.0137cm.)

§ 145. Stresses In Thick Spherical

Vessels
Fig. 364
Figure 364 shows an element cut
from the wall of a thick spherical
vessel. The element has internal radius r and external radius r-f-dr;
stresses acting on the element are also shown in the figure. From the
equations of equilibrium and joint deformation we get
B B
(25.12)

Constants A and B may be determined from the boundary conditions

at the internal and external surfaces of the vessel at r= r%and r= ru res­
pectively, where rx and r4 are the external and internal radii of the
vessel.
For example, if the vessel is subjected to an external pressure pi and
an internal pressure p„ constants A and B may be determined from
454 Resistance Under Compound Loading [Part VII

the following conditions:

£
^ = •4 + — = — pt at the internal surface

Q r^A = — pL at the external surface

wherefrom

Therefore

(P* Pi) r*{r]— r|)

(25.13)
+■(Pz Pi) 2r»(r|—rj)

§ 146. Analysis of Thin-walled Vessels

If the thickness of the cylinder wall, t=r-t—rt, is small compared
to radii rl and r2f then from formula (25.10) we get

which is the same as obtained earlier (§29).

A general formula can be derived for calculating stresses in thin-
walled vessels which represent surfaces of rotation and are subjected to
internal pressure p symmetrical about the axis of rotation.

Fig. 365

Let us cut from the vessel (Fig. 365) an element by two meridian sec­
tions and two sections perpendicular to the meridian. Let dsm and dst
be the dimensions of the element along the meridian and perpendicu­
lar to it, and let us denote by pm and p, the radii of curvature of the
meridian and of the section perpendicular to it; let t be the wall thick­
ness.
Ch. 25) Thick-walled and Thin-walled Vessels 455

From symmetry, the faces of the element will be acted upon by nor­
mal stresses om in the direction of meridian and a< in the perpendicu­
lar direction. The corresponding forces acting on the faces of the ele­
ment will be OfljdSff and Otdsmt.
Since a thin shell, just like a flexible
string, has resistance only against
tensile loading, these forces act
along the tangents to the meridian
and to a section normal to the
meridian.
Along the normal to the surface
of the element forces atdsj= ac= bc
(Fig. 366) give resultant aby which
is equal to
ab= 1>cddt = o t

Similarly force gives a resul- ^

tant a^/istdsjn^- in the same dircc-
Pm
tion. The sum of these forces must balance the normal pressure acting
on the element:
pdsmdsi = oMdst d$m~ - + <Ji dsmdst
wherefrom
I P (25.14)
Pm ^ Pt I
This basic equation, correlating a,* and at in thin-walled vessels
having a surface of rotation, was derived by Laplace.
As we had assumed that the stresses are distributed (uniformly)
over the section, the problem is statically determinate; the second
equilibrium equation can be obtained by considering the equilibrium
of the lower portion of the vessel cut by a parallel circular section.
Let us consider a vessel subjected to hydrostatic loading (Fig. 367).
Let us describe the meridian curve in the system of x and y coordinate
axes with the origin at the apex of the curve. Assume that the section­
ing plane passes at a height y from point 0. Radius of the correspond­
ing parallel circle is x.
Each force couple Or^dSft acting on diametrically opposite elements
dst of the section will give resultant be equal to
bc = 2 ab cos 0 = 2omdstt cos0
The sum of these forces acting on the whole circular section will be
2 nxomt cos 0; it will balance the liquid’s pressure p —y(h—y) at
456 Resistance Under Compound Loading fPart V ll

this level and weight Py of the liquid in the cutoff portion:

2n xcml cos 0 —nx*p -J- Py
wherefrom
px
*« = -2t cos 6 2jixt cos 6 (25.15)

Knowing the equation of the meridian curve, we can find 0, x, and Pv

for all values of y and, consequently, de­
termine am. We can then determine or*
from equation (25.14).
For instance, for a conical reservoir hav­
ing apex angle 2a and filled with a li­
quid of specific weight y to a height h
we have
p^csoo, jc = p ta n a
= y ynx'y «By yn tf tan* a
p~*y(h—y), 0*=a
x _ y ta n a
P* = cos a cos a
Therefore
y (h —y ) y t ana y,nya tan* g
« 2/ cos a + Ian a / cos a
_ y (/i—y ) y ta n a y*y tan a
2 / cos a 1 6 / cos a
V irtana / 2 \
2/ cos <
PPty(h—y)ytana
0| I Tcosa
For a spherical vessel of radius r0 sub­
jected to internal pressure p0. from symme­
try we have a. Now as pm= p (= /’o, equation (25.14) gives
2o Pt or o = IPafp
“ r F

If the meridian curve has a discontinuity of angle 0, the equilibrium

of the thin shell at the point of discontinuity can be achieved only
if a reaction acts at this point of the shell. Such a reaction can be made
to appear with the help of special rings capable of taking the load
that occurs due to unbalanced stresses o m on both sides of the point of
discontinuity.
Cli. 26\ Design for Permissible Loads 457

CH A PTER 26

Design for Permissible Loads.

Design for Limiting States
§ 147. Design for Permissible Loads.
Application to Statically Determinate Systems
In the methods explained above for designing under tension or com­
pression statically determinate as well as indeterminate structures
we proceeded from the fundamental strength condition (rmax^l<rl (§§ 4
and 18). According to this condition, the dimensions of the structure
should ensure that the maximum
stress in the critical section does not
exceed the permissible value.
Let us view the problem from a
different angle (§ 4). We require
that the load acting on the whole
structure should not exceed a permis­
sible value. This condition may be
expressed as follows:

The. permissible load is the £th

part of the load at which the struc­
ture ceases to function properly and
no longer serves the purpose for
which it has been designed. The Fig. 368
latter is generally called the ultimate
load and sometimes the breaking load in the broader sense of the word
(destruction of the structure means that it stops functioning properly).
Let us consider a system consisting of two steel rods AB and AC
(Fig. 368) loaded with a force P. By the conventional method of de­
sign we determine forces ATi and N? according to the formula

(from the equilibrium of point A). Hence the cross-sectional area of

each rod must be
^ [o] 2 [o] cosa
By the method of permissible loads we have
P < [P ]
Taking for the whole structure the same factor of safety k which we
had assumed in the method of permissible stresses, we get I F ) = ^ .
458 Resistance Under Compound Loading [Part V II

The ultimate critical load, Pu, is the load at which the stresses in the
rods reach the yjej^ stress;
Pa = 2 Suv cos<x (a)
Thus, the permissible load is
2Soy cos a
k
The strength condition (a) takes the form
2S < j y cos a
k

Keeping, in mind that ^ = lol, we have

P 25 [o] cosa
wherefrom

^ ^ 2 [o] co s«

Hence, design for permissible loads gives the same results as the
design for permissible stresses. This is always true of statically deter­
minate structures with uniform stress distribution, when the material
is utilized fully over the whole section.

§ 148. Design of Statically Indeterminate Systems

Under Tension or Compression by the Method
of Permissible Loads

We get entirely different results if we apply the method of permis­

sible loads for designing statically indeterminate systems in which
the rods are made of a material capable of large plastic deformations,
for example mild steel.
Let us consider as an example a system consisting of three rods load­
ed with force Q (Fig. 369). The rods are all assumed to be made of mild
steel having yield stress a,j. Let us denote the lengths of the side bars
by lu and that of the middle bar by l3. The permissible stress tol=
= -^. As in the previous case, we assume the ratio of the cross-section­
al areas of all the bars to be known; let all the rods be of equal cross-
sectional areas 5. Solving the problem in the same way as in § 18, we
get
Q Q cos 3 a
= NX
I - f 2 cos3 a ’ r + 2 cos3 cc
Ch, 26\ Design for Permissible Loads 459

As N£>Ni, the middle rod is stressed more than the side rods;
therefore S should be determined from the formula
o^, Q
^ Iff] 7* + 2cos3a) [a]
The side rods have the same cross-sectional area; they will have a
slightly greater reserve.
Let us apply the method of permissible loads; the strength condition
may be written as
<3 <[<?]=■%•

What is the ultimate load of the structure in this case? As the struc­
ture is made of a material having a yield plateau, in analogy with
simple tension of a rod of the same material, the^ultimate load is the
load at which the whole structure starts yielding. Let us denote this
load by Q^. Until force Q is less than this value, the deformation
(lowering of point A) is possible only by increasing the load. As soon

Fig, 369

as Q attains the value QL further deformation occurs without any

increase of the load and the structure gets out of order.
Let us study the process of deformation of the system to determine
As the middle rod is stressed more than the side rods, it attains
the yield stress earlier than the side rods. Let us denote the load
corresponding to this instant by Qyi it will be
Qv = (1 -1- 2 cos3a)
where N%=Sov is the force on the middle rod corresponding to its
yield stress.
The stresses in the side rods, having the same cross-sectional area,
will not have reached the yield stress as yet, and they will continue
to be subjected to elastic deformation. For this deformation to occur,
it is necessary that the load on the side rods be increased until the
460 Resistance Under Compound Loading [Part V II

stresses do not attain the yield stress. Only then will the maximum lift­
ing capacity. QjJ, of the structure be reached.
As the yield stress <xy has been already attained in the middle rod at
load Qf/f further increase of the load does not affect it and, consequent­
ly, force N* remains unchanged. Our statically indeterminate system is

transformed into a statically determinate one consisting of two rods

AB and AC and loaded with the force Q acting at point A vertically
downwards and the known force equal to Say (Fig. 370). The struc­
ture will continue to work in this way until
Qy <Q<Q*y
Let us plot the graph of force Q versus displacement f to illustrate
the course of deformation of the given structure (Fig. 37J). As long as
Q^QU, the distance by which point A lowers is equal to the elongation
of the middle rod and is determined by the formula
Ql3
/a t ()-t-2cosa a) ES

When Q falls in the interval the displacement of point

A has to be calculated as the lowering of trie joint of the system of two
rods AB and AC loaded at point A with force (Q—S<j„). From § 18
one knows that the lowering of point A is

h
In its turn
Q— Soy
H i. N ,=
ES ’ 2cosoc

whence
( Q - S o y)tj - (Q—SOy)t^ (Q Soy)
“ “i t s cos a * n s - 2ES cos3 a ~ 2ES cos?' a
Ck. 26] Design for Permissible Loads 461

For /,* (in the second segment) we again get the equation of a straight
line, but in this case not passing through the origin of coordinates.
When force Q attains the value Q£, the stresses in the side rods reach
the yield stress, and further deformation of the system occurs without
increase in load. The displacement curve is now parallel to the x-axis.
To determine the ultimate lifting capacity Ql of the whole system
we must, for a system of two rods loaded with force (Q—Say), find the
value of Q for which the stresses in the side rods reach the yield stress
(the same problem was solved in the previous section). Substituting
Q ~ Sav for P in equation (a) of § 147, we get
(Q — S a y)„ = Q l~ ~ S o y = 2 $ o „ cosol

whencefrom
Ql = Sov (1 + 2 cosa)
The permissible load will be
r /i 1 Ql S q „ ( l+ 2 c o s tt)
JvJ— A — k

Taking into account that

we get
[Q l= S [oj (1 + 2 cosa)
Finally
Q < [Q ] = S[o] (1 + 2 cos a) and Q
[oj (1 + 2 cos a)

This value is less than the value obtained by the conventional method,
i.c. less than
Q
|o | (1 + 2 cos’ a)

At Q=4 tf, a =30°, fa 1000 kgf/cm* (steel) we get:

(1) by the conventional method
n_ 4000 _ 1 7 1 Iim i
1000 ( 1 + 2 cos’ fl0 °)

(2) by the method of permissible loads

n ___ 4000_ _ _ _ _ | i/> j
s Ba idoci7i +2 a S l n “ ' 46
Thus, in designing a statically indeterminate system from a material
having a yield plateau, the method of permissible loads is more econo-
462 Resistance Under Compound Loading [Part V II

mical than the method of permissible stresses. This is quite obvious:

in the method of permissible stresses we take as the breaking load the
force Qu at which only the middle rod attains the yield stress (the side
rods remaining understressed). In the method of permissible loads the
ultimate lifting capacity is determined from the condition
The material of all the three rods is fully employed at the load Q£.
Hence, the method of permissible loads helps us to discover the la­
tent sources of reducing the safety factor of statically indeterminate
structures, increasing their design lifting capacity and achieving great-
er uniformity of strength of all their parts. Without any difficulty
the method can be applied to the case when the cross-sectional areas
of the middle and side rods are not equal.
The theoretical considerations discussed above were experimentally
verified a number of times, and the calculated and experimental values
of the ultimate load were found to be in good agreement with each oth­
er. This assures that the theoretical premises on which the method of
permissible loads is based are correct.

§ 149. Determination of Limiting Lifting

Capacity of a Twisted Rod

The method of designing for permissible loads may also be applied

to torsion. As already explained in § 148 the result obtained by this
method in tension and compression differs from the one obtained by
designing for permissible stresses only for a statically indeterminate
system of bars, because the stresses are distributed uniformly over the
cross sections of each bar. The situation is different in torsion: the
stresses are not distributed uniformly over the cross sections.
In § 49 we determined the required dimensions of a twisted shaft
from the condition that the maximum shearing stress at points on the
contour of the cross section should not exceed the permissible shearing
stress I t ) . We conducted the analysis on the basis of permissible stres­
ses without considering the inhomogeneity of stress distribution in
the section.
In this method of analysis, as in the analysis of statically indetermi­
nate systems, under tension or compression, we do not utilize the ul­
timate lifting capacity of the ro*d to the full. In § 49 we considered as
critical the state of the material when the shearing stresses t equal the
yield stress (for steel) only at the contour of the section (Fig. 372(a)).
According to the distortion energy theory of strength, should be
equal to 0.6 oy. The twisting moment in this case will be:

2
Ch. 26\ Design for Permissible Loads 463

and the angle of twist will be

M,i " 'V V
<P< GJP 2G ~ Gr

To further increase of the angle of twist wc must increase the twist­

ing moment, because the material inside the rod is still in an elastic
state. While the deformation increases, the increase in stress at the

Fig. 372

edge of the section will stop (yielding), and at a certain value M >M t
the distribution of stresses will correspond to the diagram shown in
Fig. 372(6). The material inside the non-hatched circle of radius OB
will continue to be in an elastic state as before.
The limiting state corresponding to complete utilization of the lift­
ing capacity of the rod will be the state in which the elastic zone within
the shaft is completely absent; in such a state the stress all over the
section will be equal to the yield stress xy (Fig. 372(c)).*
The limiting twisting moment Mt may in this case be calculated as
the sum of moments of all internal forces about the centre of the circle.
For this we divide the area of the given section by concentric circles
into a number of infinitesimal rings.
In the limiting state the stresses acting at each point of the section
have a constant value equal to xv (Fig. 372(c)). The internal force act­
ing on an elementary area of radius p will be equal to (Fig. 373) xvdS,
and the moment of the internal force will be px„d5. By summing the
elementary moments of the internal forces over the area of the whole
circle, we get
dM in= xup £ dS = xup2np dp

* This manner of working in the limiting slate is only approximate. Actually,

although the stresses at the centre change sharply, they do not increase in jumps,
and at the surface they do not remain constant but increase due to work hardening
of the material.
464 Hesistance Under Compound Loading fPari VII

If we now write the equilibrium condition for the limiting twisting

moment, we find that
r
M( = $2nt(/p1dp = 0
o
wherefrom
Af, = y jr r 5Tff (26.1)

The maximum permissible twisting moment for safety factor k is:

[ M ,]= 'T = X ' " T = T - rJ W (26.2)

Therefore

r>

Simultaneously, from conventional analysis we have (§ 50)

r> i/H E
V n lx]
Hence, the design on the basis of limiting lifting capacity enables
us to reduce the shaft diameter in the ratio

K5 r ° - 91

Due to non-uniform stress distribution over the section in elastic

state, the transition to design on the basis of limiting lifting capacity
helps reduce the consumption of materials.
It should, however, be borne in mind that the above analysis holds
true only under static loading, when
the failure occurs due to yield­
ing. A vast majority of the shafts
under torsion, however, work under
varying loads, when the failure oc­
curs due to the appearance of fatigue
cracks; therefore the analysis should
be based on this factor. Obvious­
ly, the above analysis cannot be
applied in the design of shafts in a
majority of the cases. As we shall
see later, the analysis of beams un­
der bending will be entirely differ­
ent.
The results obtained are of inter­
Fig. 373 est because they can be checked in
Ch. 26\ Design for Permissible Loads 465

iractice. It has been established experimentally that xv obtained from

fmined
ormuia (26.1) in terms of the limiting twisting moment, also deter­
experimentally, is sufficiently close to 0.6cr„, which it ought to
be according to the distortion energy theory of strength.

§ 150. Selecting Beam Section for Permissible Loads

We have seen in torsion of shafts that, if the stresses are not distri­
buted uniformly over the section, then the dimensions of the sections
obtained for permissible stresses and permissible loads are different.
A similar phenomenon is observed in the bending of beams.
In the analysis based on the method of permissible stresses we de­
termine the size of the section from the condition
_ —fflwpx <:
'•’m ax
[g]
^ I'-’J

For materials having a yield zone (mild steel), [a] is taken as

where av is the yield stress, and k v the corresponding safely factor.

Thus, we consider the material of the beam in critical state when
the maximum stress in the critical section reaches the yield stress.
Me <

Let us denote the bending moment giving rise to this state by My\ it
corresponds to the attainment of maximum carrying capacity of the
material in the maximum stressed layers of the critical section. How­
ever, this state does not mean that the maximum carrying capacity of
the whole beam, as a structure, has been exhausted.
Let us consider a steel beam of symmetrical (for example, rectangu­
lar or I-section) section (Fig. 374(a) and (&)). The distribution of stres­
ses in the critical section for a moment Af„ is shown in Fig. 374(c);
the yield stress is reached only in the boundary layers, and all the re­
maining portion of the beam remains in an elastic state. Therefore,
16-3310
466 Resistance Under Compound Loading {Part V II

for further deformation of the beam we must increase the load and the
bending moment: the lifting capacity of the beam remains to be ex­
ploited fully.
As we increase the moment the yield zone spreads towards the inside
of the beam, the stress diagram appears as in Fig. 374(d) and in the li­
mit, when the material begins to flow along the complete height of the
section and the lifting capacity of the beam is exhausted, it takes the
form of two rectangles (Fig. 374(e)). The bending moment in this stale
of the beam will be the limiting one for it as a whole. Further deforma­
tion of the beam will occur without any increase in the moment; a
so-called ductile hinge will be formed in the critical section.
Let us determine this limiting moment Mev. It is equal to the sum
of moments of forces about the neutral axis, as is evident from Fig.
374(e). A force ay dA acts on the elementary area dA at a distance 2
from the neutral axis; the moment of this force about the neutral axis
is zoy dA. As the section is symmetric, it is sufficient to calculate the
sum of moments of these forces for the upper or lower half of the sec­
tion and double the result. Thus
= 2 $ OyzdA
A /i

where A is the area of the whole section. As <r„ is constant for all points
of the section, we have
Mcv =2ou $ 2d/4=2cr^Sm«
A! 2

because the integral

^ z dA =
>1/2
represents the static moment of one half of the section about the neut­
ral axis.
The strength condition may be written as

For a safety factor of ktJ we get

KU KV

Hence the strength condition becomes

or (26.3)
Ch. 261 Design for Permissible Loads 467

Therefore, when analyzing a beam of symmetrical section for per­

missible load, its dimensions should be calculated not from the section
modulus W but from the static moment of its half-section multiplied
by two. For a rectangular section of height h and width b
25
"m ix =26
—* L2 —
4
bh? , c *
~ T = 1 -5 “5" 1.5W

Putting this value in formula (26.3), we get

Mmax
1.5 loj

Thus, when designed for permissible load the required section modu­
lus of a rectangular beam is 1.5 times less than when it is designed for
permissible stresses.
For any symmetrical section the quantity 2Smax may be taken as a
product of the section modulus and a constant n which depends upon
the shape of the section:
2Smax = nW
Therefore formula (26.3) takes the form
W > Mmax
»l<rl (26.4)

For a rectangular section n=1.5, for I-sections of the type which we

are considering n varies between 1.15 and 1.17; the mean value of n
may be taken as 1.16. Thus, if we start designing steel beams of the
commonly used sections by the method of permissible loads, we may
increase their carrying capacity by 16%, which is equivalent to
increasing their permissible stress. Such an increase in the permis­
sible stress must be thoroughly investigated (during strength check) in
conjunction with other possible factors which may cause failure of
the beam.
It has been experimentally established that 1-section steel beams
never fail solely as a result of the yield stress appearing over the whole
section. More commonly the failure is due to the loss of stability of the
flange (Fig. 375) or the web. Therefore, special attention should be paid
to checking the stability of the elements of the beam when higher per­
missible stress is used upon analyzing by the method of permissible
loads.
If a repeated load acts on the beam, the possibility of failure due to
the appearance of fatigue cracks should be taken into consideration.
This requires an additional check against a failure of this nature oc­
curring in the structure.
The analysis based on permissible loads is somewhat more compli­
cated in the case of beams having one axis of symmetry, for example in
T-section beams.
16-
468 Resistance Under Compound Loading [Part V tl

Figure 376 shows such a section and the diagram of normal stress
distribution when the carrying capacity of the beam is reached. In
such beams we must first determine the location of the neutral axis;
even at this stage of working of the beam it does not pass through the
centre of gravity of the section.
At 1
m m m

■9 ■■■*

Fig. 375 Fig. 376

Let us denote the area of the compressed portion of the section by

At and the stretched portion by A 3. The condition that the sum of the
tensile and compressive stresses should be equal gives
O gA^O yAt or Ai — Ai
The neutral axis divides the area of the section into two equal por-
tions. In bending within the elastic limits the same condition brought
us to the conclusion that the static moments of the compressed and
stretched portions of the section should be equal and, therefore, the
neutral axis should pass through the centre of gravity of the section.
Here it divides the area of the section into two equal parts.
Having determined the location of the neutral axis, we see that
[M] =
ky
where and S„ are the static moments of the upper and lower halves
of the area of the section about the neutral axis. The strength condition
takes the form
(Si + Si) > (26.5)

The above discussion is valid for pure bending; the presence of

shearing forces complicates the analysis.
§151. Design of Statically Indeterminate Beams for
Permissible Loads. The Fundamentals.
Analysis of a Two-span Beam
It was established in the preceding section that the formation of a
ductile hinge is necessary to cause breakdown of a statically determi­
nate beam.
Ch. 26] Design for Permissible Loads 469

Instatically indeterminatebeams the formationof oneductile hinge

is not enoughforfull utilizationof theirbendingcapacity: il isessen­
tial that at least one more ductile hinge be formed. We shall explain
this with the help of an example.

Fig. 377

Let us consider a two-span continuous beamof uniformsection

(Fig. 377(a)). Its bending moment diagramforworkwithin theelastic
limits (Fig. 377(/>)) is the difference of the bending moment diagrams
for forceP andsupport moment M \——t^PI- Graphic subtraction ol
the diagrams isshownbydotted lines. The resultant bending moment
diagramis hatched. The maximumstressedsections arethesectionol
application of the load with a moment 1 4 r-APl lH l and
1)4 = \^P
the middle support withmoment Af,~ |.-j^P/1< P/. When the load
is increased, stresses in the beambecome equal to the yieldstress a,,
first of all in the topand bottomlayers of thesection under loadP„
and may be expressed by the relation
l.s/y I»4U«t//
U41F o„, wherefrom P, 13/
470 Resistance Under Compound Loading JPart V II

If the load is further increased, a ductile hinge is formed in this

section when the bending moment (§ 150) becomes equal to:
Mcu— 2o'1,Sinax= ayriW
However, under such a load the beam does not exhaust its maximum
lifting capacity. It transforms into a statically determinate beam with
a hinge at point D through which moment is transmitted

Fig. 378

(Fig. 378)—this beam is still capable of taking more load. When the
load is further increased, the moment at point D remains constant
whereas the moment at the support increases until it also becomes equal
to M y another ductile hinge is formed at the support, the left span
transforms into a movable system and full lifting capacity of the beam
is utilized as the load increases to P%. The bending moment diagram of
the beam for this state of loading, which is shown in Fig. 377(e), is the
diagram for breaking moments. It is the difference of the ordinates of
pc,
triangle adb with maximum ordinate in the section of load appli­
cation and triangle, aek with an ordinate at the support 1 and ^
in the section of load application. The breaking load /* is determined
P cd Mc
from the condition that segment cd equal to ----- must also be
equal to M y

4 2
wherefrom
= (26. 6)

The strength condition may be written as follows (§ 147):

P < [P ] (26.7)
where P is the load acting on the beam, and [PI is the permissible load.
To obtain [P] we divide both sides of equation (26.6) by the safety
factor k and get

1 J k ~ ki i r<§.8)
Cfi. 26\ Deziga for Permissible Loads 471

Substituting this value of [PI in equation (26.7), we obtain

^ t

wlierefrora
Pi _ Mr
W7* On 1<tJ n fo]
(26.9)

PI
where Mr=-g is the reduced bending moment in sections / and D.
Consequently, the beam section in this example may be selected in
pi
accordance with the reduced moment Mr- - ^ and the permissible
stress /dal.
It is evident from the formulas Mr= ^ and that the
ordinates of the diagram of reduced moments (Fig. 377(d)) are pro­
portional to the ordinates of the diagram of breaking moments and are
obtained from them by replacing the breaking load Pi by the actual
load P.
If we design the beam for permissible stress, we should take
= g |p / > j (Fig. 377(6)) as the reduced moment and la] as the per­
missible stress. Hence, designing statically indeterminate beams for
permissible loads has a double advantage—the permissible stress in­
creases as in statically determinate beams, and the diagram of reduced
moments “shrinks”, i.e. its ordinates in corresponding sections become
smaller.
The increase in the lifting capacity of the beam is given by the ratio
Pff X 131 7a
Py - /x G 4 rP j, “ 0 4 n

Assuming that /t=1.16 (for I-beams, $ 150), we find that the lift­
ing capacity of the beam increases by 40% if it is designed according
to the new method.
In the example discussed above the diagram of reduced moments
(Fig. 377(c)) is obtained from the condition that in the two maximum
stressed sections the bending moments are equal. Therefore in the de­
sign of beams of uniform section this method is sometimes referred to
as the method of equal moments.
Keeping this in mind, we can plot the diagram of reduced moments
(for permissible loads) graphically: first we plot the bending moment
diagram of force P for span I of the beam (adb) and then plot diagram
aek of support moments such that be—cd (this can be done by dividing
472 Reststance Under Compound Loading IPart VII

If another force P were acting at the middle of the second span,

the diagram of reduced moments would remain unaffected (Fig. 379),
i.e. no reinforcement of the beam would be required even if an addi­

tional force were applied. There would only be a change in the order of
appearance of the ductile hinges: a hinge would first be formed at the
middle support and then in the sections where the two forces act.

§ 152. Analysis of a Three-span Beam

Let us now consider a beam having an additional span in the middle
(Fig. 380(a)). The bending moment diagram of this beam for work
within the elastic region is shown in Fig. 380(b). When the load is gra­
dually increased, ductile hinges are first formed at the intermediate
support I and the centre (approximately) of the middle span (Fig.
380(c)). However, the beam is still capable of taking further loads un­
til a third hinge forms at support 2. The final diagram of breaking
moments is shown in Fig. 380(d). The limiting value of the moment is:

u 2 8 16

and the reduced moment is:

The required section modulus is:

r —
^ Tpjf ” 16* 1<t]
Thus, three ductile hinges must be formed before the middle span
fails, and the moments in all the three hinges wilt be equal. If all the
spans of the beam were loaded, it would be essential to check the pos­
sibility of failure for each span by plotting the corresponding diagrams
of reduced moments (by equating their values in the critical sections)
CA. 26] Design for Permissible Loads 473

and select W for the maximum value of Mr thus obtained. As a con­

crete example let us load the three-span beam (Fig. 381) by a uniform­
ly distributed force q in the middle span and concentrated forces P=2ql

Fig. 380

acting at the centre of each side span. The reduced moments Mr for
all three spans are shown in the diagram with dotted lines; their values
are as follows:
M, = £ , = = g

The section should be selected for the moment Mr= y ; the section
modulus should be d is ­
similarly, we can analyze a beam with any number of spans by as­
sessing the possibility of failure in each span separately.
The method of designing continuous beams explained above employs
a number of approximations and restrictions. Firstly, it is valid only
for static loads. Secondly, the physical picture of failure of the beam
is much more complicated even for static loads than the highly sim-
plihed concept of ductile hinge formation, which we employed in the
above discussion. Plastic deformation is not limited to a particular
section but covers the whole beam length. In addition, the maximum
lifting capacity of the beam can be restricted not only by its plastic
deformation, but also by the violation of its stability as a whole or the
474 Resistance Under Compound. Loading IPart V ll

flange plates and web separately. Therefore, if this method is applied

in actual design, greater attention should be paid to the stability of
the beam even though the loading may be static.

Experiments on the failure of statically indeterminate beams under

static loading reveal that, if failure due to lack of stability is prevent­
ed, the breaking load calculated theoretically by the above method
concurs well with the experimentally determined value.

§ 153. Fundamentals of Design by the Method

of Limiting States
Design by the method of limiting states is outside the scope of the
basic course on strength of materials. It is compulsory only in design
of building structures and is not yet used in mechanical engineering.
However, keeping in mind that the method is ultimately based upon
strength of materials, we give below the fundamental concepts of de­
sign of building structures by the method of limiting states to enable
those studying this method to coordinate its methods and terminology
with that of the strength of materials.
Two groups of limiting states are considered below. The first group
deals with limiting states that appear due to loss of load carrying ca­
pacity or because the structure is not fit for working. The second group
deals with those that appear because the structure is not fit for normal
functioning.
The main limiting states of the first group are failure, loss of stabi­
lity, excessive opening of cracks, etc.
The limiting states of the second group include conditions that hin-
Ch. 26l Design for Permissible Loads 475

der normal functioning of the structure or reduce its service life due
mainly to impermissibly large displacements (deflections, angles of
rotation, etc.).
The rated, strength, Rr, which is specified by design standards on the
basis of control conditions and statistical variation of strength, is
the main parameter characterizing the resistance of materials to the
action of external forces. We may choose as rated strength the yield
stress, ultimate strength, fatigue strength, critical stress and other
similar characteristics of the material, which in the course of strength
of materials are called critical and are denoted by pCT.
The possible harmful deviation of strength characteristics from the
rated strength is taken into account by the safely factor of material, kt
by which the rated strength is divided.
The numerical value of k depends on the properties of materials and
their statistical variation. When calculating the load carrying capa­
city of structures, the value of k is not taken less than 1.1.
By the design strength of material, R, we mean the strength which is
considered while designing a structure and is obtained by dividing
Rr by k :
(26.10)

The coefficient of operating conditions, m, takes care of the special

features of a systematic nature that arise during actual functioning
of materials, elements and joints and structures, but which are not
directly reflected in the design procedure.
Coefficient m takes into account the effect of temperature, humidity
and corroding effect of the atmosphere, length of time during which
forces act and some other factors.
The reliability and capital investment factors in design of buildings
and structures are accounted for in a number of cases by the coefficieik
of reliability, kr The numerical values of coefficients k, m, and kr are
established by standards.
Thus, formula (26.10) for design strength R with suitable coeffi­
cients m and kr when necessary may be written as follows:
(26.11)

In the courses of strength of materials design strength R is known as

permissible stress and is denoted by [oj or [t 1. However, R contains
a more detailed break-up of the safety factors and does not provide
for safety against overloading. When a structure is designed by the
method of limiting states, the safety factor against overloading is cal­
culated by a special method, which will be discussed later.
The loads are defined mainly by their rated values, denoted by Pr.
These are specified by standards for various structures.
476 Resistance Under Compound Loading IPart V!1

The probable harmful deviation of loads from the rated values due
to variation of loads or changes in conditions of normal functioning is
taken account of by the coefficient of overloading, n. Coefficient n is
the safety factor against overloading. The overloading coefficients may
be different for different loads even if the latter are applied to the
structure simultaneously, for instance, for permanent and temporary
loads. This is the difference between this design method and the one in
strength of materials in which the safety factor against overloading is
the same for all loads simultaneously acting on a structure and is taken
into account by a common safety factor.
The loads used in actual design are obtained by multiplying the
rated values with the corresponding overloading coefficients «, and
are known as design loads.
If the different values of the overloading coefficient for permanent
and temporary loads are taken into account, then, for instance, the
design bending moment M due to simultaneous action of permanent
and temporary forces on a beam may be deli ned as
M = /tjAf^erm + «4/Wtcinp (26.12)
Structures should be designed by considering the possible unfavour­
able combinations of loads (for example, simultaneous loading of a
bridge by a train, breaking forces and wind, or simultaneous loading
of power transmission line towers by wind and one-sided tension due
to snapping of wire in the adjacent span). The probability of such com­
binations is taken into account by the coefficient of combination, ne-
The values of coefficients n and nc and. the recommendations regarding
their application are available in standards.
Thus, if the standards require that a coefficient of combination for
temporary loads be included in the design of a structure, formula
(26.12) for the design moment becomes
M = rt, Alperu, -J- n.entAJ[<anp (26.13)
If the standards do not contain instructions for the accounting
of inelastic deformations, it is permitted to determine forces in sta­
tically indeterminate systems on the assumption that deformations
of the structure remain in the elastic region. The strength condition,
for instance, in bending, can then be written as follow's:

Taking into account formulas (26.11) and (26.13), we get

pe«n-{- flgKtAffetup ^ Rr
IP' ^ kfh
If the strength is to be checked for limiting states of the second
group, we must determine the elastic deformation or displacement
(elongation, twist angle, deflection) due to normal load. The defor­
mation thus found must not exceed the permissible value aid down
in the standards.
 PART VIII
Stability
of Elements of Structures

C H A PT E R 2 7

Stability of Bars Under Compression

§ 154. Introduction. Fundamentals of Stability
of Shape of Compressed Bars
In all previous discussions we determined the cross-sectional dimen­
sions of bars from their strength condition. However, a bar may fail
not because of insufficient strength but because it docs not retain its
designed shape. This changes the nature of stressed stale of the bar.
The most typical example is that of a bar compressed by axial forces
P. Until now we have checked the strength of bars by the following
condition:
° = - X < ta J. where =^ or
This condition is based on the assumption that the bar works under
axial compression right up to the moment of its failure due to atJ or au.
However, even the simplest of experiments shows that it is not always
possible to load the bar up to its yield stress or ultimate strength.
If we subject a thin wooden scale to axial compression, it may fail
due to bending. At the time of failure the compressive force acting on
the scale will be considerably less than the force which the scale can
withstand before its ultimate strength is reached. The scale fails be­
cause it does not retain its designed shape of a rectangle but bends, giv­
ing rise to bending moments due to forces P and, consequently, to ad­
ditional bending stress: the scale loses its stability.
Therefore, for safe working of a structure it is not enough for it to
have sufficient strength; it is essential that all its elements are stable
and their deformation under the action of external forces is within such
limits that the nature of their work remains unaffected. Hence, in a
number of cases, in particular in bars under compression, the strength
check must be substantiated by a check of stability. Before we carry
out such a check it is necessary to get closely acquainted with the
conditions which lead to the bar losing its stability.
478 Stability of Elements of Structures [Part VU I

Let us consider a bar sufficiently longer than its cross-sectional size,

hinged at both ends (Fig. 382) and loaded by a gradually increasing
axial force P. We notice that the bar remains straight as long as P is
small. If we try to bend it to one side by applying a momentary hori­
zontal force, it comes back to its original shape upon the removal of
the external force causing deflection after doing a few oscillations.

As force P increases, the bar takes longer to return to its original

stable position; finally force P may attain a value at which the bar
fails to straighten when it is slightly bent to one side. If we try to
straighten the bar without removing force P, we find that it is inca­
pable of remaining straight. In other words, at a particular value of
P, called critical force Pvt the straight-line shape ceases to be.stable for
a bar under compression.*
The transition to critical value is sudden; a small decrease of the
compressive force from the critical value is enough to make the
straight-line shape stable again.
On the other hand, if the compressive force P is slightly higher than
the critical value, the straight-line shape becomes extremely unstable.
In this case a small eccentricity of the applied force or the non-unifor­
mity of the bar material is enough not only to bend the bar but also to
increase the curvature of bent bar due to a continuously increasing
bending moment; the process of bending comes to a stop either when a
new position of equilibrium is achieved or when the bar completely
breaks down.
Thus, for all practical purposes we can consider critical force Pc
equivalent to a load which “cripples” the compressed bar and creates
a condition in which normal working of the bar becomes impossible.

• Investigations reveal that instability begins at values of P which exceed the

critical force only by a second-order quantity.
Ch. 27] Stability of Bars Under Compression 479

It must be kept in mind that “failure” of the bar due to a force greater
than the critical occurs only when there is no obstacle to bending.
Therefore failure may be avoided if buckling is prevented by a side
support which restricts further bending.
Usually such a possibility is remote—the critical compressive force
should practically be considered as the lower limit that causes “fail­
ure” of the bar.
The loss or stability under compression may be explained by the
analogy following from mechanics of solid bodies (Fig. 383). Let us
roll a cylinder on an inclined plane ab which changes into a small
horizontal platform be and then an inclined plane cd of opposite in­
clination. The cylinder remains stable as long as we lift it along plane
ob holding it with a support perpendicular to the inclined plane. Equi­
librium of the cylinder is immaterial when it rolls on platform be.
As soon as it reaches point c its equilibrium becomes unstable—the
slightest push to the right is enough to make it go rolling down.
The physical picture of loss of stability in the compressed bar de­
scribed above can be actually reproduced in any laboratory with very
elementary equipment *. This description is not a theoretical or ideal
picture o f working of a compressed bar, but a real one showing how
actually a bar works when it is acted upon by compressive forces.
The loss of straight-line stable state by a compressed bar is some­
times referred to as axial bending, because it manifests itself in the form
of considerable bending of the bar under axial compression. Therefore,
instead of check on stability the term “check on axial bending” is still
quite prevalent, although it is not very appropriate, because we are
basically interested not in the check on bending but in the check on
stability of the straight-line shape of the bar.
Having established the concept of critical force as a “crippling” load
which puts an end to the normal functioning of the bar, we can easily
derive a condition for checking the stability of bars identical to the
strength condition.
Critical force P e gives rise to “critical stresses” in the bar, which are
denoted by o<,= . Critical stresses in the compressed bar are stresses
at which the bar fails. Therefore, to ensure stability of a straight bar
compressed by forces P, it is essential that the strength check (o=
•=£-<[al) should be accompanied by a stability check

< r= 4 < [a j (27.1)

• See N. M. Belyaev, Laboratory Experiments on Strength of Maieriats, Gos-

lekhtzdat, Moscow. 1951 (111 Russian), §85.
460 Stability of Elements of Structures fPart VIII

where fo*1 is the permissible stress for stability equal to critical stress
divided by a factor of stability:

Before we explain how stabil ity is to be checked, we must show how

to determine oc and how to select k a.

