METAL FORMING AND THE

FINITE-ELEMENT METHOD
 OXFORD SERIES ON ADVANCED MANUFACTURING

SERIES EDITORS

J. R. CROOKALL
MILTON C. SHAW

1. William T. Harris. Chemical Milling: the Technology of Cutting

Materials by Etching (1976)
2. Bernard Crossland. Explosive Welding of Metals and its Applications
(1982)
3. Milton C. Shaw. Metal Cutting Principles (1984)
4. Shiro Kobayashi, Soo-Ik Oh, Taylan Altan. Metal Forming and the
Finite-Element Method (1989)
METAL FORMING AND THE
FINITE-ELEMENT METHOD

SHIRO KOBAYASHI
SOO-IK OH
TAYLAN ALTAN

New York Oxford

OXFORD UNIVERSITY PRESS
1989
 Oxford University Press
Oxford New York Toronto
Delhi Bombay Calcutta Madras Karachi
Petaling Jaya Singapore Hong Kong Tokyo
Nairobi Dar es Salaam Cape Town
Melbourne Auckland
and associated companies in
Berlin Ibadan

Copyright © 1989 by Oxford University Press, Inc.

Published by Oxford University Press, inc.,
200 Madison Avenue, New York, New York 10016
Oxford in a registered trademark of Oxford University Press
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of Oxford University Press.

Library of Congress Cataloging-in-Publication Data

Kobayashi, Shiro.
Metal forming and the finite-element method /
Shiro Kobayashi, Soo-Ik Oh, Taylan Allan.
p. cm. — (Oxford series on advanced manufacturing;
4) Bibliography: p. Includes index.
ISBN 0-19-504402-9
1. Metal-work—Mathematical models. 2. Finite-element method.
I. Oh, Soo-Ik. II. Allan, Taylan. III. Title. IV. Series.
TS213.K56 1989
671'.072'4—dc!9 88-11995
CIP

135798642
Printed in the United States of America
on acid-free paper
 PREFACE

The application of computer-aided engineering, design, and manufactur-

ing, CAE/CAD/CAM, is essential in modern metal-forming technology.
Thus, process modeling for the investigation and understanding of
deformation mechanics has become a major concern in research, and the
finite-element method (FEM) has assumed increased importance, particu-
larly in the modeling of forming processes.
There are many excellent textbooks on the principles and fundamentals
of metal forming, but only a few describe the application of FEM to the
analysis and simulation of metal-forming processes.
The main purpose of this book is to present the fundamentals and
applications of FEM in metal-forming analysis and technology. The book
is primarily written for graduate students and researchers. However, it
should also be useful to practicing engineers who have a good background
in FEM and who are interested in applying this technique to the analysis of
metal-deformation processes.
In the application of FEM to metal forming, there are two formulations,
namely, flow formulation and solid formulation. Flow formulation assumes
that the deforming material has a negligible elastic response, while solid
formulation includes elasticity. Despite recent advances, the application of
solid formulation to the analysis of metal-forming problems remains
limited. On the other hand, flow formulation has found applications in a
wide variety of important forming problems. This book, therefore, is
mainly devoted to the applications that are based on flow formulation
(purely plastic and viscoplastic). However, recent advances achieved in
solid formulation have made it applicable to the analysis of some forming
problems. In order not to neglect these investigations, comparisons of
solutions based on both formulations, solid and flow, are presented in
Chapter 16.
The book begins with a general background on the subject in Chapter 1.
The description of metal-forming processes is given in Chapter 2, and
Chapter 3 details important technological aspects of these processes.
Chapters 4 and 5 present the theory of plasticity and methods of analysis as
applied to metal forming. The FEM formulations are described in
Chapters 6 and 7, and the applications of the method to the analyses of
various forming processes are presented in Chapters 8 through 11. Chapter
vi Preface

12 presents a thermo-viscoplastic analysis and Chapters 13, 14, and 15

include developments in the areas of deformation of porous materials,
three-dimensional problems, and preform design. The book concludes with
Chapter 16, in which further developments are discussed, along with the
outline of solid formulation and comparison of the results by both solid
and flow formulations.
Although this book primarily deals with metals, some of the principles
and solution techniques should be applicable to deformation analyses of
other materials, such as polymers and composites.
Sincere thanks are due to a number of individuals. First of all, we wish
to express our appreciation to Professor E. G. Thomsen, Professor
Emeritus, University of California at Berkeley, who helped us to devote
our careers to research in metal forming. We also thank Professor M. C.
Shaw, Arizona State University, for his encouragement and support in
writing this book, and Professor W. Johnson, Emeritus Professor, Univer-
sity of Cambridge, for his critical comments during the preparation of the
manuscript.
The senior author wishes to thank his former graduate students in the
Department of Mechanical Engineering, University of California at
Berkeley, who have contributed to the advances in the application of FEM
to metal forming.
The contents of this book are largely the results of research supported
by the Air Force Wright Aeronautical Laboratory, the National Science
Foundation, and the Army Research Office, and their support is
acknowledged.
We also thank Mr. and Mrs. Joe Bavonese for typing the manuscript.

Berkeley S. K.
Columbus S. O.
May 1988 T. A.
 CONTENTS

Symbols, xiii

1. Introduction, 1
1.1 Process Modeling, 1
1.2 The Finite-Element Method, 3
1.3 Solid Formulation and Flow Formulation, 4
1.4 Metal Forming and the Finite-Element Method, 5
References, 6

2. Metal-Forming Processes, 8
2.1 Introduction, 8
2.2 A Metal-Forming Operation as a System, 8
2.3 Classification and Description of Metal-Forming Processes, 11
References, 24

3. Analysis and Technology in Metal Forming, 26

3.1 Introduction, 26
3.2 Flow Stress of Metals, 28
3.3 Friction in Metal Forming, 30
3.4 Temperatures in Metal Forming, 33
3.5 Impression and Closed-Die Forging, 35
3.6 Hot Extrusion of Rods and Shapes, 36
3.7 Cold Forging and Extrusion, 39
3.8 Rolling of Strip, Plate, and Shapes, 41
3.9 Drawing of Rod, Wire, Shapes, and Tubes, 45
3.10 Sheet-Metal Forming, 47
References, 52

4. Plasticity and Viscoplasticity, 54

4.1 Introduction, 54
4.2 Stress, Strain, and Strain-Rate, 54
4.3 The Yield Criteria, 58
4.4 Equilibrium and Virtual Work-Rate Principle, 61
viii Contents
4.5 Plastic Potential and Flow Rule, 63
4.6 Strain-Hardening, Effective Stress, and Effective Strain, 66
4.7 Extremum Principles, 68
4.8 Viscoplasticity, 70
References, 72

5. Methods of Analysis, 73
5.1 Introduction, 73
5.2 Upper-Bound Method, 74
5.3 Hill's General Method, 78
5.4 The Finite-Element Method, 83
5.5 Concluding Remarks, 88
References, 88

6. The Finite-Element Method—Part I, 90

6.1 Introduction, 90
6.2 Finite-Element Procedures, 90
6.3 Elements and Shape Function, 94
6.4 Element Strain-Rate Matrix, 101
6.5 Elemental Stiffness Equation, 108
References, 110
7. The Finite-Element Method—Part II, 111
7.1 Numerical Integrations, 111
7.2 Assemblage and Linear Matrix Solver, 115
7.3 Boundary Conditions, 117
7.4 Direct Iteration Method, 121
7.5 Time-Increment and Geometry Updating, 123
7.6 Rezoning, 126
7.7 Concluding Remarks, 129
References, 129

8. Plane-Strain Problems, 131

8.1 Introduction, 131
8.2 Finite-Element Formulation, 131
8.3 Closed-Die Forging with Flash, 133
8.4 Sheet Rolling, 137
8.5 Plate Bending, 141
8.6 Side Pressing, 148
References, 149

9. Axisynunetric Isothermal Forging, 151

9.1 Introduction, 151
9.2 Finite-Element Formulation, 151
 Contents ix
9.3 Compression of Solid Cylinders and Heading of Cylindrical Bars,
153
9.4 Ring Compression, 159
9.5 Evaluation of Friction at Tool-Workpiece Interface, 163
9.6 Forging and Cabbaging, 165
References, 172

10. Steady-State Processes of Extrusion and Drawing, 174

10.1 Introduction, 174
10.2 Method of Analysis, 174
10.3 Bar Extrusion, 176
10.4 Bar Drawing, 178
10.5 Multipass Bar Drawing and Extrusion, 183
10.6 Applications to Process Design, 186
References, 187

11. Sheet-Metal Forming, 189

11.1 Introduction, 189
11.2 Plastic Anisotropy, 190
11.3 In-plane Deformation Processes, 192
11.4 Axisymmetric Out-of-plane Deformation, 195
11.5 Axisymmetric Punch-Stretching and Deep-Drawing Processes,
201
11.6 Sheet-Metal Forming of General Shapes, 206
11.7 Square-Cup Drawing Process, 210
11.8 Nonquadratic Yield Criterion, 217
References, 220

12. Thermo-Viscoplastic Analysis, 222

12.1 Introduction, 222
12.2 Viscoplastic Analysis of Compression of a Solid Cylinder, 223
12.3 Heat Transfer Analysis, 225
12.4 Computational Procedures for Thermo-Viscoplastic Analysis,
227
12.5 Applications, 229
12.6 Concluding Remarks, 240
References, 242

13. Compaction and Forging of Porous Metals, 244

13.1 Introduction, 244
13.2 Yield Criterion and Flow Rules, 245
13.3 Finite-Element Modeling and Numerical Procedures, 246
13.4 Simple Compression, 249
13.5 Axisymmetric Forging of Flange-Hub Shapes, 253
x Contents
13.6 Axisymmetric Forging of Pulley Blank, 256
13.7 Heat Transfer in Porous Materials, 259
13.8 Hot Pressing Under the Plane-Strain Condition, 262
13.9 Compaction, 266
References, 272

14. Three-Dimensional Problems, 275

14.1 Introduction, 275
14.2 Finite-Element Formulation, 276
14.3 Block Compressions, 278
14.4 Square-Ring Compression, 284
14.5 Simplified Three-Dimensional Elements, 287
14.6 Analysis of Spread in Rolling and Flat-Tool Forging, 289
14.7 Concluding Remarks, 295
References, 296

15. Preform Design in Metal Forming, 298

15.1 Introduction, 298
15.2 Method for Design, 298
15.3 Shell Nosing at Room Temperature, 301
15.4 Plane-Strain Rolling, 305
15.5 Axially Symmetric Forging, 309
15.6 Hot Forming, 315
15.7 Concluding Remarks, 318
References, 320

16. Solid Formulation, Comparison of Two Formulations, and

Concluding Remarks, 321
16.1 Introduction, 321
16.2 Small-Strain Solid Formulation, 321
16.3 Large Deformation: Rate Form, 323
16.4 Large Deformation: Incremental Form, 326
16.5 Comparison with Rigid-Plastic (Flow) Solutions, 327
16.6 Concluding Remarks, 334
References, 335

Appendix. The FEM Code, SPID (Simple Plastic Incremental

Deformation), 338
A.1 Introduction, 338
A.2 Program Structure, 339
 Contents xi

A.3 Input and Output Files, 340

A.4 Input Preparations, 340
A.5 Description of the Major Variables, 342
A.6 Program Listing, 343
A.7 Example Solution, 364

Index, 371
This page intentionally left blank
 SYMBOLS

A Cross-sectional area H Height

A Function of relative H0 Initial height
density for porous H1 Final height
materials
H(e) Work-hardening function
A
0 Initial cross-sectional
H Time derivative of height
area
H Increment of height
Aj
N area contribution of the
jth element to node N /, Linear invariant of stress
tensor
B Function of relative
density for porous /2 Quadratic invariant of
materials stress tensor
B Breadth /3 Cubic invariant of stress
tensor
B0 Initial breadth
/I Linear invariant of
B Strain-rate matrix
deviatoric stress tensor
C Constant
h Quadratic invariant of
Cr Class of functions with deviatoric stress tensor
continuous derivatives of
/3 Cubic invariant of
all orders up to and
deviatoric stress tensor
including r
J Jacobian of coordinate
C Volumetric strain-rate
transformation
vector
K Penalty constant
C Heat capacity matrix
K Stiffness matrix
D Diameter
K, Heat conduction matrix
D 0 Initial diameter
L Coefficient of anistropy
D Effective strain-rate
coefficient matrix L1, L
2, L] Area coordinate
E Young's modulus Lijki Small-strain moduli
E(eij) Work function -Lijkl Constitutive moduli
E Energy-rate M Coefficient of anisotropy
E tj Lagrangian strain M Gradient matrix of shape
F Coefficient of anisotropy function vector N
F(a ij) Function of stresses N Coefficient of anisotropy
F, Traction N Shape function matrix
G Shear modulus P Load
G Coefficient of anisotropy P Effective strain-rate
H Coefficient of anisotropy matrix
XiV Symbols

p. Element of strain-rate Tw Workpiece temperature

matrix B T Time derivative of
Q Heat flux vector temperature
R Roll radius T Coordinate
R Relative density of transformation matrix
porous materials UD Die or roll velocity
Ra Average relative density U0 Entrance velocity in
of porous materials rolling
R0 Initial relative density of U1 Exit velocity in rolling
porous materials UP Punch velocity
R0 Initial radius V Volume
Ri Internal radius of rings V0 Initial volume
and tubes
Vb Volume of base metal in
Re Radius of extruded or porous materials
drawn bars
K Volume of void in porous
Rn Radius of neutral point in materials
ring compression
AF Volume change
RD Die corner radius
W Width
RP Punch radius
W0 Initial width
S Microstructure
S Surface
wa Average width
W Time derivative of width
Sc Surface of tool-
workpiece contact wp Total plastic work per
unit volume
SD Surface of discontinuity
Plastic work-rate per unit
WP
SF Surface where traction is volume
prescribed
W0 Work-rate per unit
S1 Internal surface volume in reference state
Su Surface where velocity is xa Element of strain-rate
prescribed matrix B
S, Surface where heat flux is Y Yield stress in uniaxial
prescribed tension
T Thickness Y0 Initial yield stress
T Temperature Yb Yield stress of base metal
Ta Nodal-point temperature in porous materials
Tb Temperature of base YR Apparent yield stress of
metal in porous materials porous materials
Td Die temperature Ya Element of strain-rate
Environmental matrix B
Te
temperature Z« Element of strain-rate
Apparent temperature of matrix B
TR
porous materials a Height-to-diameter ratio
Ts Surface temperature c Specific heat
 Symbols xv

Cd Specific heat of die e Unit tangent vector

material m Friction factor
Cb Specific heat of base m Strain-rate exponent
metal in porous materials
m Coefficient of anisotropy
cu Specific heat of void in
n Strain-hardening
porous materials
exponent
CR Apparent specific heat of
n Coefficient of anisotropy
porous materials
n Unit normal to the
d Punch depth in sheet-
surface
metal forming
P Pressure
e Engineering strain
Pa Average pressure
e Engineering strain-rate
PD Die pressure in drawing
f Coefficient of anisotropy
First Piola-Kirchhoff
Frictional stress Pa
fs stress
f Nodal-point force vector
qf Heat generated through
f(o Yield function friction
g Coefficient of anisotropy qn Heat flux across surface
g(o Scalar function of stress Sq
invariants q Shape functions
h Heat transfer coefficient r r-Value in sheet forming
hc Heat transfer coefficient r
x, r
45, ry r-Values in the rolling,
at tool-workpiece 45°, and transverse
contact surface directions, respectively
Wlub Heat transfer coefficient r Heat generation-rate
of lubricant
sij Second Piola-Kirchhoff
h Coefficient of anisotropy stress
h(o Scalar function of stress t Time
invariants
Af Time-increment
k Shear yield stress
ui Velocity component
ky Apparent shear yield
ui( ) Velocity component at
stress of porous materials
the orth node
k1 Thermal conductivity
MO Initial velocity
kR Apparent thermal
Relative sliding velocity
conductivity of porous «J
meterials un Velocity component
normal to a surface
kb Thermal conductivity of
base metal in porous u, Velocity component
materials tangent to a surface
I Gage length in tensile AM Velocity discontinuity
test v, Relative sliding velocity
I Coefficient of anisotropy at nodal point
10 Initial gage length in V Velocity vector at nodal
tensile test point
XVI Symbols

V0 Initial velocity vector at •n Natural coordinate

nodal point ria ^-Coordinate of aih
Av Velocity corrections of node
nodal values e Angle
w
i Virtual velocity K Heat generation
w, Weight factors efficiency factor
Xa> Jai ^ex x, y, z-Coordinates of A Lagrangian multiplier
crth node A Proportionality factor
a Die semi-angle (rate) in flow rules
a Deceleration coefficient d)i Proportionality factor
ft Coupling coefficient in (infinitesimal) in flow
temperature calculation rules
Y, Y', Y" Viscosity coefficients V Coefficient of friction
Sh Radial displacement in v Poisson's ratio
bore expanding £ Natural coordinate
<50 Radial displacement in £» ^-Coordinate of ath
flange drawing node
<5a|8 Kronecker delta n Plane of zero mean stress
e Emissivity in stress space
e Effective strain n Functional
Eb Effective strain of base 6jt Variation of functional ir
metal in porous materials dn{i) <5jr-value at ;'th element
EN Effective strain value at 6nD Term due to deformation
node N energy-rate in dn
Eij Strain-rate 6jTP Penalty term in dn
Eij Infinitesimal strain djTsF Term due to traction in
ev Volumetric strain-rate dji
EH" Plastic strain-rate diti Term that includes
Lagrangian multiplier in
V Elastic strain-rate
dn
£ Effective strain-rate
5nSc Term due to friction in
Eb Effective strain-rate of djt
base metal in porous
P Density
materials
Po Initial density
-ER Apparent effective strain-
rate of porous materials Pb Density of base metal in
porous materials
Limiting strain-rate
EO
Pd Density of die material
C Natural coordinate Apparent density of
PR
£* -Coordinate of orth porous materials
node Pv Density of void in porous
n Function of relative materials
density in porous a Stephan-Boltzman
materials constant
 Symbols xvii

Oij Cauchy stress *(*) Bulge function in simple

Deviatoric stress compression
°n'
o Effective stress, flow 0(F) Strain-rate sensitivity
stress function
om Mean stress W,j Rate of rotation
T,i Kirchhoff stress
T/ Shear traction in Hill's
method
This page intentionally left blank
METAL FORMING AND THE
FINITE-ELEMENT METHOD
This page intentionally left blank
 1
INTRODUCTION

1.1 Process Modeling

In the late 1970s and early 1980s the use of computer-aided techniques
(computer-aided engineering, design, and manufacturing) in the metal-
forming industry increased considerably. The trend seems to be toward
ever wider application of this technology for process simulation and
process design.
A goal in manufacturing research and development is to determine the
optimum means of producing sound products. The optimization criteria
may vary, depending on product requirements, but establishing an
appropriate criterion requires thorough understanding of manufacturing
processes. In metal-forming technology, proper design and control re-
quires, among other things, the determination of deformation mechanics
involved in the processes. Without the knowledge of the influences of
variables such as friction conditions, material properties, and workpiece
geometry on the process mechanics, it would not be possible to design the
dies and the equipment adequately, or to predict and prevent the
occurrence of defects. Thus, process modeling for computer simulation has
been a major concern in modern metal-forming technology. Figure 1.1.
indicates the role of process modeling in some detail.
In the past a number of approximate methods of analysis have been
developed and applied to various forming processes. The methods most
well known are the slab method, the slip-line field method, the visioplas-
ticity method, upper- (and lower-) bound techniques, Hill's general
method, and, more recently, the finite-element method (FEM).
In the slab method, the workpiece being deformed is decomposed in
several slabs. For each slab, simplifying assumptions are made mainly with
respect to stress distributions. The resulting approximate equilibrium
equations are solved with imposition of stress compatibility between slabs
and boundary tractions. The final result is a reasonable load prediction
with an approximate stress distribution.
The slip-line field method is used in plane strain for perfectly plastic
materials (constant yield stress) and uses the hyperbolic properties that the
stress equations have in such cases. The construction of slip-line fields,
although producing an "exact" stress distribution, is still quite limited in
1
 PROCESS MODELING OUTPUT

Geometrical parameters Process Analysis Determination

Tool geometry and Optimization of process
Workpiece geometry geometry and
Loads, energies,
process
stresses, strains,
performance
temperature, metal
Process Parameters condition
flow (geometrical
Die and tool motion
change)
Temperature
Lubrication
Material

Material P a r a m e t e r s Primary product Secondary product Equipment

Workhardening requiremements requirements
Capacity
Strain rate sensitivity Forming limits Tolerance, surface limitations
Anisotropy property etc.

Temperature

CONSTRAINTS!
INPUT

FIG. 1.1 Block diagram for process design and control in metal forming.
 Introduction 3

predicting results that give good correlations with experimental work.

From the stress distributions, velocity fields can be calculated through
plasticity equations.
The visioplasticity method originated by Thomsen et al. [1] combines
experiment and analysis. After the velocity vectors have been determined
from an actual test, strain-rates are calculated and the stress distributions
are obtained from plasticity equations. The method has helped to obtain
reliable solutions in detail for processes in which the experimental
determination of the velocity vectors was possible.
The upper-bound method requires the guessing of admissible velocity
fields, among which the best one is chosen by minimizing total potential
energy. Information leading to a good selection of velocity fields comes
from experimental evidence and experience. This method, with ex-
perience, can deliver a fast and relatively accurate prediction of loads and
velocity distributions.
Hill [2] has given a general method of analysis for metal-working
processes when the plastic flow is unconstrained. The method is based on a
criterion of approximation derived from the interpretation of the virtual
work-rate principle. The method was applied to the analysis of compres-
sion with barreling, spread in bar drawing, and thickness change in tube
sinking.
The upper-bound method and Hill's general method are outlined in
more detail, with illustrative examples in relation to the finite-element
method, in Chap. 5. For further reference, the books that provide a wealth
of solutions to many metal-forming problems using the above methods are
listed in the References [3-21].
These methods have been useful in predicting forming loads, overall
geometry changes of deforming workpieces, and qualitative modes of
metal flow, and in determining approximate optimum process conditions.
However, accurate determination of the effects of various process para-
meters on the detailed metal flow became possible only recently, when the
finite-element method was developed for the analyses. Since then, the
finite-element method has assumed steadily increased importance in
simulation of metal-forming processes. Among the books cited, however,
only a few discuss the application of the finite-element method to
metal-forming processes.

1.2. The Finite-Element Method

The finite-element technique, whose engineering birth and boom in the
1960s was due to the application of digital computers to structural analysis,
has spread to a variety of engineering and physical science disciplines in
the last decade.
The basic concept of the finite-element method is one of discretization.
The finite-element model is constructed in the following manner [22]. A
number of finite points are identified in the domain of the function, and the
4 Metal Forming and the Finite-Element Method
values of the function and its derivatives, when appropriate, are specified
at these points. The points are called nodal points. The domain of the
function is represented approximately by a finite collection of subdomains
called finite elements. The domain is then an assemblage of elements
connected together appropriately on their boundaries. The function is
approximated locally within each element by continuous functions that are
uniquely described in terms of the nodal-point values associated with the
particular element. The path to the solution of a finite-element problem
consists of five specific steps: (a) identification of the problem; (b)
definition of the element; (c) establishment of the element equation; (d)
the assemblage of element equations; and (e) the numerical solution of the
global equations. The formation of element equations is accomplished
from one of four directions: (1) direct approach; (2) variational method;
(3) method of weighted residuals; and (4) energy balance approach.
The basis of finite-element metal-flow modeling, for example, using the
variational approach is to formulate proper functionals, depending upon
specific constitutive relations. The solution of the original boundary value
problem is obtained by the solution of the dual variational problem in
which the first-order variation of the functional vanishes. Choosing an
approximate interpolation function (or shape function) for the field
variable in the elements, the functional is expressed locally within each
element in terms of the nodal-point values. The local equations are then
assembled into the overall problem. Thus, the functional is approximated
by a function of global nodal-point values. The condition for this function
to be stationary results in the stiffness equations. These stiffness equations
are then solved under appropriate boundary conditions. The basic mathe-
matical description of the methods, as well as the solution techniques, are
given in several books (for example, References [23, 24, 25]).
The main advantages of the finite-element method are: (1) the capability
of obtaining detailed solutions of the mechanics in a deforming body,
namely, velocities, shapes, strains, stresses, temperatures, or contact
pressure distributions; and (2) the fact that a computer code, once written,
can be used for a large variety of problems by simply changing the input
data.

1.3 Solid Formulation and Flow Formulation

In the analysis of metal forming, plastic strains usually outweigh elastic
strains and the idealization of rigid-plastic or rigid-viscoplastic material
behavior is acceptable. The resulting analysis based on this assumption is
known as the flow formulation [26]. In other applications, phenomena
associated with elasticity cannot be neglected. In the so-called solid
formulation [27], the material is considered to behave as an elastic-plastic
or elastic-viscoplastic solid.
The original problem associated with the deformation process of
materials is a boundary-value problem. For the deformation process of
 Introduction 5

rigid-viscoplastic materials the boundary-value problem is stated as

follows: at a certain stage in the process of quasistatic distortion, the shape
of the body, the internal distribution of temperature, the state of
inhomogeneity, and the current values of material parameters are sup-
posed to be given or to have been determined already. The velocity vector
u is prescribed on a part of surface Su together with traction F on the
remainder of the surface, SF. Solutions to this problem are the stress and
velocity distributions that satisfy the governing equations and the boundary
conditions.
In the solid approach, the boundary value problem is stated such that, in
addition to the current states of the body, the internal distribution of the
stress also is supposed to be known and the boundary conditions are
prescribed in terms of velocity and traction-rate. Distributions of velocity
and stress-rate (or displacement and stress-increment) are the solutions to
the problem.
The solid formulations of the finite-element method for metal-forming
problems have been based on the use of the Prandtl-Reuss equations for
elastic-plastic materials. The formulation is given in the rate form and
assumes the infinitesimal theory of deformation. In analyzing metal-
forming processes, however, the elastic-plastic finite-element method with
infinitesimal formulation has severe drawbacks. The large amount of
rotation involved in metal forming rules out infinitesimal analysis, and
large-deformation analyses also have some difficulties in reproducing
observed phenomena, such as folding in compression of solid cylinders
(see Chap. 9). Furthermore, the nature of elastic-plastic constitutive
equations requires short time steps in nonsteady-state analysis, a require-
ment that is severe when the body goes from elastic to plastic
deformations.
A simplified solution to this problem is to neglect the elastic portion of
deformation and treat all plastic deformation as a flow problem. In
general, this makes an infinitesimal analysis feasible and large steps
possible.