§ 155. Euler’s Formula for Critical Force

For determining critical stress oc we must first calculate the critical
force, P ef i.e. the minimum axial compressive force which a slightly
bent compressed bar can withstand and yet remain in equilibrium.
This problem was first solved by Leonhard Euler in 1774.
Let us point out that the problem is entirely different from ail pro*
blems discussed in previous sections of this book. Until now we have de­
termined deformation of a bar when the external forces acting on it

Fig. 384

are known. Our problem is just the opposite: we assume a certain de­
flection of the bar’s axis and then determine the axial compressive
force P at which the assumed deflection occurs.
Consider a uniform straight bar hinged at both ends. One of the sup­
ports permits axial displacement of the corresponding end of the bar
(Fig. 384). The dead weight of the bar is negligible. Let us load the
bar with an axial compressive force P = P C and impart to it a slight
deflection in the plane of minimum rigidity; the bar remains in equi­
librium in the bent state because P —P c.
The imparted deflection is assumed to be very small. Therefore, the
problem may be solved by using the approximate differential equation
of the deflected axis of the bar (§ 82). Selecting the origin of coordinates
at point A and directing the coordinate axes as shown in Fig. 384, we
get from equation (15.7)

E l S = M (4

Consider a section at a distance x from the origin of coordinates. The

ordinate of the deflected axis in this section is y and the bending mo-
Ck. 27] Stability of Bars Under Compression 481

merit is:
M{x) = - P y
When compared with Fig. 384 the bending moment is found to be
negative although the ordinates are positive for the selected direction
of the y-axis *. (Had the bar been bent with its convexity downwards,
the moment would have been positive, the ^/-ordinate would have
been negative and M (x)——Py.)
Consequently, differential equation (15.7) may be written as fol­
lows:
E J ^^-P y (27.2)
P
Dividing both sides of the equation by EJ and denoting by k~, we
may rewrite the above equation as follows:

a J + tfv -o (27.3)
The general solution of this equation is:
y = a sin k x -r bcoskx (27.4)
This solution contains three unknowns:constants of integration a
and 6 and k-= because the critical force is not known.
The boundary conditions of the bar give us two equations:
at point A: x = 0 deflection y = 0
at point B: x — t deflection y — 0
It ensues from the first condition (since sin kx—0 and cos £.v=l) that
0= 6
Thus, the bent axis is a sine curve having equation
i/= a s in kx (27.5)
Upon substituting the second equation
y = 0 and x = l
we obtain
0 — a^\nkt (27.6)
It follows that either a or kl must be zero.
If a is equal to zero, then from equation (27.5) we see that the de­
flection of the bar is zero in all sections, i.e. the bar remains straight.
• If the ^-axis is directed downwards, a positive bending moment /W(x) will
correspond to a positive deflection y. However, in this case the curvature will be
negative and <Pyldxl<$. Hence, the signs in equation (27.2) will remain the same.
482 Stability of Elements of Structures [Part V l f f

This contradicts the assumptions made at the very beginning of this

derivation. Hence, sin k l = 0 , and k l may have the following values
/W= 0 , n , 2 n , 3 j i , .. ,,/m (27.7)
where n is an arbitrary integer.
This yields and since k —

-E T = F n' or (27.8)
In other words, a load capable of holding a slightly deflected bar in
equilibrium can, in theory, have a number of values. However, as we
are interested in determining the
minimum axial compressive force at
which axial bending may occur, we
must take
According to the first root, «=0,
the critical force Pr must be zero,
but this contradicts the given data.
Therefore this root is ruled out and
the minimum value of n is taken as
n=*l. This gives
(27.9)

(Here J is the minimum moment of

inertia of the bar cross section.)
Fig. 385 This is known as Euler’s formula
for compressed bars hinged at the
ends. This value of the critical I >ree (27.9) corresponds to bending of
the bar along a sine curve with ane half-wave (formula (27.5)):
y — asin — (27.10)

Higher values of the critical force correspond to bending the bar

along a sine curve with two, three, etc. half-waves (Fig. 385):
4jz*EJ 2nx
Pc = £/=asin T
D 9n*EJ
(27.11)
“c J2 » 0 = asin -2yi

Hence, the greater the number of inflection points in the sine curve
of the deflected axis of the bar, the greater the critical force must be.
Intensive investigations show that the equilibrium modes determined
by formula (27.11) are not stable; stable equilibrium modes can be
CA. 27] Stability of Bars Under Cotnpression 483

achieved only if we place Intermediate supports at points B and C

(Fig. 385).
Thus, we have solved the problem which we had set before our­
selves: the critical force for our bar is calculated by the formula
n*EJ
Pc n r
and the deflected axis is represented by the sine curve
. Jtx
y — a sm —

The constant of integration, a, has remained undetermined; its phys­

ical nature will become clear if we substitute x =■£- in the equation of
the sine curve. The condition yx=i/t (deflection at the middle of the
bar) yields
ym»x = f = a
This means that a is the deflection of the bar at its middle point.
Deflection / remains undetermined, because the deflected bar can re­
main in equilibrium in various deflected positions from the straight-
line shape for a single value oi critical force P, provided the deflections
are small.
Deflection f should be small enough to enable us to use the approxi­
mate differential equation of the deflected axis, i.e. should be
negligible as compared to unity (§ 82).
Having found the critical force, we can immediately determine
critical stress by dividing P 0 by the cross-sectional area, A. As the
critical force was determined by considering the deformation of the
bar which is not much affected by local weakenings in the section, the
moment of inertia used in the expression for P c is Thus
while calculating critical stress or writing the condition of stability
we must consider not the weakened but the total area A t of unweakened
section. Then
Pc n 2E Jt n s£ / 2 _ :
° c ~ At ~ I*At ~ ~ T ' ~ < kl/i) * '~ ' K*
(27.12)

We find that critical stress in a bar of a given material is inversely

proportional to the square of the ratio of its length to the minimum
radius of gyration of its section. This ratio, is called the flexi­
bility oi the bar and plays an important part in all stability checks of
compressed bars.
It is evident from formula (27.12) that the critical stress may be ex­
tremely small in long and thin bars; it may be less than the permissi-
 m Stability of Elements of Structures [Part VIU

ble stress fol. For steel with ultimate strength of 4000 kgf/cm4 the per­
missible stress may be assumed equal to )<rJ=1600 kgf/cm4; for a bar
having flexibility and modulus of elasticity of the material
£=2x10* kgf.'cm* the critical stress will be
=877 kgf/cm2 < 1600 kgf/cm

Had the cross-sectional area of the compressed bar of given flexibi­

lity been determined from the strength condition, the bar would have
failed due to loss of stability of its straight-line shape.

§ 156. Effect of Constraining the Bar Ends

Euler’s formula was obtained by integrating the approximate differ­
ential equation of the deflected axis of the bar with ends constrained
by a particular method (hinged). This implies that the expression fos
determining the critical force is valid only for bars with hinged endr,

P
F ig 38G F ig 3 8 7 F ig . 388

and changes when the method of constraint is different. If the ends of

the bar are hinged this type of fixation will be referred to as the basic
method of constraint. All other methods of constraint will be discussed
by comparing them to the basic method.
If we repeat the derivation for a bar rigidly fixed at one end and
loaded at the other by an axial compressive force P (Fig. 386), we shall
obtain a different expression for the critical force and, consequently,
for the critical stress. Leaving it to the reader to derive such an expres­
sion himself, we proceed to explain how the expression for the critical
force can be obtained in this case with the help of the following simple
considerations.
Ou 27] Stability of Bars Under Compression 485

Suppose the bar retains its equilibrium under critical force P when
it bulges slightly along the curve AB. Comparing Figs. 382 and 386,
we observe that the deflected axis of the bar rigidly fixed at one end is
in the same conditions as the upper half of a bar of double length
hinged at both ends.
This means that the critical force in a bar of length I which is fixed
at one end and free at the other is the same as in a bar of length 2 1
hinged at both ends:
p *ZEJ n%EJ (27.13)

If we consider a bar in which both ends are rigidly fixed and are in­
capable of rotation (Fig. 387), we observe that the middle portion of
length of the deflected bar works under the same conditions as a bar
which is hinged at both ends (points of inflection C and D may be con­
sidered as hinges, because the bending moment at these points is zero).
Thus, the critical force in a bar of length I which is fixed at both
ends is equal to the critical force in a bar of length in which the
ends are fixed by the basic method of constraint:
4n*£J
/> = (27.14)

Formulas (27.13) and (27.14) may be combined with »l then » formula

for critical force in the basic method of constraint P c= and the
generalized formula may be written as follows:
n*EJ
Pe = W (27.15)

Here p, is the coefficient of length with the following values:

when both ends are hinged (basic case) p= l
when one end is free and the other
rigidly fixed p= 2
when both ends are rigidly fixed |ut= 1/2
For the bar shown in Fig. 388 which is rigidly fixed at one end and
hinged at the other, coefficient p is approximately found to be equal to
-4 = « 0 .7 f and the critical force is:
yT
486 Stability of Elements of Structures [Part V I I I

The product pI is called the reduced (free) length; with the help of
the coefficient of length a bar with arbitrarily constrained ends can be
reduced to a bar in which the ends are constrained by the basic method:
while calculating flexibility the actual length of the bar must be re­
placed by the reduced length nl. The concept of reduced length was first

Fig. 389

introduced by Prof. F. Yasinskii * of the St. Petersburg Institute of

Railway Engineers.
Formula f^.I2) for critical stresses in bars with hinged ends may be
generalized for other types of constraints by introducing in the denomi­
nator the reduced flexibility
and (27.12')

The values of coefficient p for some types of constraints are given in

Fig. 389.
in practice, however, we rarely find constraints exactly in the form
as they have been considered here (Fig. 389). Cylindrical hinges are
generally used instead of hinged supports. Such bars may be consid­
ered as simply hinged if they buckle in a plane perpendicular to the
axis of the hinges. If, however, the bar bends in the plane of axes,
then the ends should be considered rigidly fixed (with the reserva­
tions, discussed below for rigidly fixed ends).
In structures we often find compressed bars, which are riveted or
welded at the ends to other elements quite often with the help of cover
plates. Such a constraint cannot, however, be considered rigid, because
the elements to which the compressed bars are secured are themselves
not absolutely rigid. Incidentally, even a slight rotation of the fixed
end in the fixation is enough to render it more close to a hinged con-

• Proceedings of the Conference of Raikwy Engineers, St. Petersburg, 1892 (in

Russian).
Ch. 27] Stability of Bars Under Compression 487

00
«/
JO
(0
H
Values of Critical Force
488 Stability of Elements of Structures [Part VI11

straint rather than to a rigidly fixed one. Therefore, it is inadmissible

to consider such a bar as one with rigidly fixed ends. Only when we are
quite sure about the reliability of the fixation can a small (about 10-
20%) decrease of the free length of the bar be permitted.
Finally, we have bars which rest on adjoining elements with the
whole area of their end faces. Such bars include wooden posts, indepen­
dently standing metal columns which are secured to the foundation by
bolts, etc. If the end face is properly designed and secured to the foun­
dation, such bars may be considered rigidly fixed at the end. This
group of bars also includes large columns with cylindrical hinges
when they are designed for buckling in the plane of the hinge axis.
Generally, it is difficult to ensure uniform contact between the end
face of the compressed column and the supporting foundation. There­
fore, the load carrying capacity of such columns is only marginally
greater than that of columns with hinged ends.
The formula for critical loads may be obtained in a form close to
that of Euler’s formula (27.15) even for bars of non-uniform sections
and bars being acted upon by several forces. Derivations for a few cases
of practical interest, which have been obtained by the theory of elasti­
city, are given in Table 18.

§ 157. Limits of Applicability of Euler’s Formula.

Plotting of the Diagram of Total Critical Stresses
It would seem that the results obtained in the preceding section were
enough to check the stability of compressed bars; the coefficient of
stability, k 5, remains to be determined. This, however, is far from
true. The very first study of numerical values obtained by Euler’s
formula confirms that the formula gives proper results only within cer­
tain limits.
For example, if we calculate the critical stress according to formula
(27.12) for a steel bar (E—2 X 10* kgf/cms) of flexibility X=50, we
obtain
_ JI* £ 3>14*x2xl0" ______________ _
®e=*^r = ----- go?----- -- 8000 kgt/cm2

This is almost twice the ultimate strength of steel; the bar will
cease to work even before the critical stress is achieved. We thus see
that for low flexibility bars Euler’s formula gives exaggerated values
of critical stresses and forces. What are the reasons for this? Figure
390 shows the relation between ac and X. The curve is a hyperbola,
which is known as “Euler's hyperbola”. While using this curve, it
should be kept in mind that formula (27.12), which it represents, was
obtained by integrating the differential equation oi the deflected axis,
i.e. it was derived on the assumption that the stresses in the bar are
less than the limit of proportionality when it loses its stability.
Ch. 27] Stability of liars Under Compression 489

Consequently, we cannot use the critical stresses calculated by Eul­

er’s formula if' they exceed the limit of proportionality of the given
material. In other words, Euler’s formula is valid only when it satis­
fies the following conditions:
o,s^arp or (27.17)

If we express X through equation (27.17), the limit of applicability

of Euler’s formula will change:
} /iS (27.17')

By substituting the values of modulus of elasticity and limit of

proportionality of the given material, we can find the minimum flexi­

bility at which Euler’s formula can still be applied. For steel the limit
of proportionality is a«=2000 kgf/cma; therefore, as is evident from
formula (27.17'), Eulers formula can be used for bars of this material
only when ______

h > Y ^ ! r a m

i.e. when is greater than 100.

For another type of steel, o„«3000 kgf/cm* and Euler’s formula is
applicable when X>85; Euler’s formula is applicable for cast iron
when X>80, for pine wood when X>110, etc. If we draw a horizontal
line in Fig. 390 with an ordinate orp=2000 kgf/cma, it will cut Euler’s
hyperbola into two parts; only the lower portion of the diagram, which
is valid for thin and long bars, can be used, because such bars become
unstable at stresses less than the limit of proportionality.
490 Stability of Elements of Structures [Port V W

The theoretical solution obtained by Euler can be applied to a very

limited category of bars, namely thin and long bars having high flexi­
bility. In structures we often find bars having low flexibility. Attempts
to apply Euler’s formula for calculating critical stresses in such bars
led to catastrophic results, and experiments on compression of bars
aiso show that, if the critical stresses exceed the limit of proportiona­
lity, the actual critical force is considerably less than the value ob­
tained by Euler's formula.
Let us note that for axial compression of the bar is accompa­
nied not only by elastic but also by plastic deformation; additional
bending stresses appear at the instant when the bar loses its stability
(when the bar axis becomes curved). When the load is removed, the
bar fails to straighten, as it does when compressed within the elastic
limits.
Keeping in mind all these factors, we see that it is necessary to find
methods for calculating critical stresses in those cases when they ex­
ceed the limit of proportionality, i.e. for bars which have a flexibility
less than that determined by equation (27.17'), for example, for low
carbon steel bars with flexibility A,=*0 to A.= 100.
A. A theoretical solution of the problem of stability of compressed
bar subjected to critical stresses exceeding the limit of proportionality
of the bar material was first attempted by F. Engesser (1889), who
obtained the following formula Identical to Euler’s, (27.12):

(27.18)

Here Ex is a variable modulus of elasticity, which is determined as the

tangent of the angle of slope which the tangent at a point beyond the
proportionality limit makes with the compression test diagram
(Fig. 391(g)) The langenlial modulus E i~ depends upon the. type of
diagram as well as on the magnitude of critical stress ac at the instant
when the bar loses stability.
Tt was Yasinskii who pointed out that formula (27.18) was incorrect
as it did not take into account the fact that when a bar lost stability
and its axis became curved it experienced additional stresses not only
of compressive but also of tensile nature. Conceding to the validity of
the critical remarks of Yasinkii, who pointed out the necessity of ac­
counting for load relaxation on the convex side of a bent bar, Engesser
(1895) and independently Th. Karman (1909) came to the conclusion
that in formula (27.18) the tangential modulus should be replaced by
reduced modulus £*, which took into account load addition on the con­
cave side with modulus E\ and load relaxation on the convex side with
modulus E (load relaxation, as is well known, follows Hooke’s law).
The formula for critical stresses exceeding the limit of proportionality
Ck. 27) Stability of Bars Under Compression 491

(for bars having flexibility \ > ^ p) may be rewritten as

n3E» <27.18')
Cc = x3

The reduced modulus of elasticity may be calculated from the fol­

lowing expression derived from the conditions of equilibrium of addi*
S e ttitn m - n

(b)

tional stresses (load addition—load relaxation) and on the basis of

the hypothesis of plane sections:
£ * = £|/ l + £/a (27.19)
J

where is the moment of inertia of the concave half of the section

about the neutral axis, J z is the moment of inertia of the convex half
of the section and J is the moment of inertia of the whole section about
the central axis.
On account of the difference between moduli E and £i in expression
(27.19), the neutral axis dividing the concave half of the section
from the convex half does not pass through the centre of gravity of the
section (Fig. 391(b); its location (and this means that the areas that
experience additional loading and load relaxation) depends upon the
shape of the section as well as upon the critical stress, which we want
to determine. For different sections, assuming a particular value of
Ei and using the method of successive approximations, wc can find the
location of the neutral axis and calculate the moments of inertia Ji
and and then the reduced modulus of elasticity E* as a function of
 492 Stability of Elements of Structures [Part V U i

moduli E and £ t. Foi a rectangular section, lor instance,

A ll E j
(27.19')
{V E + V E IY
It is evident from equations (27.19) and (27.19') that when the criti­
cal stress does not exceed the limit of proportionality of the material
(i.e. when the deformation is within the elastic limits and £ i= £ ),
the reduced modulus of elasticity £ * = £ .
For materials having a well defined yield plateau Et and E* tend to
zero as the critical stress approaches the yield stress. This implies that
in such cases the critical stresses cannot exceed the yield stress of the
material.
The Engesser-Karman formula did not find application in practical
design since the determination of E*, which depends upon the critical
stress, raises serious difficulties in computation and also because it
gives exaggerated values of critical stresses as compared to experimen­
tal results.
The application of Euler's formula to the inelastic region of defor­
mation became possible only after F. Shenly published his work in
which he suggested a new approach to the analysis of stability of com­
pressed bars (1946) subjected to elastico-plastic deformation. Looking
upon the loss of stability as a dynamic process under the action of a
continuously increasing compressive force, Shenly, in fact, returned
to the original formula of Engesser (27.18) with tangential modulus of
elasticity £ t, which had earlier been rejected (if the curvature is small
at the moment when the bar loses its stability, an increase of force P
by AP balances the load relaxation on the convex side due to addi­
tional compression).
The transition to generalized formula (27.18) considerably simpli­
fied the calculation of critical stresses for bars in which loss of stability
is accompanied by plastic deformation. At present, theoretical values
of critical stresses for low and medium flexibility of different materi­
als have been calculated on the basis of experimental data on E\ cor­
responding to different values of critical stresses ar greater than op,
and with the help of modern computational techniques. These values
are in good agreement with the results obtained from experimental
research.
B. The first experimental investigations on stability of compressed
bars were conducted with the aim of checking Euler's formula. It
was found perfectly valid for long (flexible) bars but was observed to
give a large discrepancy with experimental results for short bars (as
is evident from theoretical considerations). On the basis of these expe­
riments, often not conducted with due care, various empirical formulas
were proposed for determining critical stresses, though without suf­
ficient justification in most of the cases. However, the quality of ex­
perimental investigations improved as new apparatus was developed.
Ch. 27] Stability of Bars Under Compression 493

Extensive experiments that covered a wide range of materials and

were distinguished by an extreme thoroughness were conducted in
1896 by L. Tetmajer. The results of these experiments were processed
by Yasinskii who compiled a table of “breaking” (critical) stresses de­
pending upon flexibility for bars from commonly used structural mate­
rials.
Alt experiments show that short bars having stability X=30-40 lost
their load carrying capacity not due to insufficient strength but be­
cause of the compressive stresses rising to a dangerous value a* which
was critical for the given material (recall that a d=o„ for ductile state
of material and <xd= o u for brittle state). It may therefore be assumed
that for bars with low flexibility critical stresses for ductile materials
are practically equal to the yield stress av and for brittle materials to
the ultimate strength oru.
For medium flexibility bars, which find maximum application in
structures, it was experimentally established that they lose their load

carrying capacity due to loss of stability of straight-line form under

stress ae greater than the limit of proportionality ap but less than <r<j.
In such bars, the variation of critical stresses as a function of flexi­
bility follows an almost linear law. For instance, the empirical
formula of Tetinajer-Yasinskii is:
oc= a — b \ (27.20)
where a and b are coefficients that depend upon the material and are
selected such that for flexibility ?v=A,1jni, ac—ap and for low flexi­
bility c r is close to
This data is used for plotting the total critical stress diagram shown
in Fig. 392 for low-carbon steel having a limit of proportionality
op=2000 kgf/cin* and yield stress <Tl/=2400 kgf/cm4. The diagram con­
sists of three parts: Euler’s hyperbola when 100 (on the right),
horizontal line for a^ 40 when a c« o „ (on the left) and an inclined
494 Stability of Elements of Structures [Part VIU

straight line (27.20) when 40<X<!00 (joining points m and n). In

o^der to avoid sharp inflections in the <xc versus %curve at points m
and nt we may use the empirical formula due to Johnson which recom­
mends a parabolic variation of critical stress in the inelastic zone
oc= o d— a\* (27.20')
Here, when X=0 then <Jc= odt and when then o ^ o P; more­
over, coefficient a is selected such so as to ensure smooth conjunction
of the parabola (27.20') with Euler’s hyperbola (a common tangent).
For example, for structural steel having yield stress or„=2800 kgf/cmS
a<=0.09.
Hence, either by using the general theoretical formula (27.19) or
combining Euler's formula with experimental results, we can plot the
total critical stress diagram for bars of different materials and deter­
mine critical stress oc for any flexi­
bility.

§ 158. The Stability Check of

Compressed Bars
We noted in § 154 that the follow­
ing two checks should be carried
out for bars under compression:
the strength check
a — -j < M* where [<f] = ^

the stability check

*= where [ o ,]= £

Having plotted the total critical stress diagram for bars of any flexi­
bility (§ 157), we can also plot the permissible stress diagram for sta­
bility for the given material by reducing the <vordinates k, times:

We only have to choose a proper value for coefficient k t. Bearing in

mind a number of errors, which are unavoidable in axial compression
(initial curvature, eccentricity, etc.) and seriously influence the load
carrying capacity of the bar, the factor of safety for stability is taken
greater than the safety factor for strength, fa. For steel this coefficient
varies between 1.8 and 3.5, for iron between 5.0 and 5.5 and for timber
between 2.8 and 3.2.
Figure 393 shows the diagram of permissible stresses for stability
Ch. 27\ Stability of Pars Under Compression 495

and the safety factor for stability for low-carbon steel having yield
stress of<=2400 kgf/cm9.
In order to establish a relation between the permissible stress for
stability* [a J , and permissible stress for strength, (oj, let us lake their
ratio:
[o£l D a£»o

Denoting
©- (27.21)
1!
we get
[a,] = (F[or] (27.22)
where <p is the reduction coefficient of the permissible stress for com­
pressed bars.
If we have the versus X curve for a given material, know o 0 —ov
or <7o=aru and select the safety factors for strength, k6l and stability,
kg, we can put together a table for q> as a function of flexibility.
Table 19
Coefficient <p
Structu ral steel?
F le x ib ility S teel Cast
. lU c t jk iron W ood
X= T C 3 8 /2 3 0 4 4 /2 9 C -4 6 /3 3

0 1.00 1.00 1.00 1.00 1.00 1.00

10 0.988 0.987 0.986 0.97 0.97 0.99
20 0.970 0.968 0.965 0.95 0.91 0.97
30 0.943 0.935 0.932 0.91 0.81 0.93
40 0.905 0.892 0.888 0.87 0.69 0.87
50 0.867 0.843 0.837 0.83 0.57 0.80
<50 0.820 0.792 0.780 0.79 0.44 0.71
70 0.770 0.730 0.710 0.72 0.34 0.60
80 0.715 0.660 0.637 0.65 0.26 0.48
90 0.055 0.592 0.563 0.55 0.20 0.38
100 0.582 0.515 0.482 0.43 0.16 0.31
110 0.512 0.440 0.413 0.35 0.25
120 0.448 0.383 0.350\ 0.30 0.22
130 0.397 0.330 0.302 0.26 0.18
140 0.348 0.285 0.256 0.23 0.16
150 0.305 0 250 0.226 0 . 2! 0.14
160 0.19 0.12
170 0.17 0.11
180 0.15 0.10
190 0.14 0.09
200 0.13 0.08
 496 Stability of Elements of Structures [Part VIII

Table 19 contains data on coefficient <p for structural steels recom­

mended by the latest Soviet standards for designing metal structures.
The table also contains the reduction coefficients of main permissible
stresses* for improved quality steels, iron and timber (pine). The de­
sign standards used in construction recommend that coefficient q> for
timber should be calculated by the following formulas:
if flexibility 75, then q>=l —

if flexibility X> 75, then <p= ^

The values of cpobtained from these formulas are quite close to the
tabulated values.
This table helps us to select the cross-sectional area of the com­
pressed bar. The cross-sectional area depends upon fa j, which in its
turn depends upon «p and flexibility X, i.e. upon the area and shape of
the cross section. Therefore, the cross-sectional area is determined by
successive approximations in the following order.
We select the shape of the section and define its dimensions. Next,
we calculate the minimum radius of gyration and the flexibility. We
find coefficient <p from the table and calculate the permissible stress
for stability |o,l=(p|o|. We now compare the actual stress <r=£- with
lo j; if the condition
»= (27.23)

is not satisfied, or is satisfied with a big margin, we change the dimen­

sions and repeat the calculations. Obviously, the section finally se­
lected must also satisfy the strength condition
* < [o ]
In actual calculations the stability condition is sometimes written
as follows:

<27-23')
In the left-hand side or, represents the design (reduced) stress.
The order of calculations will be elaborated on in the following
example.

• The steels accepted In the standards are characterized by ultimate strength

(numerator) and yield stress (denominator) in kgf/imn*. Standards do not permit
the use of steel bars having flexibility X>I50 in structures as load carrying elements
subjected to compression. In design by the method of limiting states (see Chapter 26)
coefficient cp is considered a coefficient by which the rated Toad should be reduced.
Ch. 271 Stability of Bars Under Compression 497

Find the cross-sectional dimensions of an iron pipe column hinged

at both ends and subjected to a compressive force P = 85 If, if the ratio
of internal diameter to the external is diD=0.6. The column is 1=
=4.8 m long. The main permissible stress under compression for iron
is Iol=l200 kgf/cm*. .
Let us express area A of the section and its radius of inertia i
through diameter D:
A = n(D* ~ dl)c=±(Di — 0.36D*) = 0.503D* (a)

In stability condition (27.23) we know neither the cross-sectional

area, A, nor the reduction coefficient for permissible stress, <p. There­
fore it is essential to assume a value for one of these quantities. In the
first approximation let us assume q>=0.5. We get
P 85 000
A> 142 cm*
<pl*J 0.5X1200

In the first approximation the diameter is (a) 142/0.503=

= 19 cm. The radius of inertia is (b) /i=0.291Di=0.29l X 19=5.5 cm.
Flexibility X1= p f/i,= g - =87.5. For flexibility between X=80 and
X=90 we find from the table by interpolation that <p=0.215. The de­
sign stresses from formula (27.23') are
P 85000
= 2800 kgf/cm* > [o]

The section does not satisfy the stability condition. Therefore, in

the second approximation let us assume that diameter £>*=25 cm.
Cross-sectional area J4*=0.503D!=0.503x25*=314 cm*. Radius of
inertia /s=0.291 X25=10.2 cm. Flexibility Xi=//t *=480/10.2=47and
<p=0.654. Thus

°' = { a = = 4 L0 kgf/cm’ « 1 2 0 0 kgf/cm’

When diameter is Df= 19cm the stresses are considerably greater

than the permissible, and when it is £>*=25 cm they are much less.
Let us trv in the third approximation a section of diameter D=22 cm.
In this'case A=0.503X 22*=245 cm*, *=0.291x22=6.4 cm, X=
17-3810
498 Stability of Elements of Structures [Part VIII

*480/6.4=*75, q>=0.30 and the design stresses are:

<*r = p S S , “ 1150 kgf/cma < 1200 kgf/cm*

The calculations may be terminated here as understressing is less

than 5%.

§ 159. Selection of the Type of Section

and Material
A. As the flexibility of Ihe bar and consequently the minimum radius
of gyration plays the most important part in resistance to axial bend­
ing (loss of stability), it is important to take into account not only the
cross-sectional area but also its shape.
The most economic solution of the problem is obtained if the sect ion
of a certain cross-sectional area has the minimum radius of gyration of
the maximum possible magnitude. To achieve this, we try to select a
section in which the minimum and maximum radii of gyration are
equal, i.e. a section in which the moments of inertia about all central
axes are zero and consequently the inertia ellipse transforms into an
inertia circle. Such a bar will have uniform resistance to buckling in
all directions.
If the free length of the bar <§ 156) is different for deflection in the
two principal planes, then the principal moments of inertia should be
such that coefficient <p is the same in both the cases.
It is now essential to obtain the maximum possible central moments
of inertia for the given cross-sectional area. This is possible if the mate­
rial of the section is located as far away from the centre of gravity as
possible. Both these conditions are, for example, fully satisfied by a
pipe section (Fig. 394(a)), which is therefore often used in compressed
bars and columns.
The lower limit of wall thickness of such sections is determined either
by casting limitations (as in cast iron) or by the condition that local
deformation (warping) of the section should not occur when the bar is
working.
To prevent such local deformations and to ensure that the bar re­
tains its designed shape, the pipe-section is reinforced with the help of
plates placed at a particular distance from one another, which increase
the rigidity of the thin-walled section (Fig. 394(b)). As a matter of
fact, proper use of reinforcing plates is extremely important in de­
signing compressed bars.
Some sections which have excellent bending resistance in one plane,
for example beams, are found to be of no use as compressed bars.
Examples are: an I-beam, a section consisting of two channel sections
placed in such a way that their webs touch each other (Fig. 394(c)).
Ch. 27] Stability of Bars Under Compression 499

These sections are disadvantageous when used as compressed bars on

account of the large difference in the values of their p rin cip al moments
of inertia. This drawback may be overcome by moving apart the chan*
nel sections, as shown in Fig. 395(c). To ensure that the sections work
together as a single unit, they are
joined by means of a network (Fig.
395) of fixing plates.
Trouble-free working of such com­
posite sections can be guaranteed
only by providing a sufficiently
strong fixation (by network of plates
and angles), which ensures that
all load carrying elements function
simultaneously. Thus, if we loin
two strong channel sections weakly, 2T
they will not work as a single unit.* fa (b) ( 0

Each half of the sections will work

independently and its stability will
be considerably less than that of a
section in which both halves operate Fig. 394
as a single unit.
Insufficient attention paid to the design of reliable fixation of com­
posite sections has been the cause of serious catastrophes in the past.

In designing a composite section, distance b by which the two halves

of the section should be moved apart is determined from the condition
that the moments of inertia about principal axes y and z be almost

* The analysis and methods of determining the dimensions of network plates

for composite compressed bars are given in the course on metal structures. Also see
N .M . Belyaev, Strength of Materials, Nauka, 1965 (in Russian), §212.
17*
600 Stability of Elements of Structures [Pari VIII

equal. Usually, however, the moment of inertia about the axis per­
pendicular to the network plane is taken a little higher, because the
network cannot ensure simultaneous working of the two halves as well
as of a section which is one rigid unit.
We will show below how the stability of a composite bar subjected
to compression can be improved by rationally placing the elements of

Fig. 396

section. Suppose we have to design a section from two channel sections

(steel C-38/23) for a 4-m long column that is hinged at both ends and
is subjected to compression.
We shall compare the maximum permissible (from the point of view
of stability) compressive force acting on the column made up of two
channel sections No. 30 (see Appendix). Let us consider two versions:
in the first the two channel sections are attached alone their height
back to back so as to form an I-section (Fig. 396(a)); in the second ver­
sion the two channel sections are attached by a network and placed in
such a way that the section has identical moments of inertia about the
two principal axes of inertia (Fig. 396(7;)). The main permissible stress
under compression is assume! to be [cri = 1600 kgf/cm1.
For the first version (channel sections attached without any gap)
the minimum moment of inertia of the section is:
Jm\a = Jz = 2 (327 -|- 40.5 x 2.52s) =1168 cin*

The area of the composite section is A ^2x4Q .5=81 cm4. Radius of

inertia of the section is i = \ rJ !A —]/'{168/81—3.8 cm. Flexibility of
the bar >.=p/7=400 3.8=105.
Interpolating from Table 19 the value of <p between 100 and X—
*=110, we obtain <p=0.547. The maximum compressive force which can
Ch. 27| Stability of Bars Under Compression SOI

be safely applied to the column is:

Pt = [o] A y = 1600 x 81 x 0.547 = 70 800 kgf 71 tf
For the second version (when the channel sections are apart) Ihe mo*
ment of inertia of the composite section (Fig. 396(6)) is:
J v = /* = 2 x 5 8 1 0 = 11 620 cm4
The radius of inertia i= V \ 1620/81 = 12 cm, the flexibility X=//*=
—400/12=33.3 and the coefficient <p=0.931.
The permissible load on the bar may in this version be taken P*=
=[<rM<p=1600 X 81 X0.931 = 120 900 kgf«121 tf, i.e. 1.7 times grea­
ter than in the first. Hence, a rational approach in selecting the shape
of a section enabled us to raise the load carrying capacity of the com­
pressed bar by 70%.
Simultaneous functioning of both halves (channel sections) of the
composite section is possible only if they are reliably secured to each
other by a network or plates (Fig. 395(a) and (6)). Distance a between
the securing elements should ensure that neither of the channel sec­
tions bend in the plane of its minimum flexibility. This condition can
be satisfied only if the flexibility of each half (in this example each
channel section) does not exceed the flexibility of the column over
length a:
K .s lmln
For one channel section the minimum radius of gyration is
=2.84 cm. Therefore
a — Xyimj„ = 33.3 x 2.84 = 94.6 cm
This means that the distance between the securing plates should not
be more than 94.6 cm.
Distance b between the channel sections (Fig. 396(6)) may be de­
termined from the condition where

Here J*u and J\ are moments of inertia of one channel section about
the axes passing through its centre of gravity, and A 0 is the cross­
sectional area of one channel section. Therefore
5810— 327
Y 40.5
= 11.6cm

Since b=2(c — y0) and from the specifications ^<>=2.52 cm, we get
6=2(11.6—2.52)=18.2 cm.
B. The material of compressed bars is selected from the following
considerations. As long as the critical stress does not exceed the
502 Stability of Elements of Structures [Part VIII

limit of proportionality, the resistance of a bar to buckling is deter­

mined by only one mechanical property, the modulus of elasticity, E.
In bars of medium and particularly low flexibility the critical stress
depends to a considerable extent upon the yield stress or ultimate
strength of the material. These con­
siderations should be taken into
account while selecting a material
for compressed bars of high as well
as low flexibility.
For example, there is no sense in
using special high-strength steels
for long and thin-walled bars, be­
cause the modulus of elasticity is
approximately the same for all
grades of steel. On the other hand,
it is advantageous to use high-
grade steels in bars having critical
stress higher than the limit of pro­
portionality, because in such bars
Fig. 397 the increase of yield stress results
in an increase of the critical stress,
thus improving the resistance of the bar against buckling.
Figure 397 shows approximate diagrams depicting critical stress
as function of flexibility for structural steels: low-carbon steel C-38/23
and stronger steel C-46/33, which have yield stresses equal to 2300
and 3300 kgf/cma, respectively.
It is evident from the diagrams that for highly flexible bars (k great­
er than 100) the critical stresses, which are limited by Euler’s hy­
perbola (27.12), are the same for both steels, as the latter have identical
moduli of elasticity E upon which crc depends; the permissible stresses
for stability are considerably higher for steel C-46/33 in comparison
with steel C-38/23.
It follows from the above discussion that there is no advantage
in using high-strength steels for bars of high strength subjected to
compression in structures. Also, considerable saving of material can
be achieved by using stronger steels for bars of low flexibility.

§ 160. Practical Importance of Stability Check

The check on stability is of great importance for an engineer. It

can be said with authority that sudden failure of most of the struc­
tures occurs only due to loss of stability of its compressed elements.
Engineers know about a large number of cases of catastrophic failure
of structures in the past; yet, somehow, they fail to appreciate the
actual reasons of this. This clearly shows that often engineers do not
Ch. 27) Stability of Bars Under Compression 503

pay sufficient attention to seemingly unimportant but actually very

important aspects in the working of compressed bars.
Loss of stability is all the more dangerous because it occurs sud­
denly. Deformation is not noticeable and fails to arouse suspicion
right up to the moment when the compressive force becomes critical.
In addition, as mentioned earlier, a number of factors—eccentricity
of loading, initial curvature, local overslrcssing of the m aterial-
can further considerably reduce the resistance of compressed bars
to buckling, although the same factors have almost no effect on the
working of other elements of the structure.
Special attention should be paid to the reliability of joints of parts
in compressed bars made of composite sections. Neglect of this
factor was the cause of tragic accidents in the past, especially in case
of large bridges.
At present, an engineer has at his disposal all means to prevent
such mistakes provided he designs properly and the designed structure
is manufactured accurately. The theory of analysis of stability check
has been worked out quite soundly, as has been already discussed
(§§ 154-159).
The stability check of machine parts is slightly different. Here
particular attention must be paid to the value of permissible stress
[ol. While selecting its value, it should be borne in mind that such ,
machine parts as connecting rods, plungers, etc., are subjected to
dynamic loading. Therefore in the formula of permissible stress for
stability, namely
[<rrf] = (f [oJ
[ol should imply the permissible stress for strength under dynamic
loading (see Part IX, Chapter 29).
In § 154 there was a mention about the analogy between sudden
increase in deformation when stresses exceed the critical stress. This
analogy leads to the idea lhat in statically indeterminate structures
failure may not occur despite the stresses achieving critical value,
especially if the stresses are below the limit of elasticity.
Examples of such situations are lattice trusses of old bridges which
are still working under present conditions although the load they
have to take now is considerably higher. Part of the bars in these
trusses may find themselves loaded up to the critical stress limit
and yet remain in the elastic state of deformation. The load of these
bars is taken up by the opposite bars working under tension. When
the load is removed, the bars return to their original positions.
Other instances of such loading can be found in aviation engineering
and ship building, where we have to cope with buckling of not only
bars but also beams, plates and shells. Thus in exceptional cases we
may allow the stresses in a compressed element to reach the critical
limit provided that the stresses do not exceed the elastic limit, the
504 Stability of Elements of Structures [Part VJI1

structure is statically indeterminate and the load of elements thus

overstressed is taken up by other elements.
We shall discuss below a few more complicated problems on sta­
bility check.