1.4 Metal Forming and the Finite-Element Method

The application of the finite-element method to metal-forming problems
began as an extension of structural analysis technique to the plastic
deformation regime. Thus, early applications of the finite-element method
to metal-forming problems were based on the plastic stress-strain matrix
developed from the Prandtl-Reuss equations. Hydrostatic extrusion,
compression, and indentations were analyzed using this matrix and the
infinitesimal variational formulations.
An analysis method in the area of metal-forming application, in many
cases, can be justified only by its solution reliability and computational
efficiency. This realization has led to the development of numerical
procedures based on the flow formulation. Initial applications of the
6 Metal Forming and the Finite-Element Method
rigid-plastic finite-element method to metal-forming processes were mainly
in the analysis of compression and other simple processes. Since those
early days, many developments of the numerical techniques have occurred,
as well as the continuous growth in the field of applications. Although
advances have been made in recent years, the application of solid
formulation to metal-forming problems is limited. On the other hand, flow
formulation has found applications for a wide variety of important forming
problems.
A most important improvement was the inclusion of the effects of
strain-rate and temperature in material properties and of thermal coupling
in the solution. This development has extended the finite-element analysis
into the warm and hot working range. A further important step in the
development of analysis procedures was the development of a user-
oriented general-purpose program. The natural course of development of
the technique has been in the analysis of two-dimensional and axially
symmetric problems, and most recent developments emphasize the ap-
plication of the finite-element method to three-dimensional problems. A
further development is the unique application of the finite-element method
to preform design in metal forming. Extensive references on the use of the
finite-element method to metal-forming applications can be found in the
References [28, 29].

References
1. Thomsen, E. G., Yang, C. T., and Bierbower, J. B., (1954), "An Experimen-
tal Investigation of the Mechanics of Plastic Deformation of Metals," Univ.
California Pub. Engg., Vol. 5.
2. Hill, R., (1963), "A General Method of Analysis of Metal-Working Proc-
esses," /. Mech. Phys. Solids, Vol. 11, p. 305.
3. Hill, R., (1950), "The Mathematical Theory of Plasticity," Oxford University
Press, London.
4. Prager, W., and Hodge, P. G., Jr., (1951), "Theory of Perfectly Plastic
Solids," Chapman and Hall, London.
5. Hoffman, O., and Sachs, G., (1953), "Introduction to the Theory of Plasticity
for Engineers," McGraw-Hill, New York.
6. Unksov, E. P., (1961), "An Engineering Theory of Plasticity," Butterworths,
London.
7. Johnson, W., and Mellor, P. B., (1973), "Engineering Plasticity," Van
Nostrand and Reinhold, London.
8. Ford, H., and Alexander, J. M., (1963), "Advanced Mechanics of Materials,"
Longmans Green, London.
9. Alexander, J. M., and Brewer, R. C., (1963), "Manufacturing Properties of
Materials," Van Nostrand, London.
10. Thomsen, E. G., Yang, C. T., and Kobayashi, S., (1963), "Mechanics of
Plastic Deformation in Metal Processing," Macmillan, New York: Macmillan-
Collier, London.
11. Kalpakjian, S., (1967), "Mechanical Processing of Materials," Van Nostrand,
Princeton, NJ.
12. Avitzur, B., (1968), "Metal Forming and Processes," McGraw-Hill, New
York.
 Introduction 7

13. Johnson, W., and Kudo, H., (1962), "The Mechanics of Metal Extrusion,"
Manchester University Press, Manchester, UK.
14. Johnson, W., Sowerby, R., and Haddow, J. B., (1970), "Plane-Strain
Slip-Line Fields," American Elsevier, New York.
15. Johnson, W., Sowerby, R., and Venter, R. D., (1982), "Plane Strain Slip Line
Fields for Metal Deformation Processes," Pergamon Press, Oxford.
16. Blazynski, T. Z., (1976), "Metal Forming," Wiley, New York.
17. Rowe, G. W., (1977), "Principles of Industrial Metalworking Processes,"
Edward Arnold, London.
18. Slater, R. A. C., (1977), "Engineering Plasticity," Wiley, New York.
19. Hosford, W. F., and Caddell, R. M., (1983), "Metal Forming; Mechanics and
Metallurgy," Prentice-Hall, Englewood Cliffs, NJ.
20. Altan, T., Oh, S. I., and Gegel, H., (1983), "Metal Forming; Fundamentals
and Applications," American Society for Metals, Metals Park, Ohio.
21. Boer, C. R., Rebelo, N., Rystad, H., and Schroder, G., (1986), "Process
Modelling of Metal Forming and Thermomechanical Treatment," Springer-
Verlag, Berlin.
22. Mote, C. D., Jr., (1980), "Introduction to the Finite Element Method,"
Lecture Note, University of California at Berkeley.
23. Zienkiewicz, O. C., (1977), "The Finite Element Method," 3rd Edition,
McGraw-Hill, New York.
24. Strang, G., and Fix, G. J., (1973), "An Analysis of the Finite Element
Method," Prentice-Hall, Englewood Cliffs, NJ.
25. Huebner, K. H., (1975), "The Finite Element Method for Engineers," Wiley,
New York.
26. Zienkiewicz, O. C., (1984), Flow Formulations for Numerical Solutions of
Forming Processes, "Numerical Analysis of Forming Processes," edited by J.
F. T. Pittman et al., Wiley, New York, p. 1.
27. Nagtegaal, J. C., and Veldpaus, F. E., (1984), On the Implementation of
Finite Strain Plasticity Equations in a Numerical Model, "Numerical Analysis
of Forming Processes," edited by J. F. T. Pittman et al., Wiley, New York, p.
351.
28. Kobayashi, S., (1982), "A Review on the Finite Element Method and Metal
Forming Process Modeling," /. Appl. Metal Working, Vol. 2, p. 163.
29. Kobayashi, S., (1985), "Metal Forming and the Finite Element Method—Past
and Future," Proceedings of the 25th Int. Conf. Mach Tool Des. Res., April,
Birmingham, p. 17.
 2
METAL-FORMING PROCESSES

2.1 Introduction
In metal forming, an initially simple part—a billet or sheet blank, for
example—is plastically deformed between tools (or dies) to obtain the
desired final configuration. Thus, a simple part geometry is transformed
into a complex one, in a process whereby the tools "store" the desired
geometry and impart pressure on the deforming material through the
tool-material interface.
The physical phenomena constituting a forming operation are difficult to
express with quantitative relationships. The metal flow, the friction at the
tool-material interface, the heat generation and transfer during plastic
flow, and the relationships between microstructure/properties and process
conditions are difficult to predict and analyze. Often, in producing discrete
parts, several forming operations (preforming) are required to transform
the initial "simple" geometry into a "complex" geometry, without causing
material failure or degrading material properties. Consequently, the most
significant objective of any method of analysis is to assist the forming
engineer in the design of forming and/or preforming sequences. For a
given operation (preforming or finish-forming), such design essentially
consists of (1) establishing the kinematic relationships (shape, velocities,
strain-rates, strains) between the deformed and undeformed part, i.e.,
predicting metal flow; (2) establishing the limits of formability or produci-
bility, i.e., determining whether it is possible to form the part without
surface or internal defects; and (3) predicting the forces and stresses
necessary to execute the forming operation so that tooling and equipment
can be designed or selected.
For the understanding and quantitative design and optimization of
metal-forming operations it is useful (a) to consider a metal forming
process as a system and (b) to classify these processes in a systematic way.

2.2 A Metal-Forming Operation as a System

A metal-forming system comprises all the input variables relating the billet
or blank (geometry and material), the tooling (geometry and material), the
conditions at the tool-material interface, the mechanics of plastic deforma-
tion, the equipment used, the characteristics of the final product, and
8
 Metal-Forming Processes 9

Billet(l) Process (2,3,4,5) Product (6)

FIG. 2.1 Illustration of metal forming system using closed-die forging as an example: 1,
billet; 2, dies; 3, interface; 4, deformation mechanics; 5, forming machine; 6, product; 7,
environment [2].

finally the plant environment in which the process is being conducted [1].
Such a system is illustrated in Fig. 2.1, using impression die forging as an
example [2].
The "systems approach" in metal forming allows study of the effects of
process variables on product quality and process economics. The key to a
successful metal-forming operation, i.e., to obtaining the desired shape
and properties, is the understanding and control of metal flow. The
direction of metal flow, the magnitude of deformation, and the tempera-
tures involved greatly influence the properties of the formed components.
Metal flow determines both the mechanical properties related to local
deformation and the formation of defects such as cracks or folds at or
below the surface. The local metal flow is in turn influenced by the process
variables, which are discussed below.
Material Variables
For a given material composition and deformation/heat-treatment history
(microstructure), the flow stress (or effective stress), and the workability
(or formability) in various directions (anisotropy), are the most important
material variables in the analysis of a metal-forming process.
For a given microstructure, the flow stress is expressed as a function of
strain, strain-rate, and temperature. To determine the actual functional
relationship, it is necessary to conduct torsion, plane-strain compression,
and uniform axisymmetric compression tests. Workability or formability is
the capability of a material to deform without failure; it depends on (1)
conditions existing during deformation processing (such as temperature,
rate of deformation, stresses, and strain history), and (2) material variables
10 Metal Forming and the Finite-Element Method

(such as composition, voids, inclusions, and initial microstructure). In hot

forming processes, temperature gradients in the deforming material (for
example, due to local die chilling) also influence metal flow and failure
phenomena.

Tooling and Equipment

The selection of a machine for a given process is influenced by the time,
accuracy, and load-energy characteristics of that machine. Optimum
equipment selection requires consideration of the entire forming system,
including lot size, conditions at the plant, environmental effects, and
maintenance requirements, as well as the requirements of the specific part
and process under consideration.
The tooling variables include (1) design and geometry, (2) surface finish,
(3) stiffness, and (4) mechanical and thermal properties under conditions
of use.

Friction
The mechanisms of interface friction are very complex. One way of
expressing friction quantitatively is through a friction coefficient n, or a
friction shear factor m. There are various methods of evaluating friction,
i.e., estimating the value of \n or m. Tests most commonly used are the
ring and spike tests for massive forming and the plane-strain-draw and
stretch-draw tests for sheet forming [2].

Deformation Mechanics
In forming, material is deformed plastically to generate the shape of the
desired product. Metal flow is influenced mainly by (1) tool geometry, (2)
friction conditions, (3) characteristics of the stock material, and (4)
thermal conditions existing in the deformation zone. The details of metal
flow influence the quality and properties of the formed product and the
force and energy requirements of the process. The mechanics of deforma-
tion, i.e., the metal flow, strains, strain-rates, and stresses, can be
investigated by process modeling. Some analysis methods for process
modeling are outlined in Chap. 1 (see Section 1.1), and process modeling
by the finite-element method is the main subject of this book.
Product Properties
The macro- and microgeometry of the product, i.e., its dimensions and
surface finish, are influenced by process variables. The processing condi-
tions (temperature, strain, and strain-rate) determine the microstructural
variations taking place during deformation and often influence final
product properties. Consequently, a realistic systems approach must
include consideration of (1) the relationships between properties and
microstructure of the formed material and (2) the quantitative influences
of process conditions on metal flow and resulting microstructures.
 Metal-Forming Processes 11

TABLE 2.1 Classification of Massive Forming Processes [3]

Forging Rolling Extrusion Drawing

Closed-die forging Sheet rolling Nonlubricated hot Drawing

with flash Shape rolling extrusion Drawing with
Closed-die forging Tube rolling Lubricated direct hot rolls
without flash Ring rolling extrusion Ironing
Coining Rotary tube Hydrostatic extrusion Tube sinking
Electro-upsetting piercing
Forward extrusion Gear rolling
forging Roll forging
Backward extrusion Cross rolling
forging Surface rolling
Hobbing Shear forming
Isothermal forging (flow turning)
Nosing Tube reducing
Open-die forging
Orbital forging
P/M forging
Radial forging
Upsetting

2.3 Classification and Description of Metal-Forming Processes [3]

The metal-forming processes may be classified into two broad categories:
1. Massive forming processes (Table 2.1)
2. Sheet-metal forming processes (Table 2.2)
In both cases, the surfaces of the deforming material and of the tools are

TABLE 2.2 Classification of Sheet-metal Forming Processes [3]

Bending and Straight Flanging Deep Recessing and Flanging

Brake bending Spinning (and roller flanging)
Roll bending Deep drawing
Surface Contouring of Sheet Rubber pad forming
Contour stretch forming Marform process
(stretch forming) Rubber diaphragm hydroforming
Androforming Shallow Recessing
Age forming Dimpling
Creep forming Drop hammer forming
Die-quench forming Electromagnetic forming
Bulging Explosive forming
Vacuum forming Joggling
Linear Contouring
Linear stretch forming
(stretch forming)
Linear roll forming
(roll forming)
12 Metal Forming and the Finite-Element Method

in contact, and friction between them has a major influence on the process.
In massive forming, the input material is in billet, rod, or slab form, and a
considerable increase in the surface-to-volume ratio occurs in the formed
part. In sheet forming, a sheet blank is plastically deformed into a
three-dimensional object without any significant changes in sheet thickness
and surface characteristics.
Processes that fall under the category of massive forming processes
(Table 2.1), have the following distinguishing features:
• The workpiece undergoes large plastic deformation, resulting in an
appreciable change in shape or cross section.
• The portion of the workpiece undergoing permanent (plastic) defor-
mation is generally much larger than the portion undergoing elastic
deformation; therefore, elastic recovery after deformation is
negligible.
The characteristics of sheet-metal forming processes (Table 2.2) are:
• The workpiece is a sheet or a part fabricated from a sheet.
• The deformation usually causes significant changes in shape, but not
in cross section, of the sheet.
• In some cases, the magnitudes of permanent plastic and recoverable
elastic deformations are comparable; therefore, elastic recovery or
springback may be significant.
Most significant metal-forming processes are listed in Tables 2.1 and 2.2
[3]. Selected massive and sheet forming processes are described in Figs 2.2
through 2.29 [2-10].

FIG. 2.2 Closed-die forging without flash. A billet with carefully controlled volume is
deformed (hot or cold) by a punch in order to fill a die cavity without any loss of material.
The punch and the die may be made of one or several pieces.
FIG. 2.3 Closed-die forging with flash. A billet is formed (hot) in dies (usually with two
halves) such that the flow of metal from the die cavity is restricted. The excess material is
extruded through a restrictive narrow gap and appears as flash around the forging at the die
parting line.

FIG. 2.4 Coining opeation. Coining is a closed-die forming operation, usually performed
cold, in which all surfaces of the work are confined or restrained, resulting in a well-defined
imprint of the die on the workpiece. It is also a restriking operation used to sharpen or change
an existing radius or profile.

FIG. 2.5 Forward extrusion forging. A punch compresses a billet (hot or cold) confined in a
container so that the billet material flows through a die in the same direction as the punch.

13
FIG. 2.6 Backward extrusion forging. A moving punch applies a steady pressure to a slug
(hot or cold) confined in a closed die, and forces the metal to flow around the punch in a
direction opposite the direction of punch travel.

(A) (B)

FIG. 2.7 Robbing (A) in a container and (B) without restriction. Robbing is the process of
indenting or coining an impression into a cold or hot die block by pressing with a punch.

14
FIG. 2.8 Nosing. Nosing is a hot or cold forming process in which the open end of a shell or
tubular component is closed by axial pressing with a shaped die.

FIG. 2.9 Open-die forging. Open-die forging is a hot forming process in which metal is
shaped by hammering or pressing between flat or simple contoured dies.

15
FIG. 2.10 Various stages in orbital forging processes. Orbital forging is the process of
forming shaped parts by incrementally forging (hot or cold) a slug between an orbiting upper
die and a nonrotating lower die. The lower die is raised axially toward the upper die, which is
fixed axially but whose axis makes orbital, spiral, planetary, or straight-line motions.

FIG. 2.11 Radial forging of a shaft. This hot or cold forming process utilizes two or more
radially moving anvils, or dies, for producing solid or tubular components with constant or
varying cross sections along their lengths.

16
FIG. 2.12 Upsetting with flat-heading tool. Upsetting is the process of forming metal (hot or
cold) so that the cross-sectional area of a portion, or all, of the stock is increased.

FIG. 2.13 Schematic of the rolling process for sheet and plates. Sheet and plate rolling is a
hot or cold forming process for reducing the cross-sectional area of the stock with the use of
rotating rolls. In general, the rolled material elongates and spreads simultaneously while the
cross-sectional area is reduced.

FIG. 2.14 Roll passes for rolling an angle (L) shape. Shape rolling is a cold or hot forming
process for reducing as well as shaping the cross section of the metal stock by passing it
through a series of rotating sets of rolls with appropriately shaped grooves.
17
FIG. 2.15 Principles of ring rolling. Ring rolling is a process whereby a hollow circular blank
(cold or hot) is formed into a ring. A main roll presses on the outside diameter of the blank,
which is supported by a mandrel on the inside diameter. Shaped cross sections are obtained
by appropriate contouring of the mandrel and the roll. The height of the ring is controlled by
auxiliary rolls.

FIG. 2.16 Rotary tube piercing, a, Double-conical working rolls; b, guide roll; c, billet; d,
conical piercer point. A hollow is formed by peripherally rolling a cylindrical hot billet over a
conical piercer point. The billet is driven by a pair of cone-shape rolls, set askew to the
longitudinal axis of the billet. The frictional load between the rolls and the billet causes the
billet to rotate and forces it to advance longitudinally over the piercer point.

18
 (a) (b)

FIG. 2.17 Shear forming from a plate, (a) Start, (b) partially or completely formed part.
Shear forming is a process for hot or cold seamless shaping of dished parts by the combined
forces of rotation and pressure. This process differs from spinning principally in that it
reduces the thickness of the formed part.

DIRECT EXTRUSION INDIRECT EXTRUSION

FIG. 2.18 Tooling and metal flow for direct and indirect extrusion process. The product of
desired cross section is obtained by forcing a heated billet through a die without lubricating
the billet, the container, or the die. In the direct extrusion process, the product is extruded in
the direction of ram movement. When the product is extruded in a direction opposite that of
ram travel, the process is called indirect extrusion.

19
FIG. 2.19 Hot extrusion setup using glass lubrication. The heated billet is forced through a
die, using some form of lubrication, to obtain a product of desired cross section. Glass is the
most widely used lubricant for extruding long lengths from steels and high-temperaure alloys
on a production basis.

(a) (b)

FIG. 2.20. Drawing of (a) rod or wire and (b) tube. Drawing is the process of reducing the
cross-sectional area and/or the shape of a rod, bar, tube or wire (cold or hot) by pulling
through a die.

20
FIG. 2.21 Schematic of ironing. Ironing is the process of smoothing or thinning the wall of a
shell or cup (cold or hot) by forcing the shell through a die with a punch.

FIG. 2.22 Tube sinking process. Tube sinking is the process of sizing the outside diameter of
a tube by drawing the tube (cold or hot) through a die without supporting the tube internally
with a mandrel.

A B C D

FIG. 2.23 Typical brake-bending operations. (A) Air bending; (B) air rounding, (C) die
bending; (D) die rounding. Brake bending is a forming operation widely used for forming flat
sheets into linear sections, such as angles, channels and hats. There are two typical
brake-forming setups: air bending and die bending. In air bending, the workpiece is
supported only at the outer edges so that the length of the ram stroke determines the bend
angle of the part. In die bending, the sheet is forced into a female die cavity of the required
part angle.
21
 Workpiece
Bending roll

Driven rolls

Support rolls
FIG. 2.24 Roll bending (three-roll forming). Roll bending gives a curvature to a sheet, bar,
or shaped section by bending it between two or three cylindrical rolls that can be adjusted.

(a) Roller profiles for (b) Various passes for roll

one pass forming "HAT" section

FIG. 2.25 Roll forming. This process is used to produce long components of various cross
sections. The sheet metal is formed by passing it through a succession of progressively shaped
power-driven contoured rolls.

22
 1C) Id)
FIG. 2.26 Various spinning operations, (a) Hollow shape forming; (b) bulging, (c) reducing,
(d) threading. Spinning is the process of shaping seamless dished parts by the combined
forces of rotation and pressure. Spinning does not result in any change in thickness.

FIG. 2.27 Deep drawing. (A) First draw; (B) redraw; (C) reverse draw. In deep drawing, a
sheet blank (hot or cold), usually subjected to a peripheral hold-down pressure, is forced by a
punch into and through a die to form a deep recessed part having a wall thickness
substantially the same as that of the blank. This process is used to produce cylindrical or
prismatic cups with or without a flange on the open end. Cups or tubes can be sunk or
redrawn to increase their length and to reduce their lateral dimensions.

23
24 Metal Forming and the Finite-Element Method

FIG. 2.28 Rubber-pad forming, (a) With rubber punch; (b) with rubber pad. This is a
forming operation for producing shallow parts. A rubber pad is attached to the press slide
and becomes the mating die for a punch, or group of punches, which has been placed on the
press bed or plate. The rubber pad is confined in a container (pad holder), and the entire
slide with attached pad holder is forced against the tools, usually by hydraulic pressure. As
the slide descends, the pliable but virtually incompressible rubber fills the space between the
slide and the dies and forces the metal to take the exact contours of the dies.

BEGINNING OF FORMING COMPLETE FORMING

FIG. 2.29. Rubber-diaphragm hydroforming. In this process, the blank is held between a
diaphragm, which closes the ram pressure chamber, and a blank holder. A male punch works
against the diaphragm, and the metal is shaped by balancing the pressure of the ram chamber
against the pressure of the press base chamber on which the punch is mounted.

References
1. Altan, T., Lahoti, G. D., and Nagpal, V., (1981), "Systems Approach in
Massive Forming and Application to Modeling of Forging Processes," /. Appl.
Metal Working, ASM, Vol. 1, No. 2, p. 29.
2. American Society for Metals, (1961), "Metals Handbook", Eighth Edition,
Vol. 1 (Properties and Selection of Metals) and Vol. 4 (Forming), American
society for Metals, Metals Park, OH.
3. Altan, T., Oh, S. I., and Gegel, H., (1983), "Metal Forming: Fundamentals
and Applications," ASM International, Metals Parks, OH.
4. International Institution for Production Engineering Research, (1962), Dic-
tionary of Production Engineering, Vol. 1 (Forming and Drop Forging), Vol. 3
 Metal-Forming Processes 25

(Sheet Metal Forming), and Vol. 5 (Cold Extrusion and Upsetting), Verlag,
W. Girardet, Essen.
5. Lahoti, G. D., and Altan, T., (1976), "Design of Dies for Radial Forging of
Rods and Tube," Technical Paper MF76-390, Society of Manufacturing
Engineers, Dearborn, MI.
6. Aluminium, American Society for Metals, (1967), Vol. 3 (Fabrication and
Finishing), edited by K. R. Van Horn, American Society for Metals, Metals
Park, OH, p. 81.
7. Geleji, A., (1967), "Forge Equipment Rolling Mills and Accessories,"
Akademiai Kiado, Budapest.
8. Beyon, R. E., (1956), "Roll Design and Mill Layout," Association of Iron and
Steel Engineers, Pittsburgh.
9. Sachs, G., (1951), "Principles and Methods of Sheet-Metal Fabricating,"
Reinhold, New York.
10. Lange, K., (1972), "Lehrbuch der Umformtechnik/Textbook of Forming
Technology" (in German), Springer-Verlag, Berlin.
 3
ANALYSIS AND TECHNOLOGY
IN METAL FORMING

3.1 Introduction
The design, control, and optimization of forming processes require (1)
analytical knowledge regarding metal flow, stresses, and heat transfer, as
well as (2) technological information related to lubrication, heating and
cooling techniques, material handling, die design and manufacture, and
forming equipment.
The purpose of using analysis in metal forming is to investigate the
mechanics of plastic deformation processes, with the following major
objectives.
• Establishing the kinematic relationships (shape, velocities, strain-
rates, and strains) between the undeformed part (billet, blank, or
preform) and the deformed part (product); i.e., predicting metal flow
during the forming operation. This objective includes the prediction of
temperatures and heat transfer, since these variables greatly influence
local metal-flow conditions.
• Establishing the limits of formability or producibility; i.e., determin-
ing whether it is possible to perform the forming operation without
causing any surface or internal defects (cracks or folds) in the
deforming material.
• Predicting the stresses, the forces, and the energy necessary to carry
out the forming operation. This information is necessary for tool
design and for selecting the appropriate equipment, with adequate
force and energy capabilities, to perform the forming operation.
Thus, the mechanics of deformation provides the means for determining
how the metal flows, how the desired geometry can be obtained by plastic
deformation, and what the expected mechanical properties of the pro-
duced part are.
For understanding the variables of a metal-forming process, it is best to
consider the process as a system, as illustrated in Fig. 2.1 in Chap. 2. The
interaction of most significant variables in metal forming are shown, in a
simplified manner, in Fig. 3.1. It is seen that for a given billet or blank
material and part geometry, the speed of deformation influences strain-
rate and flow stress. Deformation speed, part geometry, and die tempera-
ture influence the temperature distribution in the formed part. Finally,
26
 Analysis and Technology in Metal Forming 27

FIG. 3.1 Simplified illustration of the interactions between major process variables in metal
forming

flow stress, friction, and part geometry determine metal flow, forming
load, and forming energy.
In steady-state flow (kinematically), the velocity field remains un-
changed, as is the case in the extrusion process (Fig. 3.2B); in nonsteady-
state flow, the velocity field changes continuously with time, as is the case
in upset forging (Fig. 3.2A) [1].

(A) (B)

FIG. 3.2 Metal flow in certain forming processes. (A) Non-steady state upset forging; (B)
steady-state extrusion [1].
28 Metal Forming and the Finite-Element Method

The state of deformation in a plastically deforming metal is fully

described by the displacements, velocities, strains, and strain-rates. There
are several approximate methods for analyzing metal-forming problems.
They are briefly outlined in Section 1.1 of Chap. 1, and the upper-bound
method and Hill's general method are further illustrated in Chap. 5.
Details of some of these methods are given in the metal-forming books
listed in Chap. 1, and the forming processes, with emphasis on technologi-
cal aspects, are described in Reference [1]. It is to be noted that every
method of analysis requires as input (1) a description of the material
behavior under the process conditions, i.e., flow stress data, and (2) a
quantitative value to describe the friction, i.e., the friction factor m, or the
friction coefficient /j,. These two quantities themselves—flow stress and
friction—must be determined by experiment and are difficult to obtain
accurately. Thus, in addition to simplifications and approximations as-
sumed in the methods, any errors in flow stress measurements or
uncertainties in the value of the friction factor are expected to influence
the reliability of the results of analysis. Johnson and Sowerby [2] reviewed
recent analytical researches into drawing, extrusion, rolling, forging, and
sheet-metal forming in the context of the limitations imposed by tech-
nological considerations.