CHAPTER 28
More Complicated Questions of Stability
in Elements of Structures
§ 161. Stability of Plane Surface in Bending
of Beams
Buckling occurs not only in axial compression of bars. For exam­
ple, buckling may occur in an I-beam under bending (Fig. 398).
The lower flange of such a beam represents a bar rigidly fixed to
the web and subjected to axial compression. The constraint does not
permit the flange to buckle in the web plane. However, for particular

Fig. 398

beam dimensions the flange may buckle on one side, causing rotation
of the sections w.r.t. one another and giving rise to torsion of the
beam (Fig. 398). Instead of bending in the plane of maximum rigidity,
as intended by designer, the beam starts working in unsymmetric
bending, which results in a sharp increase in deformation ultimately
leading to total failure.
Stability of the beam depends upon its cross-sectional dimensions
and its free length. This length is restricted by providing constraints
between the beams. Serious accidents may occur if insufficient at­
tention is paid to these side constraints in long beams, although
their height may be small (for example, failure of the Tarbes-bridge
in France).
Buckling is also dangerous for thin shells under compression,
i.e. for elements in which one dimension is considerably less than the
other. A thin and wide flange in an 1-section will warp under com­
pression; a web which Is not sufficiently strengthened by stiffening
angles will buckle.
Ch. 28] Complicated Questions of Stability 505

Let us determine for a beam the approximate critical load at which

the plane shape of the beam becomes unstable leading to complete
failure due to side buckling if the load Is further increased. Let us
consider a simply supported beam of thin rectangular section (Fig. 399)
acted upon by a lateral force P.
Let us assume that force P has reached its critical value and the
beam is buckling slightly to one side, as shown in the top view as

Fig. 399

well as the section in Fig. 399; only ends A and B remain in the
original positions due to the constraints applied to them.
with buckling the potential energy of the deformed beam should
increase due to bending on one side and torsion (the energy of bending
in the vertical plane remaining constant). The potential energy of
the external force should decrease, because of the lowering of its
point of application.
Let us denote the potential energy of side bending by Uu of torsion
by £/*, and the work done by the load in lowering by Up. As at the
critical force the transition from plane shape to buckled shape is
accompanied by transformation of energy of the load into potential
energy of deformation of the beam, we may assume that

tf .+ V .- V , (28.1)

Potential energy of buckling (w.r.t. axis zO is (§ 63):

Here the bending moment in an arbitrary section at a distance x

from the left support will be (assuming the angle of rotation, q,, to
be small)
p p
506 Stability of Elements of Structures [Part Vl ll

Substituting the expression for bending moment, we get

(28.2)

The potential energy of torsion may also be expressed through <f.

The work done by the twisting mo­
ment over a length dx is:

Keeping in mind that

and

we obtain

The total potential energy of torsion

Fig. 400
due to budding of the beam is:

Let us first determine the vertical displacement of the point of

application of force P (Fig. 400) in order to calculate the work done
by force P in lowering. Here, it is relevant to point out that there are
two reasons for the displacement of point 0 to 0,: rotation of the
section about point 0 by an angle <p and displacement in the di­
rection of axis O y\.
Since lowering of point 0 is not possible due to rotation, obviously
the cause of lowering of the point of application of force P is buckling
of the beam from plane xOz.
The vertical displacement of point 0 can be found by Castigliano's
theorem from expression (28.2):

2c j . \ JC* ^ 1 dx (28.4)

Wherefrom work done by force P (which is equal to critical force Pe

when buckling starts) in causing displacement 8 is:
Ch. 281 Complicated. Questions of Stability 50?

Now substituting the values of (Ju U3 and Up in the original equation

(28.1), we get
l/J //* tfa
**»•&+oaJ $ )'dx=wrS
o o 0
wherefrom
m //*
^ j A '* (•$•)’&
o o
Denoting rigidity in bending by E Jz=Clt and torsional rigidity by
G/*=Ca, we get the following expression for critical force:

s (fe y * *
P } - 4 C ,C ,^ t;--------- (28.5)
J Jt^p* dx
o
Under the integral sign we have two variables, <p and x, which are
interrelated because qj changes along length x, i.e. <p=<p(*). The law
of variation of tp as a function of x is not known. Using the method
of approximate solution, we assume a value of <p which relates it to x
in such a way that the conditions of constraint at the ends are satis­
fied. Let us assume that
<P= £ s in ~ (28.6)

We see that <p=0 at x= 0, that cp=cpraJX at x = — and that <p=0 at

x= l. Thus, the function vanishes at the supports and is maximum
at the middle of the span. In other words, the function satisfies the
boundary conditions.
Substituting the values of <p and its first derivative in the integral
in equation (28.5), we obtain
//3

cos* -y- dx
D3 _ &r n . .. ®
* C ~ **- 1^2----- //.,

m*

Upon integration we get

(28.7)
508 Stability of Elements of Structures [Purl VIII

The exact value of critical force for a simply supported rectangu-

lar beam (Fig. 399) is given by the formula *

The error of approximate solution is about 1.5%. The critical force

depends upon the product of rigidity under bending Ci=EJt and
torsional rigidity Cz=GJt.
If the load is uniformly distributed along the axis, then the critical
value of the distributed force will be

<28-8>
For a cantilever of length I which is loaded by a uniformly dis­
tributed force we get
<2 8 9 >

For a cantilever being acted upon by a concentrated force at its

free end we have

P'-tjfvTZ. <2 8 - 1 0 )

The formulas for critical force in an 1-bcam are the same as for a
rectangular. The difference is that the coefficient before V CiCa is
not a constant quantity but depends upon the resistance of flanges
to buckling and is determined by the ratio

-a w
where h is the height of the I-beam. Thus, for an I-beam
= i I' C A (28. 11)

where p has to be determined individually for each value of a. The

values of coefficient p have been determined for various types of loads
and are given below in Table 20.
It is evident from the table that the values of coefficient p approach
its values for a rectangular section as ///i is increased. A ta=100,
P almost coincides with its numerical value for a rectangular beam.
While studying the slability of plane shape during bending it is
essential that the normal stresses due to bending should not exceed

• S. Timoshenko, Theory of Elastic Stability, Toronto, 1961.

Ch. 28) Complicated Questions of Stability 509

Table 20
Coefficient p in Formula (28.11)

C, / 1 \*
fix P. 8» 8. 8. 8.

1 2 3 I *o 4

0.1 31.0 86.4 143.0 16 5.08 18.3 30.5

1.0 y.7f. 31.9 53.0 20 4.85 18 1 30.1
2.0 8.03 25.6 42.6 32 4.50 17.9 29.4
4.0 6.73 21.8 36.3 50 4.35 17.5 29.0
G .O 6.19 20.3 33.8 70 4.10 17.4 28.8
8.0 5.87 19.6 32.6 90 4.04 17 2 28.6
12.0 5.36 18.8 31.5 too 4.04 17.2 28.6

Column 2 is for a cantilever loaded at tlie free end.

Column 3 is for a simply supported beam with a force acting at the middle
of its span.
Column 4 is for a simply supported beam loaded with a uniformly d istrib ­
uted force.

the permissible stress for stability, where k 9 is the safety

factor.
Knowing the critical force for each type of load on the beam, we
can easily determine the critical stress:
_ _Mumx
C~~~Wwy
where M^ is the maximum bending moment due to the critical
force and Wu is the section modulus in the web plane.
The results obtained above hold good only when the critical stress
under buckling does not exceed the limit of proportionality of the
material.
If Oc> j p, then the formulas derived above give exaggerated values
of the critical stresses, just as Euler’s formula gives overstated values
for compressed bars of low flexibility. Data collected by experimental
studies should be used for determining the actual critical stresses
under buckling of beams when <T0>cr,,. Prof. Yasinskii suggests that
the analogy with compressed bars can be successfully employed by
assuming that the ratio between the actual stresses and the stresses
determined by the formulas derived in this section is the same as
between the actual stresses and critical stresses in compressed bars
when cr^Up.
Let us consider the following example. A simply supported I-beam
No. 60 (see Appendix) with a span /—6 m is loaded by a uniformly
510 Stability of Elements of Structures f Part VIII

distributed force of intensity <7=9 tf/m. Check the strength and sta­
bility of plane shape of the beam if the permissible stress [al=
= 1600 kgf/cma and the safety factor for strength and stability are
both equal to k =1.7.
The dimensions of the section (Fig. 401) and its geometrical charac­
teristics, as obtained from the specifications, are: A=60 cm, 6=19 cm,
6j=1.2cm , 6*=/= 17.8 m m «l.8cm , /ij=60-—-
—2x1.8=56.4 cm, /„ = 7 6 800 cm4, / , =
h =1725 cm4, U%=2560 cm3.
The torsional moment of inertia has been
calculated from formula (9.38'):
where t) = 1.2 for an I-beam. Substituting the
numerical values, we obtain
A - T tM + 2 6 ©
=0.4 (56.4X 1.2* + 2 x 1 9 x l.8 s) = 127 cm*
Let us check the beam section for strength
and stability of plane form.
(a) Strength check:
^ A n * if
^max ga § —40.Dtf*rn
= 405x10* kgf-cm
Fig. 401 Afmax_ 405X104
w"«* Wy ~~ 2560
= 1590 kgf/cm* < 1600 kgf/em*
(b) Check for stability. The critical load that leads to instability
of the uniplanar state of bending may be calculated from formula
(28.11):

where C ,= £ /, is the rigidity under bending, Ct —QJt is the torsional

rigidity, and p is a coefficient that depends upon the ratio of rigidi­
ties and llh .
Assuming that the modulus of elasticity for steel £ = 2 x 104 kgf/cm1
and shearing modulus from formula (6.38) G= ^ =»^ , we get

= p ^ : |/8 4 100= ^x5.56 x 290= 1610(1

Ch. 28\ Complicated Questions of Stability 511

As coefficient 0 depends upon the ratio

127 /ooo\*
Cx \ h ) lJe\h ) 2.6x I725^'§5'J ; 2.84
by using Table 20 (Column 4) and interpolating between a = 2 and
<x=4 we find (5=39.4, and the critical load is:
Qe = (<?/), = 1610 x 39.4 = 63 400 kgf
In order to avoid buckling the permissible load should be taken as

[Q] —% = = 37 300 kgf

For the chosen value of A=1.7 the intensity of the uniformly dis­
tributed load should not exceed
IQ | 37300
<7= — 6250 kgf/m

Hence, reliable functioning of the beam can be ensured only if the

given intensity of 9 tf/m is reduced by about one and a half or by
providing side supports which prevent buckling of Ihe beam (if the
beam design permits this).
Let us see how the value of critical force changes when side con­
straints are applied. Applying side constraints is equivalent to
reducing the free length of the beam by two and using other values
of coefficients a and (5:
C# /o.5/‘\* 127 X0.25 / 600\* n „
a £ p r] 2.6X 1725 \ "0?T/ “ U/ 1

From Table 20 wc obtain (5=0.57 by graphic interpolation between

a=0.1 and a= 1.0. The critical load is:
E 2 x l0 #x290
<?c=P /C ,C ,= 0 -5 7 5 25/a T X = 0.57 0.25 x 600* 367 000 kgf
(0.5/)*

This corresponds to critical stress

307000x 600
°c 8x2560
10300 kgf/cm*

The value of orf is considerably greater than the limit of proportion­

ality of the material; therefore, formula (28.11) cannot be applied.
The value of critical stress obtained above actually corresponds to
the yield stress of the material. Therefore, in this case it is sufficient
to carry out the strength check only.
512 Stability of Elements of Structures [Part VIII

§ 162. Design of Compressed-bent Bars

While studying the combined effect of axial and lateral forces in

Chapter 21 we used the principle of superposition of forces and added
the stresses due to tension or compression to the stresses due to bend­
ing. The strength condition in this case is:

= (21.1)

Assuming that the axial force N —± P does not participate in

bending, in the formula we used bending moment A f-,x caused only
by lateral forces. However, we have already seen while studying the
stability of bars by Euler’s method (§ 155) that in the case of buckling
the axial compressive force P creates an additional bending moment
M ’—Pf, which gives rise to additional stresses and displacements
due to additional bending of the
f bar (Fig. 402). The maximum stress
in the critical section may be deter­
mined by the formula
I ^max j Pf (28 12)
if ~w~ w
T
w where / is the maximum deflection
due to the lateral and axial com­
pressive forces. If the axial force is
tensile, it decreases the curvature
of the buckled bar and reduces stress
this case is of little interest
(the third term in formula (28.12)
will be deducted).
Fig. 402 From a comparison of formulas
(21.1) and (28.12) we see that by
applying the principle of superposition of forces (Chapter 21) we
neglect the additional bending moment Pf due to axial forces and
the stress Pj!W. Strictly speaking, the principle of superposition of
forces cannot be applied at ail if both axial and lateral forces act on
the bar. By neglecting the third term in formula (28.12) we intro­
duce a serious error except when the bar is sufficiently rigid and
deflection / is small in magnitude. However, if we ignore the bending
caused by axial forces in flexible bars, this may lead to serious errors
while determining the stresses.
In order to avoid such errors it is essential to take into account
the bending moment due to axial forces P by determining deflection f
caused by the combined action of both axial and lateral forces.
For the compressed-bent bar shown in Fig. 402 the differential
Ch. 28] Complicated Questions of Stability 513

equation of the elastic curve is:

E Jl = Mc- P y (28.13)
Here M 0 = y - x —2j- is the bending moment due to the lateral
forces. Dividing (28.13) by EJ and substituting we obtain

i f - r P y ^ Y j ^ x — qjr )
The general solution of this equation may be expressed
y = Ct sin kx -fC t cos kx -\-y*
After choosing the particular solution y* ^here y* = ^ Y j ' X
X (Jp + lx — and determining the constants of integration
Ci and C%from constraint conditions at the bar ends (y=0 at x= 0
and y —0 at x= l) we may calculate y and find deflection / at x = li 2 .
However, if several forces are acting on the bar, this approach
leads to cumbersome calculations, because for different portions M*
has different expressions and the elastic curve consists of a number
of conjugated curves. In such cases it is simpler to solve the problem
by an approximate method. The idea lying at the root of this method
is that the shape of the elastic curve is defined beforehand with the
condition that it must satisfy the boundary conditions; this makes
it much easier to solve the problem.
Suppose a beam is loaded by lateral forces P t, Pa Pa and axial
compressive force P (Fig. 403). Bearing in mind that a sinusoidal
elastic curve was obtained while solving Euler’s problem, we assume
that in our problem also the elastic curve due to lateral forces is si­
nusoidal:
y0= f0s in ^ (28.14)

It can be easily seen that this equation for the elastic curve satisfies
the boundary conditions: at both supports, at x = 0 and x = l , deflec­
tion 0. At the supports the bending moments are also zero (Af«=
= E Jyl——EJjf sin ™ becomes zero for x = 0 and *= /); /o is the
maximum deflection of the beam due to lateral forces acting at right
angles to its axis.
Let us rewrite equation (28.13) by substituting EJyl for M t:
EJy" — EJy%— Py
or
E JyH+ Py = — E J ^ /„ sin~
514 Stability of Elements of Structures [Part VI11
n
Dividing by EJ and substituting ^= A *, we get

—jr /.s in S ! (28 . 15)

It will be easiest to look for the solution in the form y= f s illy ,

i.e. assume that under the combined action of axial and lateral

forces the beam bends along the same sine curve. After substituting
in equation (28.15), we obtain
nx a - e . nx
^ jJ s in ^ -M » /s in ^ = 7rfoSinT
or
iw*
«2 f= h
wherefrom
/- (28.16)

Substituting k*=Pi(EJ) and keeping in mind that expression ^

may be represented as Euler’s critical force, we get
/= (28.17)

Coefficient C accounts for the effect of axial forces on deflection:

C l _ Pc (28.18)
F_ P e —P
Pc
It is clear from formula (28.18) that deflection f should theoreti­
cally become infinite when compressive force P attains its critical
value. Note that critical force Pc enters the formula formally as a
substitution for n^EJIl2, where J is the moment of inertia of the
section about the neutral axis when the beam is acted upon by lateral
forces. This means that J is not Jmia, as the beam is usually placed
in such a way that the moment of inertia of its section is maximum
about the neutral axis.
Ch. 28\ Complicated Questions of Stability 515

Let us apply the solution obtained here to particular examples,

for instance to a simply supported beam subjected to bending by a
uniformly distributed force q and compressed by forces P (Fig. 402).
As already established,
= M0 + P / = £ + PftC (28.19)
5a/4
where / 0= 334^ (see Chapter 15); therefore

Multiplying and dividing the fraction inside the parentheses by

ii4, we obtain for this fraction
5.a«P/* 1.028P
4W E J ^ Pe

and the expression for becomes

'W » .« = T ( 1 + i r r c ) = ? <:‘ (28-2t>)

For practical purposeswe may assume that 1.028P=P,and after
substituting coefficient C=p ~ p in formula (28.20) we get

= = (28.20')

For the assumed approximation (1.028«1) coefficient C i=aC is

found to be equal to C, i.e. the same as in deflection. Some authors
suggest that we can always assume that C i—C on the basis of the
assumption that bending moments are proportional to deflections.
Formula (28.12) for normal stresses acquires the form
+ ( 28. 21)
l~ r .
p
When p - is small, coefficient Ci is close to unity and formula (28.21)
coincides with (21 . 1).
Let us note that when the beam is symmetrically loaded by lateral
forces, approximate formulas (28.17) and (28.21) give results that
are very close to the exact solution. The results in the case of unsym-
metric loading are slightly poorer, yet they are quite acceptable for
practical calculations (discrepancy does not exceed 5-7%). If all
forces act in one direction, the deflection ft may be considered maxi­
mum at the middle of the span.
51G Stability of Elements of Structures [Part VIIt

It is evident from the formulas derived above that there exists

a non-linear relationship between deflections and stresses and the
forces applied: if ail the forces are increased, say, n times, the stresses
increase more than n times due to the increase in the value of coef­
ficient Cj. This means that the strength condition crmJx^ [a ] ceases
to be valid.
Therefore, in order to ensure sufficient strength the compressed-
bent bars should be designed for permissible loads. Let us derive the
strength condition for the beam discussed above.
Let us assume that in our beam the maximum stresses become equal
to yield stress when all the forces are raised k0 times. Formula (28.21)
may be rewritten as follows:
koP
T

where k0P and kjj are limiting loads. To go over to permissible loads
we divide this equation by the safety factor k0. The equation then
becomes
1 _Oy
WI
Pc

Here j- =fol is the main permissible stress under compression.

The strength condition may be written as follows:

W —s ^ M (28.22)
~T7
The effect of axial forces on stress in the given bar is taken into ac­
count by coefficient

c »“ r s ~ r ^ t r (28'23)
Pc
We now impose a restriction on deflection by writing the rigidity
condition:
/» ..= /.—^ = / . (28.24)
T?
where 1/1 is the permissible deflection and k f is the safety factor
against deflection.
Besides checking the strength and rigidity of the bar in the plane
of bending it is necessary to check its stability in the plane of mini-
Ch. 28\ Complicated Questions of Stability 517

mum rigidity when it is subjected only to compressive forces P (§ 158)

and also check the stability for plane surface in bending (§ 161).
Let us consider one more example. Suppose that a simply supported
beam is subjected to bending by a force P» acting at the middle of
its span and is compressed by an axial force P. From formulas (28.17)
and (28.18) the deflection at the middle of the span is:

l-tfi. or t = $ § j — '-p- (28.25)

1 r.
The maximum absolute normal stresses in the critical section are:
P M 0+ P f
^max — S W (28.26)
Off
_ p \ p °l \ p p °p r p ./ V 21/1
/ , 4, \PI*
l~ 1i \
5 w 4 8 £ 7 1=3"5" ' 411?I

Substituting
)2EJ n*E J
*0.822/2 ~ 0.822
we obtain
^ p . p »l ( i * 0.822P P. \ , na 0_.
tfaiix— 5 i 4 j ^ H f~c Ft. — P J (28.27)
After transformations, we get
_ p j _ p 9i pe- o . m p
< w — s . ? c— p (28.28)
Off
n _ p » Mq
°jaax “5" r

Here Ci is slightly less than C. For example, when P =0.5P C, C=»

=»P PJ_p — 2 while C i~ 1.822. Calculations are the same as in the
preceding example.

§ 163. Effect of Eccentric Compressive Force

and Initial Curvature of Bar
A. In case of eccentric application the axial compressive force leads
to eccentric compression, which, as shown earlier (Chapter 21) re­
sults in axial compression and bending (§ 162). By using the results
obtained in § 162 we can take into account the effect of initial eccen­
tricity e of the axial compressive force P (Fig. 404(a)).
518 Stability of Elements of Structures [Part V l t l

According to formula (28.17) the maximum deflection in this case is

f = / 0C
where f , is the deflection due to bending caused by moment M0=Pe
and is equal to (Chapter 15)
, _ W pel*
8 EJ= TeJ

and coefficient C that accounts for the effect of axial forces on deflec­
tion (28.18) is:
/■>_ 1___ Pe
Pc
Consequently

i=M r r r (28’29)
Pc

and the maximum compressive stress according to formula (28.12) is:

®max — s - r w t w

Substituting the value of /, (28.29), we obtain

* P , Pe , P Pci* 1 _ P , Pe / , , PI* I \
°m#x “ T + W ' W &EJ l S W f 1 8EJ 4 _ P _ \
(28.30)
Pe \ P eJ

After opening the brackets, we find that

■ W - T + T f f C ,, where C .- f r ^ T (28-31)

B. If there is an initial curvature in the bar compressed by forces P

(Fig. 404 (fr)), the eccentricity of the point of application of force P
is assumed to be known: in the middle of the span it is equal to
and the total deflection is (28.17):
f —f£* or —

As in the earlier case of eccentric application of compressive forces,

deflections rise sharply only when force P approaches the critical
value, Pc- In both cases (A and B) the Euler critical force should be
considered dangerous. Therefore, irrespective of eccentricity and
initial curvature, the stability of a bar against buckling should be
checked as in axial compression. The strength check is different,
Ch. 28) Complicated Questions of Stability 519

because in these cases bending should be considered in addition to

compression (see § 122).
Hence, when highly flexible bars (ormax<cr2,) are subjected to axial
compression, they lose stability upon the compressive force attaining
the critical value determined by Euler’s formula. Euler’s critical
force P = P Cshould be taken as the breaking load. Neither the eccen-
trie application of compressive force nor the presence of initial cur­
vature have any influence upon
the breaking load in these bars. p S ftt , p
The above conclusion does not Ai > A
hold well for bars of small and
medium flexibility. At critical i -
~~ i
stresses exceeding the limit of
proportionality the above-men­
tioned factors considerably re­
duce <fc. This has been observed
experimentally and confirmed by
theoretical attempts at calculat­
ing critical deformation. Experi­
ments reveal that eccenlricity of
application of force considerably
affects the stability of bars of
small and medium flexibility, it also affects the stability of long
bars, but to a much smaller degree.
The additional factors, which have been discussed in this section,
compel us to increase the stability factor for bars of small and medium
flexibility and select a value which is slightly greater than the sa­
fety factor in Ihe case of long bars. For proper evaluation of the in­
fluence of ivc-n tricity and initial curvature on the strength and sta­
bility of compressed bars, we must get an idea about the numerical
values of e and yo.
fn accurately manufactured bars we may expect an initial deflec­
tion yb which is the of the length; if the manufacturing ac­
curacy is not high, the initial deflection may be twice as large.
If the centering is proper, eccentricity may be the of the
length. Further, we must take into account the tolerances of the
cross-sectional dimensions; it may be assumed that they are equi­
valent to an eccentricity of ^ and an initial deflection of the same
magnitude. Here h is the cross-sectional dimension in the plane of
possible buckling. In composite sections we must provide for an
i.
additional eccentricity of about ^ on account of the possible
difference between the areas of individual elements.
520 Stability of Elements of Structures [Part V I I f

Thus, lor a solid section we may assume the following minimum

values of e and y0:
_ I . it . ___/ . h
e ~~ 750 “T 45 3,10 ^ 0— 1000 ”*"40

Besides eccentricity and initial curvature, there are a number of

other factors which affect the stability of compressed bars much
more than the strength of beams and elements under tension. These
factors include work hardening, initial stresses due to manufacturing
inaccuracy of various parts, local defects in castings and knots in
timber. In steel structures the effect of all these additional factors
is taken into account by an increased (by 10-20%) stability factor (see
§ 153).
In conclusion let us note that in this section we discussed only a
few problems in which critical force was determined at the instant
when the bar crosses over from the existing state of equilibrium to a
new one. it was assumed that bending was the only source that caused
instability. However, it is known that loss of stability may occur in
other forms too, in particular as bending plus torsion and pure torsion *
(in case of axial compression of thin-walled bars).
Instability is characteristic not only of bars. The theory of sta­
bility deals with many complicated problems of stability of com­
plete structures and their individual elements—arches, frames, shells,
plates, etc. Of special interest arc problems of stability of such struc­
tures and their elements when subjected to dynamic loading **
and also investigations on stability in the process of eiasto-plastic
deformation and visco-elastic deformation (see Chapter 32).
It is impossible to solve these problems if a static approach is
adopted towards problems of stability as problems of equilibrium in
one or the other form. In all these problems deformation should be
studied in time, i.e. the stability of movement must be investigated.
Many complicated problems of stability are solved at present pre­
cisely on the basis of these principles. The reader can get acquainted
with them in the special literature.***

* See N. M. Belyaev, Strength of Materials, Nauka, 1965 (it) Russian), §213.

•• The problem of dynamic stability of prismatic bars subjected to variable
loading was first solved in 1924. See N. M. Belyaev, Selected Works on Engineering
Structures, Leningrad, 1924 (in Russian).
• • • See, for instance, I. 1. Goldcnblat, Modern Problems of Vibration and Stability
of Engineering Structures, Stroiizdai, 1948 (in Russian): V. V. Bolotin, Dynamic
Stability of Elastic Systems, Gostekhizriat, 1956 (in Russian), A. S. Vuhtiir, Sta­
bility of Deformable Systems, Nauka, 1967 (in Russian).
 PART IX
Dynamic Action of Forces

' CHAPTER 29

Effect of Forces of Inertia.

Stresses due to Vibrations
§ 164. Introduction
Until now we were solving the fundamental problem of strength
of materials: determining cross-sectional sizes and selecting proper
material for elements of structures by assuming the loading to be
static.
It was explained in § 2 that loading may be considered static if
there is no mechanical movement of the parts when pressure is trans­
ferred from one part to the other or when both parts are acted upon
by body forces. Under such loading each element of the structure
remains in equilibrium under the action of external forces and stresses.
The constancy of movement is characterized by constant velocity
of the parts under consideration and complete absence of acceleration
of these parts. Tf acceleration is experienced by the body or the parts
contacting with it, the loading is said to be dynamic. For instance,
the earth pressure on a bulkhead is an example of static loading,
because neither the bulkhead nor the earth mass move, their velo­
cities are constant and equal to zero.
Similarly, the force exerted on the rope by a load which is lifted
by it may be considered static provided the load is raised with a con­
stant velocity. On the other hand, the force exerted will be dynamic
if the load is raised with acceleration. The connecting rods of steam
and interna! combustion engines experience dynamic loading, be­
cause their individual elements have different velocities. Two other
examples of constructions working under dynamic loading are the
foundation of a machine with rotating parts mounted eccentrically
w.r.t. the axis of rotation {the foundation in this case experiences
centripetal acceleration) and the foundation and piston rod of a
steam hammer (in the process of forging the hammer block comes
to a stop in a very short period on account of very strong retardation).
Even these examples are enough to make it clear that in practice
we come across various types of acceleration which bodies under
522 Dynamic Action of Forces [Part tX

consideration or bodies contacting with them have to experience.

The acceleration may be constant in direction as well as magnitude
or only in direction; it may also be reversible.
Under variable and reversible stresses the bodies fail due to a gra­
dually increasing crack, and the failure is said to occur due to fa­
tigue. Jf there is a sharp change in the velocity of an element of struc­
ture w.r.t. the force being exerted on it by adjacent elements, i.e.
if the element experiences shock loading, the material of the element
may behave as if it were brittle, although it is ductile under static
loading. Therefore, in conducting a strength check for elements of
structures subjected to dynamic loading, it is important to study not
only the effect of external loading on the magnitude of stresses in the
element but also its effect on the nature of resistance of the element
material.
The effect of acceleration of elements of structures on the stressed
state of the material may be accounted for as follows. If a body moves
with acceleration, it is being acted upon (experiencing) by forces (pres­
sure) from other bodies. From the law of equal reactions the body
under consideration acts upon the other bodies with forces equal in
magnitude and oppositely directed, namely the forces of inertia.
This logic is also applicable to each element of the body moving with
acceleration; the elements act on the contacting elements with a
force equal to the force of inertia.
Thus, when elements of structures move with acceleration, they
experience additional stresses which are equivalent to static stresses
caused by forces of inertia; each element of the structure gives rise
to stresses in the adjacent elements, as if the latter were acted upon
by forces equal to the corresponding forces of inertia.
Here we must differentiate between three situations. If the mag­
nitude and location of the external forces acting on the element under
consideration does not depend upon the deformation of the element,
i.e. if the deformation does not change the nature of motion of the
body, then its acceleration is determined from the methods of kine­
matics of solid bodies, and the dynamic action of external forces is
reduced to the addition of a static load corresponding to the inertial
forces. This method is applicable to a majority of situations of prac­
tical importance (except shock loading).
If the acceleration changes, in the process, this invariably gives
rise to vibrations in the element under condsideration. The vibrations
in their turn may cause resonance that results in a sharp increase in
deformation and stresses. These stresses may be very high and must
be added to the stresses obtained by considering the inertial forces
as an additional static load.
Finally, there may be cases (shock loading) when the acceleration
and consequently the corresponding forces of inertia depend upon
the deformability of the element under consideration. In such cases
Ch. 29) Forces of Inertia. Stresses due to Vibrations 523

the mechanical properties of the material must be taken into account

while calculating the inertial forces. The method of strength check
in each of the above cases will be explained through the following
examples.

§ 165. Determining Stresses in Uniformly

Accelerated Motion of Bodies
We shall begin the study of strength check under dynamic loading
with the simplest case when points on the element of structure under
consideration move with constant acceleration without causing vi­
brations. As an example we shall study
the uniformly accelerated lifting of load M Sr
Q suspended from a steel cable of cross-
sectional area 5. The specific weight of
the cable material is y, the load is lifted
with a constant acceleration a cm/see2
(Fig. 405). We shall determine the stres­ Tf
ses in an arbitrary section at a distance X
x from the lower end of the cable. Let us i
cut the cable at this section and study
the equilibrium of the lower cutoff por­ D u
tion. This portion moves upwards with
acceleration a, which means that be­
sides the force balancing Its weight it is m
acted upon from the upper portion by a Fig.
force equal to its mass times accelera-
tion a, i.e. Q-l-yS* a, where g is the acceleration of gravity.
s.
From the law of equality of action and reaction the upper portion
will experience a similar force acting downwards. Thus the dynamic
stresses o D acting in the sectioned plane on the lower portion will
balance not only the static load Q+ySx but also the additional force
Q ~J$,Xa. To determine these stresses we must study the equilib­
rium of the lower portion under the action of aBt static load Q+ySx,
and the force of inertia --" y — a acting downwards (Fig. 405). We find
that
____Q+yS* , Q +yS*___ Q + V $ x f « , a \
°i>— s— + - * s - “— r - l 1+7 j
Ratio - $ * represents static stress <r. in the section oT cutting;
therefore
O n = < f .( l + f ) (29.1)
524 Dynamic Action of Forces [Part IX

i.e. the dynamic stress is equal to the static stress multiplied by

coefficient l + ~ . This coefficient is called the dynamic coefficient
and is denoted hy K d-
OD=KDax (29.2)
This form of the formula for dynamic stresses shows why we paid
special attention to calculating the stresses under static loading:
in a large number of cases dynamic stresses may be expressed through
static stresses by multiplying the latter with the appropriate dynamic
coefficient.
The strength condition may be written
&D m a x = <T.v max 0 + 7 )-* * * « » «
wherefrom
r°i _ i° i
i + fL Kd
g
Thus in a number of cases dynamic analysis may be replaced by
static by simply dividing the permissible stress with the dynamic
coefficient Ki>. This is done when it is difficult to obtain the dynamic
coefficient theoretically and we have to be satisfied with the value
of the dynamic coefficient determined experimentally. This method
is employed, for example, in taking into account the dynamic nature
of temporary loads acting on bridges.

§ 166. Stresses in a Rotating Ring (Flywheel Rim)

As a second example we shall determine the stresses in a uniform
ring rotating at a high speed (Fig. 406(a)). The flywheel rim works
under similar conditions, provided we neglect the effect of spokes.

00 (b)
Fig. 406

Lot S be the cross-sectional area of the ring, y the specific weight

of its material, n its number of revolutions per unit time, m its an­
gular velocity of rotation and D the mean diameter of the ring.
Ch. 29] Forces of Inertia. Stresses due to Vibrations 525

Let us isolate an element of length ds from the ring. When the

ring rotates, this element moves along a circular path with constant
angular velocity <o. Angular acceleration e is zero. Therefore tangen­
tial acceleration of the element is w ,~ tD /2=0; normal (centripetal)
acceleration of the element is wn=it>*Di2 and is directed toward
the centre of the ring. In order to determine o D, the force of inertia
must be applied to each and every element of the ring. This force is
directed away from the centre and is equal to

wn J ds= Y ^ T ' ds = qds

where q is the intensity of the inertial force per unit length of the
rim. Thus, the ring experiences stresses as if it were loaded by a radial
force of 'intensity q per unit length (Fig. 406(b)). Force P stretching
the rim is (§ 19):
p _ O q
F ~ 2
Stress Cjy is:
P Dif DSy uPD vtfS
»D — $ ~ l S ~ 2 g S 2 = 4g ~ g

where i/-wD.'2 is the linear velocity of points on the surface of the

ring. Thus, stresses in the rim depend only upon the specific weight
of the rim material and the linear velocity of points on the rim sur­
face. Let us solve the following problem to get an idea about the
approximate value of these stresses:
n = 360 rpm, Z>= 4m, Y ^ .b g f/c m 3
Angular velocity is:
2nn 2«x360
a )= iKT= “ « r“ = I2stsec- *
The stress is:
y(o3Ds 7 .5xl44n3x 16x10*
4 x l 0 3X<Wl
= 435 kgf/cma

§ 167. Stresses in Connecting Rods

Let us check the strength of connecting rod AB joining two wheel
axles of a steam engine (Fig. 407); to the driving wheel Ot is trans­
mitted a torque from the steam engine. The connecting rod is secured
to the wheels at points A and B with the help of cylindrical hinges,
AO* and BOi are both equal to r, diameter of the wheels is D, length
of the connecting rod is I and the steam engine moves with a constant
velocity v.
526 Dynamic Action of Forces tPart IX

As the connecting rod is in movement, the first step in checking

its strength is to establish whether the motion is with acceleration,
i.e. solve a clear-cut problem of kinematics. The connecting rod is
in relative motion w.r.t. the steam engine, and the engine imparts
to it translational motion of velocity v.
As the translational motion is of constant velocity, the acceleration
can appear only in the relative motion. Now in relative motion of
the connecting rod two of its points A and B move identically, de-

Fig. 407

scribing circles of radius r in a single plane. The relative motion of the

connecting rod may therefore be considered plane-rectilinear and it
may be safely concluded that all points of the connecting rod have
the same velocity and acceleration as points A and B.
Point A moves with the second wheel describing a circle of radius r.
If the steam engine is moving with uniform velocity, the angular
velocity of rotation © of the wheels must also be constant. This means
that angular acceleration must be zero and hence the tangential
acceleration of point A must also be zero, i.e. wt—0. Point A expe­
riences only centripetal acceleration directed from A towards Oa
and equal to©*r. Any other point on the connecting rod, say, point K,
experiences the same acceleration parallel to 0*4.
To check the strength of the connecting rod, the load due to inertial
forces must be added to its dead weight. The inertial force per unit
length of the connecting rod is:
_ 1X|Sy,,, _ 5 y
<?= 8 H
this force of inertia acts parallel to 0*4 but is directed opposite to
the direction of acceleration.
In the position of the connecting rod shown in the diagram, bending
caused by its dead weight is opposite to that caused by the forces of
inertia. The connecting rod finds itself critically loaded in the lower­
most position A\B% when both the forces act in the same direction.
The total load qo per unit length of the connecting rod will in this
case be
?o = T S + f »,r = Ts ( l + - i )
Ch. 29) Forces of Inertia. Stresses due to Vibrations 527

The connecting rod should be analyzed as a beam hinged at points A

and B and loaded by a uniformly distributed force qD. The maximum
bending moment at the middle of span will be

8 8 + g )

The maximum stress in this section will be

n _Mpijix__ s yf* / 1 i \
v Z>roax — ^ ~ fl? "g" V 1 IT /

Example. Analyze two following shapes of the connecting rod, (o)

rectangular cross section and (b) I-section (Fig. 408), for the data
given below:
<*>=30sec\ y — 7.86 gf/cm*, r = 50cm, / = 150cm
In this example
y P f t , U*r\ 0.00786X150® / , , 50 x30* \ ___
------- 5------- ( l + - - 9gr - j = 1036kgf/cm

For the rectangular section

S = 10x4.5 = 45 cm8, W = 1 ^ 1 ° ! . = 75 cm*

•^- —-^■ = 0 .6 cm"1, 1036 = 622 kgf/cm8

For the I-section:

S = 1 0 x 4 .5 —2 x 6 x 1.5 = 2 7 cm®, 5x6>= 64.2 cm*

4 “= = 0 -42 cm "1, c r ^ , = 0.42x1036 = 435 kgf/cm8

Hence, despite the decrease in the section modulus (almost by 15%)

the maximum stresses in the second case are less by 1.5 times due
to considerable decrease in the weight of the connecting rod.
Besides bending, the connecting rod is subjected to tension or
compression due to force P with which wheel Oj acts on wheel 0*.
In position A 1 B 1 the connecting rod experiences compression. Neglect­
ing the effect of deflection on the bending moment we may write the
strength condition as follows:

•‘- “ T + f i r O + T ^ M
In addition to the strength check the connected rod should also
be checked for stability by considering it as simply supported in the
528 Dynamic Action of Forces {Part IX

plane of bending caused by qD, and as a bar fixed rigidly at one end
in a perpendicular plane. While calculating flexibility the maximum
value of the radius of gyration should be used in the first case and the
minimum in the second.
We may similarly design connecting rod AB hinged at end A with
the crank OA that is rotating about point 0 with angular velocity o>
(Fig. 409). If the crank rotates with constant angular velocity, point
A of the connecting rod experiences only centripetal acceleration,

Fig. 408 Fig. 409

whereas point B experiences only tangential acceleration. All other

points of the connecting rod between points A and B experience
both accelerations. Limiting ourselves to forces of inertia arising in
the connecting rod due to centripetal acceleration only, let us study
the position of the connecting rod when it is perpendicular to the
crank and, consequently, when the centripetal acceleration of point A
is perpendicular to the crank axis. Let us assume that the centripetal
inertial force q is perpendicular to the crank at all points and changes
linearly along the length of the connecting rod from q=q 9 at point A
to q—0 at point B. The greater the length of the connecting rod as
compared to the crank the higher is the accuracy of this assumption.
The connecting rod may be considered as a simply supported beam
hinged at points A and B. The bending moment is maximum at x = -^ ~
V3
(x is measured from point B) and is equal to (see § 59)

Since (°2r and oi . we get

g, / 8_____ SyI*tcAr
9 Yr3 W ~ 9 V T W g
Ou 291 Forces of Inertia. Stresses due to Vibrations 529

§ 168. Rotating Disc of Uniform Thickness

The problem of determining stresses and deformation in shafts

and discs rotating at high speeds is of considerable interest. Due to
their high speeds of rotation, the steam turbine shafts and discs ex­
perience large centripetal forces.
The stresses caused by these forces
are distributed symmetrically about
the axis of rotation of the disc.
Let us study a simple problem on
analysis of discs of uniform thick­
ness. The analysis of such discs
lies at the base of several approxi­
mate methods employed in analyz­
ing discs of an arbitrary shape. We
shall use here some results obtained
while deriving the formulas for the
analysis of thick-walled cylinders
(§ 144). Let us assume that stres­
ses <Jr and a t remain constant over
the width of the disc of unit thick­
ness; we shall consider axial stress
at to be equal to zero.
Let ns write the conditions of equilibrium of element AB isolated
from the disc by two meridian sections and two concentric cylindri­
cal surfaces (Fig. 410). In this case, besides the forces acting on ele­
ment AB, we must also take into account inertial force
vr x l x < W 6
d/ = co*r
e
which acts from the centre of the disc towards its periphery. Equa­
tion (25.1) derived in § 144 may be replaced by the following relation:
dar ya*r*.... a 0 (29.4)
or—ot + r dr 8