3.2 Flow Stress of Metals

The yield stress of a metal under uniaxial conditions, as a function of
strain, strain-rate, and temperature, can also be considered as the flow
stress (or the effective stress). The definition of the effective stress as a
representative stress under combined loading is given in Chap. 4.
The flow stress o is important because in metal-forming processes the
forming loads and stresses depend on (1) part geometry, (2) friction, and
(3) the flow stress of the deforming material. The flow stress of a metal is
influenced by:
• Factors unrelated to the deformation process, such as a chemical
compositon, metallurgical structure, phases, grain size, segregation,
and prior strain history
• Factors explicitly related to the deformation process, such as tempera-
ture, degree of deformation, and rate of deformation. The degree of
deformation and rate of deformation under general loading are
measured by the effective strain £ and the effective strain-rate e,
respectively, and their definitions are also given in Chap. 4
Thus, the flow stress a can be expressed as a function of temperature T,
strain e, strain-rate e, and microstructure S:

In hot forming of metals at temperatures above the recrystallization

temperature, the influence of strain on flow stress is insignificant, and the
 Analysis and Technology in Metal Forming 29

influence of strain-rate (i.e., rate of deformation) becomes increasingly

important. Conversely, at room temperature (i.e., in cold forming), the
effect of strain-rate on flow stress is negligible, and the effect of strain on
flow stress (i.e., strain hardening) is most important. The degree of
dependence of flow stress on temperature varies considerably among
different materials. Therefore, temperature variations in a forming opera-
tion can have quite different effects on load requirements and on metal
flow for different materials. For instance, a drop of approximately 100°F in
the hot forming temperature (from 1700 to 1600°F) would result in a 40%
increase in flow stress for titanium alloy Ti-8Al-lMo-lV. The increase in
flow stress that would result from the same temperature drop, 100°F within
the hot working range (from 2200 to 2100°F), would be only about 15% for
AISI type 4340 steel [3].
To be useful in metal-forming analyses, the flow stresses of metals must
be determined experimentally for the strain, strain-rate, and temperature
conditions that exist in metal-forming processes. The methods most
commonly used for obtaining flow stress data are tensile, uniform
compression, and torsion tests. The compression test is particularly simple,
and therefore it is very widely used. In this test, the flat platens and the
cylindrical sample are maintained at the same temperature, so that die
chilling, with its influence on metal flow, is prevented. To be applicable
without errors or corrections, the cylindrical workpiece must be upset
without any barreling; i.e., the state of uniform deformation in the
workpiece must be maintained. Barreling is prevented by using adequate
lubrication, e.g., Teflon or machine oil at room temperature and, at hot
working temperatures, graphite in oil for aluminium alloys, and glass for
steel, titanium, and high-temperature alloys [4]. The load and displace-
ment or specimen height are measured during the test. From this
information the flow stress is calculated at each stage of deformation, or,
for increasing strain, at a strain-rate given by the ratio of the instantaneous
ram speed to specimen height.
At room temperature the flow stresses of most metals (except that of
lead, for example) are only slightly strain-rate dependent. Therefore, any
testing machine or press can be used for the compression test, regardless of
its ram speed.
At hot working temperatures, i.e., above the recrystallization tempera-
ture, the flow stresses of nearly all metals are very much strain-rate
dependent. Therefore, whenever possible, these temperature range com-
pression tests are conducted on a machine that provides a velocity-
displacement profile such that the constant-strain-rate condition can be
maintained throughout the test. Mechanical cam-activated presses called
plastometers or hydraulic programmable testing machines (MTS, for
example) [5] are used for this purpose. In order to maintain nearly
isothermal and uniform compression conditions, the test is conducted in a
furnace or a fixture. The specimens are lubricated with appropriate
lubricants—for example, oil-graphite for temperatures up to SOOT and
30 Metal Forming and the Finite-Element Method

FIG. 3.3 Flow stress vs. strain, and strain-rate vs. strain, for type 403 stainless steel at 1800,
1950 and 2050°F (tests were conducted in a mechanical press where e was not constant) [4].

glass for temperatures up to 2300T. The fixture and the specimens are
heated to test temperature and then the test is initiated. Examples of
high-temperature uniaxial flow stress data are given in Figs. 3.3 and 3.4.

3.3 Friction in Metal Forming

Friction conditions at the die-material interface greatly influence metal
flow, formation of surface and internal defects, stresses acting on the dies,
and load and energy requirements. There are three basic types of
lubrication that govern the frictional conditions in metal forming [5, 6].
1. Under dry conditions, no lubricant is present at the interface and
only the oxide layers present on the die and workpiece materials may
act as a "separating" layer. In this case friction is high, and such a
situation is desirable in only a few selected forming operations, such
as hot rolling of plates and slabs and nonlubricated extrusion of
aluminium alloys.
 Analysis and Technology in Metal Forming 31

FIG. 3.4 Flow stress vs. strain, and strain-rate vs. strain, for Waspaloy at 1950, 2050 and
2100°F (tests were conducted in a mechanical press where e was not constant) [4].

2. "Hydrodynamic" conditions exist when a thick layer of lubricant is

present between the dies and the workpiece. In this case the friction
conditions are governed by the viscosity of the lubricant and by the
relative velocity between the die and the workpiece. The viscosities
of most lubricants decrease rapidly with increasing temperature.
Consequently, in most practical high-speed forming operations, such
as strip rolling and wire drawing, the hydrodynamic conditions exist
only within a certain regime of velocities, where the interface
temperatures are relatively low [6].
3. "Boundary" lubrication is the most widely encountered situation in
metal forming. Increases in temperature at the interface and the
relatively high forming pressures do not usually allow the presence of
a hydrodynamic lubrication regime. Boundary lubrication, on the
other hand, does not lend itself to reliable analysis.
Consequently, most of the knowledge on metal-forming lubrication
is empirical, with very little analysis-based information.
32 Metal Forming and the Finite-Element Method
In most forming applications, the lubricity of a lubricant is the single
most significant factor, since it directly determines the interface friction. In
order to evaluate the performances of various lubricants and to be able to
predict forming pressures, it is necessary to express the interface friction
quantitatively, in terms of a factor or a coefficient. The friction shear
stress, fs, is most commonly expressed as

p being a compressive normal stress to the interface, or as

k being the shear strength of the deforming material, where 0 < ra < 1.
Studies in forming mechanics indicate that eq. (3.3) adequately repre-
sents the friction condition in bulk forming processes while eq. (3.2) is
commonly used for representation of friction in sheet-metal forming. A
reason for this is that the compressive normal stress at the interface in
sheet-metal forming is much smaller in magnitude, in comparison with that
in bulk deformation processes. For various forming conditions, the values
of m vary as follows:
• m = 0.05-0.15 in cold forming of steels, aluminium alloys, and
copper, using conventional phosphate-soap lubricants or oils
• m = 0.2-0.4 in hot forming of steels, copper, and aluminum alloys
with graphite-based (graphite-water or graphite-oil) lubricants
• m= 0.1-0.3 in hot forming of titanium and high-temperature alloys
with glass lubricants
• m = 0.7-1.0 when no lubricant is used, e.g., in hot rolling of plates or
slabs and in nonlubricated extrusion of aluminium alloys
In determining the friction factor m for hot forming, in addition to
lubrication effects, the effects of die chilling or heat transfer from the hot
material to colder dies must be considered. Therefore, the lubrication tests
used for determining friction factors must include both lubrication and
die-chilling effects. Consequently, in hot forming, a good test must satisfy
as well as possible the following requirements.
• The specimen and die temperatures must be approximately the same
as those encountered in the actual hot forming operation.
• The contact time between specimen and tools under pressure must be
approximately the same as in the forming operation of interest.
• The ratio of the newly generated deformed surface area to original
surface area of the undeformed specimen must be approximately the
same as in the process investigated.
• The relative velocity between deforming metal and dies should have
approximately the same magnitude and direction as in the forming
process.
Lubricity, as defined by the friction factor m, is most commonly
 Analysis and Technology in Metal Forming 33

measured by using the ring test. In the ring test, a flat ring-shaped
specimen is compressed to a known reduction. The change in internal and
external diameters of the forged ring is very much dependent on the
friction at the die-workpiece interface. If friction were zero, the ring
would deform in the same way as a solid disk, with each element flowing
radially outward at a rate proportional to its distance from the center.
With increasing deformation, the internal diameter of the ring is reduced if
friction is high, and is increased if friction is low. Thus, the change in the
internal diameter represents a simple method for evaluating interface
friction [5,6].

3.4 Temperatures in Metal Forming

In metal-forming processes, both plastic deformation and friction contrib-
ute to heat generation. Approximately 90-95% of the mechanical energy
involved in the process is transformed into heat. In some continuous
forming operations such as drawing and extrusion, performed at high
speeds, temperature increases of several hundred degrees may be in-
volved. A part of the generated heat remains in the deformed material,
another part flows into tooling, while still a further part may flow into the
undeformed portion of the material. The temperatures developed in the
process influence lubrication conditions, tool life, and the properties of the
final product, and, most significantly, determine the maximum deforma-
tion speed that can be used for producing sound products without
excessive tool damage. Thus, temperatures generated during plastic
deformation greatly influence the productivity of metal-forming processes
[5].
The magnitudes and distribution of temperatures depend mainly on:
• The initial material and die temperatures
• Heat generation due to plastic deformation and friction at the
die-material interface
• Heat transfer between the deforming material and the dies and
between the material and the environment (air or coolant)
In actual forming operations there is temperature gradient in deforming
material and in the dies. The temperature distributions encountered in
forming operations for producing discrete parts, such as die forging,
upsetting, and deep drawing, are quite different from the temperature
increases found in quasicontinuous deformation processes such as wire
drawing, rolling, and extrusion. In forming operations of the former type,
e.g., in cold forging, the metal flow is kinematically nonsteady state.
Deformation takes place in a relatively short period of time, i.e., from
several milliseconds to a fraction of a second. The deforming material is in
contact with the dies during this short period. After the part has been
formed and removed from the die, the dies can cool off during a
considerable period of time, until the next part is loaded into them.
34 Metal Forming and the Finite-Element Method

H Q -H. INCH

DISPLACEMENT. H Q -H, MM

FIG. 3.5 Load vs. displacement curves obtained in closed-die forging of an axisymmetric
steel part (dimensions in inches) at 2012°F in three different machines with different initial
velocities, Vpi [5].

In continuous forming operations, e.g., wire drawing, the metal flow is

nearly steady state. The deforming material is continuously in contact with
the die and there is a cumulative temperature increase that significantly
influences die life, production rate, and the quality of drawn material.
The influence of temperatures in metal-forming operations is most
dramatic in hot forming operations, where the contact time under pressure
between the deforming material and the dies is the most significant factor
influencing temperature conditions. This is illustrated in Fig. 3.5, where
the load-displacement curves are given for hot forging of a steel part using
different types of forging equipment [5]. These curves illustrate that owing
to strain-rate and temperature effects, for the same forging process,
different forging loads and energies are required by different machines.
For the hammer, the forging load is initially higher owing to strain-rate
effects, but the maximum load is lower than for either hydraulic or screw
presses. The reason for this is that in the presses the extruded flash cools
rapidly, whereas in the hammer the flash temperature remains nearly the
same as the initial stock temperature.
Thus, in hot forming, not only the material and the formed shape but
also the type of equipment used (rate of deformation and die-chilling
effects) determine the metal-flow behavior and the forming load and
energy required for the process. Surface tearing and cracking or develop-
ment of shear bands in the formed material often can be explained by
 Analysis and Technology in Metal Forming 35

excessive chilling of the surface layers of the formed part near the
die-material interface.

3.5 Impression and Closed-Die Forging

In impression or closed-die forging, two or more dies are moved toward
each other to form a metal billet, at a suitable temperature, in a shape
determined by the die impressions. In most practical hot forging opera-
tions, the temperature of the workpiece materials is higher than that of the
dies. Metal flow and die filling are largely determined by (1) the flow stress
and formability, (2) the friction and heat transfer at the die-material
interface, and (3) the complexity of the forging shape. For a given metal,
both flow stress and forgeability (formability in forging) are influenced by
the metallurgical structure, temperatures, strains, strain-rates, and stresses
that occur during deformation.
The main objective of forging process design is to ensure adequate flow
of the metal in the dies so that the desired finished part geometry can be
obtained without defects and with prescribed properties. Metal flow is
greatly influenced by part or die geometry. Often several operations
(preforming or blocking) are needed to achieve gradual flow of metal from
an initially simple shape (cylinder or round-cornered square billet) into the
more complex shape of the final forging.
The steps involved in designing a forging process are seen in Fig. 3.6. In
the finish-forging die, the flash geometry is selected such that the flash is
encouraged to restrict metal flow into the "flash gutter," outside of the die
cavity, as seen in Fig. 3.7 [3]. This results in an increase in the forging
stresses. Therefore, it is necessary on one hand to decrease the flash
thickness or increase the flash width, while on the other maintaining the
die stresses at a level such that the dies are not damaged. Computerized
upper-bound and slab techniques can be used for this purpose.
Design of blocker and preform geometries is the most critical part of
forging die design. The blocker operation has the purpose of distributing
the metal adequately within the blocker (or preform) to achieve the
following objectives:
• Filling the finisher cavity without any forging defects
• Reducing the amount of material lost as forging flash
• Reducing die wear by minimizing metal movement in the finisher die
• Providing the required amounts of deformation and grain flow so that
desired forging properties are obtained
Traditionally, blocker dies and preforms are designed by experienced
die designers and are modified and refined by die try outs. The initial
blocker design is based on several empirical guidelines. These guidelines
depend on the material used and on the forging machine utilized.
At present, computer-aided design (CAD) of blocker cross sections can
be carried out using interactive graphics. However, this method still
36 Metal Forming and the Finite-Element Method
CAD/CAM PROCEDURE OF
FORGING DIE DESIGN

FIG. 3.6 Outline of a CAD/CAM procedure for forging die design and manufacture.

employs the same empirical relationships listed above, but stored in a

quantitative manner in the computer memory.
As illustrated in Fig. 3.6, the ultimate process and die design in forging
requires nonisothermal flow simulation via 2D and 3D FEM techniques.
Thus, dies can be designed better and the need for experimentation is
reduced.
3.6 Hot Extrusion of Rods and Shapes
Extrusion is used to produce long and straight sections of constant cross
section. There are basically three variations of extrusion, depending on the
lubrication technique used. In the nonlubricated extrusion process (Fig.
3.8A), a flat-face die is used, and the material flows by internal shear and
causes a "dead-metal zone" to form in front of the extrusion die. In
 Analysis and Technology in Metal Forming 37

FIG. 3.7 Metal flow and load-stroke curve in impression-die forging: (A) upsetting, (B)
filling, (C) end, (D) load-stroke curve.

lubricated extrusion (Fig. 3.8B), a suitable lubricant is present between the

extruded billet and the extrusion tooling, i.e, the container and the die.
The third and most recently developed technique is hydrostatic extrusion
(Fig. 3.8C), in which a fluid film between the billet and the tooling exerts
pressure on the deforming billet. Hydrostatic extrusion is used only in
unusual applications for extruding special alloys, composites, or clad
materials, where adequate lubrication cannot be easily provided by
conventional lubrication techniques [5].
Metal flow during extrusion varies considerably, depending on the
material, friction, and the shape of the extruded section. In extrusion of
aluminium alloys, temperatures developed during the process, greatly
influence the maximum ram speed. This is especially true in extrusion of
hard aluminium alloys. Temperatures are determined by simultaneous (a)

(A) (B) (C)

FIG. 3.8 Schematic illustrations of the nonlubricated (A), lubricated (B), and hydrostatic (C)
extrusion processes.
38 Metal Forming and the Finite-Element Method
heat generation due to plastic deformation and friction, (b) heat transfer at
billet-tooling interface and heat conduction within the billet and the
tooling, and (c) heat transported with the extruded product.
In hot extrusion of aluminum and copper alloys, container lubrication is
not used and the dies are of the "flat-face" type, with the die opening
imparting the desired section geometry to the extrusion. In extrusion of
steels, titanium alloys, and other high-temperature materials, glass- or
graphite-based lubricants are used. The dies have some sort of "smooth
entry" design to provide for easy metal flow and to avoid severe internal
shear, or formation of a dead-metal zone, during extrusion. "Smooth
entry" dies are also used successfully for extruding composite materials. In
these applications, internal shear, which occurs in extrusion with flat dies,
must be avoided in order to maintain the integrity and the uniformity of
the composite structure.
In today's industrial practice, the design of extrusion dies, whether of
the "flat-face" or the "smooth entry" type, is still an art rather than a
science. Die design for a new extrusion is developed from previous
experience and through costly experimentation and in-plant trials. Thus,
process and die development may require relatively long periods of time
and may tie up extrusion presses that should preferably be used for actual
production. A scientific design of the extrusion process could use FEM to
establish the following variables [5]:
• Optimum number of shaped orifices in the die; as seen in Fig. 3.9, in
aluminum extrusion several sections are often extruded simul-
taneously [7]
• Location of the orifices relative to the billet axis for uniform flow
through each orifice
• Orientation of the orifices
• Modification of the shape of the orifices to correct for thermal
shrinkage and die deflection under load

FIG. 3.9 Schematic illustration of a flat-face die for extrusion of "T" sections [7].
 Analysis and Technology in Metal Forming 39
• Determination of die bearing lengths for balancing metal flow to avoid
the twisting and bending of the extrusion emerging from each orifice
• Determination of the extrusion load: quite difficult since the extruded
sections are usually nonsymmetrical, resulting in a complex 3D metal
flow in the deformation zone
Some semiempirical analytical techniques have been developed for
computer-aided design of extrusion dies. However, these techniques are
approximate and could be vastly improved by developing FEM-based
computer programs for metal-flow simulation.

3.7 Cold Forging and Extrusion

Cold extrusion is a special type of forging process in which cold metal is
forced to flow plastically under compressive force into a variety of shapes.
These shapes are usually axisymmetric with relatively small nonsymmetri-
cal features, and, unlike impression die forging, the process does not
generate flash. The terms cold forging and cold extrusion are often used
interchangeably and refer to well-known forming operations such as
extrusion, upsetting or heading, coining, ironing, and swaging. Several
forming steps are used to produce a final part of relatively complex
geometry, starting with a slug or billet of simple shape, as shown in Fig.
3.10 [8].
In warm forging, the billet is heated to temperatures below the
recrystallization temperature, for example, up to 700-800°C (1292-1479°F)
for steels, in order to lower the flow stress and the forging pressures. In
cold forging, the billet or the slug is at room temperature when
deformation starts.
In cold forging, the tool stresses are quite high, in the order of
250-350 ksi (1724-2413 MN/m2). Consequently, the prediction of the
forming load and stresses is quite important for die design and machine
selection. In addition, the distribution of the strains in each deformation
stages is significant, since it determines the hardness distribution as well as
the formability of the part. Both these variables, strains and stresses, are
influenced by the following process parameters:
• Area reduction. The extrusion load increases with increasing reduction
in cross-sectional area because the strain increases with reduction.
• Die geometry (angle, radii). The die geometry directly influences
material flow, and therefore it affects the distribution of the effective
strain and flow stress in the deformation zone. In forward extrusion,
for a given reduction, a larger die angle increases the volume of metal
undergoing shear deformation and results in an increase in forming
load. On the other hand, the length of the die decreases, which results
in a decrease in die friction load. Consequently, for a given reduction
and given friction conditions there is an optimum die angle that
minimizes the extrusion load.
£

FIG. 3.10 Schematic illustration of forming sequences in cold forging of a gear blank [8]. Left to right: sheared blank, simultaneous forward rod and
backward cup extrusion, forward extrusion, backward cup extrusion, simultaneous upsetting of flange, and coining of shoulder.
 Analysis and Technology in Metal Forming 41

• Extrusion velocity. With increasing velocity, both the strain-rate and

the temperature generated in the deforming material increase. These
effects counteract each other, and consequently the load in cold
extrusion is not affected significantly by the extrusion velocity.
• Lubrication. Improved lubrication lowers the container friction force,
and the die friction force, resulting in lower extrusion loads.
• Workpiece material. The flow stress of the billet material directly
influences the loads necessary for homogeneous and shear deforma-
tions. The prior heat treatment and/or any prior work hardening also
affect the flow stress of a material. Therefore, flow stress values
depend not only on the chemical composition of the material but also
on its prior processing history.
• Billet dimensions. In forward extrusion, an increase in billet length
results in an increase in container friction load. In backward extru-
sion, the billet length has little effect on the extrusion load.
The quantitative influence of the process variables discussed above upon
formability, product properties, and die stresses can be best predicted by
using FEM-based codes. While there have been analyses of forging
operations by FEM, practical applications are still needed for improved
process and die design.

3.8 Rolling of Strip, Plate, and Shapes

Most engineering metals, such as aluminum alloys, copper alloys, and
steels, are first cast into ingots and are then further processed by hot
rolling into "blooms," "slabs" and "billets." These are known as semi-
finished products because they are subsequently rolled into other products
such as plate, sheet, tube, rod, bar, and structural shapes.
The primary objectives of the rolling process are to reduce the cross
section of the incoming material while improving its properties and to
obtain the desired section at the exit from the rolls. The process can be
carried out hot, warm, or cold, depending on the application and the
material involved. The literature on rolling technology, equipment, and
theory is very extensive because of the significance of the process
[9,10,11]. Many investigators prefer to divide rolling into cold and hot
rolling processes. However, from a fundamental point of view, it is more
appropriate to classify rolling processes on the bases of the complexity of
metal flow during the process and the geometry of the rolled product.
Thus, rolling of solid sections can be divided into the following categories,
as illustrated in Fig. 3.11.
1. Uniform reduction in thickness with no change in width. This is the
case in strip, sheet, or foil rolling where the deformation is only in
the directions of rolling and sheet thickness and there is no
deformation in the width direction (plane-strain deformation). This
type of metal flow exists when the width of the deformation zone is at
least about 10 times the length of that zone (Fig. 3.HA).
42 Metal Forming and the Finite-Element Method

FIG. 3.11 Four types of metal flow in rolling: (A) strip, (B) plate, (C) simple shape, (D)
complex shape (broken and solid lines illustrate the sections before and after deformation,
respectively).

2. Uniform reduction in thickness with an increase in width. This type of

deformation occurs in rolling of blooms, slabs, and thick plates. The
material is elongated in the rolling (longitudinal) direction, is spread
in the width (transverse) direction, and is compressed uniformly in
the thickness direction (Fig. 3.11B).
3. Moderately nonuniform reduction in cross section. In this case the
reduction in the thickness direction is not uniform. The metal is
elongated in the rolling direction, is spread in the width direction,
and is reduced nonuniformly in the thickness direction. Along the
width, metal flow occurs only toward the edges of the section.
Rolling of an oval section in rod rolling, or rolling of an airfoil
section (Fig. 3.11C), would be considered to be in this category.
4. Highly nonuniform reduction in cross section. In this type of
deformation, the reduction is highly nonuniform in the thickness
direction. A portion of the rolled section is reduced in thickness
while other portions may be extruded or increased in thickness (Fig.
3.11D). As a result, in the width (lateral) direction metal flow may be
toward the center. Of course, in addition, metal flows in the
thickness direction as well as in the rolling (longidutinal) direction.
The strip rolling process is schematically illustrated in Fig. 3.12. A very
large number of books and papers have been published on the subject of
strip rolling. The most rigorous analysis was performed by Orowan [12]
and has been applied and computerized by various investigators. More
recent studies consider elastic flattening of the rolls and temperature
conditions that exist in rolling [13,14]. The roll separating force and the
roll torque can be estimated with various levels of approximation by the
 Analysis and Technology in Metal Forming 43

FIG. 3.12 Schematic representation of strip rolling (strip has constant width).

slab method, the upper-bound method, or the slip-line method of analysis.

Most recently, computerized numerical techniques have been used to
estimate metal flow, stresses, roll separating force, temperatures, and
elastic deflection of the rolls.
As seen in Fig. 3.12, because of volume constancy, the velocity of the
deforming material in the x, or rolling, direction must steadily increase
from entrance to exit. At only one point along the roll-strip interface is
the surface velocity of the roll equal to the velocity of the strip. This point
is called the neutral point or neutral plane, indicated by N in Fig. 3.12.
The interface frictional stresses are directed from the entrance and exit
planes toward the neutral plane because the relative velocity between the
roll surface and the strip changes its direction at the neutral plane. This
influences the distribution of the rolling stresses.
In the rolling of thick plates, metal flow occurs in three dimensions, as
can be seen in Fig. 3.11B. The rolled material is elongated in the rolling
direction as well as spread in the lateral, or width, direction. Spread in
rolling is usually defined as the increase in width of a plate or slab as a
percentage of its original width. The spread increases with increasing
reduction and interface friction, decreasing plate width-to-thickness ratio,
and increasing roll-diameter-to-thickness ratio. In addition, the free edges
tend to bulge with increasing reduction and interface friction. The
three-dimensional metal flow that occurs in plate rolling is difficult to
analyze. Therefore, most studies of this process have been experimental in
44 Metal Forming and the Finite-Element Method

FIG. 3.13 Schematic illustration of five different roll-pass designs for a steel angle section
[18].

nature, and several empirical formulae have been established for estimat-
ing spread [15]. Attempts were also made to predict elongation and spread
theoretically [16, 17].
Rolling of shapes, also called caliber rolling, is one of the most complex
deformation processes (Figs. 3.11C and D). A round or round-cornered
square slab is rolled in several passes into (a) relatively simple sections
such as rounds, squares, or rectangles, or (b) complex sections such as
L,U,T,H, or other irregular shapes. For this purpose, certain intermediate
shapes or passes are used, as shown in Fig. 3.13 for rolling of angle
sections [18]. The design of these intermediate shapes, i.e. roll pass design,
is based on experience and differs from one company to another, even for
the same final rolled section geometry. Relatively few quantitative data on
roll pass design are available in the literature.
At the present stage of technology, the process variables are considered
in roll pass design by using a combination of empirical knowledge, some
calculations, and some educated guesses. A methodical way of designing
roll passes requires not only an estimate of the average elongation, but
also the variation of this elongation, within the deformation zone. The
deformation zone is limited by the entrance, where a prerolled shape
enters the rolls, and by the exit where the rolled shape leaves the rolls.
The deformation zone is cross-sectioned with several planes. The roll
position and the deformation of the incoming billet are investigated at each
of these planes. Thus, a more detailed analysis of metal flow and an
improved method for designing the configuration of the rolls are possible.
It is evident that this technique can be drastically improved and made
extremely efficient by use of computer-aided techniques and, ultimately,
by FEM.
In recent years, most companies that produce shapes have computerized
their roll pass design procedures for rolling rounds [19,20] or structural
shapes [21, 22]. In most of these applications, the elongation per pass and
the distribution of the elongation within the deformation zone for each
pass are predicted by using an empirical formula. If the elongation per pass
 Analysis and Technology in Metal Forming 45

is known, it is then possible, by use of computer graphics, to calculate the

cross-sectional area of a section for a given pass, i.e., reduction and roll
geometry. The roll geometry can be expressed parametrically, i.e., in
terms of angles, radii, etc. These geometric parameters can then be varied
to "optimize" the area reduction per pass and to obtain an acceptable
degree of "fill" of the roll caliber used for that pass.
The above discussion illustrates that, except in strip rolling, the metal
flow in rolling is in three dimensions, i.e., in the thickness, width, and
rolling directions. Determinations of metal flow and rolling stresses in 3D
rolling, i.e., shape rolling, are very important in designing rolling mills and
in setting up efficient production operations. However, the theoretical
prediction of metal flow in such complex cases is nearly impossible at this
time. Numerical techniques are being developed in an attempt to simulate
metal flow in such complex rolling operations. The FEM-based computer
codes offer excellent potential for predicting metal flow and stresses in 3D
rolling operations.