Equation (25.4) of the same section (equation of joint deformation)

remains valid in the presentcase also, i.e.

o,) (»-5)

Substituting the value of (<rr—a,) from equation (29.4) in equation

(29.5), we get
^ = (29.6)
dr dr g r '
1 S -3 aio
530 Dynamic Action of Forces [Part IX

Differentiating equation (29.4) with respect to r and substituting

from equation (29.6), we get the following linear differential
equation:
<Par
r 7Pr
or
p l f r - 0

Upon integration we obtain

i- -^ya>
■ * + r‘ 8g
*r* (29.7)

It ensues from equations (29.4) and (29.7) that

-
(Ji= °_ r +i r_-dor , yoPr* A B I -h3n
1 f + ! — - = A — ?;----- (29.8)

In formulas (29.7) and (29.8), A and B are constants of integra­

tion, which must be determined from the conditions at the disc sur­
face. In determining the constants we shall study the following two
cases: (a) disc with a central hole, and (2) solid disc. Let us assume
that ends of the disc are free of external forces.
For the disc with a central hole, stress ar must be zero at r—rx
as well as at r=r* (Fig. 410). When the conditions at the disc surface
are applied to formula (29.7) we get the following equations:
A + j i - 2 Q L y m 'n = 0
and

wherefrom
A= ^ ?<*>’ (r? -KS), B -— y® M i

Substituting the values of A and B in formulas (29.7) and (29.8),

we get

( '* + '• —■
r‘— T r - )
and

°*= J s r ( p + i*) ( ' ■i +fs + + 3t‘) r ' ]

Ck. 29) Forces of Inertia. Stresses due to Vibrations 531

Assuming for brevity that

we may write the equations obtained above as follows:

(29.9)

= p J1 + a* ( l mP*] (29.10)

Let us point out that ar becomes zero at p = l and p ^ a , i.e. at the

internal and external peripheries of the disc. It is positive for values
of p between 1 and a and, as is not difficult to prove, becomes
maximum at
p = V ra=* j / ^ ^
At this value
(<U»« = P ( l - « ) a (29.11)
Stress o t is also positive for all values of p and becomes maximum
at the internal periphery of the disc, where p = a :
(<*/)»«=P t2 + (1~ m) a*J (29-12)
From a comparison of equations (29.11) and (29.12) we can easily
notice that (o()mix is always greater than (<r,)max. Therefore irre­
spective of whether we check the strength of the disc by the theory
of maximum shearing stresses or the distortion energy theory, the
strength condition will be
(o,)„„ = ^ v»’n [2 + (1 - m ) a»J < [o-j (29.13)

Figure 411 shows curves depicting the change in values of o?=o,/p

along the disc radius for values of a between 0 and I and for p=>0.3.
We note that the maximum values of a,, (29.12), (at the internal
periphery of the disc) do not change much with the value of the
hole’s radius, i.e. with a (curve ab). At a « 0 , i.e. when the radius
of the central hole is very small, there is a sharp change in the value
of a t at the hole edge due to stress concentration (curve acd). Under
these conditions
M „,,= = 2 ,> = 2 ± K y«>V; (29.14)

In a very thin circular ring, where and a « l #

(29 15)

IB*
532 Dynamic Action of Forces [Part IX

which is the same as obtained in § 166. In this case the maximum

value of ai, <29.15), is only 20% greater than (ot)ntx for a disc with
a very small hole (29.14).
It is evident from formulas (29.9) and (29.10) that stresses ar and a,
increase sharply with the increase in the peripheral velocity u=<or,.
It should be noted that besides velocity v and mechanical properties
of the material p and y these stresses depend only upon dimension­

Fig. 411

less quantities p and a. Henceor and o» will be equal in two geomet­

rically identical discs having same p’s. This property enables us to
replace the actual testing of large discs by testing of small models
in the laboratory.
In a solid disc, a r and a, are equal at the centre, where r= 0. A com­
parison of formulas (29.7) and (29.8) indicates that this condition
can be satisfied only if the constant of integration B is equal to zero.
The other constant, A , can be found from the following condition:
at r —ri, i.e. at the external surface of the disc, stress crf must be zero.
Therefore

Substituting the above value of A and B^O in formulas (29.7) and

(29.8), we obtain
<V= P ( 1 - P 2) (29.16)
Ch. 29] Forces of Inertia. Stresses due to Vibrations 533

and
or, = /? (1 — /tip4) (29.17)
The corresponding curves showing variation of <t?= j and 0 /=
along the disc radius are given in Fig- 411 (curves fk and fed). Both
stresses are positive for all values of p and increase towards the
centre. At p=0
(0r)».x = (<*/)m.x = P = ^ (29-18)

Thus, in a disc with a very small central hole, stress o, at the edge
of the hole is two times greater than at the centre of a solid disc on
account of stress concentration (see formula (29.14)).
The above discussion was based on the assumption that the edges
of the disc are free of external loading. This assumption generally
does not correspond to reality. Usually the disc is mounted on the
shaft in the hot state or by a hydraulic press with an interference
fit, which ensures that deformation of the disc hole due to centripetal
forces is always less than the deformation of the opposite sign in­
curred during mounting, i.e. the disc sits tightly over Ihe shaft in
normal working. The external periphery of the disc is usually fitted
with a rim for mounting turbine blades; during rotation the rim gives
rise to additional centripetal forces which are transmitted to the disc.
Thus the internal and external peripheries of the disc are subjected
to uniformly distributed tensile or compressive forces. The stresses
caused by these forces may be computed by the formulas derived in
the analysis of thick-walled cylinders (formulas (25.9), § 144). Upon
adding the stresses obtained from formulas (25.9), (29.9) and (29.10)
we can draw a complete diagram depicting distribution of stresses in
a rotating disc.

§ 169. Disc of Uniform Strength

The formulas derived in the preceding section and the curves drawn
in Fig. 411 show' that there is considerable variation in the values
of <jr and or* along the radii of discs of uniform thickness. The most
non-uniform distribution of stresses occurs in discs of uniform thick­
ness with a central hole. The design of such discs is based on the
maximum stress <r, at the inner edge of the disc, which imposes re­
strictions on the limiting value of maximum velocity. For achieving
high velocity of rotation the discs have to be made of variable thick­
ness which decreases from the centre towards the periphery. The most
economical shape of the disc is one in which the same stress acts on
all points of the disc. Such discs are known as discs of uniform strength.
While designing such discs it is assumed that the stresses remain
534 Dynamic Action of Forces [Part IX

constant over the thickness of the disc; this generally gives a small
error in the calculated stress value.
The basic formulas for designing discs of variable thickness can
be derived as before by consid­ ?J
<srV‘d8 *i{SrirdB) ering the equilibrium of an ele­
ment abed. (Figs. 410 and 412) of
the disc. Let us denote by z the
variable thickness, which is a cer­
tain function of the radius, r.
Faces ad and be of the element
cut by meridian sections are act­
CfZdr ed upon by forces o*z dr mak­
ing an angle d6 with each other,
face dc of the element is acted
upon by a radial force OfZrdQ di­
rected towards the centre and
face ab is acted upon by a radial
Fig. 412 force a,zr d0+d (a^r d8) directed
from the centre towards the outer
surface of the disc. To these forces we must add the force of inertia
due to the mass of the element,
td r n » 2 &
8

acting from the centre towards the periphery.

Projecting all the forces enumerated above on the radius, we get
the following differential equation for the equilibrium of a disc of
variable thickness:
d{crzrd&)—atzdrd 0 -i-zdr d& =0

or
± (rzo r) - z o t + z & - ^ = 0

If z=const, the above equation transforms into equation (29.4) de­

rived in the preceding section.
In a disc of uniform strength stresses ar and cr, are constant at all
points and are equal. Equating their value to the permissible stress
la), we can write the following equation of equilibrium:

or
\_&z y<o8r 2 nr
2 dr Io]g
C7». 29] Forces of Inertia, Stresses due to Vibrations 535

where

Upon integrating the above equation, we get

z — Ce-nr%
where C is a constant of integration. If the disc does not have a cen­
tral hole then from the condition z=z 0 at r—0, it ensues that C=a0.
The thickness of the disc at the centre {z0) is determined from con­
ditions at its outer surface.
A solid disc of uniform strength can be used even at very high
peripheral velocities. However, from the point of view of convenience
of manufacturing, discs of variable thickness with central holes are
generally used. These discs, which in shape are close to discs of uni­
form strength, provide the most advantageous distribution of stresses
along the radius. The methods of analysis of such discs are discussed
in special courses.

§ 170. Effect of Resonance on the Magnitude

of Stresses
In the first two problems discussed in §§ 165 and 167, the acceler­
ation was assumed to be fixed in direction w.r.t. the element on
which it was acting; in the last example the acceleration was continu­
ously changing its direction through 360° during one rotation of the
wheel. In this case the srtesses and deformations periodically changed
their sign resulting in vibrations of the body.
A similar situation will arise if the beam is loaded with a machine
which has a rotating load having eccentricity w.r.t. the axis of rota­
tion (Fig. 413). The force of inertia of the rotating load will give
rise to stresses and deformations in the beam which periodically change
their sign. The beam will begin to vibrate with a period which is
equal to the period of rotation of the load. These vibrations are known
as forced vibrations. If the period of forced vibrations is the same as
the period of natural vibrations of the beam, then resonance occurs
and the amplitude of vibrations increases sharply with the passage
of time. The amplitude of vibrations is in fact restricted by frictional
forces and resistance of the atmospheric medium. However, despite
these restrictions the amplitude may assume large values, which far
exceed the deformations the beam would have experienced under the
same acceleration of constant direction.
There was a case when due to resonance the angle of twist of a shaft
increased six-fold as compared to the angle before resonance. This
happened with the crankshaft of a motor of the airship Graf Zeppelin
in its very first flight across the Atlantic ocean.
536 Dynamic Action of Forces [Part IX

Thus, if resonance is not curbed at the very outset but is allowed

to continue for some time, it results in a gradual growth of deforma­
tion and a corresponding increase in stresses ultimately leading to
failure. Therefore, at the design stage it is essential to prevent reso­
nance in structures which are subjected to variable acceleration of
constant period.
Since tneperiod'of theexciting forces is generally given, the design­
er can control only the period of natural vibrations of the structure:

Fig. 413

the period of natural vibrations should be selected in such a way that

it differs considerably from the period of the exciting forces.
Questions concerning the determining of period, frequency and
amplitude of natural and forced vibrations are discussed in theo­
retical mechanics.* Therefore, below (§ 171) we shall apply without
proof the results of theoretical mechanics in determining stresses and
checking the strength of elements of structures subjected to vibration.

§ 171. Determination of Stresses in Elements

Subjected to Vibration
A. An elastic system disturbed from its stable state of equilibrium
begins to vibrate. The vibrations occur near the position of elastic
equilibrium in which the loaded system experiences static deformation
6 g and a corresponding static stress p a (o* or x„ depending upon the

nature of deformation). In a system subjected to vibration, to the

static deformation is added dynamic deformation which depends
upon the type of vibrations and their amplitude. This results in a
change in the value of pH. Hence, while checking the strength of a
vibrating system it is essential to determine the dynamic deformation
and the corresponding stresses in addition to the static deformation
and stresses.
In a number of cases the nature of vibrations of the system can be
completely defined by one quantity (one coordinate). Such systems
are known as systems with a single degree of freedom; the examples
of such systems are a light stretched or compressed spring with a weight
suspended at its end performing longitudinal oscillations, a beam of

* See, for instance, L. G. Loitsyanskii and A. I. Lurye, A Course of Theoretical

Merhanirx Owtpkhizdfti IftRS fin Piwsiont Port II.
Ch. 29) Forces of Inertia. Stresses due to Vibrations 637

small (as compared to Q) dead weight (shown in Fig. 414) performing

oscillations in a direction perpendicular to its own axis, etc.
In vibrating systems having one degree of freedom, the total de­
formation of an arbitrary section may be obtained by adding the static
and dynamic deformations. Obviously, the strength check should

Fig. 414

be carried out for the section where the total deformation is maxi­
mum. In the simplest cases the total deformation is obtained by add­
ing the maximum static deformation 6flllwx and the maximum am­
plitude A of the vibrations:

A = « . » . „ (\ 1 + T -m ax
- )/ = Kofi, (29.19)

As long as the system deforms within the elastic limits, the stresses
are directly proportional to strain. Therefore

p * (l -* -°s* m
*ax
- )/ = K dP, (29.20)

K b — 1 + 7'■f“^— (29.21)
m ax
is the dynamic coefficient under vibration. The strength condition in
this case is as follows:
Pd = K dPs < [ p\ (29.22)
Thus, as in the previously discussed problem, in which forces of
inertia of constant direction were considered, the determination of
dynamic stresses and strength check under vibration can be replaced
by the determination of static stresses and dynamic coefficient
Since K n depends upon A , we must know how to determine the
maximum amplitude of vibrations under different situations.
It is well known that the differential equation of an oscillating
load Q performing natural vibrations, may be written in the form of
an equilibrium equation which takes into account the forces of iner­
tia in addition to the external force (load Q) and the force of elastic
638 Dynamic Action of Forces [P art I X

resistance of the system:

| / + P - ( ? = i / + / )l = i / + a = 0 (29.23)
® 9 9

Here x is the coordinate which completely determines the location

of load Q during vibration (see, for example, Fig. 414), P is the total
elastic resistance of the system to vibrations, P — Q=Pt is the
restoring force (an additional elastic force which appears in the sys­
tem due to the displacement by dx of the point of application of
force Q on account of vibration) which in the first elastic limit may
be considered proportional to coordinate x (F,=cx), and c is a propor­
tionality constant which is equal to the force required to cause unit
static deformation of the system in the direction of force Q. If the
static deformation due to load Q is 6q, then c= £ - .
After solving equation (29.23) we get the following formulas for
calculating frequency a)0 and period of the natural vibrations /0:
to- /f= V i ■

Hence, natural vibrations of a weightless body are equivalent to

simple harmonic motion with a frequency (period) equal to the fre­
quency (period) of oscillation of a simple pendulum which is equal
in length to the static deformation of the system due to load Q. For
instance, if the load stretches a prismatic bar,

In the case of a simply supported beam loaded by a force Q acting

at the middle of its span,

B. If in addition to force Q and force of elastic resistance P the system

is acted upon by an exciting force F and force of resistance of the at­
mospheric medium, R , then the differential equation of vibration
may be written in the form of an equilibrium equation similar to
equation (29.23):

= | * " + « - F + /? = 0 (29.24)

In a sufficiently large number of cases the resistance of the atmo­

spheric medium, R, may be considered directly proportional to the
CH. 29] Forces of Inertia, Stresses due to Vibrations 639

velocity of the vibrating body: R= rx'. If the exciting force varies

according to the sine law
F = H sin at
where H ^ F ^ x and g> is the frequency of the exciting Force, then
equation (29.24) may be written as follows:
—X? + r x '+ cx = H sin ait
8
or
x* -{- 2 nx’ -}-ojjlx » ^ sin <ot (29.25)

Here n = ^ is the damping coefficient and o>0 is the frequency of

natural vibrations which occur in the system even when the exciting
force, F, and the force of resistance, R, are absent.
After solving equation (29.25) we get the following expression for
amplitude A of tne forced vibrations in the presence of damping:
ff
A
— Y ( ^ —<*>*)* 4n*6>*
s

(29.26)

Here
gH gH 6 q h

is the static deformation of the system due to maximum exciting

force F (Fmi%—H). The ratio of amplitude A of the forced vibrations
to deformation 8 H is called the amplification factor of vibrations and
denoted by 0:

Therefore formula (29.21) for dynamic coefficient K d may now be

written as
(29.28)
* 'i max °Q
540 Dynamic Action Qf Forces [Part IX

The amplitude of natural vibrations has not been accounted for

in the above expression, because it can have appreciable effect only
iit the beginning of vibrations; in presence of a resisting medium it
sharply decreases with the passage of time.
Figure 415 contains curves depicting the variation of amplification.
factor |i as a function of 77 for various values of the damping coef-
, i\ u°
ficient n ratio —J . If the frequency of the exciting force is close to
the frequency of natural vibrations, i.e. if ^ - « I , and if the damping
coefficient is not large, then the denominators in formulas (29.26)
and (29.27) for determining A and p will be very small and the am­

plitude and amplification factor will be very large (Fig. 415). Under
such circumstances even a small exciting force will result in high
stresses (on account of resonance).
With the increase in damping resonance becomes less effective. Jt
should, however, be noted that damping can considerably decrease
the amplitude of forced vibrations only under near resonance con*
at. 291 Forces of Inertia. Stresses due to Vibrations 541

ditions (0.75^-^^1 .2 5 ); the effect of damping is imperceptible if

the ratio — is outside this range.
It is evident from formulas (29.26), (29.27) and (29.28) and Fig. 415
that if frequency to of the exciting force F is very low, then the am­
plitude of vibrations tends to 6 H, the amplification factor tends to
unity and the maximum stress can be calculated as the static stress
due to load Q and maximum value of exciting force F (Fmax^-H).
If the frequency of the exciting force is very high, then the ampli­
tude of vibrations and the amplification factor tend to zero and force
Q may be considered as a fixed load. The maximum stress in the
system is in this case equal to the static stress due to load Q.
This is a factor of great practical importance; it is employed in
the design of various types of dampers, seismographs, vibrographs
and other instruments. In machine design the shock absorbers pro­
tecting the foundation from vibrations are selected in such a way
that the frequency of natural vibrations of the machine mounted on
the absorbers is considerably less than the frequency of the exciting
force.

§ 172. The Effect of Mass of the Elastic System

on Vibrations
A. If the vibrating system carrying load Qhas a sufficiently distrib­
uted mass (meaning thereby that the number of the degrees of free­
dom is large), then the. simplified calculations discussed in the pre­
ceding section will give a considerable error. In such cases the differ­
ential equations of motion should be written with the mass of the
system being taken into account. Instead of solving such problems
from equilibrium conditions, on the basis of which equations (29.23)
and (29.24) were obtained, it is more convenient to solve them using
the law of conservation of energy.
Assuming that the energy imparted to a system in disturbing it
from its stable state of equilibrium is equal to the sum of kinetic
and potential energies of the load and elastic system and is constant
for natural vibrations, we get the following equation:
U + T = const (29.29)
This equation shows that vibration is accompanied by continuous
transformation of one type of energy into another without any loss.
When the elastic system occupies one of the extreme positions, where
the velocity of vibrations is zero and consequently the kinetic energy
is zero (7*=0), the potential energy of the load and system is maxi­
mum, 0 = Um3X. On the other hand, when the system is in equi­
librium, 0 and T*=Tmax.
542 Dynamic Action, of Forces | Part IX

It should be noted that the principle applied in deriving equation

(29.29) is applicable only to systems with a single degree of freedom,
because the law of conservation of energy does not take into account
the heat transfer, which occurs in systems with a big number of de­
grees of freedom. Hence, the problem of vibration of systems with
more than one degree of freedom is reduced to the fundamental prob­
lem discussed in § 171, and we can approximately determine only
one (principal) frequency of the
natural vibrations.
Let us discuss a few examples on
application of equation (29.29).
B. As the first example we shall
k study the vibrations of load Q sus­
h
f*7—$ —*■
pended from an end of a prismat­
ic bar of length /, cross-sectional
el area A and specific weight v (Fig.
416). If the suspended load is dis­
turbed from the state of equilibri­
um and left to itself, it starts per­
Q— forming longitudinal vibrations
____ 1 about the position of equilibrium.
Let us write down expressions for U
Fig. 416 and T for the vibrating load-bar
system.
Potential energy of the system changes by U = U (—Ut w.r.t. the
potential energy in equilibrium; here U« is the potential energy of
the system at the initial moment (in equilibrium) and Ut is the po­
tential energy at instant t.
Let us denote the potential energy of load Q at the initial moment
by Uq', potential energy of the bar at the same moment Is equal to
where A/, is the static deformation of the bar due to load Q.
Hence
u ,= u 9 + 2 g *

At instant I when the load has lowered by a distance x and the bar
has received additional deformation x, the potential energy of the
load decreases by Qx, whereas the force of elastic resistance and static
deformation of the bar increase by times. Consequently,

M s+ x
u ,= u 9 - Q x + I q ^ ± m ,
Uq- Q x + 9^Al,-+ Q x + 2Qx*
AI, U |
Ch. 29] Forces of Inertia. Stresses due to Vibrations 543

and
U = Ut— U 0 =-Qj- (29.30)

Kinematic energy of the system is the sum of the kinetic energy

Ti of the load and Tt of the bar. Kinetic energy of the load is 7 \—
— While calculating the kinetic energy of the bar, we must
bear in mind that at instant t the load and consequently the lower
face of the bar are moving with velocity x \ whereas the velocity of
the upper face of the bar is zero. The velocities of intermediate
sections will be within these two extremal values.
Let us assume that displacement of bar sections w.r.t. the fixed
end follows the same law as in static tension, i.e. it is directly pro­
portional to the distance of the section from the fixed end. Thus
(Fig. 416), if the lower face gets displaced by xt then the section at a
distance £ from the fixed end must get displaced by x \ , and the
velocity of this section will be . The kinetic energy of an element
of length d% cut at a distance £ from the fixed end will be dT2^

^ * '* ( f ) ’•
Kinetic energy of the whole bar will be the sum of quantities dTa
over the bar’s length:

T - h ± £ x '*
ia~ j \i) 3
o
Thus, the kinetic energy of the bar is equal to the kinetic energy
of a concentrated load of mass i.e. it is equal to the kinetic
energy of a load whose mass is ‘/, of the bar and which moves with
the same velocity as the bar. The total kinetic energy of the load-bar
system is:
ja i\
(q 3 )

Substituting T and U (29.30) in equation (29.29) and differentiating

the last with respect to t, we get

i («+T
-r)*"+¥■*-7 (c+sr)*"+Tkx- °
or
til Qq . yA l
544 Dynamic Action of Forces {Part IX

Here Al f is the static deformation due to load < 2 + ^ . The differ­

ential equation obtained above by taking into consideration the
mass of the vibrating bar differs from equation (29.23) only in the
factor before x and both equations become identical once the mass
of the bar is ignored. Therefore, the correction due to mass of bar,
which must be introduced in the calculations of the preceding section,

Fig. 417

consists in determining static deformations, required for calculating

the frequency of natural vibrations, not for load Q but for a load <2
plus one-third of the weight of the bar. Thus, the weight of the vi­
brating bar reduces the frequency of natural vibrations and increases
their period. The quantity ^ is called the reduced mass of the bar.
C. As the second example we shall study a simply supported beam,
loaded by force Q at the middle of its span (Fig. 417).
Let us denote the maximum static deflection of the beam due to
load Q by and the variable deflection of the middle
section due to vibrations by z. Let us assume that the additional de­
flection of the beam due to vibrations varies along its length In the
same manner as due to the static load <2; the variation occurs according
to the following equation (see § 85):

y = 4%’j 31,11r <31‘x - 4 * a> (29- 31)

Thus, if with respect to the position of static equilibrium z is the

additional displacement of the middle section due to vibrations,
then the displacement of the section at a distance x from the left
support will be
i/ = - ( 3 /4x —4x3)

The velocity of vibration of the centre of gravity of this section will be

Ch. 291 Forces of Inertia. Stresses due in Vibrations 545

The kinetic energy of an element of length dx of the beam will be

<*rs = y A d x 4*V
~W \ $ )
and the kinetic energy of the whole beam will be
i/s
T a= 2 l d z ' ajl j (3/**—4jc’)ad* = | g ^ Z'* (29.32)
8 o
Kinetic energy of load Q is:

Since potential energy of bending is calculated by the formula

As the middle section gets displaced by z from the position of static

equilibrium, Therefore

Substituting the values of U and T=Ti-\-Tt in equation (29.29)

and differentiating it with respect to t t we get
48LV e z = z '+ fz = 0
2" 4 P It

It is evident from the above expression that the beam should be

17
considered weightless and ^ =0.486 of its weight should be added
to Q to account for its mass while determining the frequency and
period of natural vibrations; the quantity is called the re­
duced mass of the beam.
Let us point out that if the deflected beam be approximately con­
sidered as corresponding to the sine curve y - f sin j , then the re­
duced mass will be not but j — , which is sufficiently close
to the actual value,
546 Dynamic Action of Forces [Part IX

The reduced mass thus determined has been obtained on the assump-
tion that the mass of the beam is small as compared to Q, because we
have neglected the effect of mass of the beam on its deflection. Equa­
tion (29.31) of the deflected beam axis corresponds to a situation
when it is loaded by a single concentrated force acting at the middle
of its span.
D. Let us now consider the other extreme case, when the mass of the
beam is very large in comparison with Q or when the vibrating beam
is loaded by a continuous uniformly distributed force of intensity q
(which includes the weight of the beam). The equation of the deflected
beam is as follows (see § 86):
y= — ^ ( ^ - 2 / * » + * « ) = _

where / is the deflection at the middle of the span.

The kinetic energy of an element of length dx at a distance x from
the left support may be expressed through velocity z‘ of the middle
section by the following formula:

‘" ' - it *'’

Total kinetic energy of the beam is:
/
Si*'* 3968
2g 7875

Potential energy of the beam

i . t
n B J (* EJ C r 192 z . . . .1 * .3072572*
u — r ] U S )* ” -rj is -jr-

Substituting these expressions for U and T in equation (29.29) and

differentiating it with respect to t, we get:
48EJg
2 = 0

$ « ') p
The reduced mass of the beam in this case is:
31 _ £
§3 g
0.492 —
B
Thus, the equation of the deflected beam axis does not have much
effect on the period of natural vibrations, as long as the general na­
ture of deflection does not change.
Ch. 29] Forces of Inertia. Stresses due to Vibrations 547

If a simply supported beam deflects along a curve that has no points

of discontinuity,*then the curve may be assumed to be a sine curve
of half wave length y = f sin™ , and the reduced mass of the beam
may be considered equal to ^ 2 / .
Thus, while determining the first frequency of natural vibrations
of a system of distributed mass, the system may be assumed to be
weightless and its reduced mass added to the mass of the concentrated

Fig. 418

force acting on the system; the “reduction” method holds well even
in such cases of loading when Q=0.
Example. A non-uniform bar of length I carrying load Q at one end
(Fig. 418) rotates with angular velocity © about an axis to which its
other end is fixed. The distance between the centre of gravity of load
Q and the axis of rotation is r. Find a relation between the cross-
sectional area Ax and distance x of the section from the free end if
stresses in all the sections are equal to lo}. Specific weight of the
material is y.
Each point of the bar with abscissa * experiences centripetal accel­
eration ©*(/ — x). Therefore, to determine the stresses all elements
of the bar must be loaded by forces of inertia acting away from the
centre and equal to the mass of element multiplied by ©*(/ — x).
An element of length dx cut by two adjacent sections with abscissas
x and x+ dx and cross-sectional areas Ax and Ax+dAxt is acted upon
by the force of inertia &&£(/ — *)©*, where g is the acceleration
of gravity.
The size of the sections should change in such a way that this force
of inertia give rise to stress lol on the area dAx (see analysis of uni­
form strength bars under tension and compression, § 25). We obtain
the following differential equation for Ax:
[a]dAx = - ^ - ( l - x ) ^

After separating the variables and integrating, we get

548 Dynaniic Action of Forces [Part IX

Constant C can be determined from the boundary condition at x=0:

Ax—At (cross-sectional area of the end faces of the bar). This area
depends upon the force of inertia of load Q, which stretches the end
face element:

Substituting In A0=C in the expression for A xt we get

ln A* ~ {2l~ ~ + ,n A•’ or A*= A»WP [ i f R ■

$ x]

C H A PT E R 3 0

Stresses Under Impact Loading

§ 173. Fundamental Concepts
Impact takes place when the velocity of the element under consid­
eration or of elements adjoining it changes in a very short period
of time.
In piling, a heavy load falls on the upper face of the pile from a
certain height and drives it into the soil; the drop weight comes to a
stop instantaneously, causing impact. A similar phenomenon takes
place during forging; both the forged part and the hammer head
experience impact as the latter comes to a sudden stop when it hits
the part to be forged. During impact high pressures are created be­
tween the colliding bodies. The velocity of the falling body changes
over a short period and in particular cases fails to zero as it comes
to a stop. This means that the hammer head is subjected to a large
acceleration from the forging in a direction opposite to that of its
movement, i.e. the hammer head experiences reaction P D which is
equal to the product of its mass and the acceleration.
Denoting this acceleration by a, we can write reaction P D—~ u ,
where Q is the weight of the falling body. In accordance with the law
of equality of action and reaction a force of the same magnitude acts
on the forging in the opposite direction (Fig. 419). These forces give
rise to stresses in both bodies. Thus, the forging experiences stresses
as if it wore being acted upon by the force of inertia of the hammer
head; these stresses may be calculated by considering (§ 164) the force
of inertia P» as a static load acting on the structure. The chief dif­
ficulty is how to compute this force. We do not know the duration of
impact, i.e. the lime during which the velocity falls to zero. There­
fore, acceleration a and consequently force Py remain unknown.
Hence, although computation of stresses under impact loading is a
Ch. SOJ Stresses Under Impact Loading 549
particular part of the general problem of taking into account forces of
inertia (§ 164), a different method based on the law of conservation
of energy has to be employed to calculate P D, stresses and defor­
mation.
During impact there is a sudden transformation of one type of
energy into another: kinetic energy of the moving body is transformed
into potential energy of deformation. By expressing this energy as a
function of force P ti (stress or defor­
mation (§ 98)), we can determine
these quantities. □
Engineering problems are gener­
ally solved by the theory of elastic
impact, which makes use of the
following main assumptions:
(1) The kinetic energy of the
5 m
striking body completely changes
into potential energy of deforma­
tion of the body which is hit; we
ignore the. energy that is spent on
deforming the striking body and
the base on which the hit body is mmmm
placed.
(2) The distribution of stresses Fig. 4J9
and strains over the volume of the
hit body remains the same as under static loading; here we ignore the
change in distribution of stresses and strains at the point of collision
and also the stresses and strains arising from high-frequency vibra­
tions which appear in the whole volume of the body due to impact.
The first assumption usually leads to a higher safety factor being
specified, as the hit body is assumed to be in worse conditions than
it really is; the second assumption does not add to the safety factor
for the more stressed parts of the hit body.

§ 174. General Method of Determining Stresses

Under Impact Loading
A. Imagine that a rigid body A of weight Q whose deformation may
be neglected falls from a certain height H and hits another body B
which rests on an elastic system C (Fig. 420). As a particular case we
may consider a load falling on the face of a prismatic bar, the other
end of which is rigidly fixed (longitudinal impact) or a load falling
on a simply supported beam (bending impact).
Elastic system C undergoes deformation during a very short period
of time. Let us denote by 6© the displacement of body B (whose own
deformation may be neglected) in the direction of impact. In the
particular cases enumerated above displacement 8^ represents axial
550 Dynamic Action of Forces [Part IX

elongation &tD in the case of longitudinal impact and deflection

f D of the section of impact in bending impact. As a result of the im­
pact, system C experiences stresses pD (<*d or *d, depending upon the
type of deformation).
Assuming that kinetic energy T of the falling body is completely
transformed into potential energy
of deformation, UD, of the system,
we may write
T = Ud (30.1)
By the time of completion of de­
formation the falling body covers
a distance therefore its ki­
netic energy can be expressed in
terms of the work WD done by it:
T = WD= Q (H + d D) (30.2)
Let us now calculate UD. If the
deformation is static, potential en­
ergy U» is numerically equal to half
the product of the acting force and
corresponding deformation (§ 98):
(30.3)

Static deformation 8, of the section of impact may be calculated

by Hooke’s law and in general is written as
or Q = c6.

Here (see § 171) c is a proportionality factor (sometimes also know

as rigidity of the system); it depends upon the properties of material,
shape and size of the body, type of deformation and location of the
section under impact. Thus, in simple tension or compression 64=
= A /4= ^ a n d c = in bending of a simply supported beam loaded
at the middle of itsAii span by a concentrated force Q, static defor-
ioc /
mat ion 6a= / 8 m#x= and c = ^ —; etc.
Thus, formula (30.3) may be rewritten as follows;
U ,= T Q 6 .= |- 6 ;

This formula is based on two assumptions: (a) Hooke’s law must

be applicable, and (b) force Q, stress pe and corresponding deforma­
tion 6, must increase gradually from zero to a finite value.
Ch. 80\ Stresses Under Impact Loading 551

Experiments on determination of the modulus of elasticity in

bars subjected to vibrations within the elastic limits show that
Hooke’s law remains valid and the modulus of elasticity remains un­
affected by the dynamic nature of loading. Of the nature of increase
of stresses it must be said that although the increase is fast it is not
instantaneous even in the case of impact loading: dD increases grad­
ually during a very short period of time from zero to a finite value,
and the increase in stresses p D runs parallel with the increase in de­
formation.
Reaction of system C to the falling weight Q (let us call it P D)
appears as a result of the development of deformation 6 D. It in-
erases with 6D from zero to a finite maximum value and, if stress
p D does not exceed the limit of proportionality, is related to it by
Hooke’s law:

where c is the same proportionality constant, which retains the same

meaning under impact loading also.
Thus, both conditions necessary for the validity of formula (30.3)
are satisfied by impact loading too. Consequently, it may be consid­
ered that the formula for UD under impact loading must be the same
as obtained by loading system C with a static force of inertia P D, i.e.

Ud = T />i >6o = T 6b “ ^ : 6“ (3°.4)

(here, as before, we consider c—Q.16,). Substituting the values of T

and Uj> in equation (30.1), we get
Q (ff+ 6 ® )= ^ « > (30.5)
or
6o—26,6^—2//6s = 0 (30.6)
wherefrom
8i) = 8 ,± K 6 J + 2ff6J
or, keeping the positive sign before the square root to determine the
maximum deformation in the direction of impact, we obtain

«o=*6. ( ' + ] / l + ^ r ) = K j > 8 , (30-7)

Since according to Hooke’s law stresses are proportional to de­
formation,
Ps ( l 4- Y , + X ‘) ~ KdPs (30.8)
552 Dynamic Action of Forces [Part IX

and ______
P » = c (l + (30-9)
It is evident from these formulas that impact strain, stress and force
depend upon static deformation, i.e. upon the rigidity and longitu­
dinal dimensions of the body under impact. This statement wifi be
proved below with the help of individual examples. Constant

K o=l + J / h ~77 (30.10)

in this case represents the dynamic coefficient. Substituting H in
formula (30.10) by where v is the velocity of the body under im­
pact at the beginning of impact, we get
(30.11)
Besides, since

where To=QH is the energy of the body under impact at the beginning
of impact, the expression for the dynamic coefficient may also be
written as foilows:

/Ca” i + Y i (3tu2)
B. If in formulas (30.7) and (30.8) we put H = 0, i.c. if wc apply force
Q instantaneously, then and Pd= 2 p»; if force Q is applied
suddenly, then the deformation and stress are two times the defor­
mation and stress due to a statically applied load of the same mag­
nitude.
On the other hand, if height H (or velocity u) from which the load
falls is large as compared to 6«, then 1 may be neglected as compared
Off
io j - in the radicand in formulas (30.7) to (30.11). The expressions
for 6j) and may be written as follows:

8 „ = 8 s( l + Y i Y ) ■ PD=P* ( { + Y t :) (3 n l3 )
Of f
If ratio is very large, then the first term in the parentheses may
also be neglected and the expressions are written as follows:

»'’ = p‘ Y ^ (30.14)
Ch. 30] Stresses Under Impact Loading 553

The dynamic coefficient in this case is:

* - > = / ! f30-' 5)
2 H
It should be noted that unity in the radicand be ignored if -^ ^ lO
(the error of the approximate formula will not exceed 5%), but unity
in the radicand can be neglected only for very high values of the
ratio For example, in order to ensure that the error of approxi-
®s oH
mate formulas (30.14) and (30.15) does not exceed 10%, the ratio
must be greats than 110.
Formulas &d= K j£ , and p D=KnP» in which K d is expressed in
terms of (30.12), may also be used for solving the problem on
collision between bodies moving with a certain velocity, for deter*
mining the stresses in the cylinder of an internal combustion engine
due to a sharp increase in gas pressure on account of ignition of fuel,
etc. On this basis these formulas may be considered as general for­
mulas for impact analysis.
Generalizing what has been said above, we can suggest the follow­
ing method of determining stresses under impact. Applying the law
of conservation of energy, we must (1) calculate kinetic energy T
of the body under impact; (2) calculate potential energy Uu of the
bodies experiencing impact, when they are loaded by the inertial
forces (the potential energy may be expressed through stress (cD, Tj>)
in a particular section, through deformation (elongation, deflection)
or through the force of inertia P n of the body under impact; and (3)
equate UD and T and from this equation determine either the
dynamic stress directly, or first determine deformation and then
applying Hooke’s law find stress or force PJ} and finally calculate the
corresponding dynamic stress and deformation.
The method outlined above is based on the assumption that the
kinetic energy of the body under impact is fully transformed into
potential energy of deformation of the elastic system. This assumption
is not very accurate. Kinetic energy of the falling body is partially
transformed into heat and partially into the energy spent on inelastic
deformation of the foundation on which, the elastic system rests.
In addition, if impact occurs at a high velocity, then the defor­
mation of the body suffering this impact does not get enough time
to spread over the whole body, and local stresses of considerable
magnitude, which sometimes exceed the yield stress of the material,
appear in the region of impact. For example, if a steel beam is hit by
a lead hammer, then a large portion of the kinetic energy is trans­
formed into the energy' of local deformation. A similar phenomenon
554 Dynamic Action of Forces [Part IX

may occur even at low velocity of Impact if the body suffering impact
is very rigid or heavy.
All the situations discussed above pertain to a high value of ^ .
It may be stated that the method of analysis described above is ap­
plicable until ^ does not exceed a certain value. Accurate inves­
tigations confirm that the error does not exceed 10% if ^ -d O O .
Now, since this ratio can be expressed in terms of (see earlier
discussion) it may be stated that the above method is applicable
until the energy of impact does not become more than 100 times the
potential energy of deformation due to static loading of the elastic
system suffering impact by a force equal to the weight of the impacting
body. Consideration of the mass of the body under impact (see § 178)
helps in somewhat increasing the limits of applicability of this method
in such cases when the impacting body has a big mass.
A more accurate theory of impact is given in the theory of elas­
ticity.

§ 175. Concrete Cases of Determining Stresses

and Conducting Strength Checks Under Impact
A. The formulas derived in § 174 show that qualitative changes may
occur due to a quantitative change in the period of the force acting
on a body.
Let us study some simple cases of deformation under impact loading.
In this study we shall determine the dynamic coefficient with the help
of formulas (30.10) and (30.12) and the approximate formula (30.15).
We shall determine 5^, p D and P D by the following relations:
Pd ^ K oP^ Pd ^ K dQ
In the case of an axial tensile or compressive impact (Fig. 421),
Q_ Q*l _ o\A l U$EA
6, = Als Ps A ’ Us = Wa “I F " * 5 21
Dynamic coefficient K d may be calculated from one of the following
expressions:

i+ / '+ x - - i+ / i + iW -

= 1+ / ,+ ^ - = I+ | / I+ ^ T <30I6>
Having calculated K d we can easily determine oD and P D and
Ch, 301 Stresses Under Impact Loading 555

The approximate formula for determining stresses in this case is

as follows:

Kd = ~ Y ^ ^ . o0 = o, K „ = Y ~ ^ - {XA7)
It should be noted that under static as well as dynamic loading
the stresses in the compressed bar depend upon the compressive force
and the cross-sectional area of the bar.
If load Q is applied statically to the bar, then the force transmitted
to the bar is equal to Q and does not depend upon its material and
size. In the case of impact loading P D, wnich gives rise to stresses in
the bar, depends upon the accelera­
tion with which the body suffering
impact resists the impacting body,
i.e. P D depends upon the time n o
during which the velocity of the
impacting body changes. This pe­ 1H____
riod depends upon the axial dynam­
ic deformation A/D and upon the
pliability, of the bar material. The
Alp
r~ T I
greater the pliability, i.e. the smal­
ler the modulus of elasticity E and
the greater the bar length /, the v m m 1r n
longer is the duration of impact and
the smaller are acceleration and
force P D‘
Thus, if stress distiibution is Fig. 421
uniform in all the sections of the
bar, dynamic stresses decrease with the increase in cross-sectional
area and pliability (i.e. increase in length and decrease in modulus
of elasticity E). Only due to this reason springs placed between im­
pacting bodies are able to damp the impact. The formulas derived
earlier express the same idea. For example, formula (30.17) with
certain approximation expresses the idea that in longitudinal impact
the stresses depend not upon the cross-sectional area, but upon the
volume of the bar.
Having determined the dynamic stress from formulas (30.8) and
(30.16) or (30.17), we can now write the strength condition as fol­
lows:
<JD^ \?d\ (30.18)
where lo^l is the permissible normal stress under impact, which for
a ductile material is equal to The safety factor k D may
be considered equal to the primary safety factor &0 under static load­
ing (i.e.» 1.5-1.6; § 16), because the dynamic nature of loading has
55$ Dynamic Action, of Forces |Port IX

already been accounted for in formulas (30.16) and (30.17). However,

keeping in mind the not too accurate theoretical basis of their deri­
vations, a slightly higher value, up to 2, of the safety factor is iin-
ployed. In addition, a better material is generally used in such cases
(more uniform material having better plastic properties).
B. In bending, static deformation which represents static deflec-
tion /, of the beam in the section of
impact, depends upon the type of
loading and constraints.
Thus, in a simply supported beam
of span / experiencing impact at
its middle from a weight Q, falling
from height H (Fig. 422(a)), we
get:

■9} and
Fig. 422

In a cantilever experiencing impact at its free end from a falling

weight Q (Fig. 422(6))

and
IJ - Q / t mix QSP
* 2 liCJ

Substituting the values of 6s= f#m8X and U» in the expression for Kr>*
we first determine K d and then through it the dynamic stresses and
deformation. For example, o Dm;iX for a simply supported beam can
be calculated by the following formula:

(* + (3o.i9)

Strength condition (30.18) may, in this case, be written as

Ch. 30\ Stresses Under Impact Loading 657

In case of impact on a simply supported beam (Fig. 422(a)) the

approximate formulas for calculating ffnmax and fo max are as follows:

(30.21)
and
(30.22)

Identical expressions for f D,nix and ODmax can be obtained in case

of impact on a cantilever (Fig. 422(b)). Keeping in mind that

zmix
and
J »I
) ~X

we can modify formula (30.22) as follows:

- . / 67\E (30.23)
Oom*x V Al
From the approximate formula (30.23) it is obvious that the dynamic
stresses in bending depend upon the modulus of elasticity of the beam
material, volume of the beam, shape of cross section (ratio I f * j
and the type of loading and constraints (in this particular example,
the radicand contains 67\>; in beams loaded and constrained in a
different manner the numerical constant before To will be different).
Thus, in a rectangular beam of height h and width b, the dynamic
stress will be the same irrespective of whether the beam is placed on
the thin or flat face and its magnitude is (according to the approxi­
mate formula): _____
- "D m a x =
"i/*
Y — JT "

because in both cases

h b

It should be recalled that under a similar static load the maximum

stress in a beam placed on its flat face is y times more than the stress
558 Dynamic Action of Forces [Part IX

in a beam placed on its narrow face. Obviously, the above statement

is true only if the impact occurs within the elastic limits.
The resistance of beams to impact loading also depends upon their
section modulus and rigidity. The greater the pliability (deformabil-
ity) of a beam, the greater is the kinetic energy of impact which it
can absorb at the same permissible stress. Maximum deflection oc­
curs in a beam in which the maximum stress is the same in all sections,
i.e. in beams of uniform strength. Such beams are capable of with­
standing greater deflection than uniform beams having the same
permissible stress; this means that uniform strength beams can ab­
sorb greater amount of impact energy. Precisely for this reason, springs
are made in the shape of uniform strength beams.
C. Let us now study the problem of determining stresses under a
twisting impact. If a rotating shaft is suddenly stopped by applying
brakes at one of its ends and the other end is acted upon by force T%
of the flywheel which twists the shaft, then stresses in such a shaft
can be determined by the method explained above. The shaft is twis­
ted by two force couples (the force of inertia of the flywheel and
the frictional force of the brakes) each of moment M.
In this example
st A1 M
“ 9a *= » Ps = ’*s m« = - ^
and
U - - *sm e*W pl <?sG JP
2GJp 6
2 jp “* 21
Therefore

Y v j - V W <30-24)
and ' P

Pd - To«™= K d-**m„“ ».* j / l t “ =2

(30.25)
because
j - nr* tv — nr* jp — 2 _ 2
P 2 ’ wp 2 ’ "W p ~ nr* ~ A
Keeping in mind that kinetic energy T, of the flywheel is:

where J 0 is the moment of inertia of the flywheel, and to its angular

velocity, we may write

Y ^ r <30-26)
Ch. 30] Stresses Under Impact Loading 559

It should be noted that even under twisting impact the maximum

dynamic stresses depend upon the modulus of elasticity and volume
of the shaft.

§ 176. Impact Stresses In a Non-uniform Bar

It was explained in § 175 that the volume of the bar should be
increased to reduce the stresses due to longitudinal impact. However,
this is true only when the cross-sectional area of the bar does not change
along its length—the stresses are equal in all sections. The situation
will be entirely different if various
portions of the bar have different
areas of cross section (Fig. 423).
n s,
\
We know (formulas (30.16) or (■iff* ■d, *3
(30.17)) that dynamic stress in lon­ r
gitudinal impact depends upon the 1
cross-sectional area of the bar as fi­
well as its pliability (deformabil- ll J-
ity). The maximum stresses in a l
necked bar (Fig. 423(a)) must, for
example, be determined for the mi­
nimum cross-sectional area (at the
neck) taking into consideration the vm. mm?.
compressibility of the bar, which m to ( 0
depends upon the deformation of
the whole bar and not only the Fig. 423
neck portion.
Stresses in this case may be brought down in two ways. Firstly,
by increasing the cross-sectional area of the neck portion (if this is
permissible from design considerations) by using a bar of diameter
di (Fig. 423(b)); in doing so we increase the cross-sectional area and
to a smaller extent decrease the compressibility of the bar. There is
a slight increase in the force of inertia, but the cross-sectional area
of the neck portion increases by a higher degree, thus resulting in
an overall reduction in stress.
However, this (first) method cannot generally be applied because
the design of structures demands that the neck be retained. In such
cases the strength of the bar is increased by reducing its cross-sec­
tional area in the thick portion, thus increasing its pliability. If
we reduce the diameter of the whole bar to d t (Fig. 423(c)), we auto­
matically increase its compressibility and consequently decrease
the dynamic force P D as well as the dynamic stress. Thus, a reduction
in the magnitude of stresses may be achieved by two methods, both
of which make the stresses uniform: by increasing the volume with
the addition of material at the neck or by decreasing the volume
with reduction in cross-sectional area of the thick portion.
560 Dynamic Action of Forces fPorf IX

These conclusions can be easily checked analytically. Let us deter­

mine the maximum dynamic stress in each of the three bars shown in
Fig. 423(c), (b), and (c), caused in each case by a longitudinal impact
of energy T 0 =QH. For the bar shown in Fig. 423(c) let At be the
cross-sectional area of the thick portion and A s the cross-sectional
A I
area of the neck; let -r- = q and - r — p . We shall calculate thestres-
ses by approximate formulas (30.14) and (30.17). According to
formula (30.14) the maximum dynamic stress in the bar shown in
Fig. 423(a) is:

PD—aa — as

Since
Qti i Q
liAt ~ EAl |£ - [ p + 9 (< -/0 ]
we find that

Stresses in the uniform bars shown in Fig. 423(b) and (c) may be
calculated from formula (30.17):

° ' - V W = V W ^
Since [p+q(l—p)1<<?<l, we find that oax r cx r ft. For instance,
if —0.8 a n d 0.1, then <7=0.64 and p—0 ,1; after computing
we get cr0=1.52ff& and o e=0.82o„=I.25or6. Thus a neck which re­
duces the diameter by 20% over one-tenth of the total length of the
bar results in a 50% increase in stresses; if the bar is made of a uni­
form section corresponding to the minimum diameter, the stresses
reduce by 20%.
Although 'these calculations have been done on the basis of approx­
imate formulas, the relation established between a„, ab> and a 0
is quite close to the relation which we would have obtained by using
the accurate formula (30.8) for a not very low value of impact energy
T0.

§ 177. Practical Conclusions from the Derived Results

The results of the preceding computations are of great practical
importance. First of all they show that the nature of resistance of
bars considerably differ from their resistance to static deformation.
Ch. 30) Stresses Under Impact Loading 501

Under static compression, thickness of a portion of the bar does no*

affect the stresses in sections of the remaining portion; under impact
it increases these stresses. Reduction of cross-sectional area over a
small length results in a sharp increase in stresses throughout the bar.

fa) (b) (c)

Fig. 424

The endeavour should be to reduce the stresses by increasing the

pliability of the bar by increasing its length, adding a shock-absorber,
using another material of lower modulus of elasticity and using a
uniform cross-sectional area along the bar length. Generally, the
most effective way is to reduce the bar to a uniform diameter equal to
the minimum.
Therefore, while designing bars working under impact loading,
it is essential to have a uniform section all along the bar length;

Fig. 425

greater thickness of some portions is permissible over a small length,

but necking is highly undesirable even over a very small length. If
a sufficiently strong bar cannot be designed under such conditions,
then it is essential either to increase the length of the bar or to increase
its cross-sectional area uniformly.
As an example let us study a bolt transmitting tensile impact
from one part of the structure to another. The design shown in Fig.
424(a) has poor impact resistance, because the threaded length of
the bolt having smaller diameter acts as a neck. The greater part of
the impact energy is absorbed by the threaded portion of the bolt.
The chances "of la ilure are high.
In a properly designed bolt the impact energy should be absorbed
more or less uniformly by the whole bolt; this can be achieved by
making the bolt diameter uniform over the whole (or almost the whole)
19—3310
562 Dynamic Action of Forces [Part IX

length and equal to the minimum diameter of the thread. For this
we may either machine the bolt shank (Fig. 424(6)) or drill a hole
in it (Fig. 424(c)).
As an example of increasing impact resistance of bolts by increasing
their length we may study the design shown in Fig. 425(a) and (6).
The cylinder cover of a boring tool is sometimes subjected to strong
impact from the boring tool. Small bolts securing the cover to the
cylinder according to Fig. 425(a) fail easily. Failure can be prevented
by increasing their length as shown in Fig. 425(6).

§ 178. The Effect of M ass of th e E la stic System

on Im pact

Let us study how the mass of the body subjected to impact affects
the impact stresses. As an example we shall consider impact in bend­
ing (Fig- 422). Weight Q drops on beam A B and at the moment of
impact has a velocity vw= V 2 g H , at the same instant the beam has
a velocity (it Is stationary). On account of impact all elements
of the beam will acquire a certain velocity (different for each element)
in a short time while the weight will correspondingly slow down.
At the point of impact the weight and the beam material in imme­
diate vicinity have identical velocities equal to v m. Medium velocity
t/m may be found from Carnot's theorem:
_ MttOw (30.29)
0 -f-ccQo

Here Q and Mw are the weight and mass of the striking body, Q«
and M0are the weight and mass of the body subjected to impact (beam),
and a is the mass reduction coefficient (less than unity) which has
to be introduced to account for the fact that not all parts of the body
suffering Impact move after impact with the same velocity, vm (see,
for example, Fig. 422). For tension and compression a = y ; If the
beam is subjected to bending as shown in Fig. 422 (a), then a —17/35«
etc.* It is evident from (30.29) that vm< vw—V 2 gH\ the grea­
ter mass Mo of the body suffering impact the less is vm as compared
to i/lP. Kinetic energy that remains in the beam-weight system after

* Detailed derivation of the expression for coefficient a for the beam shown in
Fig. 422(d) is given in $231 of N. M. Belyaev, Strength of Materials, Nauka, Edi­
tions 7-14 (in Russian).
Ch, 30) Stresses Under Impact Loading 583

impact is:
QM&m AlwPm
Y (Mw+ aA !#) (M.+aAf.)*
T«
1t + F (30.30)
l+ « -^
i.e. l+ P times less than the kinetic energy of the weight if the lat­
ter strikes a weightless beam. Hence, if the mass of the body suffer­
ing impact is taken into consideration, the dynamic coefficient should
be calculated not by formulas (30.10), (30.11) and (30.12), but by the
formula

K o=l + V 1 + l/* (i+17 = 1 + V X+

(30.31)

i.e. if the mass of the body suffering impact is taken into account,
the design stresses due to impact are reduced.
As an example of analysis of a complicated structure under im­
pact, let us study the impact load Q at the middle of a beam which

is constrained by a fixed hinged support at end A and another hinged

support at point B mounted at the middle of the second beam (Fig. 426).
The first beam has span It, moment of inertia J t and modulus of
elasticity £ , and the respective quantities for the second beam are
J i, and £. The maximum dynamic stresses occur in the outer fibres
of the middle sections of the beams (first as well as second). Our aim
is to determine these stresses.
We shall solve this problem by multiplying the static stresses due
to load Q in the first (AB) and second (CD) beams with the dynamic
coefficient ______
*£.= ' + V l + T T

\9*
5ti4 Dynamic Action of Forces | Part IX

The static deflection of the first beam in the section of impact is

determined by the deformation of the whole structure and is equal
to /o=*/i+y/s, where is the maximum static deflection of the first
beam due to force Q, and /* is the corresponding deflection of the
second beam due to force Since
Qi\ and /
h 48EJ, 96EJ2
we get
f - Qfi I Qft Qt
/o 4 8 EJ, 192EJ„ ™ 48£^,
and
%LJ,H
/Q>=1 +

The maximum stresses in the first and second beams are

°Ui =* KDah -o Kd and oo, = KpPs,= K q

As the potential energies due to impact accumulated by the first

beam (UD), the second beam {Un) and both beams combined (UD=
= UDt+U »= To) are proportional respectively to i/v U9t, and Ut
(the square of the dynamic coefficient serves as the constant of pro­
portionality), we get

_ /.
(a)
Uo ~
fi+ ifft
h
Us Y Qf ' + T Qfi
and

Up. T '« fpj

ilfl 7\ t; I » (b )
* jQ ft+ ~Q h y /i
u

Deflection /„ represents the total pliability of the whole structure at

the point of impact, deflections f 01 and f ot represent fractions of the
total pliability which depend upon the deformation of the first and
second beams separately. It follows from formulas (a) and (b) that
U o .~ T .ljL and U„, = T , l j j

Thus, distribution of impact energy between the beams is directly

proportional to their pliability as a fraction of the total pliability
Ch. 80\ Stresses Under Impact Loading 565

at the point of impact. If the dimensions of the beams are selected

in such a way that /oi =/o 2, then V D= UD= ^ T 9. Had there been
a rigid support in place of the second beam, the total impact energy
would have been absorbed by the first beam; the second beam helps
in damping the impact on the first.
The same effect would have been observed if instead of rigid sup­
ports we had used very pliable supports made of rubber spacers or
helical springs for constraining the beam ends.

§ 179. Impact Testing for Failure

It was pointed out earlier that dynamic action of force is distin­
guished not just by the fact that stresses (within elastic limits) under
dynamic loading are different from stresses under static loading.

The material itself reacts to dynamic loading in a different way than

to a load which increases gradually. This is especially noticeable in
impact loading.
Experiments on failure of specimens under impact loading show
that the tension test diagram in this case is completely different from
the tension test diagram under static loading. Figure 427 shows the
tension test diagrams for mild steel under static and dynamic loads;
the curve of Impact loading is distinguished by a sharp increase in
the yield stress and by a displacement of the maximum load towards
the left. This shows that the velocity of impact also affects the me­
chanical properties of the material. There are cases when materials
having excellent plastic properties under static loading behave as
brittle materials under dynamic loading. Therefore, materials for
elements subjected to dynamic loading are selected after conducting
an impact test. In impact test specimens of the material are subjected
to impact failure under tension, but more often under bending, and
the energy required for breaking the specimens acts as a pointer to
the properties of the material. Impact test under bending is most
commonly employed.*

* For details see N. M. Belyaev, Laboratory Experiments on Strength of Mate­

rials, Gostekhizdat, 1951 (In R ussian), §89.

I ? * —3 3 1 0
566 Dynamic Action of Forces fPart IX

If T is the energy spent on breaking the specimen and A is the

cross-sectional area of the specimen in the section of failure, then
the impact strength oi the specimen material is obtained by dividing
T by A:
T
a/—-jk g f m/cm*
To reveal the properties of the specimen material during an impact
test, the specimen is given a particular shape—a cut is made in the
section of impact. Cuts of various shapes shown In Fig. 428 can be
made; the one shown in Fig. 428(b) is generally used at present.

Fig. 428
I
The idea behind making the cut is to subject the specimen mate­
rial to dynamic loading under the most unfavourable circumstances.
The cut creates considerable weakening of the sections in the middle
of the span, causing a sharp increase in bending stresses over a small
length of the specimen.
We have already seen (§ 176) the strong effect which any local weak­
ening of the section can have on the stresses. Almost all the energy
of impact is absorbed by a small volume of material around the weak­
ened section, causing a sharp increase in the dynamic stresses. In
addition, the cut also gives rise to a local increase of stresses at its
base, which are similar in. nature to local stresses at the edges of
holes (§15).
Figure 429 shows distribution of stresses in the section of a beam
weakened by a cut. Curve a shows the diagram of stresses Oj in a sec­
tion without a cut; curve b shows the distribution of normal stresses
in the section with a cut without taking into account the local stres­
ses; finally, curve c shows the complete picture of variation of normal
stresses o, under bending.
We see that jusl the decrease in the height of the section increases
the stresses 2.25 times; if the local stresses are also taken into ac-
Cft. 501 Stresses Under Impact Loading 567

count, the coefficient of stress concentration comprises 5.22 w.r.t.

the parent beam and 2.32 w.r.t. the beam of reduced height.
Generally, the local stresses result in the working of the material
in three-dimensional stressed state; these conditions are not con*
ducive to development of plastic deformation, and the material fails
as it it were brittle.
Thus, in the example under consideration, in addition to normal
stresses Oi in sections perpendicular to the specimen axis there act
tensile stresses o2 in sections parallel to the axis. The distribution
T en srn C ffliirtsse h

curve for these stresses is also shown in Fig. 429. Besides, inside the
specimen acts the third principal stress, o8, also tensile. Thus, the
material near the base of the cut is subjected to three-dimensional
tension, under which plastic deformation is very difficult. If the
yield stress of the material under tension, o„, and stresses o, and <r»
are less than o^ for example 0.2o», the beginning of yielding in a
three-dimensional stressed state is determined by the third theory
of strength according to the following equation:
a,—0 ,«»ow, or Of—0.2cr, = 0.8o,<*<jv
Hence, the material at the base of the cut can undergo plastic
deformation only (or values of o, greater than o„, namely at o, equal
to 1.25o„. Due to such restrictions to plastic deformation the mate­
rial may start behaving like a brittle material. This further aggravates
the effect of Impact loading.
Thus, the cut helps in a clear classification between materials
which are more sensitive to adverse action of impact and those which
are less sensitive. Here the various types of cuts show a varying degree
of effect; for example, a sharp cut enhances the adverse effect of im­
pact more than a round cut. Therefore, impact strength of various
materials can be compared only if the specimens are alike.
The drawback of specimens shown in Fig. 428 is that the base of
the cut falls in the stretched zone, where failure starts. Obviously,
the strength of such a specimen depends upon the quality of the cut;
566 Dynamic Action of Forces [Part IX

on the other hand, the cut makes it impossible to test specimens

which must retain the contours of the actual part (this is sometimes
of great importance).
The specimen (Fig. 430) prepared in the Strength Testing Labora­
tory of the Leningrad Institute of Railway Engineers is free of these
drawbacks, when this specimen
fails, the cut develops parallel to
the direction of impact. Also, in
this specimen almost all the energy
of impact is concentrated in the
weakened zone, but the failure is
more close to the real one.
There is a sharp difference be­
tween such broken specimens of duc­
tile and brittle materials; consider­
able plastic deformation may be
observed in the stretched zone for
materials having low sensitivity to
impact. However, failure of brittle
materials occurs almost without
Fig. 430 an permanent deformation.

§ 180. Effect of Various Factors on the Results

of Impact Testing
As a rule impact testing is carried out at room temperature on at
least four identical specimens. This restriction on the minimum num­
ber of specimens to be tested is necessary to keep a check on random
errors of manufacturing and testing, which may sometimes seriously
affect the impact strength of the specimen.
As an example, Table 21 shows values of impact strength for a
number of materials at room temperature; the tests were conducted
on the type of specimen shown in Fig. 428(6).
It can be easily noticed that impact strength is greatly affected
by a number of factors, namely shape of the specimen, velocity of
impact and most of all temperature of the specimen.
Impact strength of specimens of the same material falls as the tem­
perature. of testing is reduced. In some materials (mild steel) the fail
is very sharp; in hardened and alloy steels (chrome-nickel) the fall
is comparatively smooth. In Fig. 431 curves a, b, c show the varia­
tion of impact strength; the curves were obtained at the Strength
Testing Laboratory of the Leningrad Institute of Railway Engineers
(LI R E).
Before testing, the specimen is brought to the required tempera­
ture in a trough which can be heated by placing on an electric heater
and cooled with the help of liquid air. Curves a, b, c shown in Fig. 431
Cfi. SO] Stresses Under Impact Loading 569

Table 21
Impact Strength for Some Materials
H eat treatm ent

an n ealin g hardening and

T ype o( (tee) and its ch em ical tem pering
com p osition

ob at 0(1 at
O cg t/m m ') (kgf -m/cra*) (k g l/m m 'j (kgi -m /em *)

Carbon steels (carbon content)

< 0.15 35-45 >25 36-50 >25
0.15*0.20 40-50 >22 45-65 >20
0.20-0.30 50-60 >20 55-75 > 15
0.30-0.40 60-70 > 16 70-85 > 12
0.40-0.50 70-80 > 12 80-95 > 8
0.50-0.00 80-90 > 10 90-105 > 5
0.60-0.70 85-95 >8 > 100 >3
> 0.7 >95 >6 > 105 >2
Alloy steels
Nickel steel: C 0.20, Ni 3.0 50-58 25-20 70-80 24-18
Chrome-nickel steel: C 0.3, >65 >7 75-90 >20
Ni 2 .5-3.0, Cr 0.5-0.8
Chrome-nickel molybdenum 65-70 13 95-100 20-16
steel: C 0.25-0.35, Ni
2.5-3.5,Cr 0.8-1.2,Mo 0.3-0.5

make it clear that a reduction in temperature may cause a sharp

reduction in impact strength and thus cause brittle fracture of parts
of structure. This phenomenon has been often observed in practice;
cold brittleness of rails, rims and other elements used in railway
transport has often been the cause of a number of problems.
A very important point in this context is that for quite a few ma­
terials (Fig. 431, curve a) the transition from plastic failure, having
high impact strength, to brittle failure takes place in a small temper­
ature interval. For instance, a material having good impact strength
at room or nearly room temperature may experience brittle failure
even with a small reduction in temperature. Therefore, results of the
usual impact tests at room temperature cannot be considered sufficient
for assessing the resistance of material to dynamic loading; it is nec­
essary to obtain a more complete picture of impact strength as a
function of temperature (Fig. 431, curves a, b, c).
The more to the left is the “critical1' interval of fall-off of impact
strength, the lower is the sensitivity of the material to temperature
changes under dynamic loading and the greater is its reliability.
Variation of the shape of the specimen can, to some extent, be uti­
lized to replace testing at various temperatures. Experiments show
670 Dynamic Action of Porta {Part IX

that in wider specimens the “critical” temperature interval shifts

towards the right, i.e. towards higher temperatures. Therefore, if
the usual impact test at room temperature gives satisfactory results,
then tests may be conducted on wider specimens to check whether
the temperature of conducting the experiment is close to the critical
interval. If brittle fracture oc­
curs after this test, it means
that the temperature of exper­
iment is close to the critical
interval.
Finally, it should be noted
that ductile properties of a
material may be seriously af­
fected by residual stresses,
which appear in the material
after quenching, cold rolling,
or rolling at low temperatures
resulting in strain-hardening
ol the material. As a rule, these
stresses cannot be assessed
by simple tension test. Resid­
ual stresses generally occur in
a three-dimensional stressed
state of the material; this may
result in brittle fracture. Situations like this have been observed in
manufacturing large I-beams having thin Ranges. In the course of
our experiments an I-beam No. 50 failed due to brittle fracture when
it was dropped on earth on a frosty day. Tests under static loading
and the chemical and metallurgical analysis revealed that the mate­
rial of the beam was of good quality. Only under dynamic loading
at different temperatures was cold brittleness observe! in specimens
cut near the flange edge—in the maximum hardened region. As for
the effect of chemical composition of steel on brittleness, the impact
strength decreases, as can be seen from Table 21, with an increase in
carbon content, i.e. with an increase in ultimate strength and decrease
in plastic properties of steel. Phosphorus has an extremely adverse
effect on impact strength, especially at low temperatures. Therefore
the percentage of phosphorus in steel should be restricted in every
possible way if the steel is to be used in the manufacture of elements
working under impact.
Ch. S I | Strength Check for Variable Loading 571

CHAPTER 31

Strength Check of Materials

Under Variable Loading
§ 181. Basic Ideas Concerning the Effect of Variable
Stresses on the Strength of Materials
The resistance of materials to loads which systematically change
in magnitude or magnitude and sign is considerably difTerent from
the resistance of the same materials to static and dynamic loads.
Consequently, the problem of checking strength of materials under
variable loads requires to be studied in a special chapter.
It is well known that machine parts subjected to cyclic loads some­
times fail suddenly, without any noticeable permanent deformation,
al stresses which they reliably withstand under static loading. The
attention of engineers was first directed to this problem by the obser­
vation that machine elements manufactured from materials showing
under static tests excellent plastic properties—elongation, contrac­
tion and impact strength—failed without any noticeable plastic
deformation, as if they were made of some brittle material.
When the engineers first started studying the causes of such fail­
ures (first half of the nineteenth century), they still did not have a
clear idea about the structure of metals; it was assumed at that time
that plastic metals have a “fibrous” structure, whereas brittle metals
have a “crystalline” structure. As the failure of elements generally
occurred not immediately but after a certain, considerably long,
period of operation of the machine, it was assumed that under vari­
able stresses the metal gets “fatigued” and changes its structure from
fibrous to crystalline, becoming brittle in the process.
The surface of fracture itself seemed to support this hypothesis. As
a rule, the surface had two distinct zones: a smooth, ground surface
(the surface in which the crack developed gradually) and a coarse
grained surface (surface of final failure of the section weakened by
the crack). Figure 432 shows the surface of failure of a wagon axle;
we can see the outer ring-shaped smooth surface and the inner coarse­
grained surface, which is characteristic of brittleness.
However, at the beginning of the twentieth century, the study of
the microstructure of metals under microscope proved that the hy­
pothesis was not correct. It was observed that metals have a crystalline
structure in the ductile state. Observations showed that when the ele­
ment is subjected to variable stresses, no principal changes in the
microstructure or in mechanical properties occur The materials
of the piston rods of steam engines and of the wagon axles retained
their structure irrespective of how long they worked.
This precludes any talk of “recryslalllzation” from one form to
another under the action of variable stresses.
572 Dynamic Action of Forces [Part IX

The nature of failure undervariableloading wasdiscovered at the

beginningof this century. Numerous experiments revealedthat under
the action of variable stresses, a crack appeared in the metal and
gradually penetrated into its interior. Under variable deformation
the edges of the crack approach each other at one instant and move
away at another; this explains the ground, smooth surface in the
zone of failure. Asthefatiguecrack develops, thecross section weak­
ens more and more, and finally
a chance impact is enough to
cause complete failure, whichoc­
curs when the strength of the
weakened cross section becomes
insufficient.
The fatigue crack is a sharp
cut on the surface of cross sec­
tion, similar to (hecut madeon
specimens used inimpact testing
The base of the crack finds
itself in a three-dimensional
stressedstate, which isconducive
to brittle failure under impact
(see§179). Thisexplains the pres­
ence of coarse-grained structure
in the zone of failure, caused by
brittle fracture.
Thus,brittlefailure under variable loadingis caused not because
thematerialchanges itsmicrostructure and becomes brittle, hut
becauseof the appearance of a fatigue crack inthe three-dimensional
stressedstate, which is conducive to brittle failure without any plas­
tic deformation.
The failure undervariable loading isof a localizednatureanddoes
not involve the material of the whole structure. Therefore, if acrack
is seen to develop under variable loading, in many cases it is not
necessary to change the whole structure: it suffices toreplace the
damaged part andremedy the factors whichcaused the crack.
The theoryof failure discussedabove is nowaccepted by engineers
throughout the world. Consequently, the term“fatigue ot material”
has last its physical meaning: while describing the process of failure
under dynamic loadingwe must talk not about failure due to fatigue
but about failure due to gradual growth of crack.
However, onaccount of its brevity andwidespreaduse intechnical
literature, the term“fatigue of materials” is still in vogue, although
it expresses a different meaning: fromnowon whenever we speak
about “fatigue of material”, it will mean failure due to gradual de­
velopment of a crack.
Our aimnow is to expose the factors which cause the crack, and
C k. s t \ Strength Cheek for Variable Loading 573

lay down such rules for design of machine elements and structures
and strength check that guarantee safe working under variable loads.
This problem (s very important, especially in machine building,
where most often we have to deal with cyclic stresses- It can safely
be assumed that approximately 90% of the total failures of machine
parts occur due to development of the fatigue crack. These failures
are very dangerous and often result in serious accidents, because it
is not always possible to notice the developing hair-thin fatigue crack
in time. Failures of wagon and engine axles are chiefly caused by these
cracks and invariably result in derailment, accompanied by tragic
consequences. Similar failures have been observed in aircrafts as well
as in other branches of machine building.

§ 182. Cyclic Stresses

Stresses in the parts of machines and structures may change either
due to change in the magnitude of load (for instance, stresses in the
connecting rod and the cylinder wall of an internal combustion en­
gine change due to change in pressure of gas mixture inside the cyl­
inder) or due to change in position of the part under action of a con­
stant load (for instance, if the constant weight of a wagon acts on
the axle which is rigidly connected with the wheels, the bending
stresses at any point of the axle’s cross section vary as the point
changes its location due to rotation of the axle).
The variation of stresses in parts of machines or structures may be
unstable (for example, changes in stresses acting in a part of bridge
due to moving trams, automobiles, pedestrians) or steady (for example,
change in stresses acting in the connecting rod and cylinder wall of
an internal combustion engine, rotating wagon axle, transmission
shaft, etc.). From among the various types of steady variable stresses,
cyclic stresses are the most important; besides, these stresses are the
most widely investigated. A single repetition of stress (from mini­
mum to maximum and back) is known as the cycle of variation of
stress', if such a cycle is continuously repeated during functioning of
a part, the stresses acting in the part are called cyclic stresses. Cyclic
stresses act in wagon axles, shafts, connecting rods, turbine blades
and many other parts of machines and structures.
Figure 433 depicts various types of cyclic stresses in “stress p (o or t )
versus time t” coordinates: (1) constant sign cycle (Fig. 433(a)), in
which stresses vary only in magnitude; (2) fluctuating (zero base)
cycle (Fig. 433(d)), in which stresses vary between zero and a certain
maximum value; (3) constant stress (Fig. 433(c)); (4) alternating
cycle (Fig. 433(b)), in which stresses change in magnitude and in
sign (all the cycles enumerated above are known as asymmetrical
cycles)', (5) symmetrical cycle (Fig. 433 (a)), an alternating cycle in which
the upper and lower ‘limits of stress variation have the same abso-
574 Dynamic Action of Forces {Part IX

lute values. The curves which describe the variation of stresses in

time may considerably differ in appearance; variation of stresses in
machine parts often follows the sinusoidal law.
The maximum absolute stress in the cycle is denoted bv / w (amax,
W ) . while the minimum is denoted by pmla (onlB, The ratio

Fig. 433

of minimum stress to maximum with the signs taken into account Is

known as the cycle characteristic, or the coefficient of asymmetry of
cycle, This coefficient varies between —1 and +1;
in Fig. 433 the coefficient of asymmetry is given for all the cycles.
The half of the sum of maximum and minimum stresses of a cycle
(taking their signs into consideration) is known as the constant compo­
nent of cycle, or mean cycle stress:

The hall o! the difference of maximum and minimum stresses (also

taking their signs into consideration) is known as the variable com­
ponent of cycle, or the amplitude of stresses in the cycle:
n _Pmax"-Pmii) __ *— r _
ra j rm tx (31.2)
Ch. 3l\ Strength Cheek for Variable Loading 575

Thus, any cycle of stress variation may be obtained by superimposing

a symmetrical stress cycle pmtx= —Pmi,,—pa on a constant stress
P max = P m l0 ~ P m *

§ 183. Strength Condition Under Variable Stresses

Experiments show that gradually developing cracks appear only
under variable stresses, oscillating systematically between extreme
values.
It is also known that a large number of elements of machines and
structures exhibit good resistance to variable loads over a long du­
ration, provided the stresses remain within certain limits. Hence,
just the fact that the stresses are variable is not enough to cause a
crack—for the crack to appear it is essential that the maximum value
of the variable stresses should exceed a particular value, which is
known as endurance strength, or endurance limit. Endurance limit
pr represents the maximum value of a periodically changing stress
which the material can withstand for a practically infinite period of
time without fatigue cracks appearing in it.
The endurance limit for a variable stress cycle wilLbe denoted by
pn or, or Tr with a subscript representing the cycle characteristic:
P - 1 is the endurance limit of an asymmetrical cycle of characteristic
/*=—1; po.s is the endurance limit of an asymmetrical cycle of charac­
teristic r = 0.2; etc.
Thus, the possibility of failure due to gradual development of a
fatigue crack is subject to the following two conditions: (1) periodic
oscillation of the variable stresses between two extreme values, and
(2) the maximum value of the actual stresses in the element of
structure exceeding the endurance limit of the material.
The strength condition in this case must-express the fact that the
maximum actual stress must be less than the endurance limit pr
and ensure a certain margin of safety:

where kr is the safety factor.

At present the endurance limit can be determined only experimen­
tally. It depends mainly upon the (a) material (steel, iron, non-fer­
rous metals); (b) nature of deformation (bending, torsion, etc.); and
(c) degree of asymmetry of the cycle, i.e. the interrelation between
the extreme values of the variable stress.
A few additional factors affecting the endurance limit (corrosion,
dimensions of elements) will be discussed separately (§§ 186 and 187).
As for the maximum dynamic stress pmtx, experiments show that
contrary to failure under static loading the fatigue cracks in brittle
as well as ductile prismatic bars appear not due to the maximum
876 Dynamic Action of Forces | Part IX

design stress, pnn (for example, in bending oBlx® ^ , but due to

the local stresses (§ 15) which occur at places of deviation from the
prismatic shape {cuts, scratches, holes, transition from a thin portion
to a thick portion, etc.)
These local stresses pt are considerably greater than the maximum
stress and may be expressed by the following formula:

here a e is the coefficient of stress concentration; its value depends

upon the nature of deviation from the prismatic shape.
In the next sections we will explain how to determine the en­
durance limit and the coefficient of stress concentration.

§ 184. Determination of Endurance Limit

in a Symmetrical Cycle
Of maximum interest is the determination of the endurance limit
In a symmetrical cycle (pm—0), because this value is minimum. The
endurance limit also varies depending upon the type of deformation-
bending, axial deformation (tension and compression) and torsion.
Endurance limit in bending is determined on machines in which
a round specimen is loaded through ball bearings or as a cantilever
with a force acting at one end or as a simply supported beam acted
upon by two symmetrical equal forces; the specimen rotates at 2000-
3000 rpm. During each rotation the maximum stressed portion of
the specimen material undergoes a symmetrical change of stress from
maximum compressive to an equal maximum tensile, and vice versa.
The number of cycles of the specimen is determined by its rpm, which
can be registered by a special counting device.* The contour oi the
specimen must be smooth, ruling out any possibility of occurrence
of local stresses. The experiment for determining the endurance
limit is carried out as follows. A batch of specimens consisting of
6-10 pieces is prepared from the material to be tested; the specimens
are numbered /, 2, 3, . . . .
The first specimen is placed in the machine and loaded in such a
way that a particular value of maximum normal stress o' is obtained;
this value is generally taken 0.5-0.6 of the ultimate strength of the
material. The machine is put into operation, and the specimen ro­
tates experiencing variable stresses oscillating between + o ' and —o'
until ultimate failure. At this moment a special device switches off
the motor, the machine stops and the counting device indicates the
number of cycles, Nu required to break the specimen at stress o'.

* For details see N. M. Belyaev, laboratory Experiments on Strength of Mate­

rials, Gostekhizdat, 1956 (in Russian).
Ch. SI] Strength Check for Variable Loading 577

The second specimen is similarly tested at a stress o ', less than o';
the third specimen is tested at stress o"<or*, and so on. The number
of cycles required to break the specimen increases respectively. Thus,
if we go on reducing the test stress for each successive specimen,
we reach a stage when the specimen does not fail even after withstand­
ing a very large number of cycles. The stress corresponding to this
stage is very close to the actual endurance limit.
Experiments show that if a steel specimen does not fail after 10?
cycles, it can practically withstand an infinite number of cycles
(10*-2xl0*J. Therefore, while determining the endurance limit of a
Ugf
mm2
kg f
mm*
ChrORe-mchtl steel
60
Chrmt-nickel sted
60 iiO
_o c^ltOJOkQf/m1
c&tSSigf/m2
20 2 0

0
I 6 6
_
8 10
0
2 6
_L
V ”59%tO-8%

* ]£ *
N
Fig. 434 Fig. 435

material, the experiment is stopped if the specimen does not fail

after 10T cycles. In a number of cases the tests are stopped after a
smaller number of cycles, but never before completing 5x10* cycles.
A similar dependence does not exist for non-ferrous metals, and to
make sure that the specimen can really withstand a very Jaqre num­
ber of cycles at the given stress the experiment is stopped only after
subjecting the specimen to 2 x 10“ and even 5x10s cycles. Therefore,
in the case of non-ferrous metals the endurance limit can be speci­
fied only for a particular number of cycles, say for 107 cycles, the ma­
terial has one endurance limit and for 3 x l0 7 cycles a different en­
durance limit.
The experimental results have to be processed graphically in order
to determine the numerical value of endurance limit. Figures 434
and 435 show two such methods. In the first method stresses cr\ cr\
. .. are laid off the |/-axis, and Nu . . . are laid off the x-axis.
The ordinate of the horizontal tangent (asymptote) to the curve gives
the endurance limit o ^ . In the second method, we lay off the quan­
tity 10*/N along the x-axis. In this case the endurance limit is deter­
mined as the segment cut off on the y-axis by the extended curve.
578 Dynamic Action of Forces, [Pari IX

because in this case the origin of coordinates corresponds to Af—oo.