3.9 Drawing of Rod, Wire, Shapes, and Tubes

Drawing is one of the oldest metal-forming operations and has major
industrial significance. This process allows excellent surface finishes and
closely controlled dimensions to be obtained in long products that have
constant cross sections. In drawing, a previously rolled, extruded, or
fabricated product with a solid or hollow cross section is pulled through a
die at a relatively high speed. In drawing of steel or aluminum wire, for
example, exit speeds of several thousand feet per minute are very
common. The die geometry determines the final dimensions, the cross-
sectional area of the drawn product, and the reduction in area. Drawing is
usually conducted at room temperature using a number of passes or
reductions through consecutively located dies. At times, annealing may be
necessary after a number of drawing passes before the drawing operation is
continued. The deformation is accomplished by a combination of tensile
and compressive stresses that are created by the pulling force at the exit
from the die, by the back-pull tensile force that is present between
consecutive passes, and by the die configuration.
In wire or rod drawing, the section is usually round but could also be
shaped. In cold drawing of shapes, the basic contour of the incoming shape
is established by cold rolling passes that are usually preceded by annealing.
After rolling, the section shape is refined and reduced to close tolerances
by cold drawing, as shown in Fig. 3.14 [23]. Here again, a number of
annealing steps may be necessary to eliminate the effects of strain
hardening, i.e., to reduce the flow stress and increase the ductility.
In tube drawing without a mandrel, also called tube sinking, the tube is
initially pointed to facilitate feeding through the die; it is then reduced in
outside diameter while the wall thickness and the tube length are
increased. The magnitudes of thickness increase and tube elongation
46 Metal Forming and the Finite-Element Method

FIG. 3.14 Cold rolled (round to triangle) and cold drawn shape, requiring numerous
annealing steps [23].

depend on the flow stress of the drawn part, die geometry, and interface
friction.
Drawing with a fixed plug is widely known and used for drawing of
large-to-medium-diameter straight tubes. The plug, when pushed into the
deformation zone, is pulled forward by the frictional force created by the
sliding movement of the deforming tube. Thus, it is neccesary to hold the
plug in the correct position with a plug bar. In drawing of long and
small-diameter tubes, the plug bar may stretch and even break. In such
cases it is advantageous to use a floating plug. This process can be used to
draw any length of tubing by coiling the drawn tube at high speeds of up to
2000 ft/min. In drawing with a moving mandrel, the mandrel travels at the
speed at which the section exits from the die. This process, also called
ironing, is widely used for thinning of the walls of drawn cups or shells, for
example, in the production of beverage cans [24] or artillery shells [5].
The principle of a can ironing press is illustrated in Fig. 3.15. In the
figure, the press is horizontal, and the ram has a relatively long stroke and
is guided by the hydrostatic bushing (A). The front seal (B) prevents
mixing of the ironing lubricant with the hydrostatic bushing oil. With the
ram in the retracted position, the drawn cup is automatically fed into the
press, between the redraw die (D) and the redraw sleeve (C). The redraw
sleeve centers the cup for drawing and applies controlled pressure while

FIG. 3.15 Schematic illustration of multiple-die ironing operation for manufacturing bever-
age cans [24].
 Analysis and Technology in Metal Forming 47
the cup is drawn through the first die (D). As the ram proceeds, the redrawn
cup is ironed by passing through the carbide dies (E), which gradually
reduce the wall thickness. The ironed can is pressed against the doming
punch (T), which forms the bottom shape of the can. When the ram starts
its return motion, the mechanical stripper (G), assisted by the air stripper
(F), removes the can from the ironing punch (H). The punch is made of
carbide or cold-forging tool steel. The stripped can is automatically
transported to the next machine for trimming of the top edge of the can
wall to a uniform height.

3.10 Sheet-Metal Forming

The products made by sheet-metal forming processes include a large
variety of shapes and sizes, ranging from simple bends to double
curvatures with shallow or deep recesses. Thus the scope of sheet-metal
forming is very broad. Sachs [25] described systematically the principles of
the numerous sheet-metal forming methods, emphasizing their similarities
and differences. Some important processes are described and significant
variables are discussed in the References [26-28].
Yoshida [29] proposed a classification of general press-forming processes
based on the governing deformation mechanisms. The basic mechanisms
are stretching, drawing, and bending. Depending upon the shape and the
relative dimensions of the blank and the tool, one or more basic
mechanisms is predominantly involved in sheet metal deformation.
The limits of sheet-metal forming are determined by the occurrence of
defects, such as wrinkles and ruptures in the blank. Some defects observed
in cup drawing are shown by the sketches in Fig. 3.16 [28]. The limiting
drawing ratio, which is the ratio of the maximum possible blank dimension
to the dimension in a drawn part, is a measure of the limit in the drawing
range. The occurrence of wrinkles further restricts the drawing range.
Wrinkling may form either in the flange over the die surface or in the
blank around the die shoulder. Wrinkling over the flat face of the die can
be avoided by applying the blank holding force, and the possibility of
wrinkling of the blank around the die shoulder can be reduced by taking
the radius of the die corner large enough compared with the blank
thickness.
In stretching of sheet metals over the punch, the rupture of the blank
occurs over the punch profile. Necking precedes eventual rupture.
Therefore, the forming limit is governed by the condition of instability,
and the site of necking initiation depends on the friction condition at the
punch-workpiece interface. In flange stretching (or hole expansion),
fracture occurs at the edge of the flange. Necking, however, does not
always initiate from the periphery of the flange but, for some materials,
starts at some distance from the edge. The stress and strain fields in the
above stretching processes are nonuniform, and the analysis for the
instability condition must take this nonuniformity into account.
48 Metal Forming and the Finite-Element Method

WRINKLES

FRACTURE

EARING
FIG. 3.16 Some defects observed in cup drawing [28].

Bending operations are involved in all complex stamping. In bending, in

contrast to most stretching operations, there is a severe gradient of stress
throughout the thickness of the material. On the outside of the bend the
stress is tension and on the inside it is either compression or a reduced
level of tension. The severity of the tensile strains depends on the radius,
angle, and length of the bend. Fracture occurs on the tensile side by
thinning and fracture.
Formability of sheet metals, as the measure of the ability of the metal to
undergo the desired shape change without failure, is often evaluated by
simple tests such as a tensile test. The parameters obtained by simple
tension, namely, degree of anisotropy, the workhardening rate of stress-
strain relationship, the maximum uniform elongation, and the maximum
nominal stress, are related to formability. However it is not possible to
evaluate accurately the formability of the materials in terms of these
parameters. Thus, for complete assessment of the formability, the direct
methods, such as the Ericksen test, Swift cup test, and Fukui conical cup
test [30] have been used for determination of formability.
An important development in representation of the formability of sheet
metals is the forming limit diagram [31]. In this diagram the major and
minor surface strains at a critical site are plotted at the onset of visible,
localized necking in a deformed sheet, and the locus of strain combinations
 Minor Strain (%)

FIG. 3.17 Forming limit diagram and deformed rectangular strip and circular blank with
circular cutoff [34].

49
50 Metal Forming and the Finite-Element Method
that will produce failures in an actual forming operation can be drawn.
Figure 3.17 shows the forming limit diagram for different materials.
Experimental methods are used to construct the diagram. The Nakajima
method [32] uses a hemispherical punch and rectangular blanks with
various widths and lubrication conditions. Analogously to the Nakajima
method, Hasek [33] used hemispherical punch stretching of circular blanks
with various circular cutoffs. The method of Hasek has an advantage of
eliminating or reducing the wrinkling phenomena that occur at the edges
of rectangular strips in the Nakajima method. In Fig. 3.17, deformed
specimens (FEM simulation [34]) in the above two methods are also
shown.
Friction at the punch-workpiece interface is an important factor for
formability, and it changes the strain-path of a critical site of the sheet in
the forming limit diagram of Fig. 3.17. Material parameters, such as
strain-hardening and strain-rate sensitivity, are important for formability.
Another important factor in sheet-metal forming is the anisotropy
(direction-dependent properties) of the sheet metal. Earing in deep
drawing of cups is due to anisotropy (see Fig. 3.16). Anisotropy also
influences formability. For example, the limiting drawing ratio in cup
drawing increases with increasing r-value, which is the ratio of the width
strain to the thickness strain in uniaxial tension and is a measure of
anisotropy in the thickness direction [26].
In sheet-metal forming, considerations of residual stresses and spring-
back are particularly important. Consequently elastic properties of sheet
metals cannot be neglected in the analysis of problems in which these
considerations are of major concern.
Although simple analytical technique considers only the stress com-
ponents in the plane of the sheet (plane-stress situation), complexity of
product shapes and deformation in forming make it difficult for research in
analytical methods to advance technology that would be useful in
metal-stamping plants [35].
Among many areas of research need, die design and process design are
of particular technological importance and two specific problems, namely,
design of drawbeads in stamping operations and design of multistage
forming, are given as examples.
In a die-formed part the presence of sloping walls that are not confined
between the die and the punch introduces the danger of buckling or
wrinkling. Generally, this tendency to wrinkle increases with increasing
unsupported area and with decreasing metal thickness. To avoid wrinkling,
large parts made from thin metal must be formed with a considerably
higher hold-down force. In addition, beads are frequently added to the
mating die and hold-down surfaces, either all around or only in the areas
where wrinkling would otherwise develop. A drawbead is shown in Fig.
3.18 in the drawing of a box-shaped part [25]. Beads retard and control the
flow of sheet metal into the die cavity. Design of drawbeads includes the
determination of their orientation, configuration, and location.
 Analysis and Technology in Metal Forming 51

FIG. 3.18 Draw beads in drawing a box-shaped part [25].

In most sheet-metal parts, the shape is formed almost entirely in the first
operation. Subsequent processes are generally trimming, piercing, flang-
ing, and some minor restriking of detail. There are instances, however, in
which the formability of the metal is such that it is impossible to reach the
final form in one operation, and a number of intermediate shapes are
required. An example is shown in Fig. 3.19 for forming an automobile

FIG. 3.19 Multistage forming processes for an automobile wheel center [36],
52 Metal Forming and the Finite-Element Method

wheel center [36]. The problem is to determine the number of preforming

operations and the configuration for each preform.
The two examples of design in sheet-metal forming described present a
challenge to the power of numerical analysis to make a significant
contribution to the advancement of technology.

References
1. Lange, K., (editor), (1972), "Study Book of Forming Technology," (in
German), Vol. I, II and III, Springer-Verlag, Berlin.
2. Johnson, W., and Sowerby, R., (1978), "Metal Forming Processes: Analysis
and Technology," ASME, AMD, Vol. 28, p. 1.
3. Allan, T., and Boulger, F. W., (1973), "Flow Stress of Metals and Its
Application in Metal Forming Analyses," Trans. ASME, J. Engr. Ind., Vol.
95, p. 1009.
4. Douglas, J. R., and Allan, T., (1975), "Flow Stress Determination for Melals
al Forging Rates and Temperalures," Trans. ASME, J. Engr. Ind., Vol. 97, p.
66.
5. Allan, T., Oh., S. I, and Gegel, H., (1983), "Melal Forming: Fundamenlals
and Applications," ASM International, Metals Park, OH.
6. Schey, J. A. (editor), (1983), "Metal Deformation Processes: Friction and
Lubrication," Marcel Dekker, New York, 1970; superseded by Schey, J. A.,
"Tribology in Metalworking: Lubrication, Friction and Wear," American
Sociely for Melals, Melals Park, OH.
7. Billhardt, C. F., Nagpal, V., and Allan, T., (1978), "A Computer Graphics
System for CAD/CAM of Aluminum Exlrusion Dies," SME Paper MS78-957.
8. Sagemuller, F., (1968), "Cold Impact Exlrusion of Large Formed Parts,"
Wire, No. 95, p. 2.
9. Dieter, G. E., (1961), "Mechanical Metallurgy," McGraw-Hill, New York,
Chapter 29, p. 488.
10. Geleji, A., (1967), "Forge Equipmenl, Rolling Mills and Accessories,"
Akademiai Kiado, Budapesl.
11. Larke, E. C., (1957), "The Rolling of Strip, Sheet and Plate," Chapman and
Hall, London.
12. Orowan, E., (1943), "The Calculation of Roll Pressure in Hoi and Cold Flal
Rolling," Proc. Inst. Mech. Eng., Vol. 150, p. 140.
13. Alexander, J. M., (1972), "On Ihe Theory of Rolling," Proc. R. Soc. London,
Series A, Vol. 326, p. 535.
14. Laholi, G. D., Shah, S. N., and Allan, T., (1978), "Computer Aided Analysis
of the Deformations and Temperalures in Strip Rolling," Trans. ASME, J.
Engr. Ind., Vol. 100, p. 159.
15. Ekelund, S., in H. Neumann, (1963), "Roll Pass Design," (in German), VEB
Deulscher Verlag, Leibzig, p. 48.
16. Oh, S. I., and Kobayashi, S., (1975), "An Approximate Melhod for
Three-Dimensional Analysis of Rolling," Int. J. Mech. Sci., Vol. 17, p. 293.
17. Neumann, H., (1969), "Design of Rolls in Shape Rolling," (in German), VEB
Deulscher Verlag, Leibzig.
18. Schulza, A., (1970), "Comparison of Roll Pass Designs Used for Rolling
Angle Sections," (in German), Stahl and Eisen, Vol. 90, p. 796.
19. Neumann, H., and Schulze, R., (1974), "Programmed Roll Pass Design for
Blocks," (in German), Neue Hutte, Vol. 19, p. 460.
20. Suppo, U., Izzo, A., and Diana, P., (1973), "Eleclronic Computer Used in
Roll Design Work for Rounds," Der Kalibreur, Vol. 19, p. 3.
 Analysis and Technology in Metal Forming 53