The second method is now more commonly used.
The endurance limit can be similarly determined under axial load­
ing (tension and compression) and torsion; special testing machines
(pulsators *) are used for this purpose.
An enormous amount of experimental data is now available on
determination oi endurance limits of various materials. A greater
part of the experimental studies pertain to steel, because steel is the
most commonly used material in machine building. The results of
these experiments show that for all grades of steel the endurance
limit is related by a definite law to the ultimate tensile strength ou.
For rolled and forged steels the bending endurance limit under a
symmetrical cycle comprises (0.40-0.60)ou; lor cast steel the en­
durance limit varies between (0.40-0.46)ou.
Thus, with sufficient practical accuracy, we may write the following
relation for all grades of steel:
ot, «=0.4o„ =»fJro0
If the specimen is subjected to axial loading under a symmetrical
cycle (variable tension and compression), then the corresponding
endurance limit ol, is found to be less than the endurance limit under
bending; the ratio between the two endurance limits may be taken
equal to 0.7, i.e.
o t.^ O T o t.
The reduction may be explained by the fact that In tension and
compression all sections of the specimen experience equal stresses,
and in bending maximum stresses occur only at the outer fibres (the
remaining material remains underloaded and thus somewhat impedes
the emergence of fatigue cracks); besides, there is always bound to be
some eccentricity in the application of axial loads.
Finally, the torsional endurance limit under a symmetrical cycle
comprises 0.55 of the bending endurance limit. Thus, under a sym­
metrical cycle we get the following values for steel:
o t.-0 .4 0 o . )
oti =* 0.7ot. ■*0.28ob I (313)
at, ™0 55at, = O.22o0 j
These relations can be employed for obtaining formulas for the strength
check.
In the case of nonferrous metals we get a more flexible relation
between endurance limit and ultimate strength; the empirical for-

• Ibid.
C h . $ /] Strength Check tor Variable Loading 679

mula is
o^., = (0 24-0.50) oK
While using expression (31.3), it should be borne in mind that the.
endurance limit of a material depends upon a large number of factors
(§ 187); the relations given in expression 131.3) were obtained on
specimens of small diameter (7*10 mm) having a polished surface
and no sharp changes of shape along the length.

§ 185. Endurance Limit in an Unsymmetrica! Cycle

The equipment required for determining endurance strength under
an unsymmetrical cycle is much more complicated than the equip­
ment used for symmetrical cycles.
A special spring capable of stretching and compressing the specimen
should be added to the simple testing machine discussed earlier, in
which the specimen only rotates. Quite often we have to employ
even more complicated machines, which are capable of exerting
axial load on the specimen (tension, compression) under different
extreme values of the variable stresses.
However, we now have at our disposal sufficient experimental
data to obtain a graphical or analytical relation between the en­
durance limit and the cycle characteristic r, i.e. £ssla.
PjJlBX
Let us remind the reader about the notations used here: p„ repre­
sents the ultimate strength of the material, p„ the yield stress, pr
the endurance limit corresponding to a cycle of characterisi ic r,
p_, the endurance limit in a symmetrical cycle, pmtx and pmiu the
upper and lower extreme values ol the cycle, pw— the
mean stress in the cycle, pg= Pm-^ Pm)fl the amplitude of oscflla
tions of the cycle, 2pg the double amplitude of the cycle, and
^PmiJPnax the characteristic of the cycle.
The values of pmax, pmln, pa and p m which correspond to working
of the material at the endurance limit will be denoted by a subscript r.
Pr max* P/mim Prm' Pra
(the maximum absolute value of p ,max or p ^,,, must coincide with p,>.
The results of experiments for determining endurance limit under
different cycles are conveniently represented in the form of diagrams.
The simplest among these diagrams is the diagram in the pm- and
/^-coordinates (Hay’s diagram) shown in Fig 436. On this diagram
the values of pm are laid off on the x-axis to a certain scale and the
values of pa are laid off on the y»axis in the same scale. Curve ABCD
has been plotted on the basis of experiments for determining the en­
durance limit under different cycles of variable stresses. For deter-
580 Dynamic Action of Forces IPart fX

mining with the help of this diagram the endurance limit pr for a
cycle having coefficient of asymmetry r we draw from the centre of
coordinates 0 a straight line OS at an angle p, so that (see (31.1)
and (31.3))
tan0 pu _ 1—' (31.4)
Pm 1+'
and extend it until it intersects curve ABCD at point C; from this
point we then drop a normal CE on the abscissa. The sum of segments
CE and EO, which are respectively equal to Pac and pmc* gives the
endurance limit
PrC—Pmix^PaC + PmC (31.5)
Thus, point A having ordinate OA=pa—p^± and abscissa pm—0
represents the endurance limit under a symmetrical cycle: r*=—I,

P=n/2, while point D having ordinate pa= 0 and abscissa OD=pm=

=P*i=Pu represents the endurance limit under static loading (r—+1,
p=0), which is equal to the ultimate strength of the material. If we
draw a straight line OT at 45° inclination through the centre of coor­

dinates, point B, where this line intersects curve ABCD, represents

the endurance limit under fluctuating (zero base) load, because or­
dinate BF=*paB of point B is equal to abscissa OF—pmB, i.e. from
(31.1) and (31.2) I + r = l —r, and r = 0.
Ch. Sll Strength Check }or Vur table trading 581

Figure 437 depicts Hay’s diagram lor common low-carbon steel,

and Fig. 438 shows the same diagram lor grey iron having the follow­
ing properties: a ue=78 kgl/nim*, ou(= 22 kgf/mm*, o_i=<7.3 kgf/mm*,
and a0c=*46 kgf/mm* (oCfl is the endurance limit under fluctuating
cycle in compression). This diagram proves that under variable loading
too cast iron has a much higher strength in compression to that in
tension.
The endurance limit should be looked upon as a critical stress
similar to ultimate strength under static loading, because a stress
slightly greater than the endur­
ance strength can cause failure
within a practically feasible num­
ber of variations of the load.
Therefore, curve ABCD in Fig.
436 represents a curve of crit­
ical (limiting) stresses for mate­
rials that do not have a yield
zone. If the material is ductile
and the critical stress for it un­
der static loading is yield stress
py, then it can be easily estab­
lished from Fig. 439 that line AGH
should be taken as the line of criti­
cal stresses. In the endurance lim­
it diagram ABD (Figs. 436 and 439), If we cut segments OH and ON
equal to yield stress pv on the x• and y-axis, respectively, and join
points N and H by a straight line NH, then the sum of the abscissa
and ordinate of any point on this line is always equal to pa (for in­
stance, forpoint K, OLJt-LK=pm^ p a H a OL+LH—p„, as LK—LH).
Hence, straight line NH determines the critical limit of stresses under
static loading, while curve ABD determines if under variable loading.
In the region in which straight line NH lies above curve ABD (from
the axis of ordinates to point G, the point of intersection of these
lines) the critical limit of stresses is determined by curve i4G. But
where line NH lies below curve ABD, the critical limit of stresses
is determined by the straight line NH. In Figs. 436 and 439 the
critical stress lines are hatched.
On account of the fad thal usually we determine experimentally
pyt pu, and /?_i while the endurance limits for other values of r are
generally not determined, straight lines AD (Fig. 436) or AH (Fig.
439) have often to be accepted as the critical stress lines in the absence
of experimental data. The critical stress curve is sometimes replaced
by a straight line in order to simplify calculations; let us note that
such an approximation adds to the existing safety factor.
A fairly large number of formulas were proposed for establishing
the analytical relationship between endurance limit, ultimate strength
582 Dynamic Action of Forces [Part IX

and the cycle's characteristic. Those deserving attention are:

(31.6)
Pr=“ >“ '+ < !+')4. Pa

pr (l+Oa9, Pa (31.7)

(31.8)

where q\=p,^lpu. Coefficients «i and n%in the last formula have differ­
ent numerical values depending upon the material. For low-carbon
steel tti=0 and n*=l, and a parabolic relationship exists between
pra and p,m. On the other hand, for steels with high ultimate strength
rti^ l, rti=0, and expression (31.8) is represented by a straight line.
As already mentioned, endurance limit pr depends not only upon
the material, nature of deformation and type of cycle, but also upon
the shape of part and the condition of its surface, upon its dimensions,
and so on. From among the factors listed the endurance limit pr is
affected most by the shape of part and the condition of its surface.
As these factors are equally important in static loading also, they
deserve a detailed discussion.

§ 186. Local Stresses

Uniform distribution of stresses over the section of the part under
tension or compression and linear variation of normal stresses over
the section of a beam subjected to bending or shearing stresses along
the radius of section of a shaft subjected to torsion are valid only
for uniform prismatic rods which are free of internal or external flaws
and damages and only in sections that are sufficiently far away from
the point of application of load. The distribution of stresses is violat­
ed at the point of application of load and also where the part has
holes, recesses, transitions from one dimension and shape to another,
internal and external flaws and damages, non-uniform structure of
material, etc.
For instance, in a plate (Fig. 440(a)) stretched by forces P acting
along its axis the normal stresses in section mn located sufficiently
far from the point of application of force are distributed uniformly.
In section mi«i, where there is a small circular hole in the plate, the
distribution of stresses will be different. Near the edge of the hole the
stress will be considerably (about three times if the hole is small)
higher than in section mn. However, the high stresses act only on a
small area of section mmi near the hole; in the remaining area of the
section the stresses are approximately the same as in section mn.
These high stresses are known as local stresses pt (a, or t {); the sources
C h. SJ] Strength Check for Variable Loading 583

of local stresses (holes, recesses, damages, etc.) are called factors

(sources) of stress concentration. The ratio between the maximum
local stress p/mM to the nominal stress pnom, i.e. to the stress at the
same point in the absence of stress concentration source, is known as
the coefficient of stress concentration, a c:
q a max (31.9)
c Pnom

Even in a uni-axially loaded element the local stresses create a

three-dimensional stressed state. In Fig. 440 it is shown that besides
stresses in sections perpendicular to the bar axis additional normal
stresses of smaller magnitude appear around the hole in a plane nor­
mal to the first (o*).
The coefficient of stress concentration depends chiefly upon how
fast the prismatic shape changes. If the transition frun a large di­
ameter to a smaller diameter takes place sharply, at right angles, the

Fig. 440

maximum value of a e is obtained. If the transition is smoothened

by making a fillet of certain radius, the value of a e becomes smaller
and may even become equal to unity (Fig. 441 (a) and (b)).
Figure 440 depicts a few examples of stress concentration due to
holes and recesses in parts subjected to tension or bending.
 584 Dynamic Action of Forces [Part IX

The following methods are employed for determining the coefficient

of stress concentration. In a number of cases (for example, in tension
and bending of bars with holes and necks) it is possible to determine
the local stresses by the theory of elasticity. Another popular method
is the experimental determination of local stresses in a planar stressed
transparent model (madeof glass, celluloid or bakelite) under polar*
ized light; from the colour of
various portions oi the model we
can find the difference between
principal stresses at various
points and then through some
additional measurements and
computations determine the prin­
cipal stresses.
The experimental methods in­
clude determining local stresses
with the help of lacquer coatings
or meshes * (quadratic or circu­
lar of small diameter), which
are applied to the surface of the
specimen, and studying brittle
(gypsum) models.
If we tesl the strength ol a
material (Fig. 441 (a) and (b)) on
two specimens, one with local
stresses and the other without, we
find that au of the first sped men is less than ou of the second specimen
(of course, using the same modulus of section of diameter d); the ratio
of these two values of ultimate strengths gives us the required coeffi­
cient of stress concentration, <xc.
However, a more reliable method of obtainirg a r is by determining
the endurance limit of two specimens, one with local stresses and the
other without. The first specimen gives a lower (on account of local
stresses) endurance limit o', as compared to the second a’r% the ratio
determines coefficient a c. It was noticed that the value of the
stress concentration coefficient differs with the method employed,
although the factor causing local stresses was the same in each case.
The first two methods—one based on the theory of elasticity and
the other on optical study—give almost equal values of a c. This is
obvious, because in both cases the results pertain to isotropic elastic
materials. Also, the values of a c determined from endurance tests
are found to be quite close to the values obtained by the first two
• See. for example, G. A. Smirnov-Alyaev, Strength of Materials Under Plastic
Deformation, Mashgiz, 1961 (in Russian).
C/r. 3i] Strength Check for Variable Loading 585

methods for some materials (chrome-nickel, high strength carbon

steel) and considerably less for other materials (tnild steel). It was
found that the coefficient of stress concentration depends not only
upon the shape of the specimen, but also upon its material. The great­
er the ductility of materia), the lower is ils coefficient of stress con­
centration. The reasons for this have already been explained in § 16;
the plasticity of a material creates a sort of buffer, which to a certain
extent mitigates the effect of local stresses.
Thus, we have two coefficients of stress concentration: first, the
theoretical one, a c t, takes into account only the shape of the spec­
imen and is mainly determined by any of the first two methods;
the second, the actual one, a c-0 is determined by the test on endur­
ance and takes into account not only the shape but also the material
of the specimen.
As this consideration affects only the amount by which the local
stresses exceed the general, i.e. the quantities (ac ,—1) and (ae a—1),
the sensitivity of a material ‘to local stresses may be determined by
the ratio of the two, known as the sensitivity factor of the material:
i
(31.10)
9“ a c .t— 1

This factor depends upon the material; it may be equal to unity in

high grade, heat-treated alloyed steels and may be as low as 0.5 in
case of mild steels. It is found that iron is the least sensitive to local
stresses; in the case of iron q is close to zero and the actual stress con­
centration factor, a c.0, is almost equal to unity. This can be explained
by the fact that the ultimate strength of iron is strongly affected by
microscopic graphite inclusions, which are .actually nothing but
sharp cracks in the base metal. The effect of these cracks, which are a
regular feature of iron, is so strong that it almost completely makes
iron immune to the effect of other stress concentration factors.
The sensitivity factor, however, depends not only upon the mate­
rial but also upon the shape and size of the part, and q increases with
the increase in dimensions of the :body.
Sensitivity factor q for steel may be determined approximately
(without accounting for the absolute dimensions of the body) with
the help of the curves in Fig. 442, depending upon the ultimate strength
of the material (between 40 and 130kgf/mm9) and the theoretical
stress concentration coefficient a c. t (Figs. 443 and 444). This diagram
was obtained from experimental data on endurance testing of small
(diameter of 7-10 mm) specimens of various grades of steel for differ­
ent values of the theoretical stress concentration coefficient. It is
evident from the diagram that ’the sensitivity factor increases with
ultimate strength and theoretical stress concentration coefficient.
The increase of ae.t beyond 1.8 ceases to have any effect on the sen­
sitivity factor. For a highly alloyed steel having high ultimate strength
20 —3310
58G Dynamic Action of Forces (P a r t I K

(130 kgf/mm*), coefficient q may be considered equal to unity, and

®c.a!P®c. /•
Aftteh less data on the sensitivity factor for nonferrous metals are
available. For cast Electron (an alloy of magnesium with aluminium,

W 60 80 100 1W

Fig. 442

zinc and manganese) this coefficient is 0.15; in rare cases it may go

up to 0.25. The sensitivity factor is higher for rolled and stamped
Electron and varies between 0.35 and 0.50. In case of aluminium
alloys the values of this coefficient are still lower.

The curves in Figs. 443 and 444 showing the variation of a e,t as a
tunction of may be used for determining the theoretical stress
concentration coefficient in more common situations of stress con-
Ch. 31) Strength Check for Variable Loading 587

centration (holes, necks, fillets) depending upon the sharpness of

change in shape of the part under tension or compression (Fi^*443)
and pure bending (Fig. 444). The coefficients were determined for
rectangular specimens by the optical method. In round specimens
with necks and fillets the corresponding values of a r>( are found
to be somewhat less. Some values of a Cnl for round specimens are
given in Table 22.
Table 22
Coefficient of Stress Concentration

T ype of d eform ation and source of stress co n cen tra tio n a?

I. Bending and tension

1. Semicircular neck on shaft, ratio of radius of neck to dia-
meter of shaft
0.1 2.0
0.5 1.6
1.0 1.2
2.0 1.1
2. Fillet, ratio of radius to height of section (diameter of
shaft)
0.0625 1.75
0.125 1.50
0.25 1.20
0 .5 1 .10
3. Transition at right angle 2.0
4. Sharp V-shape neck 3.0
5. Whitworth tnread 2.0
6 . Metric thread 2.5
7. Hole, the ratio of hole diameter to cross-sectional dimen*
sions varies from 0.1 to 0.33 2.0
8 . Scratches on surface due to cutting tool 1.2-1.4
II. Torsion
1. Fillet, the ratio of fillet radius to the minimum shaft
diameter
0.02 1.8
0.10 1.2
0.20 1.1
2. Keyways 1.6-2.0

It should be emphasized that the curves in Figs. 443 and 444 and
Table 22 help determine the theoretical, i.e. the maximum possible,
values of the concentration coefficient. Knowing the theoretical stress
concentration coefficient, a e.t, the actual stress concentration coef­
ficient, a e.a, can be computed by the following formula:
«*.«. = l + t f K . t — 1) (31.11)
20*
588 Dynamic Action of Forces [Part IX

which ensues from expression (31.10). However, if the sensitivity

factor q is determined approximately from the curves in Fig. 442,
then the actual stress concentration coefficient, can also be
determined only approximately. It is therefore desirable to determine
a c a directly by conducting endurance tests on specimens of required
shape. The stress concentration coefficients, a c.< and a r,fl, for a wide
range of sources of stress concentration are given in handbooks, and
the methods of their determination are a subject of study in special
courses.*
Quite a few simple approximate empirical formulas were proposed
for determining the actual stress concentration coefficient of steel
versus its ultimate strength. When the part does not have sharp
angles, necks, or keyways and has a good finished (but not polished)
surface,
(31.12)

and when the part has sharp angles, necks, or scratches,

(31.13)

These formulas are valid for steel with ultimate strength between
40 and 130 kgf/mm3 and are sufficiently accurate for practical appli*
cation; ow is expressed in kgf/mm4.
While talking of local stresses, it is necessary to emphasize the
effect of surface damages on the endurance limit. Experiments show
that the endurance limit of forged parts which do not undergo subse­
quent heat treatment is less than that of parts of the same materials
in which the outer layer is machined and polished; in mild steels the
difference may be 15-20%, and in high grade steels it may be as high
as 50%.
A similar phenomenon is observed in springs made from high grade
alloy steels if the spring wire is not machined after heat treatment
(hardening and annealing). Such a surface can sometimes reduce the
endurance limit two-fold. Even notches and scratches reduce the en­
durance limit by 10-20%.
A very important cause of considerable stress concentration is the
interference fit between two parts, for example, the fit between disc
or pulley hub and shaft or axle. Numerous experiments reveal that in
interference fits the actual stress concentration coefficient may reach
1.8-2. It may be reduced by proper designing of the two parts (§ 191).
Poor surface finish can be the source of considerable stress concen­
tration in machine parts made of high strength steels. For example.

* Sec, for example. S .V . Serensen, l.M . Telelbaum. and N. I. Prigorovskii,

Dynamic Strength in Median teat Engineering, Mashinostroenie, 1975 (in l(ussian).
Ch. 31\ Strength Check for Variable Loading 589

for steels having ultimate strength between 50 and 140 kgf/mm*

milling (denoted by V ) without subsequent grinding and polishing
creates a stress concentration equivalent to a r= 1.25*2 (here and
farther on the lower value refers to steel having ou=50 kgf/mm*
while the greater value refers to steel having crtt= l4 0 kgf/cm1. Rough
grinding (denoted by W ) reduces the stress concentration coefficient
to 1.1*1.45; fine grinding and rough polishing (denoted by V V V )
correspond to a c=1.05*1.15, and only after fine polishing (denoted
by V V V V ) &r— I. Nonferrous metals and alloys are somewhat
less sensitive to the effect of surface finish on stress concentration.
The combined effect of local stresses and chemical reactions can
result in a sharp reduction in the endurance limit of elements sub­
jected to corrosion. Experiments reveal that the endurance limit
registers a sharp reduction if the tests are conducted in water or some
other fluid which can cause corrosion. However, this effect is less
pronounced in case of stainless steel parts.
Finally, the microstmclure of steel is another factor affecting the
local stresses and consequently the endurance limit. The metal is a
conglomerate of crystal grains of various sizes and arbitrary orien­
tation; therefore the actual stress distribution is to some extent non-
uniform even under simple tension. The degree of non-uniformity of
stress distribution increases with the non-uniformity of grain size.
Therefore, a fine grained homogeneous structure obtained by proper
heat treatment helps to increase the endurance limit of the material.
In conclusion, it must be emphasized once again that the higher
the strength of a steel the greater is its sensitivity to ail types of cuts
and surface damages and the higher is the quality of machining which
it requires.
The expressions of endurance limit and coefficient of stress concen­
tration derived in this section will be used in subsequent sections
for laying down rules to be followed in selecting permissible stresses.

§ 187. Effect of Size of Part and Other Factors

on Endurance Limit

The values of endurance limit in the preceding section were all

obtained for small specimens of a diameter between 7 and 12 mm.
In recent experiments endurance limits have been determined on
larger specimens having diameter between 40 and 50 mm. There are
fatigue testing machines which are capable of testing wagon axles
of diameter 150 mm or even 300 mm.
The experiments reveal that, firstly, there is a large spread in data
if a big specimen of this size is used for determining the endurance
limit by the method given in Fig. 435. Secondly, the endurance
limit, though not accurately known on account of the large spread,
 590 Dynamic Action of Forces [Part IX

is found to be less and sometimes much less than the endurance limit
determined on small specimens. This decrease in endurance limit is
more pronounced in alloyed steels; the effect of absolute dimensions
on endurance limit is less in case of carbon steels.
The experimentally established fact that the endurance limit of
parts is less than the endurance limit of small specimens tested in
the laboratory is of utmost importance, particularly so, because this
factor is not accounted for while determining the strength factor.
Unfortunately, the reduction has until now eluded a sound scicn*
title explanation; obviously, it is due to a number of factors, which
include the following:
(a) There is a greater possibility of the presence of internal sources
of stress concentration in large specimens (inclusions, bubbles, etc.);
the smaller specimens are more clear in this respect.
(b) In the process of manufacturing, the surfaces of specimens get
work hardened, and work hardening is more pronounced in specimens
of small size; it has been established experimentally that work
hardening results in an increase in endurance limit.
(c) Finally, for the same value of maximum stress (in outer fibres),
the decrease of stress as we move towards the interior of the body is
more intense in case of small specimens than in large specimens,
and their crystal grains work under less severe conditions.
All these ideas are, however, only assumptions. Experiments on
determination of actual stress concentration coefficient with speci­
mens of various sizes show that increasing the specimen size is to a
certain extent equivalent to increasing the sensitivity factor of its
material. These facts are of practical importance, because they show
that sources of stress concentration causing local stresses are actually
more dangerous than laboratory experiments on small specimens
make them to be.
Hence, while checking the strength of a material, the effect of ab­
solute dimensions on endurance limit must always be taken into ac­
count. Such a consideration can be avoided only if we determine the
endurance limit on specimens of natural size. This, however, is not
always possible. Moreover, sufficient experimental data is now avail­
able on comparative fatigue tests conducted on small (diameter 7-
10 mm) and large specimens of the same material. Making use of this
data and assessing the reduction in endurance limit due to increase
of absolute size by the scale factor, oc5, which represents the ratio of
endurance limit pr of a small specimen to pf of a larger specimen or
the element itself, \ye can approximately calculate the endurance
limit of the element if the endurance limit of the smaller specimen is
known. Since a.t= prlp,rt

(31.14)
Ch. 31] Strength Check for Variable Loading 591

The scale factor a , can be determined as a function of absolute

dimensions of the specimen or element by the curves shown in Fig. 445.
The reduction of the endurance limit of the element is expressed
through the endurance limit of a lQ-mm diam. specimen obtained
by fatigue testing. The value of a„ depends not only upon the abso­
lute dimensions of the element, but also upon its material and the
sources of stress concentration. In Fig. 445 curve / is used for deter­

mining the scale factor for carbon steel elements in absence of stress
concentration, curve 2 is used for carbon steel elements with mild
stress concentration (ac.«<2) and alloy steel elements with no stress
concentration, curve 3 is used for alloy steel elements with stress
concentration. The curves in Fig. 445 can be used for smooth speci­
mens only in bending and torsion, but can be used for specimens
with stress concentration in all states of stress.
The production processes employed in manufacturing of parts (heat
treatment and chemical heat treatment, metal cutting, roiling, drop
forging, press fits, welding of joints, etc.) also create factors that
influence the strength of materials to variable loading.
Some of these processes can lead to a reduction of the endurance
strength; on the other hand, there are methods of surface treatment
which improve the endurance strength of material. These methods
are: (a) work hardening of the surface layer of finished part by bur­
nishing with rolls or by shot preening; (b) chemical heat treatment
of the surface: nitriding, case-hardening, cyaniding; (c) hardening
of the surface layer by high frequency current (induction hardening)
or by gas flame. The strengthening effect of these processes lies in the
fact that residual compressive stresses are set up in the surface layer:
when the latter add up with the alternating stresses due to external
592 Dynamic Action of Forces fPari IX

load, we get asymmetrical sitress cycles with a compressive mean

stress pm, which 4s less dangerous (for the part.
The effect of production process -and surface treatment may be
taken into acoount >(!1>) while determining 'the endurance limit on
small laboratory specimens which are subjected to identical treat ment
before they are used for fatigue tests, .(2) (by correspondingly changing
the stress concentration coefficient or by introducing a special coeffi­
cient for production processt Kp, which may have a value greater than
unity (when the resistance of the part to variable loading decreases)
as well as less than unity.
In a number of cases the conditions in which the part functions
also considerably affect the (endurance limit of the material. The most
important are the effect of corrosion and temperature, as well as
breaks, underloading and overloading of the part during its working.
If the part works in conditions that are conducive to corrosion (for
example, if the part is under water), its resistance to variable loading
decreases and the fatigue curve plotted in p-N coordinates does
not have a portion which asymptotically approaches the horizontal,
in such a case the part oan have -only a limited endurance strength
corresponding to a definite number of cycles. The harmful influence
of corrosion may be reduced by work hardening, nitriding, oxidation,
chromium plating and some other methods of surface treatment.
At the design stage the effect of corrosion may be taken into account
by a corresponding increase in the coefficient of stress concentration.
The change in endurance limit of material must also be taken into
consideration when the part functions in conditions with high or
low temperatures. The endurance limit of metals (steel, cast iron,
nonferrous metals) slightly improves at low temperatures, and this
is true for smooth specimens as well as specimens having'stress con­
centration. The same metals suffer a drop in endurance limit, at 'first
gradually and then faster as the temperature is increased. If the
part functions temporarily in conditions with tow or Ji’igh tempera­
tures, this can be taken into account by introducing a special coeffi­
cient.
Parts of machines and structures are quite often subjected to short­
term underloading and overloading. -Breaks'in operation, underloading
and overloading for a relatively short period of lime generally have
a positive effect on the endurance limit of part material, i.e. they
increase the endurance limit; overloading (stretching over a long
period of ’time) reduces the endurance strength. Safe overloading
value for a certain period of time or safe duration of overloading
for a given load are found by plotting special curves known as damage
susceptibility curves, hut we shall not elaborate on this here. We also
note the positive influence of the so-called “(raining”: the part is
made to work during a certain number of cycles under stresses that
are just a little below the endurance strength.
C/i. J / | Strength Check for Variable Loading 593

While designing a part the effect of its operating conditions may

be taken into account by a special coefficient K„ c, which as coeffi­
cient K,, may be greater or less than unity.
The effect of the frequency of a variable stress cycle on the endu­
rance limit is usually considered when the endurance limit is being
determined. The existing fatigue testing machines give, as a rule,
about 3000 stress cycles per minute. Experimental studies showthat
variationof the number of cycles between 500 and 10000 does not
have any appreciable effect on endurance strength. Therefore, while
designing parts subjected to variable loading the special dynamic
stress coefficient KD should be used only when the cycle frequency
is less than 500 or greater than 10000 and also when the variable
load is simultaneously an impact load.
§ 188. Practical Examples of Failure Under Variable Loading.
Causes of Emergence and Development of Fatigue Cracks
Having established all aspects of failure under variable loading,
let usstudyafewpractical casesof suchfailures.
Figures 446 and 447showthe broken axle of awagon (bending ac­
companied by torsion), inwhich failure occurred due toshamchange

Fig. 446 l-'ig. 447

froma thick portion toa thin portion: instead of asmooth fillet the
transition was sharp, with rough notches on the surface left by the
cutting tool. The fatigue crack appeared at the outer surface and
developed along a ring shaped path. The material of the axle was
satisfactory; this is borne out by the extremely small area of mo­
mentary rupture.
Figure 448 shows the fracture of a non-rotating axle which bends
inthe vertical plane. The material isshaft steel withanapproximate
594 Dynamic Action of Forces | Pari IK

ultimate strength of 50kgf/mma. The crack appeared and developed

due to sharp transition (at right angles) froma square shape to a
circular shape.
Figure 449 shows the longitudinal section of the other end of the
same shaft, which has yet not ruptured; fatigue cracks developing

Fig. 448 Fig. 449

fromthe outer fibres towards the interior are clearly visible in the
region of sharp transition.
The zone of fatigue cracks and the zone of ultimate failure are
bothclearly visible inFig. 448. Pay attention to the series of curved
strips and lines on the surface of the fatigue cracks. These are the

Fig. 450 Fig. 451

traces of gradual development cf thecracks; the failure occurs appro­

ximately along the normal to these lines. Hence, by studying these
lines, we canalways point out the originof the crack; as a rule, this
is the point where thesourceof stress concentration is most effective,
Ch. 31 1 Strength Check for Variable Loading 595

resulting in the fatigue crack. The development of the crack can be

explained by the fact that high local stresses appear at its base, thus
helping the crack propagate towards the interior of the material.

Fig. 4.53

It is interesting to note that fatigue cracks did not appear in the

axles, the fracture of one of which is shown inFig. 448, which were
made fromsteel of lower strength (<ju«40 kgf mm-) although the
same shape was retained. This can
be explained by the fact that ther
steels have different sensitivity to;
local stresses. ,
Figure 450 shows the fatigue!
crack appearing inthe oil hole of a
crankshaft working under variable
(in opposite directions) torsion. ,
The cracks make an angle of 45°%
with the shaft axis and are perpen­
dicular to the principal stresses.
Figure 451 shows the beginning
of a fatigue crack on acar axle at
the location of a very small
(0.5-mm high) but very sharp re­
cess. We find that the fatigue crack'
starts developing simultaneously
at a number of points, which may Fig. 454
not necessarily be all in the same
plane. Laterall these cracks merge into a single crack.
Figures 452 and453 showtwo steamengine axles out of which the
one with smoothtransitions workedsatisfactorilyfor40 years, where­
as the other with sharp change servedonly forone year. The mate­
rial of thesecond axle was better than that of the first.
Finally, Fig. 454shows afatigue crackwhich begandue to internal
sources of stress concentration. Abubble or ahollowinclusion inthe
rail headbecame the centre of local stresses. This resulted ingradual
596 Dynamic Action of Forces [Part IX

development of the fatigue crack, weakening the section and leading

to ultimate failure. In the section of failure the fracture looks like
a silver spot.
The above examples are enough to show the salient features of
fatigue failures.
Experience shows that the principal factor causing fatigue failure
is not the poor quality of material (usually it was found to satisfy
the required standard), but improper machining, which gives rise
to considerable local stresses. Only rarely does it happen that poor
quality of the material gives rise to a fatigue crack, which would
not have occurred had the material been of standard quality. We can
cite an example when failure occurred due to a point-size sharp mark
on the axle’s surface.
Before concluding this section it is necessary to explain the phys­
ical process by which the fatigue crack appears and develops.
Displacement of crystal grains, as in static tensile loading, begins
under the action of high local stresses which are caused by one or the
other source of stress concentration and are usually much greater than
the yield stress. The only difference is that in tension the plastic de­
formation and displacement of crystal grains are caused by the same
stress. Therefore, they affect the whole volume of the specimen and
develop in a particular direction; under variable loading these de­
formations are concentrated in a small volume subjected to local
stresses and reverse their direction at definite intervals. They, there­
fore, do not have any appreciable effect on the strength of the spec­
imen as a whole, but the small portion experiencing local stresses
gradually passes through all stages of plastic deformation: which the
material of a simple pensile lest specimen has to bear on the whole.
At each cyclic change of loading, the permanently deformed por­
tion of material which falls in the zone of high local stresses gets dis­
placed in one and then the reverse direction; each displacement occurs
in a plane different from that of the preceding one, because these
displacements are accompanied by cold hardening of the material.
With the increase in cold hardening the rigidity of the permanently
deformed portion lends to the rigidity of the surrounding elastic mate­
rial, taking a greater fraction of the load upon itself. This leads to a
growth or actual maximum stresses in the small volume under consid­
eration although the mean stress (measured) in the remaining por­
tion remains constant. This reduces the load on the elastic zone,
resulting in less strain over the whole zone including the permanently
deformed portion within the elastic zone.
Hence, under variable loading there is a gradual increase of actual
maximum stresses in the overstressed portion and a corresponding
gradual decrease in its deformation. A fatigue crack does not appear
if the attenuation of deformation ceases before the factual stresses
reach the breaking point; the element works under stresses less than
Ch. $1\ Strength Check for Variable Loading 597

the endurance limit. On the other hand, if the stresses reach the break­
ing point, a crack appears. The process is repeated at the bottom of
the crack and results in gradual development of the crack; the element
works under stresses which exceed the endurance limit.
In general the physical process of failure under variable loading
does not differ much from failure under simple tensile loading. This
conclusion is supported by the latest experimental studies of failure
in both cases with the help of X rays.*

§ 189. Selection of Permissible Stresses

With the help of diagrams shown in Figs. 436 and 439 we can de­
termine the critical stress for any type of cycle. Let us now study the
order in which the permissible stresses are established. For the sake
of simplicity of analysis we shall
consider AD (Fig. 436) the critical
stress line for brittle materials and
AH (Fig. 439) for ductile mate­
rials. In order to obtain the per­
missible stress the abscissa and ordi­
nate of every point on both lines
will be reduced in accordance with
the accepted values of the safety
factors; the latter will have differ­
ent values for the constant and
variable components of the stress
cycle.
Mean stress pm may be looked
upon as a certain constant static
stress. It is known that under sta­ Fig. 455
tic loading the critical stress for
brittle materials is the ultimate strength, whereas for ductile mate­
rials it is the yield stress. Factors like the manufacturing process
and operating conditions, dimensions of the part and its surface
finish, etc., have much less an effect upon the ultimate strength and
yield stress as compared to the endurance limit. The effect of these
factors can be taken into account by slightly changing the main safe­
ty factor, k0t which usually accounts for errors in determining the
properties of material, magnitude and location of applied load, er­
rors of design, and other factors. No special provision need be made
for stress concentration in case of ductile materials that have a suf-
* See, for example, I. A. Oding and V. S. Ivanova, Mechanism of Fatigue Fail­
ure of Metals, Mashgiz, 1962 (in Russian); Q. V. Uzhik (Editor), Fatigue and En­
durance of Metals, IL, 1963 (in Russian); S. V. Serensen (Editor), Problems of
Mechanical Fatigue, Moshinostroenie. 1964 (In Russian); P. G. Forrest, Fatigueof
Metals, Pergamon Press, Oxford, 1962.
598 Dynamic Action of Forces [Part IX

ficiently large yield zone. Hence, for ductile materials the permis­
sible stress under static loading may be found as (Fig. 455)

[P+‘] = S” (31.15)
whereas for brittle materials (Fig. 455)

(3 U 6 )

where ae a is the actual stress concentration coefficient, k0l is the

main safety factor for yield stress, and is the main safety factor
for ultimate strength.
Under a symmetrical cycle of loading the critical stress is the en­
durance limit, which as a rule is less than the yield stress of material.
For a symmetrical cycle the permissible stress 1p . J is found by
dividing the endurance limit p_i by the strength coefficient, kn
which besides the main safety factor k 0 includes the coefficient of
actual stress concentration a c-a, the scale coefficient a , and, if nec­
essary, coefficients Kp and /(„ „ (coefficients of production process
and operating conditions). If the variation of loads is not smooth
but is accompanied by sharp impacts, then we must also use a dyna­
mic coefficient Kd, which lies between one and two. Hence for brittle
as well as ductile materials,

or
^P- ^ = k ^ J sKpK0 .,Kd (3Ll8)
The stress diagram plotted in pa-pm coordinates (Fig. 455) shows
the critical stress line AD (AH) and safe (permissible) line A\DX\
the last has been plotted for values Ip+il=ODt and Ip_iJ=Oi4j de­
termined from formulas (31.15) and (31.17). For finding the permis­
sible stress fprl for an arbitrary variable cycle having the coefficient
of asymmetry equal to r, we must draw through the origin of coor­
dinates 0 a straight line OM making an angle 0 [tan
with the abscissa until it intersects line AiDx(AxHt). The sum of the
abscissa pm(M) and ordinate pa(M) of point M represents the per­
missible stress lprl:
[Pr]=Pm»*(M)=pm(M )+ pa(M) (31.19)
It follows from similar triangles OAtDx and NMDX that
ODt OAt
ax t .
Ch. 31} Strength Check for Variable Loading 599

or
[p+ il fP -il
lP + i| —p m ( M ) P * ( M )

wherefrom
Pa (M) [ p « l + PmW l > - i ] = IP + J [p-i] (31.20)
Since
Pa (Af) = —2~Pniax (A f) 585 2 f
and

*
it ensues from (31.20) that
r_ t 2 [P+il f p - il (31.21)
IPrl O + r J J p - i H - t l — r ) l p + l|

The strength condition may be written as

Pmax ^ [P rl (31.22)

This general method of determining permissible stresses may be

elaborated as follows:
Given (a) type of deformation, (b) ratio of pma)l to pmU, (c) shape
of the part; (d) mechanical properties of the material (cru). Find the
permissible stress \p r\.
Solution:
(1) Calculate
Pmax-H Pmln * _ Pnm’-’ Pmln
Pm'**----- 2 ----- 3 T

(2) Find
f —£aLn.
Proax

(3) Determine the endurance limit under a symmetrical cycle for

the given type of deformation:
P- 1= f
(4) From the curves in Figs. 443 and 444 determine a CtQ, depending
upon the configuration of the part.
(5) Find q from the curve in Fig. 442.
(6) Calculate the actual coefficient of stress concentration
600 Dynamic Action of Forces [Part IX

The actual stress concentration coefficient may also be calculated

from the following formulas (§ 186);
1.2 + 0 .2 S c r !2 or <ru—40
a c.a a c.a 1.5+1.5 no*

if the technology of machining the part is known.

(7) Determine scale factor a 5 from the curve in Fig. 445, depending
upon the size of the part.
(8) Determine permissible stress in a symmetrical cycle:

(9) Find the yield stress n„=prcru.

(10) Determine the permissible stress under static loading

(11) Determine the required permissible stress

rn 1 2 [/? .n .1 ( P - i l
( i - r ) ^ +1] + ( I + r ) | p - , ]

This method of determining permissible stresses is not very accu­

rate on account of the modifications introduced in the diagrams and
also due to an insufficiently accurate accounting, for the stress con­
centration factors. If desired, more accurate breaking stress diagrams
may be used in which the dotted lines shown in Figs. 436 and 439
are not drawn. The accurate method of analysis may give much higher
values of permissible stress for cycles having characteristics approxi­
mately equal to zero if the endurance limit is close to the yield stress;
in all other situations the results obtained from the approximate and
accurate diagrams are not much different.

§ 190. Strength Check Under Variable Stresses

and Compound Stressed State
The above methods for checking the strength of materials under
variable stresses were discussed in connection with the simplest types
of deformation: tension, compression, torsion and bending. The
question that now arises is: How to apply these methods in case of
compound stressed state?
From a practical point of view the most important situation is
that of combined bending and torsion. As explained earlier in § 125,
strength check has to be carried out for an element of the material in
a bi-axial stressed state; four of its faces are acted upon by shearing
Ch. 31] Strength Check for Variable Loading 601

stresses and two by normal stresses , where W = ^ - is

the section modulus of the round shaft.
For a strength check, under static loading we employed the following
two conditions: condition (A) when- the theory of maximum shearing
stresses was applied, and (B) when the theory of distortion energy
was applied:
(A) V a* + 4 t * < [a] and (B) f o 1 + 3t* < [o]

Both formulas may be written in a general form by dividing: them

by lot:

<A) )// FF+^Zy< l and {Biy i ^ + 7 s y <1

or in the general form

where It| « ^ when the theory of maximum shearing stresses is

£
applied, and -c=-j~j= when the distortion energy theory is applied.
Thus, the strength check by both theories may be represented by a
common equation

W+W<X (31-23)

Since the fatigue crack is caused by the same physical processes

of deformation of the material which result in failure under static
loading, equation (31.23) may also be employed for checking the
strength of materials under variable loading. Stresses o and r may
be broken into the components <rm, crtt and r m, ra:
o = om-h<Ja, r — r^+ T a,

Here (al and fxl represent the permissible stresses under bending
and torsion, lo# and hjl, respectively, obtained from the simplified
Ip J -lp J diagram (Fig. 456) by taking into consideration the
stress concentration coefficient of the particular type of deformation
and the cycle characteristic, ~2^L or 7 ^ .
Y 0'mffx ‘max
602 Dynamic Action of Forces [Part IX

§ 191. Practical Measures for Preventing

Fatigue Failure
The results derived in the preceding sections enable us to find
means, which ensure sufficient strength of machine parts and struc­
tures under variable stresses.
These measures may be divided into two categories. On one hand,
we must be able to manufacture elements of machines and structures
from materials which have the greatest resistance to variable stresses.
We have seen that in this respect the requirements of the material
lead to the following two points: first, it is desirable to use a metal
which has a high ultimate strength and sufficient ductility, because
this is compatible with high endurance limit; second, the metal must
be free of all internal factors of stress concentration (this requirement
can be fulfilled by a material of uniform fine-grained structure with­
out any residual stresses (e.g. after quenching) or disruption of uni­
formity in the form of cracks, gas bubbles, non-metal!ic inclusions,
etc.).
These requirements explain why important parts subjected to
variable loading are so often manufactured from alloyed steels (chrome-
nickel, chrome-vanadium) having high ultimate strength and
fine-grained structure free of internal stresses, which is imparted by
proper heat treatment (hardening with subsequent annealing).
However, such steels are sometimes found to have microscopic
cracks (especially in chrome-nickel steels), which are known as flakes.
These flakes, just like inclusions, may sharply reduce the endurance
limit of steel in spite of its having high ultimate strength.
The second category of measures ensuring sufficient strength under
variable loading consists in careful design of the outer profile and
proper finishing of the external surface of the element. The chief
aim of the designer and technologist should be to reduce as far as
possible the coefficient of stress concentration caused by sharp chan­
ges of the profile and defective machining. Reduction of local stresses
can be achieved chiefly through smooth transitions of profiles, grooves,
cuts and fillets. It is inadmissible to provide unsmoothed transitions,
although the radius of the fillet curve may be very small. Whenever
possible the radius should be large enough to affect noticeably the
reduction in local stresses. The proper radii may be selected with
the help of Figs. 443 and 444. It should be noted that sometimes just
a small increase of the radius can rid the element from danger of fail­
ure.
We know of the failure of a large number of crankshafts of aircraft
engines in the Royal Air Force. These failures occurred at the fillet
near the mounting seat of the propeller; the failures stopped when the
fillet radius was increased by just 1/8 in-«3 mm.
This can be explained as follows. At stresses close to the endurance
Cft. 5/1 Strength Check for Variable leading 603

limit the curve depicting the relation between breaking stress and
number of cycles Is almost horizontal (Fig. 434). Therefore, if the ac­
tual stress even slightly exceeds the endurance limit, failure is ine­
vitable, because a majority of the parts undergo cyclic changes which
are large enough for the fatigue crack to appear.

TTTTF
w (to
Fig. 456

On the other hand, the fatigue crack does not appear if we make
the actual stress just a little less than the endurance limit by reducing
the coefficient of stress concentration. The sharper the change of pro­
file, the greater the difference between the rigidity of adjacent por­
tions and the sharper the change of stress, tne greater is the coef­
ficient of stress concentration. Therefore the local stresses can be
reduced not only by making the transitions smoother, but also by
decreasing the difference between the rigidities of adjacent portions
of the element in the sections where stress concentration is unavoid­
able. These considerations have recently led to the concept of crippl­
ing cuts. For example, when a gear or pulley is mounted on a shaft
with interference fit, considerable local stresses appear in the shaft
material (Fig. 456(a)).
The coefficient of stress concentration for normal stresses under
bending in a section perpendicular to the shaft axis varies between
1.8 and 2. Figure 456(a) also shows the variation of normal stresses
in the outer fibres of the shaft, on which the pulley is mounted with
an interference fit. The stresses were determined by optical method.
It is evident from the diagram that there is a sharp local increase
in stresses, especially in the compressed zone near the hub edge.
The local stresses decrease and become more uniformly distributed
if a cut is made in the hub near the contacting surface to smooth
down the sharp change in rigidity at the edge (Fig. 456(6)). The
coefficient of stress concentration falls from 2 to 1.4; if, as shown in
Fig. 457, the shaft diameter at the seat is increased in addition to
604 Dynamic Action of Forces [Part IX

the cut already made, the coefficient of stress concentration may be

reduced to 1.0*1.05. The local stresses may similarly be reduced by
bossing the thicker part near the section of sharp transition (at right
angles) (Fig. 458(a) and (6)).
In all these examples the aim of changing the profile is to smooth
the change of stress in the section of transition.
The resistance of materials to variable loading depends as much
upon the surface finish as upon the smoothness of transitions in the .

profile. Any scratches or cuts left by the cutting tool can play an im­
portant role in subsequent failure of the part. We know of the failure
of a large number of pistons of steam engine cylinders, when tapered
hubs were fitted on the worn tapered ends of the plungers to increase

their thickness. Before fitting the hub the worn out surface of the
plunger was subjected to coarse turning without any subsequent
finishing. The fatigue cracks appeared under the hub, beginning
from the source of stress concentration in the form of a scratch left
by the cutting tool. Therefore, fine finishing, nickel plating and
varnishing, if the part works in a corroding medium, are not super*
C/i. 321 Fundamentals of Creep Analysis G05

fluous luxuries but absolute necessities for safe functioning of a major­

ity of parts subjected to variable loading.
It should be noted that the questions of proper selection of mate­
rial and rules regarding proper design of parts cannot be studied in
isolation from one another. The better the material, the higher is
its ultimate strength and the higher the quality of machining which
it requires. If we use a costly alloyed steel and do not pay sufficient
attention to the reduction of focal stresses, we run the risk of bringing
to nought the advantages that accrue from the use of high quality
steel. The sensitivity factor of such a steel is much higher than that
of mild steel. This was explained in § 186.
Figures 459 and 460 show the p»-pm diagrams for mild and high
grade alloyed steels. On these diagrams lines AMB correspond to
failure due to development of a fatigue crack, and lines GN represent
failure due to plastic deformation when the stresses exceed the yield
stress.
Lines AMN, which are shaded on the diagrams, represent the curves
of breaking stresses (in the wider sense of the word). It is obvious
that the chances of failure due to development of fatigue crack are
far greater in case of alloyed steel than in mild steel. The chances
of reduction of local stresses due to plastic deformation are consider­
ably less in the first case as compared to the second. This to a large
extent explains the higher sensitivity of alloyed steels to stress con­
centration.
Summing up, we may conclude that the higher the grade of steel
the higher is the quality of finishing which it requires so that all its
properties may be fully exploited.

CH A PTER 32

Fundamentals of Creep Analysts

§ 192. Effect of High Temperatures on Mechanical
Properties of Metals
On account of the fast development of machine building, increas­
ingly vital Importance is being attached to strength analysis of ma­
chine parts working for long periods at high temperatures. Such
parts include discs and blades of steam and gas turbines, pipes and
other elements of steam generators, various parts of internal combus­
tion engines, jet engines, chemical plants, etc.
The behaviour of the materials of such parts is affected by the
absolute temperature as well as the duration for which the parts
work at the high temperatures.
The properties of metals change considerably at high temperatures;
therefore the known properties of strength and ductility at normal
006 Dynamic Action of Forces [Part fX

(room) temperature cannot serve for the analysis of elements of the

same metals working at high temperatures.
As the temperature increases, there is at first a gradual reduction
of the modulus of elasticity and the limit of proportionality; beyond
a particular temperature (for carbon steels after 300-350°C, for alloyed
steels from 350-400*0, for nonferrous metals from 50-150°C) this
reduction goes on getting steeper. Thus, for example, the modulus
of elasticity of a commonly used steel is approximately 25-30% less
at 600°C and 50% less at 800°C as compared to its value at room tem­
perature. The reduction of the modulus of elasticity and limit of

proportionality is even more pronounced in case of nonferrous metals.

The yield stress, a*, and ultimate strength, au, of carbon steels in­
crease at first with the increase in temperature and become maximum
at a temperature of about 200-250°C, the maximum value is 10-20%
greater than cru and ov at room temperature. If the temperature is
further increased, the yield stress falls sharply: at 600°C the yield
stress of steel is only 40% of its value at room temperature. In case
of nonferrous metals and their alloys the increase of temperature
is accompanied by a continuous decrease of yield stress. With increase
in temperature the endurance limit varies in a manner which is more
or less similar to the variation of ultimate strength.
Ch. 32) Fundamentals of Creep Analysis 607

The plastic properties (total relative elongation and reduction of

cross-sectional area at the moment of failure) suffer a slight setback
as the temperature is increased from 20 to 200-250°C; with a further
increase in temperature the plastic properties, as a rule, again begin
to improve. However, the plastic properties of austenitic chrome-
nickel steels, bronze, brass and nickel are adversely affected by high
temperatures. On the other hand, the plastic properties of aluminium
and magnesium improve.
The curves in Fig. 461 show the variation of strength and plastic
properties of mild steel (0.15% C) as the temperature is raised to
800°C.

§ 193. Creep and After-effect

The variation of strength and plastic properties with the increase
in temperature is of vital importance in the design of elements of
machines and structures. However, the most important factor affecting
the behaviour of metals at high temperatures is creep.
Creep signifies a continuously (may be very slow) increasing de­
formation under constant forces (or stresses) and high temperature.
In a number of metals (lead, brass, bronze, aluminium and a few
other nonferrous metals and alloys) creep may occur even at room
temperature.
The higher the temperature, the faster is the growth of deformation
due to creep. Sometimes a gradual, continuously increasing defor­
mation over a sufficiently long period of time at high temperatures
may load to failure of an element, even though the stresses induced
in it are less than not only the ultimate strength but also the propor­
tionality limit at room temperature.
For example, the diameter of a steam pipe working at high temper­
ature and pressure increases continuously; finally ii may Tail due to
rupture of its walls (this sometimes actually happens). The creep
of discs and blades of steam turbines may result in overlapping of
the gap between the blades and the turbine housing, leading to break­
age of the blades.
Creep of metals is an irreversible (permanent) deformation, which
may be studied as slow yielding. In a number of cases (especially
in a compound stressed stale) plastic deformation due to creep re­
sults not only in a change in stress but also their redistribution over
the volume of the element. The change in stress is prominent
when there is a restraint to total deformation of the body due to
certain specific features of its working. In such cases, the elastic
deformation experienced by the body during loading decreases with
passage of time; this results in the beginning of plastic deformation,
which subsequently continues to grow. It is accompanied by a reduc­
tion of stresses in the element. Such a reduction of stresses due to
608 Dynamic Action of Forces {Pori IX

gradual increase of plastic deformation at the cost of elastic is known

as after-effect.
Due to* after-effect a mildi interference fit between two* parts may
loosen so much as to- impede the normal functioning of the structure.
For example, the loosening of the flange bolts of a gas pipe or the high
pressure cylinder of a>steam turbine may ultimately lead to leakage
of gas or steam if the bolts are not tightened periodically; the loosening
of fit between the turbine disc and the shaft may lead to clearance
between the two resulting ini the coming off of the disc.
As already stated, creep may occur in some nonferrous metals and
their alloys even at room temperature. However, in steel, iron and
a number of nonferrous metals and their alloys creep begins only
when they are heated above a certain, unique for each metal, temper­
ature (carbon steels and iron above 300-3506C, alloyed steels above
35O-4O0°C, light alloys above 50-l50cC, etc.). Creep is not observed
in these metals if they are heated below the specified temperatures.
Besides, even at temperatures equal to or higher than the specified,
creep does not begin as long as the stresses remain less than a partic­
ular, specific for each metal, value. After-effect begins at approxi­
mately the same values of temperature and stress as creep.
Creep is especially prominent in.* metals. However, it occurs in a
number of olheF materials also. For example, at room temperature
creep can be observed in various plastics (celluloid, bakelite, poly-
vinil chloride plastic, etc.), concrete and cement mortar. In reinforced
concrete structures, creep, with the passage oF time, leads to redis­
tribution of stresses between concrete and reinforcement; the latter
gets slightly overloaded whereas the stresses in concrete decrease.
However, the creep of concrete and the ensuing redistribution of
stresses have almost no effect on the load carrying capacity of the
reinforced concrete structure. Creep at roam temperature also, occurs
in timber under compression, and especially under bending.
Experimental study of the phenomenon of creep began* quite re­
cently (in 1910). These studies aroused widespread interest in the
early twenties, when the first important results were published.
Creep testing presents a number of difficulties even in simple ten­
sion. These tests can be conducted only on a special apparatus ca­
pable of mainlaining a constant load and temperature and measuring
the specimen’s deformation. The creep tests must be conducted with
a high degree of. accuracy if reliable results are desired, the duration
of the tests should not differ much from the service life of the element,
and this involves testing over tens of thousands of hours. All these
factors make creep testing a complicated, time-consuming and costly
affair. On account of all these difficulties the phenomenon of creep
has until now not been studied experimentally sufficiently well even
for simple tensile loading.
Testing for creep is still more complicated and cumbersome in com-
Ch. 32] Fundamentals of Creep Analysis 609

pound stressed state. The majority df these tests were ‘conducted on

thin-walled pipes subjected to a combination .of internal pressure,
torsion and tension. However, the number of such tests conducted
until now is very small. , . . .
Scientists have not been successful in working out short tests on
creep. The reason is that the stipulated duration of testing is a nec­
essary and important condition for obtaining reliable results which
may be used for creep analysis and design. The results of Ihe accel­
erated tests can, as yet, serve only -for an approximate qualitative
estimate of the effect of high temperatures on the behaviour of metals.

§ 194. Creep and After-effect Curves

A. It was mentioned above that experimental study of creep, which
forms the basis of design of elements of machines and structures
working at high temperatures, is generally carried out on specimens
subjected to simple tension. In creep testing ithe temperature and
tensile -force acting on the specimen must remain constant over the
duration of the test. . . . ,
The elongation of the specimen is measured at -regular intervals
of time; from the readings we plot a curve 'in the coordinate system
relative elongation e versus time i, -and the curve is known as the
creep curve of Hie material. The Shape of the .curve depends upon the
material, stress and 'temperature at -which the test is conducted.
A typical creep curve for metals is shown in'Fig. 462 (curve OASCD),

m e n the specimen heated to a particular temperature T is loaded,

its deformation increases fast In the beginning (depending upon the
speed of loading) from zero to a certain value OAj (it is assumed that
loading of the specimen is stopped at a stage which corresponds to
point A on the creep curve).
 610 Dynamic Action of Forces [Part IX

This is followed by gradual increase in deformation of the loaded

specimen, and the material begins to creep. The growth of creep de­
formation is depicted by curve ABCD; the ordinates of points on this
curve (for example, the ordinate of point K) represent the sum of elas­
tic strain e<?=0i41 and creep strain ec= A tKi:
e = e, + ec
The rate of creep deformation at any point of the curve is determined
by the slope which the tangent to the curve at the point makes with
the abscissa, i.e.
de .

The whole process of creep may be divided into three successive

stages. In the first stage, represented by segment AB on the creep
curve, the deformation takes place with a non-uniform, continuously
decreasing speed; this is the zone of non-uniform, or unstable, creep.
Depending upon the material, stress and temperature, the duration
of the first stage varies from a few hours to a few hundred and even
(in exceptional cases) a few thousand hours.
The nature of creep in the first and second stages is affected mostly
by the following two factors: (1) increase of strength of the material
due to strain hardening, which occurs as a consequence of increase
in residual (permanent) deformation, and (2) removal of strain hard­
ening or decrease of strength due to high temperature. Creep can be
studied as an interaction of these two lactors, which are chiefly re­
sponsible in causing “pure” creep. This picture of creep may be com­
plicated, especially during the subsequent stages, by various internal
(for example, microstructure change and phase changes of the metal)
and external (for example, corrosion) factors.
When the strengthening effect of strain hardening is balanced by
the weakening effect of long exposure to high temperatures, the de­
crease of creep rate ceases and the second stage (segment BC) begins,
this is the stage of uniform, or stable, creep, in which creep occurs
with a minimum uniform velocity.
This velocity remains constant until a neck begins to form on the
specimen (point C on the creep curve). If the load on the specimen
remains constant, then the local reduction of the cross-sectional
area of the specimen in the third stage (segment CD on the creep
curve) is accompanied by an increase of stresses, which in their turn
result in a higher creep rate. This leads to ultimate failure of the
specimen (point D on the creep curve).
The shape of the creep curve may change considerably if the tem­
perature or stress is changed. Figure 463 shows creep curves at con­
stant temperature, T, but different fixed values of stresses o( (ci< o2<
< o a<<T4< o s). The creep curves at fixed stress or but different fixed
Ch. 32i Fundamentals of Creep Analysis 611

values of temperature T{ ( r , < T i < r a< r 4< r s) are identical to the

above curves.
At low values of stress (05 =0 ,) creep may be completely absent,
i.e. for the loaded specimen the e-/ diagram may be represented by a
straight line passing through point Ai and parallel to the abscissa.
At a somewhat higher value of stress (0 = 02) there will be a short

period of unstable creep, which will stop when the rate of creep be­
comes zero. At a still higher value of stress (cr=a3) the velocity of
creep will not be zero but will be so small that failure due to creep
will occur after a very long period, which far exceeds the service
life of the element.
At stress a = a 4 we get the creep curve shown in Fig. 462. If the
stress or temperature is further increased, the creep curves also un­
dergoes a further change: creep progresses at a faster rate, and the
straight line portion of the curve—the zone of stable creep—goes
on getting shorter till it reduces to a point {curve 5 in Fig. 463), i.e.
the zone of unstable creep directly changes into the zone of failure.
In this case the zone of stable creep is represented by an inflection
point on curve ABCD, point B&, which coincides with point C8.
The nature of failure due to creep depends mainly upon the proper­
ties of the material at the given temperature. Carbon steels at temper­
atures less than 550°C, copper, lead and some other light alloys
generally fail after large plastic deformation and neck formation.
Special heat resistant steels having good creep strength fail after
comparatively small deformation, the failure is brittle in nature
and usually begins at the location of stress concentration.
612 Dynamic Action of Forces [Part IX

B. As already stated, after-effect is the gradual reduction of stresses

in a loaded element whose total deformation is constant and equal
to the elastic deformation in the loaded state. The reduction of stresses
takes place due to gradual decrease of elastic deformation and an
equivalent increase in plastic deformation according to the following
formula:
e = e 4!-f-ec :=e2=const (32.1)
The after-effect curve is shown in Fig. 464. The process of after­
effect may be divided into two stages: in the first stage (segment AB
on the after-effect curve) the stresses decrease very fast, and this is
accompanied by a sharply decreasing after-effect rate; in the second
stage (segment B C on the curve) the reduction of stresses is consid­
erably slower and is accompanied by slowly decreasing after-effect
rate.
Depending upon the material, initial stress and temperature, the
duration of the first stage varies from several tens of hours to a few
hundred hours. The physical process
accompanying after-effect in the first
and second stages has not yet been
studied in sufficient details. There
are still very few good experimental
set-ups for studying after-effect, and
this makes it difficult to compare
the results of experiments on after­
effect with those on creep. In majo­
rity of the machines which have
been used until now for investigat­
ing after-effect it has been impossi­
ble to achieve pure after-effect.
It is generally assumed that
growth of plastic deformation in
after-effect is similar to its growth in
creep and therefore the rate of after­
effect may be calculated from the creep velocity. If this assumption
were true, there would be no need for studying after-effect separately.
However, there is another view which holds that creep rate cannot
be taken as the rate of after-effect, because these processes are basi­
cally different, the mechanism of origin and growth of plastic defor­
mation in after-effect is somewhat different from that in creep.
In after-effect the reduction of stresses in the element is caused by
the growth of plastic deformation at the cost of elastic deformation,
and the length of the element remains constant. In creep the growth
of plastic deformation is exclusively due to elongation of the element.
The total deformation in creep is considerably greater than in after­
effect; this is an important difference, because the magnitude of
Ch. 32] Fundamentals of Creep Analysis 613

deformation at higher temperatures may considerably affect creep,

giving rise to after-effect, diffusion and other processes which can
strongly influence the resistance of a material to plastic deformation.
C. In studying creep a very important requirement is to establish a
functional dependence between the main quantities which define
the creep curve (stress (<j), temperature (7), and time (/)) and creep
deformation (e or ec) or creep velocity =
Various investigators proposed a number of formulas correlating
the above quantities. The majority of these formulas were obtained
empirically, and only a few of them additionally took into account
the physical nature of the process of creep. Therefore, none of these

formulas is able to justify the experimental results oyer a wide range

of stress, temperature and time variation. Jn a majority of the cases,
analysis based on these formulas concurs well with the experimental
data only over isolated portions of the creep curve, mainly in the
zone of stable creep.
Since experimental study of unstable creep is much more difficult
than that of stable creep, the zone of unstable creep on the creep curve
has not been investigated sufficiently. Therefore, in actual creep
analysis, the zone of unstable creep is often neglected by extending
line BC (Fig. 466) till it intersects the vertical axis at point Bj, and
the total deformation due to creep (for example, tK) is calculated by
the following approximate formula:
^ ■ = eo+ e^ ei + 8c = e/ + ^ tana
= -f- tpVc (32.2)
Stable creep has been experimentally studied to a much greater
extent. From among the various relations for creep rate, proposed
by different research workers, the following have been found to give
614 Dynamic Action of Forces [Part IX

the best results compatible with experimental data:

(1) vc = ko» ')
(2 ) t/c = asinh £ j <32-3)
In these formulae k, n, a, and b are certain constants which depend
upon the properties of the material and the testing temperature. The
second relation is more compatible with experimental results than
the first, but it considerably complicates computations. Besides, the
data available on k and n of the first relation is much more than the
data available on coefficients a and b of the second relation. There­
fore, the first relation is more commonly used in creep analysis.
Coefficients in Formula (32.3)

Chemical composition in %
N o. Type of steel
C Mn Si Mo Cr Nl W N b’

1 Carbon steel 0.15 o.so 0.23

2 Carbon steel 0.43 0.68 0.20

3 Molybdenum steel 0.13 0.49 0.25 0.52

4 Chrome-molybdenum o.u 0.45 0.42 0.50 2.08

steel

5 Ch rome-mol ybden um 0.48 0.49 0.62 0-52 1.20

steel

6 Chrome-nickel steel 0.06 0.50 0.61 17.75 9.25

(18-8)

7 Steel-09 0.52 0.82 0.57 13.51 15.2 2.01

a Steel Nb 0.19 0.72 0.69 1.71 0,87 0.77

Ch. 32] Fundamentals of Creep Analysis 615

Table 23 contains data on the values of coefficients k and n for some

steels, tested for creep at various temperatures and stresses.

§ 193. Fundamentals of Creep Design

A. It is obvious that at high temperatures the most suitable operating
conditions for a part are those which correspond to the first or second
creep curve of Fig. 463 when creep deformation does not appear at
all or disappears soon after the part is loaded. However, the corre­
sponding stresses<Xi equal to the creep limit and stresses as are usually
so small in magnitude that if they were to be accepted as the upper
stress limit, this would lead to an unjustified increase in the dimen-
Table 23

V a lu e s o! coefficien ts
In form ula (3* 3)
H eat Temperature Sires* range
treatment t°C) (kgf/cm1)

Annealing 844°C 427 1410-2110 6.35 0.17x10" *®

538 280-560 3.05 0 . 12x 10- 1*
593 110-250 3.10 0.2GX10-1*
G49 30-90 2.85 0 .1 6 x 1 0 " 10
Annealing 844°C 427 1060-1690 6 .0 0 .2x l 0- «
538 210-630 3.9 0.14X10“ 14
649 30-180 1.7 0 . 12x 10- “
Annealing 844°C 482 910-1410 5.40 1 . 2 x l 0 -*»
538 560-1060 4.00 0 .6 X 1 0 -1®
593 210-420 3.55 0.23X 10-M
649 GO-120 3.10 0 . 2 x l 0 -»*
Annealing 844°C 482 970-1410 8.35 0.58x 10-30
538 460-840 4.95 0 .1 4 X I0 -”
593 280-560 6.90 0 . 1 0 x 1 0 -* »
649 140-280 3.25 0.17x10-1®
Annealing 844°C 427 1410-2110 6.35 O .H SxlO - 46
538 320-1060 3.55 0.175 x 10-1®
649 70-250 2.95 0.3G5X10-1*
Hardening 1093°C 538 880-1340 4.4 0.21X I0- !®
593 560-1060 4.3 0.17x10-1®
C49 350-840 6.1 0 .1 4 x 1 0 -10
816 110-280 4.7 0 . 2 l X l 0 ->«
Hardening I175°C 600 800-2200 3.15 0.65X10-M
650 400-1500 2.9 0 .2 9 x l0 - i‘
Normalization 850®C 500 1500-2500 4.3 0.41X10-3®
600 200-500 3.1 0.59X 10-K
616 Dynamic Action of forces IPart IX

sions of machine parts. Therefore, a small creep deformation is gen­

erally permitted in machine parts (third curve in Fig. 463).. However,
it is necessary that the total strain in the part equal to the sum
of strain due to load, a,, and strain due to creep, ec, (Fig; 465) should
not during the whole service life of the part, tv, exceed a given per­
missible strain del which depends upon the function of the part, its
operating conditions etc. For instance, the permissible strain fej
for the (pipes-of steam superheaters is .0.02, for steam pipes it tis Oj003,
while for steam-iurbine cylinders its value is 0.00:1.
Hence, for uni-axial loading 4he design equation is:
(32.4)
It follows from Fig. 465 that
(32.5)
where
= Ip tan a = ipvc = f (32.6)
is the uniform (stable) creep strain during the service life tp of the
part. Using formulas (32.5) and (32.6), equation (32.4) may be modi­
fied as follows:
(32.7)
wherefrom
(32.8)
Stress

H^r
is sometimes called the virtual creep limit for permissible total creep
strain and is denoted by cce. Thus, the design equation for permis­
sible creep strain may be written as
= (32.9)

If the elastic and .unstable creep strains of the element are negli­
gible as compared to the stable creep strain, then creep analysis
may be based on the maximum permissible stable (minimum) creep
rate. Obviously, the permissible creep rate should be determined
from the condition that creep deformation increasing with this con­
stant rate should not exceed, during the whole service life of the ele­
ment, a certain permissible ^value of deformation which does not
disrupt the normal functioning of the structure or machine. The
corresponding maximum stress, which does not give rise to a creep
rate greater than the permissible at the particular temperature,
may be considered as the limiting stress. Often this stress is referred
C/i. 32] Fundamentals, of Creep Analysis 617

to as the creep limit of the material from consideration of the minimum

permissible or uniform rate of deformation (<*«.)• Evidently csn is a
function of temperature and the minimum permissible creep rate.
Table 24
Creep Velocity

Part per hour

1 Turbine discs with tight fit 10-*

2 Bolts, flanges and cylinders of steam turbines 10 -a
3 Steam pipes, welded joints of boiler pipes Ift- 7
4 Pipes of steam superheaters io -° -io -s

As an example. Table 24 gives the approximate permissible values

of the minimum relative creep rate, lucl, for a few parts of steam boilers
and turbines.
In creep analysis based on minimum creep rate, the fundamental
equation of uni-axial stressed state of the material may be written
as follows:
vc = kon < [vc] « ( 3 2 . 1 0 )
since
* = v et p

From equation (32.10) we get:

M = = (32.11)

The strength condition in terms of stresses may be written as fol­

lows:
= (32.12)

B. It is implicit that in creep analysis from considerations of per­

missible deformation and permissible creep rate it is not enough to
ensure that the creep deformation does not exceed a permissible value
at a particular temperature during the whole service life of the ele­
ment. It is also essential to provide a certain safety factor against
the occurrence of such a failure. Hence, points K\ and /C* (Fig. 466)
on creep curves /, 2, and 3 corresponding to abscissas tpu and tp9f
and ordinates lelx, (elj, and Ul9 must lie on the segments corresponding
to the first and second stages of creep.
2t —asio
618 Dynamic Action of Forces [Part IX

This must always be checked analytically and therefore requires

special investigation. As already explained, creep analysis from con*
siderations of permissible deformation and permissible creep rate may
be replaced by an analysis based on permissible stresses, the creep
limits or ow. However, it is essential to check beforehand that
of (OfBor o j does not exceed a permissible value, which is a certain
fraction of the ultimate strength at the given temperature.
0 ui,tytfon£

It is known that at high temperatures the ultimate strength of

materials is greatly dependent upon the duration of testing; a compar­
atively small increase in duration may cause considerable reduction
in the ultimate strength. At a certain temperature (above 800°C for
mild steels) the specimen may even fail under a load which induces
stresses that are less than the limit of proportionality at room tem­
perature, provided the specimen is subjected to this load for a suf­
ficiently long time. Therefore, at present the strength of a material
at high temperatures is characterized not by the ultimate strength,
determined from short-term tests, but by the long-term strength (aar).
The long-term strength at a particular temperature characterizes the
stress which will cause failure only after a specified period. The curves
in Fig. 467 show the variation of long-term strength of chrome-mo­
lybdenum steel (0.1% C, 1.55% Si, 4.88% Cr, 0.51% Mo) as a func­
tion of time at various temperatures.
A few scientists are of the opinion that the non-uniformity of stress
distribution at the locations of stress concentration is smoothened
during creep and therefore stress concentration need not be taken into
account in creep design. It is relevant to point out that machine
parts working at high temperatures are, as a rule, manufactured from
Ch. 32] Fundamentals of Creep Analysis 619

special heat-resistant steels, which have poor tendency to creep; gener­

ally such parts fail after undergoing small deformation and the fail­
ure is brittle in nature. Consequently, in a majority of practical
cases smoothening of the stresses does not occur and it is essential
to account for stress concentration in creep design.
Therefore, while determining the long-term strength of heat resis­
tant steels, the possibility of stress concentration should be taken
into account, i.e. the experiments for determining o0t should be
conducted on specimens of corresponding shape.
If the elements of machines are subjected to the simultaneous
action of fatigue and creep, then the long-term strength should be
determined from fatigue test at the appropriate temperature. Thus,
the following important cases may
be distinguished while calculating
the strength of elements of ma­
chines and structures working under
high temperatures.
If the temperature is not high
enough to cause creep (§ 193), the
critical state is determined by the
yield stress or ultimate strength
of the material at the given temper­
ature, obtained by the usual tests.
The strength condition is:

°< m = t <32-13>
If creep is possible at the given F,g* 468
temperature (see § 193), then the
first thing to do is to establish which of the permissible stresses is
maximum for the total service life tp of the element: the permissible
stress from considerations of total creep deformation (tree) or minimum
creep rate (om), or the permissible stress from considerations of
long-term
-
^ , where k t is the long-term safety
strength [orai1 = riv
factor, which may be considered approximately equal to the usual
safety factor k (Fig. 468).
If [a)=ac (Ocs, o j c t a j , zone A in Fig. 468, the creep design
should be carried out according to formulas (32.4), (32.9), (32.10)
or (32.12). If, on the other hand, l<x]=ac> [crotJ. zone B in Fig. 468,
then creep design should be based on the formula
(32.14)

If the elements experience after-effect, then care should be taken

that the interference fit between them does not loosen beyond a per-
21*
'..0 Dynamic Action of Forces [Fart I X

missible limit: reduced stress <t, due to after-effect should not be less
than a certain minimum value which ensures the required interference
between the elements:
m in

From this condition we can specify periods after which the joints
should be retightened to the required interference by special methods
(for instance, by tightening the flange bolts of gas or steam pipes).
In conclusion it should be pointed out that this sect ion dealt main­
ly with the methods of creep analysis in simple situations (uniaxial
stressed state), which can be utilized for building up the analysis of
more complex cases. The methods of creep analysis in compound
stressed state are the subject of study in special monographs.*

§ 196. Examples on Creep Design

Example 1. Determine the frequency of tightening flange bolts
of a steam pipe to prevent leakage of steam, if the initial pull of P =
—300) kgf on each bolt cannot be reduced by more than 40%. The
temperature at which the bolts work is T ~ 4259C. The cross-sectional
area of each bolt is ^4=3 cm8; the bolts are made of mild steel having
modulus of elasticity £ r =1.77x lO* kgbcm* (at 7,=425°C); the
stable creep rate for the material may be determined from the formula
v<-*hcr,f and at
r = 425°C, fe= 2 .2 6 x !0 -« ]^ l ? and ft = 6
Solve the problem on the assumption that the flanges of the steam
pipe are absolutely incompressible; the zone of unstable creep may
be neglected.
Solution. If the flanges oi the steam pipe are absolutely incompres­
sible, then the total deformation of the bolt elongated during tight­
ening by A/0 must remain constant. During creep the elastic defor­
mation of the bolt A/e will gradually change inlo'plaslic deformation
AIp) this will lead to the reduction of stresses in the bolt. The following
condition must be satisfied
M e+ A A/0—const
or
cr + —£ r + 8/»—eo—£ t
where Oo is the initial tightening stress in the bolt, <i is the stress in
the bolt at the instant of failure U and Er is the elasticity modulus
• See, for example, I. M. Kachanov. Theory of Creep, Fizmatgiz, 1960 (in Rus­
sian); N- N. Malinin, Apnli-J Theory of Plasticity and Creep, Mashinostroenic, 1975
(in Russian).
Ch. 32} Fundamentals of Creep Analysis 621

of the material at the given temperature T . Upon differentiating the

above equation w.r.t. t, we obtain
1 da , d&p _ I d a , I da0
"Erdt ' dt ~ E t H ^ c = sCT dt

Neglecting unstable creep and substituting kan for i/c, we get the follow­
ing differential equation:
$ + £ rto”- 0
or

Upon integrating this equation we obtain

(a)
v ^ = h ^ =E^ + c
where C is the constant of integration. Since cr—<r» at f=0,

(n -I)o T 1

Substituting this value of C in equation (a), we get the following

formula correlating a and t:

1) ETkdi~lt\n~l
Substituting the numerical values given in this problem, we get
3000
“T " _ iooo
r = (i+ 2 x io -an 1^
Jl I { 6 -l)1 .7 7 X l0 « x 2 .2 6 x l0 -a6^ ~ 2 j 6 _ ‘ / j C' ‘

™ /= t ( i ooo - J £ ) .

The values of t corresponding to different values of or are given in

Table 25, column A. If the stresses in the bolts are not to decrease
by more than 40%, then the bolts must be tightened after every 5930
hours or approximately after 8.5 months.
This solution of the after-effect in bolts is approximate. Due to
pliability of the steam pipe flanges the stresses in bolts will reduce
at a much faster rate. However, if the pliability of the flanges is taken
into account, the solution becomes quite complicated. Without con­
sidering this aspect in detail we give here the final results (see Table 25,
622 Dynamic Action of Forces [Part IX

column B) obtained for the case when the elastic deformation of the
flanges is e#./=3xl0~,/ >f, and their creep rate is i/e.,= 5 x
(where P t kgf is the pull on each bolt at instant /)•
Table 25
Values of t

«F t (hours) l (hours)
a
(kg!/cm1) A B (kgf/cm1) A B

1000 0 0 600 5900 5000

900 347 335 500 15500 11900
800 1026 960 400 48300 31700
700 2475 2230

Hence, if we follow the results of the more accurate solution, re-

tightening should be carried out not at an interval of 5930 hours
but at an interval of 5000 hours or about every 7 months. This period
will be still less if we consider the zone of unstable creep during creep
design.
Example 2. A round 24-mm diameter shaft working at r=540°C,
is twisted by a constant torque Af(=20 kgf*m. The shaft is made of
alloyed steel having shearing modulus <3=6x 10? kgf/cm* (at T —
=540°C). The stable creep rate may be calculated from the formula
vc~ kxat at T=54CPC £=2.5X10“1|> cm4fl/(kgf', *hr) and n = 5.
Find the shearing stress distribution in the shaft’s cross section
and also its angle of twist after 1000 hours of working under load.
Solution. Let us assume that the hypothesis of plane sections under
torsion remains in force during creep too (this hypothesis agrees
sufficiently well with experimental data). Then two cross sections
at a distance dx will remain planes and only turn w.r.t. each other
by an angle d<p. Since the radii of the sections do not warp, we may
use the usual (§ 47) formula for determining relative shear at a dis­
tance from the shaft centre:

After loading the shearing strain begins to increase on account

of creep of the shaft’s material, the relative angle of twist ^ also
increases accordingly. The total shearing strain may be expressed
as the sum of elastic strain y e and creep strain y« Y—Yo+Yc-
The rate of growth of the total shearing strain may be written as
fol lows:

v ~ d i - - d r + s r - v* + v‘ - P d n t
Ch. 32\ Fundamentals of Creep Analysis 623

For simplification, the relatively small rate of growth of the elastic

deformation may be neglected in comparison with the large rate of
growth of creep deformation. The exact solution, with vr^=0, is much
more difficult. Besides, neglecting the zone of unstable creep and
assuming that vc—kxn, we get

wherefrom

where

® - ( t = & )* <b>
The condition representing the equality of the moments of external
(torque) and internal forces about the shaft axis may be written as
follows:
r r i
M = M t — ^ Tp d A — C 2 n p x p d p = 2o<D ^ p 2+ * d p
a o o
where r is the shaft radius. By introducing the notation

we may write
<D= i k (c)
J pC

and taking into account expression (a)

X - ^
C
We see that the distribution of shearing stresses of creep over the
cross section of the shaft is not linear.
Substituting in the expression for J pe the values r= l.2 c m and n = 5 t
we obtain:
.2^ r “ = 1.964x l.2 3-2 =3.52 cm32
The shearing stresses at a distance p from the shaft are:
624 Dynamic Action of Forces [Part IK

Table 26
Values of x

p % 0 , J e
(mm) (kgf/cm*) (kgf/cm*) (mm) (kgf/cm*) (kgf/cm*)

12 737 589 4 246 473

10 Gt 4 568 2 123 411
6 492 543 0 0 0
6 369 513

Table 26 contains values of t corresponding to different values

of p. The table also contains values of x from consideration of pure
torsion of the shaft calculated by the formula
M t p _ 2 M tp 2 X 2000
Jp nr* ~~ 3 . 1 4 X 1 . 2*
p = 614p

Creep helps in equalizing the stresses over the shaft’s cross section;
the stresses at the surface register a small decrease, whereas the stresses
near the axis increase considerably (Fig. 469).
Let us determine the angle of twist per unit length of the shaft.
On the basis of (b) and (c) we may write

Integrating this expression w.r.t. t, we get:

p
dx

since at '=<>• we “ e that C = ^ and

Integrating and putting x = l cm, we obtain

Substituting the numerical values of this problem, we get

2000X 2
2 .5 x I0“1# ( | ^ ) B1000
V* " e x 1 0 6 X 3 .1 4 X 1 . 2*
=« 0.00102 -h 0.014 80 0.0158 rad/cm
Ch. 32] Fundamentals of Creep Analysis 625

It can be easily seen that on account of creep the angle of twist In­
creased almost 16-fold as compared to the angle of twist under pure
torsion (first term of the above result). Hence, the decrease of maxi­
mum shearing stress in the shaft does not mean that its working
conditions improve due to creep.
Example 3. A 50-cm long simply supported beam 20 mm by 40 mm
is acted upon at the middle of its span by a concentrated force
%lgf/wz
0■—■ WO
■ ■ 1/90
■ ■ B■OO ■ 800
■

— ■//’•00m

—>1 b * 2 0 m

Fig. 470

—400 kgf (Fig. 470) works at T=500°C. The beam is made of mild
steel having modulus of elasticity £ = 1 .6 X 10® kgf/crn* (at 7'=500°Q.
The stable creep rate is v e= k a n, and at T =5(XrC

fe = 1 .5 x l0 -ls ( ~ ! ) " h r - ‘ and n=3

Neglecting the zone of unstable creep, find the distribution of nor­

mal stresses in the critical section and determine the maximum de­
flection of the beam after 10 000 hours of loading.
Solution. While solving this problem we shall neglect the effect
of shearing stresses and assume that the hypothesis of plane sections
under bending remains valid in creep too (this assumption agrees
quite well with the experimental results). Assuming that the defor­
mation of the beam’s fibres follows the same law in the stretched and
compressed zones, we may express the strain of a fibre at a distance y
by the relation (§ 63) e = i p where p is the radius of curvature of the
neutral surface of the beam.
In the loaded state the deformation of the beam fibres increases
gradually on account of creep of the beam’s material; the radius
of curvature of the neutral surface also increases. The total relative
deformation of an arbitrary fibre may be represented as the sum of
elastic and creep strains, i.e. e = e <l+ e c. The rate of change of totat
626 Dynamic Action of Forces [Port IX

creep strain may be written as

„ de de, . de, , d / I\
t, = 37= l f + - 3 f 57( 7 )
For simplification we neglected, as in the preceding section, the rate
of change of elastic strain. For an accurate solution of the problem
(assuming v ^ O ) consult special monographs (see footnote on page 620).
Neglecting, in addition, unstable creep and assuming t/c=Aort, we
obtain

" “ "c = I (f)

wherefrom

where

® -[« (*)r <»

>
The condition expressing the equality of moments of the external
and internal forces about the neutral axis may be written as follows:

M ^ \ a y d A — 2 <D $ y l* ~ d A
A Al / 2
Introducing the notation
r 1+—
2 J y + » d A = J.
Al/2
we get
©
II

and, if we recall (a),

^ m ~r
< * =Jzc
j~ yn

It should be noted that in creep the distribution of normal stresses

over the height of the section is not linear.
If the beam has a rectangular section,

Ai/i 0 2 n
Ch. 321 Fundamentals of Creep Analysts 627

Substituting the numerical values of the problem in the expression

for J tc, we get
2X3+ I 10
3X 2X 43
2x4 3 =
3 4-1 _4_ 8.64 cm 3
2 3 (2 x 3 + 1 ) 7X2 3

The maximum bending moment is:

M „ „ = ? = ^ = 5 0 0 0 k g f-c m

The normal stresses in the rectangular beam may be calculated

from the formula
M -- 5000 4 579t/3
a-7 Z « " = m v 3

The normal stresses computed from this formula for various values
of y are given in Table 27. For comparison, the values of normal stres­
ses under elastic deformation are given in the same table.
Table 27
a (k gl/cro1) «r (kgf/cm 'J

y (m m ) C last tc y (m m) E la s tic
Creep deform ation Creep
deform ation

20 938 729 4 188 426

16 750 677 2 94 339
12 563 615 0 0 0
8 375 537
•

Due to creep the stresses over the beam’s cross section level out.
The stresses decrease in the fibres farthest from the neutral layer, where­
as the stresses near the neutral layer increase (Fig. 471). The de­
crease in maximum stress is more pronounced in the rectangular
beam as compared to the I-beam, because the latter is comparatively
thinner near the neutral layer.
Let us determine the beam’s deflection. Expressing the curvature
of the beam by the approximate relation » we may wri*e

i d ) - & ( § ) ■ Since Irom (b) 3011 (c)

 628 Dynamic Action of Forces [Part IX
we have
d2 (dy\ b Mn
i A ‘“ )° = kn
In the case of a simply supported beam (Fig. 472), M =(P/2)x and

-P x F
d**Ut ) (d )
where
kP»
R

Integrating equation (d) w.r.t. x, we obtain

± (d y\_ _ R
Tx \d t) rt+j
or

3? “ («+ \){n+2)xn** + Cx + D
where C and D are constants of integration. Since y = 0 for all values

Pi
Z

i 1 r-X
* - J
-* -f

Fig. 472

of t when *=0, and ^ = 0 when x = *j , obviously gf = 0 at x

and |<f ( ^ ) = s ( § ) 5=0 for XS=J - Therefore
_ Dln+l
c— fT F T jzm an<> B - 0
Ch. 82] Fundamentals of Creep Analysis 629

The maximum deflection occurs in the section where x= lf2; we get

/ d y \ _______/?/»•*a kP»if" ' 5
{dl / m ax 2 " + - {n + 2) " (/j + 2) 2*‘" 4n

Integrating w.r.t. i, we obtain

kpnin+i
Urnax t+H
(«+2>2*M1+,,y»
integ
where H is the constant of integration. Since at /= 0
Pi*
{/m a x = (//m » x ^ e 48 C J g
we have
Pi*
H = ({/max^tf = 48EJt

The maximum deflection due to creep may therefore be written as

pp kPnin+%
{/m ax
4 8 £ /z ( r t f 2)2,<"+ '>./£
f Mt3 2X 4s
~ 12 ~ 12
■= 10.67 cm4

and
I I 400X503 , 1.5Xlft-»X400»X60»+a tAf>nA
I {/in ax I = 48 X 1.6 X 10° X 10 67 8.641
' (3-f-2jX22<a * •
>
= 0.0611 -f 0.3634 = 0.425 cm
Under creep, the maximum load on the beam should be determined
from consideration of permissible deformation. If, for instance, the
maximum deflection of the rectangular beam should not exceed 1/600
of its length (0.1 cm) after 10 000 hours of operation, then the maxi­
mum permissible load may be calculated from the condition
PP , kPnln+z
4&n~t 'T'(n+2)22i»+l)J%! L/J

Substituting the numerical values, we get

5(P P , 1 5 x 10- wXf»0», A nnn D, ^ n i
4*X10.67X1.6xT6*‘ + 5X8.643X2* 1 0 0 0 0 P ^ 0 l

or
P -f0.2326P»x 10-5 < 40.96
wherefrom
/> ^ 4 0 .8 kgf« 40 kgf
 Appendix
Rolled Steel Profiles
(COST 8239-72, 8240-72, 8509-72)

I-sectlon

ft b s t
P r o file A rea o f M ass o f
N o. se c tio n I m
(cin*) (kg)
(mm)

10 100 55 4.5 7.2 12.0 9.46

12 120 64 4.8 7.3 14.7 11.50
14 140 73 4.9 7.5 17.4 13.70
16 160 81 5.0 7.8 20.2 15.90
18 180 90 5.1 8.1 23.4 18.40
18a 180 100 5.1 8.3 25.4 19.90
20 200 100 5.2 8.4 26.8 21 .00
20a 200 110 5.2 8 .6 28.9 22.70
22 220 110 5.4 8.7 30.6 24.00
22a 220 120 5.4 8.9 32.8 25.80
24 240 115 5.6 9 .5 34.8 27.30
24a 240 125 5.6 9.8 37.5 29.40
27 270 125 6 .0 9.8 40.2 31.50
27a 270 135 6 .0 10.2 43.2 33.90
30 300 135 6.5 10.2 46.5 36.50
30a 300 145 6.5 10.7 49.9 39.20
33 330 140 7.0 11.2 53.8 42.20
36 360 145 7.5 12.3 61.9 48.60
40 400 155 8.3 13.0 72.6 57.00
46 450 ICO 9.0 14.2 84.7 66.50
SO 500 170 10.0 15.2 100.0 78.50
SS 550 180 11.0 16.5 118.0 92.60
60 600 190 12.0 17.8 138.0 108.00
 Appendix

T abulated v a lu e s about, axes

x-x u-y
wx Jy WU
(cm 4) (CIU*) icui) (cm') (era4) (cm 1)

198 39.7 4.06 23.0 17.9 6.49

350 58.4 4.88 33.7 27.9 8.72
572 81.7 5.73 46.8 41.9 11.50
873 109.0 6.57 62.3 58.6 14.50
1290 143.0 7.42 81.4 82.6 18.40
1430 159.0 7.51 89.8 114.0 22.80
1840 184.0 8.28 104.0 115.0 23.10
2030 203.0 8.37 114.0 155.0 28.20
2550 232.0 9.13 131.0 157.0 28.60
2790 254.0 9.22 143.0 206.0 34.30
3460 289.0 9.97 163.0 198.0 34.50
3800 317.0 10.10 178.0 260.0 41.60
5010 371.0 11.20 210.0 260.0 41.50
5500 407.0 11.30 229.0 337.0 50.00
7080 472.0 12.30 268.0 337.0 49.90
7780 518.0 12.50 292.0 436.0 60.10
9840 597.0 13.50 339.0 419.0 59.90
13380 743.0 14.70 423.0 516.0 71.10
19062 953.0 16.20 545.0 667.0 86.10
27696 1231.0 18.10 708-0 808.0 101.00
39727 1589.0 19.90 919.0 1043.0 123.00
55962 2035.0 21.80 1181.0 1356.0 151.00
76806 2560.0 23.60 1491.0 1725.0 182.00
632 Appendix

Channel Section with Sloping Flanges

ft b s 1
Area <•! M ass of 1 a
(kg)
N o. (cm*)
(mm)

5 50 32 4.4 7.0 6.16 4.84

6.5 65 36 4.4 7.2 7.51 5.90
8 80 40 4.5 7.4 8.98 7.05
10 100 46 4.5 7.6 10.90 8.59
12 120 52 4.8 7.8 13.30 10.40
14 (40 58 4.9 8.1 15.60 12.30
14a 140 62 4.9 8.7 17.00 13.30
16 100 64 5.0 8.4 18.10 14.20
16a 160 68 5.0 9.0 19.50 15.30
18 180 70 5.1 8.7 20.70 16.30
18a 180 74 5.1 9.3 22.20 17.40
20 200 76 5.2 9.0 23.40 18.40
20a 200 80 5.2 9.7 25.20 19.80
22 220 82 5.4 9.5 26.70 21.00
22a 220 87 5.4 10.2 28.80 22.60
24 240 90 5.6 10.0 30.60 24.00
24a 240 95 5.6 10.7 32.90 25.80
27 270 95 6.0 10.5 35.20 27.70
30 300 100 6.5 11.0 40.50 31.80
33 330 105 7.0 11.7 46.50 36.50
36 360 110 7.5 12.6 53.40 41.90
40 400 115 8.0 13.5 61.50 48.30
 Appendix 633

Table 15

T a b u la ted v a lu e s about axes

x-x V I/ *■>
(cm)

J w lx S* wy lv
(cm*) (cm*) (cm) (cm*) (cm*) (cm*) (cm)

22.8 9.1 1.92 5.59 6.61 2.75 0.954 1.16

48.6 15.0 2.54 9.00 8.70 3.68 1.080 1.24
89.4 22.4 3.16 13.30 12.80 4.75 1.190 1.31
174.0 34.8 3.99 20.40 20.40 6.46 1.370 1.44
304.0 50.0 4.78 29.60 31.20 8.52 1.530 1.54
491.0 70.2 5.60 40.80 45.40 11.00 1.700 1.67
545.0 77.8 5.66 45.10 57.50 13.30 1.840 1.87
747.0 93.4 6.42 54.10 63.30 13.80 1.870 1.80
823.0 103.0 6.49 59.40 78.80 16.40 2.010 2.00
1090.0 121.0 7.24 69.80 86.00 17.00 2.040 1.94
1190.0 132.0 7.32 76.10 105.00 20.00 2.180 2.13
1520.0 152.0 8.07 87.80 113.00 20.50 2.200 2.07
1670.0 167.0 8.15 95.90 139.00 24.20 2.350 2.28
2110.0 192.0 8.89 110.00 151.00 25.10 2.370 2.21
2330.0 212.0 8.99 121.00 187.00 30.00 2.550 2.46
2900.0 242.0 9.73 139.00 208.00 31.60 2.600 2.42
3180.0 265.0 9.84 151.00 254.00 37.20 2.780 2.67
4IU0.0 308.0 10.90 178.00 262.00 37.30 2.730 2.47
5810.0 387.0 12.00 224.00 327.00 43.60 2.840 2.52
7980.0 484.0 13.10 281-00 410.00 51.80 2.970 2.59
10820.0 601.6 14.20 350.00 513.00 61.70 3.100 2.68
15220.0 761.0 15.70 444.00 642.00 73.40 3.230 2.75
684

82 2$S & S£o lO

o o—
*>
2co o — —CM
H

S3 COO s S3 S3
do oo d oo o —

«4
H eoS u3 — o oo>
«co 3 3
><6 d — p-W ci
cm

d ■<** d <o

0 o oo
s s s 38 f=8
oo oo o' oo do

eo ^ $ t^Q> 88
oo OO o oo M—

lON
CO IOC7O t-*. c5 cs s s
N 0>< > o
od oo — ——

cOcp
{Ot'* a s 00- s s t9 <*M
© CM
oo •■*— «H CMCO ■>ro
•*>

s s LOV 1C
oo h* o 2§
hE 0>0>
do CO o oo
H
s s «§ o £SS
oo o — —CM CMCO

o><p N © to
.43
86

2$ •9*op O —N
E
u CNcs
X'
CO«* CO«* CO CO"* <o *•
e
E
ID 00 <
8 W CM CM
O s

»q
lO 0 0 cm CO
uinu a iy o j j d d c d

"
 635

'Z COCOb. OWN

rtO N V CM
3 »
s
oo v>
—• cn cn
3 f>eo
oncnc© CNCOCO CO^J*
8oS£
Co5 lO
0 0 8 8 8 8
*M/5 to b~ 00

£3 00 t*.

.2 6
.3 0
.21
so o —*— XOt/5 8 g g o8 2 o>o>©
<o
H

s a g S 2 8 8 8 2 8 8 2 8 8 O N V -O
t© 00® o i oaa*%
iio
8 8 8 5 8 i/5 2 $ 8 5 )

0^600 —o tpt/5 Tj< 2CO2CO-SC9Oco co

t< t>* h* 8 8 8 8 8 8 <N N N
o o o o o o —o o ~ * * ~ * —'

t^ o o w o rn — O)
T a b u la te d v a lu e s a b o u t a x es

^o>w s s a 0>00<0 "0 " 1/5 00 S5 8 ©CN t/5 00 O

------CN oi ol to CNCOlb V5 «D b- <3 >— cn cd
1/5 b! p

i/5 co -a- oo «o CNCN— a> 00

in in m K S S 2 8 8 9 3 9 NNN© <0
** *“• CNCN CNCNCN CNCNCNCNCN

2 § g 2 8 2 2 2 8 8 8 2 2 2 © N © M «t
1/5 b. oo COOM — Tj< © 1/5
CNCN 8 8 9 9 8 8 8 2

l^ J C O CO© 1/5 ■a" CO

.2 2
.2 0
.2 3

.3 9
.3 6
,3 7

8 S S 2 2 O) o> O)
•*‘tH
o CNCN CNCNCN

■
»?t
a a a 2 3 8 -N § 2 8 2 2 2 C a o o c i
<0 -*U5 1/5 0 00 t^o> — 2 2 2 8 5 ; 8 co cd 9 9

©g~ 8 8 2
m op o>
8 8 8 8 5
•£0*00
0>—*CN| 8 8 2 9 2
S=E CNCO00 c * ic o ^ NOV ■<r io v © s to to 00 © ©
*3*
10
•a COTf 1/5 05 -«r v5 CO a" 1/5 to «r t/5 to Tb 1/5 to b . OO
E
E
■o tO
§ 8 8 2 s

jaq o <D CO
tnnu o jijo /d ■<< to 1/5 0 b.
636

6.89
9.02
5.80
7.96
10.10

6.78
7.36
9.65

12.20
8.33

10.10
10.90
9.64

10.80
12.20
17.90
8.51

20.60
23.30
15.(0
«

2.17
2.19
2.23
2.27

2.43
8S222

2.55
2.47
2<9

2.51
S sS S sS jg
t- <NCMtNoi C4 w wcicsoiwco

ooooooo
69.6
98.3
113.0
127.0
83.9

102.0
119.0

169.0
137.0
93.2

194.0
145.0

219.0
* -ih oE *• ~ to «5e*e>Jel
oj S3§4CO$ Tf 5)

q>90coo*
lOOidiO acowtoiQv?
0)009)0)0)0)
*"*• apN
am• pq•
***~ — •mm* rn*rn*^

a rf Co—00U5
21.8
23.5
27.0

34.0
38.9
30.3
T a b u la te d v a lu e s about axes

43.8
48.6

50.7
54.2

112.0
60.9
86.9
99.3
74.1
<
—oov
—<cC
vN
>IN
^ s!
(N
2.90
2.89
2.86
2.87

3.09
2.91

3.46
3.08

3.49
3.50
3.48
3.11
3.11

c8 c8SooSoo
co cocococo eoeo
»I?
O O'. '£>O © <0000 © ©o o ooooooo
ooc>o« 8888 co
©•©►
o"
*5?o?qTco
co “ ui
n. <
—o
— c$55c5eo >
e*

eo8?5§j N Slflv
• • • •wa s s s s s s s
CMCMC4CMCM CMCMCMCM NcJcici COCOCOCOCOCOIN

to COCO00 — t^OCO —COO© o o o © ©©©

S?D5Sf2 si
w© 3?ssjeo
o c* —r-^© ©
7.39

11.50
10.10

■gZE
fgff*
12.80
8.78

12.30
10.80

10.60
13.90
9.38

12.80
8.63

12.30
15.60

13.80
15.60
19.20
22.80
26.30
29.70

*vtl
<8~
lO to
•>3 <n<ot^eoo> U5<01^00 ©i^«o©> <9(•*«0© C4 CD
S
•O
e
©
»n. o
8 omt
aaq io•
-uinu a||jo>Jd 00 © o
 637

a o
& t*
< .
g g
- «
US00 ~t-*-CMtS
2 2 2 £S&g}
«*■to 1<S
2 ^ 8
© "r © in © -M-
v N ft^ re o c o N
w f i eo co

s(9 O £ 8 <^5§ •$ S5 3 $ SS8 8 8 & 8 !? 3 S g

oi eo <0 <0 <0 <0 <0<0 COCO00 ■m- sj- v v»r
H

o o
ooeo to © cmto ~ 00-0- to «* 00 — —CM<e
sCO©00 — <3 *£ w — v> |$ © CO—© h- US
uou5 © t^o> © S3© 8 CO C
MM MOoi — Tf Cl
MC4C9

o> 00 O) OCh» to US ■«}■ O C O N C tM O l

te *r ■*»• •*?*»• *»•
C4 0 t CMCMCMCMCMCM CMCMCM COCOCOCOCOCO<o

N 00
Tabulated values about axes

C4 — CMUS© M" © V © © © — USF» ©

r^co 2 —— £} -N C «I S sS S S S S S

X
<Q f e © 3 ;w o o jo
68 a s CO00 CO00 t-- ▼ ^P ^P
.Hw w ir^ ' us us us <C<D<0 CO<OOO

OO
1450
1229

1662
1866

2248

0)10
1341

2061

s — © us
§ fus
^K S? S f o
CM00 to © 2
r-^ oo Kooc>

N.USIO MQ© CO— ©!£>:£ ey© rj©

oo oo oo 0oo oo F- co co CO ©© ©© as to oo
J*g coco oo oo co oo co <o ■*»•'»*■rr
•?
H
©o
(COO jN c o io ta ©CM £1 ■4" 3* C O ©'S
©©©
r-35 S
KV —M
oo © © - N®t "r
8© 3

t-- © OS © ■**•00 h-CO© M" ■M

« CO— 00
in s S b «
11 2 S 8 8 SSS; CMCMCO e o S ^ S * ? 'J iS S

h» 00 CO© O CMMr CD © O CM © —CMV to OOg

^ *■*mm
e
e o•m US
■o (N 8
■"* *«4 mm M
in
J»(l C
mM
m <D
uinu 3190^4
 to to P rofile num ­
C71 o oe ber

aa o-

a.
C l 4k >cn 8 © 5 co to to —

•— vo -4 © cn cn 4k —<i>
4k Co — © tQ 0 O v>»
IOUIOO)S^O»
« • • • • • •
8• 8• “ 4kCl tOJk
» >O''l
•
4k CO
JO 00
0 “ '»“ © 4k oi ok cn go cn o > «»
0 <0 to bo -?8-,

0&S^O> o»cn.a Co to
—">40 |s»i?
N
>0->J
9 N 51 * “ =2
cn 4k
&o to
-o 5

' J n| S v1 s | s) s| c i o> C l O C> C l O c i C l d cn n _

28 88bb82is 3h
CI b Q ■'4 -4 ~4
S — Cn VO •— CO C l ss
SK=SSgg5i
O *. cn—
to O• O COto
SO \t 82£882« 8s 7>*
o*
cn«* o^5 * «o
cn O 22S885I M c
sr
£
<
<0<000<00<0 00 oo ■ > )'l'4 N |k l, >lS n s «*
!333 8S 8!8bs282 28 aa
WB
>CO to to t o t o —
cnc5“ <o f*
cn
IO s S o S m <o >t o © <— i n VO la

fe6k> Ok ^ 4h»kfik CO CO CO CO 04 Co CO 00 CO
t o VO tp CD CO VO
SSSSSSigg kk S' • u )5 s» 3 kk 53
S'

£3$8£28S
s4k&.&.—
a a o StoSa?
I Ok
><© 09 OJ $” & & ! £ < £
82 in ’
S $9? $n f t to cn Oo
o co ox Jk t o t o to
to to
4k CO

'O-o -o-oCl ci ci
Table III (C ont.)

o>cn c i cn cn cn ctx cn cn
Sb=8S2gSS SS OC Oo
tocn
A
3

Cn^ ___S
= 2283*82 U 24kS<*oSo88 88 S"g £
4k cn © b > — b cn Ool^ ci o —*o oo to b — cn ~ a.

8C9
 Name Index

Belelyubskii, N. A. 283 Kachanov, L. M. 136, 620

Beltrami, F 145 Kachurin. V. K. 7
Belyaev, N. M. 5, 6, 7, 8, 26, 106, 148, Karman, Th. 490
188, 207. 290, 479, 499, 520, Kipnis, Ya. I. 8, ICO
562. 565, 576 Krylov, A. N. 305
Benardos, N. N. 158 Kurkin, S. A. 163
Bolotin, V. V. 520 Kushelev, N. Yu. 8
Bubnov, I. G. 353

Lam6, G. 138, 446

Loitsyanskii, L. G. 536
Castigliano. A. 340
Lurye, A. I. 536
Clapcyron, B. P. E. 335
Clebscb, R. F. A, 304
Coulomb, Ch. A. 140 Malinin, N. N. 620
Mariotte, Ed. 138
Maxwell, J. C. 349
Davidenkov, N. N. 148 Mises, R. 145
Druzhinln, S. I. 144 Molir, O. 349
MQIler<Breslau, H. F. B. 349

Engesser, F. 490
Euler, L. 480 Navier, C. M. L. 138
Navrotskii, D. I. 160
Nikolaev, G. A. 163
Pridman, Ya. B. 148
Forrest, P. G. 597 Oding, I. A. 597
Ovechkin, G. 163

Gadoiin, A. V. 440, 446

Galileo Galilei 18 Pavlov, A. P. 163
Goldenblat, I. 1. 520 Pirlet, 349
Golovin, Kh. S. 440 Poncelet, J . V. 138
Guest, J . J . 140 Prigorovskii, N, I. 588
Puzyrevskii, N. P. 305

Hencky, H. 145 Rankine, W. J . M. 138

Huber, F. 145

Saint-Venant, B. 139, 189

Ivanova, V. S. 597 Serensen, $. V. 588, 597
€40 Name Index

Shtaerman, I. Ya. 96 Vereshchagin, A. N. 349

Sinitskii, A. K 8 Vinokurov, V. A. 163
Slavyanov, N.G. 158 Vlasov, V. Z. 188
Smirnov-Alyacv, G. A. 584 Vol’mir, A. S. 520

Telelbaum, I. M. 588 Yagn, Yu. 1. 144

Timoslienko, S. P. 5, 188, 508 Yasinskii, F. 490
Tresca, H. 140

Uzhik, G. V. 597 Zhuravskii, D. I. 270

 Subject Index

absolute displacement 129 bending, pure 225

absolute elongation 33 uni-platiar 256
active force 312 unsymmetric 379
alter-effect 608 bending moment 203, 348
alternating cycle 573 diagram of 204
amorphous material 21 biaxial stress 102
amplification factor of vibrations 539 breaking away, failure by 101
amplitude of vibrations 535 breaking load 457
brittle failure, theory of 138
angle, twisting 169
angle of shear 129 brittle material 41, 52
anisotropic material 37, 56 bulk modulus 123
axes of inertia, principal 251 butt joint 159
axial compression 27
axial force 29
axial moment of inertia 233 cable, flexible 92
axial tension 27 cantilever beam 213
axis, neutral 227 capacity, lifting 471
Castlgliano’s theorem 340
centre, flexural 387
barfs), compressibility of 559 characteristic cycle 574
curved 423 circle, Mohr’s 110
with large curvature 439 moment of inertia of 240
prismatic 27 Ciapcyron's theorem 335
rigidity of 34 coefficient, damping 539
with small curvature 439 dynamic 524
beam, cantilever 213 of dynamic response 60
continuous 366 of length 485
critical section of 200 of operating conditions 475
deflection of 292 of overloading 476
equation of deflected axis of 293 for production process 593
fictitious 314 of reliability 475
riveted 289 of stress concentration 576
simply supported 207 comparison of displacements 361
statically determinate 199 complementary shearing stresses 109
statically indeterminate 199, 356 complex figure, moment of inertia of
of uniform rigidity 329 245
of uniform strength 324, 558 component constant of cycle 574
welded 290 component variable of cycle 574
beam section, angle of rotation of 293 composite stressed state 101
beam supports, reaction of 197 compound loading 378
642 Subject Index

compressibility of bars 559 damage susceptibility curve 593

compression, axial 27 damping coefficient 539
eccentric 392 dead weight 86
compressive stress 101 deflection of beam 292
deformation 21
concentrated force 19
elastic. 21. 37
condition, of joint deformation 67, 80,
lateral 36
360
local 488
of strength 30 plastic 21, 47
conditional stress 41 total energy of 126
conical spring 186 design load 476
connecting rod 525 design moment 406
conservation of energy 331 diagram, of bending moment 204
constancy of volume 50 of reduced moment 472
constant sign cycle 573 of shearing forces 204
constant stress cycle 573 stress-strain 47
constraint, redundant 357
differentiation, successive 313
contact stress 105
displacement, absolute 129
continuous beam 366
generalized 333
continuous load, intensity of 206
distortion, potential energy of, theory
core of section 396
146
crack, fatigue 572
distributed force 19
creep 607
distribution, uniform 29
stable 610 doublC'Shear rivet 153
unstable 610 ductile failure, theory of 140
creep curve 609 ductile hinge 466
creep limit 617 ductile material 41
critical force 478 dynamic coefficient 524
critical section 86 dynamic load 20
critical stale, of material 63 dynamic loading 521
critical stress 479 dynamic response, coefficient of 60
crushing of rivets 154 dynamic stress 555
crystalline lattice 21
crystalline material 21 eccentric compression 392
curved bar 423 eccentricity 392
cycle, alternating 573 eccentric tension 392
characteristic 574 elastic deformation 21, 37
component constant of 574 specific, work of 46
constant sign 573 elasticity, limit of 23, 43
constant stress 573 modulus of 34
fluctuating 573 elementary force 23
mean stress of 574 elongation, absolute 33
of stress variation 573 relative 33
zero base 573 relative residual 44
endurance limit 61, 62, 575
cyclic stress 573
energy of deformation, total 126
 Subject Index 643

energy theory of strength 145 forced vibrations 535

envelope 142 formula, Saint-Venant’s 407
equal moments, method of 471 frame 351
equation, of deflected axis 293 free torsion 187
of method of initial parameters 305
of three moments 372
generalized displacement 333
equatorial moment 233
generalized force 333
Euler’s formula 482
generalized Hooke’s law 335
externa) force 19
graph-analytic method 313
method of 344

helical spring 181

factor of safety 24, 30 hinge, ductile 466
main 63, 64 hinged support, fixed 197
factor of stress concentration 583 movable 197
failure, by breaking away 101 Hooke's law 33
due to shearing 101, 133 generalized 335
through rupture 132 hydrostatic load 455
fatigue 60, 572
fatigue crack 672, 594 impact load 20
fatigue limit 61 impact test 565
fictitious beam 314 initial parameters, method of 305
fillet weld, joint with 162 integration, successive 312
fixed hinged support 197 intensity of continuous load 206
fixed support 198 interaction, force of 19
rigidly 198 I-section 285
flexibility 483 isotropic material 37
flexible cable 92
flexural centre 387
joint, butt 159
fluctuating cycle 573
lapped 151
force(s), active 312
riveted 159
axial 29
welded 159
concentrated 19
with side fillet weld 162
critical 478
joint deformation, condition of 67, 80,
cumulative action of 335
360
distributed 19
external 19 lapped joint 151
method of 344 lateral deformation 36
generalized 333 lattice, crystalline 21
of interaction 19 law, Hooke’s 33
normal 29 generalized 335
passive 312 of complementary shearing stres­
of reaction 20 ses 109
shearing 203 conservation of energy 331
superposition of 83 constancy of volume 50
volume 19 of cumulative action of forces 335
644 Subject Index

lifting capacity 471 of external force 344

limit, of elasticity 23, 43 graph-analytic 313
endurance 61, 62, 575 of initial parameters 305
of proportionality 41, 47 of superposition of forces 223
limiting states, first group 474 Vereshchagin's 349
second group 474 modulus, bulk 123
limiting stress circle 139 of elasticity 34
load, breaking 457 reduced 490
continuous, intensity of 206 tangential 490
design 476 section 236
dynamic 20 Mohr’s circle 110
hydrostatic 4% Mohr’s strength theory 141
impact 20 moment, bending 203
permanent 19 design 406
repeated variable 20 equatorial 233
static 20 of section, static 232
suddenly applied 20 static, about neutral axis 267
temporary 19 moment of ineria, axial 233
ultimate 457 of circle 240
load area 214 of complex figure 245
load curve 214 of parallelogram 240
local deformation 488 polar 173, 250
local stress 58, 61, 157, 576, 582, 583 principal 256
long-term strength 618 multiple-shear rivet 154
less of stability 483

natural vibrations 535

material, amorphous 21 net area 31
anisotropic 35. 56 neutral axis 227
brittle 41, 52 neutral layer 227
critical state of 63 radius of curvature of 231
crystalline 21 of trapezoid 435
ductile 41 normal force 29
isotropic 37 normal stress 100
sensitivity factor of 585
strength of 18
maximum rigidity, plane of 386 octahedral plane 120
maximum shearing stresses, theory of octahedral shearing stress 121
140 operating conditions, coefficient of
maximum tensile stresses, theory of 476
138 overloading, coefficient of 47$
Maxwell and Molir theorem 347
mean stress of cycle 574 parallelogram, moment of inertia of
method, of comparison of displacements 240
361 passive force 312
of equal moments 471 permanent load 19
 Subject Index 645

permissible stress 24 redundant unknown 359

planc(s), of maximum rigidity 386 relative elongation 33
octahedral 120 relative reduction of area, permanent
principal 102 45
plane of inertia, principal 255 relative residual elongation 44
plastic deformation 21, 47 relative rigidity 36
pliability 555 relative shear 129
of structure, total 564 reliability, coefficient of 475
Poisson’s ratio 37 repeated variable load 20
polar moment of inertia 173, 250 reservoir, thin-walled 103
potential energy of distortion, theory of residual elongation, relative 44
146 resistance, to rupture 134
principal axes of inertia 251 to shear 134
principal moment of inertia 256 resonance 535
principal plane 102 rigidity, of bar 34
principal plane of inertia 255 maximum, plane of 386
principal radius of inertia 256 relative 36
principal strcss(es) 102 of system 550
trajectory of 282 torsional 177
shearing 120 rigidly fixed support 198
principle of superposition of forces 83 rivel(s), crushing of 154
prismatic bar 27 double-shear 153
production process, coefficient for 593 multiple-shear 154
product of inertia, of section 232, 247 riveted beam 289
proportionality,-limit of 41, 47 riveted joint 159
pure holding 225 rod, connecting 525
pure shear 127 thin-walled 194
pure torsion 187 rupture, failure through 132
resistance to 134
theory of 138
radius of curvature of neutral layer 231 rupture strain, total true 51
radius of inertia 256
principal 256
reaction, of beam supports 197 safety, factor of 24, 30
force of 20 main factor of 63, 64
redundant 357 Saint-Venant’s formula 407
reciprocity of displacements, theorem of scale factor 590
347 section, of beam
reciprocity of works, theorem of 347 critical 200
reduced mass 544 core of 396
reduced modulus of elasticity 490 critical 86
reduced moment, diagram of 472 product of inertia of 232, 247
reduced stress 147 static moment of 232
reduction of area, permanent relative 45 section modulus 174 , 236
redundant constraint 359 sensitivity factor of material 585
redundant reaction 359 shear, angle of 129
646 Subject Index
pure 127 strength theory, first 138
relative 129 fourth 146
resistance to 134 of Mohr 141
theory of 140 second 139
shear centre 387 third 140
shear centre line 387 stress, biaxial 102
shearing, failure due to 101 compressive 101
modulus of elasticity for 131 conditional 41
shearing force{s) 204 contact 105
diagram of 204 critical 479
shearing stress(es) 100 of cycle, mean 574
complementary 109 cyclic 573
maximum, theory of 140 dynamic 555
octahedral 12! local 58, 61, 157, 576, 582. 583
principal 120 maximum shearing 140
simply supported beam 207 maximum tensile 138
span 92 normal 100
specific work, of elastic deformation permissible 24
46 principal 102
total 46 reduced 147
spherical tensor 117 rupture, true 51
spring, conical 186 shearing 100
helical I8t octahedral 121
stability check 477 principal 120
stable creep 610 tensile 100
state(s), composite stressed 101 triaxial 103
limiting 474 uniaxial 102
statically determinate beam 199 variable 575
statically indeterminate beam 199, 356 stress circle 111
statically indeterminate problem 66 limiiing 139
statically indeterminate system 66 stress concentration, coefficient of 576
static toad 20 factor of 583
static loading 60, 521 stress deviator 124
static moment, about neutral axis 267 stress intensity 121
of section 232 stressed state, composite 101
straight-line form, loss of stability of 483 stress-strain diagram 47
strength, condition of 30 stress tensor 117
energy theory of 145 stress variation, cycle of 573
long-term 618 structure, total pliability of 564
of materials 18 successive differentiation 313
tensile, ultimate 30 successive integration 312
theory of 136 suddenly applied load 20
true ultimate 51 superposition of forces, method of 223
ultimate 23, 43 principle of 83
strength endurance 575 support, fixed 198
in unsymmetrlc cycle 579 hinged, fixed 197
 Subject Index 647

movable 197 t r u e r u p tu r e s tr e s s 51
rigidly fixed 198 tr u e u l t i m a t e s tr e n g th 51
system, rigidity of 550 T -sc c tio n 243
statically indeterminate 66 tw is tin g a n g le 169

tangential modulus of elasticity 490 u ltim a te lo a d 4 5 7

temporary load 19 u ltim a te s tr e n g th 2 3 , 4 3
tensile strength, ultimate 30 tr u e 51
tensile stresses 100 u lt i m a t e te n s ile s tr e n g th 3 0
maximum, theory of 138 u n ia x ia l s tre s s 102
tension, axial 27 u n ifo rm d is tr ib u tio n 29
eccentric 392 u n i- p la n a r b e n d in g 2 5 6
lest, impact 565 u n i t fo rce, b e n d in g m o m e n t d u e to 3 4 8
theorem, Castigllano's 340 u n s ta b le c re e p 6 1 0
Clapeyron’s 335 im s y m m e tric b e n d in g 379
of Maxwell and Mohr 347 u n s y m m e tric c y c le , s tr e n g th e n d u r a n c e
of reciprocity of displacements 347 In 5 7 9
of reciprocity of works 347
theory', of brittle failure 138
v a r ia b le lo a d , r e p e a te d 2 0
of ductile failure 140
v a r ia b le s tre s s 675
of maximum tensile stresses 138
V e re s h c h a g in ’s m e th o d 349
of maximum shearing stresses 140
v e sse l, th in -w a lle d 454
of potential energy of distortion 146
v ib r a tio n s , a m p lif ic a tio n fa c to r of 539
of rupture 138
a m p litu d e of 535
of shear 140
fo rced 5 3 5
of strength 136
n a tu r a l 535
thin-walled reservoir 103
v o lu m e fo rce 19
thin-walled rod 194
thin-walled vessel 454
three moments, equation of 372 w a rp in g 488
torque 165 w e ig h t, d e a d 86
torsion, free 187 w e ld e d b e a m 290
pure 187 w eld ed jo in t 159
torsional rigidity 177 w o rk , of e la s tic d e fo rm a tio n 4 6
total pliability of structure 564 sp e c ific 46
total specific work 46 w 'ork h a r d e n in g 590
trajectory of principal stresses 282
trapezoid, neutral layer of 435
trlaxlai stress 103 z e ro b a se c y c le 5 7 3
true rupture strain, total 51 Z h u r a v s k ii's fo rm u la 2 8 4