21. Metzdorf, J., (1981), "Computer Aided Roll Pass Design—Possibilities of

Application," (in German and French), Der Kalibreur, No. 34, p. 29.
22. Schmeling, F., (1982), "Computer Aided Roll Pass Design and Roll Manufac-
turing," (in German), Stahl und Eisen, Vol. 102, p. 771.
23. "Rathbone Cold Drawn Profile Shapes and Pinion Rods," Rathbone Corpora-
tion, Palmer, MA.
24. Brochure of the Standum Company, Compton, CA.
25. Sachs, G., (1951), "Principles and Methods of Sheet Metal Fabricating,"
Reinhold, New York.
26. Schey, J., (1987), "Introduction to Manufacturing Processes," 2nd edition,
McGraw-Hill, New York.
27. Kalpakjian, S., (1984), "Manufacturing Processes for Engineering Materials,"
Addison-Wesley, Reading, MA.
28. Johnson, W., and Mamalis, A. G., (1978), "Aspects of the Plasticity
Mechanics of Some Sheet Metal Forming Processes," Hellenic Steel publica-
tions, Thessaloniki, Greece.
29. Yoshida, K., (1959), "Classification and Systemization of Sheet Metal Press-
forming Process," Scientific Papers of the Institute of Physical and Chemical
Research, Vol. 53, No. 1514, p. 126.
30. Fukui, S., Kudo, H., Yoshida, K., and Okawa, H., (1952), "A Method for
Testing of Deep-Drawability of Sheet Metals," Report of the Institute of
Science and Technology, University of Tokyo, Vol. 6, p. 351.
31. Keeler, S. P., and Backofen, W. A., (1963), "Plastic Instability and Fracture
in Sheets Stretched Over Rigid Punches," Trans. Am. Soc. Metals, Vol. 56, p.
25.
32. Nakajima, K., Kikuma, T., and Hasuka, K., (1968), "Study on the For-
mability of Steel Sheets," Yawata Technical Report, No. 264, p. 141.
33. Hasek, V., (1978), "Untersuchung und Theoretische Beschreibung wichtiger
Einflussgrossen auf das Grenzformanderungsschaubild," Institute of Metal
Forming Report, University of Stuttgart, West Germany, p. 213.
34. Toh, C. H., (1983), "Process Modeling of Sheet Metal Forming of General
Shapes by the Finite Element Method based on Large Strain Formulations,"
Ph.D. dissertation, Department of Mechanical Engineering, University of
California at Berkeley.
35. Koistinen, D. P., and Wang, N-M, (editors), (1978), "Mechanics of Sheet
Metal Forming," Plenum Press, New York and London.
36. Duncan, J. L., and Sowerby, R., (1987), "Review of Practical Modeling
Methods for Sheet Metal Forming," Proc. 2d Int. Conf. Tech. Plasticity,
Stuttgart, p. 615.
 4
PLASTICITY AND VISCOPLASTICITY

4.1 Introduction
The theory of plasticity describes the mechanics of deformation in
plastically deforming solids, and, as applied to metals and alloys, it is
based on experimental studies of the relations between stresses and strains
under simple loading conditions. The theory described here assumes the
ideal plastic body for which the Bauschinger effect and size effects are
neglected. The theory also is valid only at temperatures for which
recovery, creep, and thermal phenomena can be neglected. The basic
theory of classical plasticity is described by Hill [1], and also in References
[2-5], in addition to the books listed in Chap. 1. A concise description of
the general plasticity theory necessary for metal forming is given in the
book by Johnson et al. [6]. In this chapter, certain important aspects of the
theory are presented in order to elucidate the developments of the
finite-element solutions of metal-forming problems discussed in this book.
First, various measures of stress and strain are introduced. Then, the
governing equations for plastic deformation and principles that are the
foundations for the analysis are described. The extension of the theory of
plasticity to time-dependent theory of viscoplasticity is outlined in Section
4.8. Particular references are made, in Sections 4.3 through 4.7, to the
books by Hill [1] and by Johnson and Mellor [7], and to the section on
general plasticity theory in the book by Johnson et al. [6].

4.2 Stress, Strain, and Strain-Rate

The basic quantities that may be used to describe the mechanics of
deformation when a body deforms from one configuration to another
under an external load are the stress, strain, and strain-rate. Various
measures of these quantities are defined, depending upon how closely
formulations represent actual situations. Although it is not possible to
provide the complete mathematical formulations in one-dimensional de-
formation, these measures are introduced for the case of simple uniaxial
tension.
Consider the uniaxial tension test of a round specimen whose initial
length is /0 and cross-sectional area is A0. The specimen is stretched in the
54
 Plasticity and Viscoplasticity 55

FIG. 4.1 Uniaxial tension, (a) Tension specimen; (b) stress-strain curves.

axial direction by the force P to the length / and the cross-sectional area A
at time t, as shown in Fig. 4.1. The response of the material is recorded as
the load-displacement curve, and converted to the stress-strain curve as
shown in the figure. The deformation is assumed to be homogeneous until
necking begins.
There are two modes of description of the deformation of a continuous
medium, the Lagrangian and the Eulerian. The Lagrangian description
employs the coordinates Xt of a typical particle in the reference (or
undeformed) state as the independent variables, while in the case of
Eulerian description the independent variables are the coordinates xt of a
material point in the deformed state. When the deformation is in-
finitesimal, where products of the derivatives of the displacements can be
neglected, we need make no distinction between the two.
In the infinitesimal deformation theory, the stresses and strain-rates (or
infinitesimal strains) are expressed with respect to a fixed coordinate
system in the material configuration at a time under consideration. In
56 Metal Forming and the Finite-Element Method
uniaxial tension, they are defined by

stress

straain-rate

infinitesimal strain ds

where the dot denotes the time derivative. The stress defined in eq. (4.1) is
called true stress or Cauchy stress. The total amount of deformation is
measured by integrating infinitesimal strain as

and is called logarithmic or natural strain.

In the Lagrangian description of finite deformation, the measures of
stress, strain, and strain-rate are expressed as follows.
Let the position of a particle in the deformed configuration at time t be
designated by

where A' is a reference position of a particle and t is the time. In uniaxial

tension let X be directed along the longitudinal axis of the specimen (see
Fig. 4.1); then

Extension is defined as the displacement gradient relative to the reference

position and is expressed by

This is the engineering strain.

The Lagrangian strain component Eu is defined by

(see Reference [8] for geometrical interpretation of eq. (4.6)). The

strain-rate components are the time derivatives of strain components given
by eqs. (4.5) and (4.6). They are

and
 Plasticity and Viscoplasticity 57

since

The Piola-Kirchhoff stress tensor is defined as force intensity acting in

the deformed configuration but measured per unit area of the reference
configuration, while the Cauchy stress is denned as force per unit area in
the deformed state. In the uniaxial tension, the engineering stress

corresponds to a component of the nonsymmetric (or first) Piola-Kirchhoff

stress tensor. Correspondence between stress and strain-rate measures is
established such that the rate of work per unit volume (W0) in the
reference configuration is the product of stress and strain rate. From eqs.
(4.7) and (4.8),

Then the stress measure corresponding to the Lagrangian strain rate En is

The stress given by eq. (4.10) corresponds to a component of the

symmetric (or second) Piola-Kirchhoff stress tensor. (See References
[9,10] for general background on finite deformation.)
For the analysis of metal forming processes, flow formulation is based on
infinitesimal deformation theory, while solid formulation considers finite
deformation. Consideration of finite deformation with flow formulation is
given in Chap. 11 in connection with sheet-metal forming, and solid
formulation is outlined in Chap. 16, where comparison of solutions
according to the two formulations are discussed. Since flow formulation is
used mostly throughout this book, further detail on stress and strain-rate
measures in the infinitesimal deformation theory is given.
The strain-rate tensor [etj], where i,j=x,y,z, is symmetric and the
tensor components are defined by
58 Metal Forming and the Finite-Element Method

where ut are velocity components and y,7 are engineering shear strain-rate
components. Using suffix notation, eq. (4.11) can be written as

where a comma denotes differentiation with respect to the coordinates that

follow.
The Cauchy stress tensor [ofi], where i, j = 1,2,3 or x, y, z, is also
symmetric and is represented by the nine components as

The stress may also be specified by the three principal components, or by

the three tensor invariants. The principal stresses (oly o2, cr3) are the roots
of the cubic equation

where Ilt I2, and 73 are quantities independent of the direction of the axes
chosen and called the three invariants of the stress tensor oti. They are
defined by the relations

The first (linear) and second (quadratic) invariants have particular physical
significance for the theory of plasticity.

4.3 The Yield Criteria

A yield criterion is a law defining the limit of elasticity under any possible
combination of stresses. It is expressed by

A function of stresses /(<?,y) is called yield function. The suitability of any

proposed yield criterion must be verified by experiment.
For isotropic materials, plastic yielding can depend only on the
magnitude of three principal stresses and not on their directions. Then any
yield criterion is expressible in the form

From the experimental fact that the yielding of a material is, to a first
approximation, unaffected by a moderate hydrostatic pressure, or tension,
it follows that yielding depends only on the principal components
 Plasticity and Viscoplasticity 59

(oi, o2', cr3') of the deviatoric stress tensor

where am = \(a\ + o2 + o3) is the hydrostatic component of the stress and

dij ( = 1 for i=j and = 0 for i¥=j) is the Kronecker delta. The principal
components of the deviatoric stress tensor are not independent, since
ai + a2 + <73' is identically zero.
The yield criterion then reduces to the form

where

Two Criteria
Two simple criteria have been in extensive use for the analysis of metal
deformation. Tresca's criterion (shear stress criterion) given in 1864 is

with o1 >CT2> o3. It may alternatively be written in the form of eq. (4.19)
in terms of J2 and J3, but the results are complicated and not useful.
Another criterion was proposed by Heuber (1904), by von Mises (1913),
and by J. C. Maxwell in a letter to Kelvin in 1856 [7]. It has been
traditionally called the von Mises criterion or the Huber-Mises criterion,
but may be appropriately called the Maxwell-Heuber-Mises criterion.
(We refer to the distortion energy criterion according to a physical
interpretation of the criterion suggested by Hencky in 1924.) The criterion
states that yielding occurs when J2 reaches a critical value, or, in other
words, that the yield function/of eq. (4.19) does not involve J3. It can be
written in the alternative forms

or

or

where A; is a parameter regulating the stress scale and depending on the

material property. In the suffix notation used in eq. (4.21a), a recurring
letter suffix indicates that the sum must be formed of all terms obtainable
by assigning to the suffix the values x, y, z (or 1, 2, 3).
The constant in eqs. (4.20) and (4.21) may be determined from simple
states, such as in uniaxial tension. At yielding in simple tension, cr, = Y
and 02 = o3 = 0. Therefore, eqs. (4.20) and (4.21) may be written as
60 Metal Forming and the Finite-Element Method
and

respectively. Parameter k in eq. (4.21) can be identified as the shear yield

stress by considering yielding in simple shear and k = Y/\/3 according to
the criterion (4.21).
It must be noted that the yield criteria defined by eqs. (4.22) must
depend on the previous process of plastic deformation (strain hardening).
If we assume that hardening occurs if, and only if, plastic work is done,
then the assumption that the yield criterion is independent of the
hydrostatic component implies that there is no volume change during
plastic deformation.
Geometrical Representation
A state of stress is completely specified by the values of the three principal
components. Then any stress state may be represented by a vector in a
three-dimensional stress space, where_the principal stresses are taken as
Cartesian coordinates. In Fig. 4.2, OS is the vector (olt o2, o3) and its
component, OP, is the vector representing the deviatoric stress
(oi, o2', <V). OP always lies in the plane n whose equation is ol + o2 +
a3 = 0. The_hydrostatic component (om, om, am) of the stress is repre-
sented by PS, which is perpendicular to the plane n.
A yield criterion, which is independent of the hydrostatic component of
stress, is represented by a curve C in the plane Jt. The yield locus
corresponding to the shear stress criterion is a regular hexagon, while it is
obvious from the relation in eq. (4.21a) that the locus of the distortion
energy criterion is a circle of radius \/2 k or V2/3 Y. By selecting the
values of the constant in eq. (4.20) and k in eq. (4.21), according to eq.

FIG. 4.2 Geometrical representation of a plastic state of stress in (tr1; a2, a3) space.
 Plasticity and Viscoplasticity 61

FIG. 4.3 Yield locii on the Jt-plane for distortion energy criterion and maximum shear stress
criterion.

(4.22), the two criteria can be made to agree with each other and with
experiment for uniaxial tension. The two loci are shown in Fig. 4.3.
For most metals the distortion energy criterion fits the data more closely
than the shear stress criterion. Furthermore, the distortion energy criterion
is continuous and convenient to use in numerical analysis. Therefore, the
distortion energy criterion is exclusively used in this book. Other yield
criteria, such as those for anisotropic sheet materials and for porous
metals, are introduced in Chaps. 11 and 13, respectively.

4.4 Equilibrium and Virtual Work-Rate Principle

Equilibrium Equations
In the rectangular Cartesian coordinate system, the equilibrium equations,
if the body force is neglected, are given by

With suffix notation, eqs. (4.23) are written as

In the notation used in eq. (4.24), a recurring letter suffix indicates the
sum, as explained for eq. (4.21a), and a comma denotes partial
differentiation as explained for eq. (4.12).
62 Metal Forming and the Finite-Element Method
Equilibrium with Tractions
The stress along the boundary surface S is in equilibrium with an applied
traction Ft (force per unit surface area). Equilibrium of the stresses is
written as

where «, is the unit outward normal to the surface. Writing eq. (4.25a), in
unabridged notation in the two-dimensional case (see Fig. 4.4), one has

where the components of the unit outward normal ny are given by

(dyldl, dxldl}.
Virtual Work-rate Principle
The virtual work-rate principle states that for the stress field that is in
equilibrium within the body and with applied surface tractions, the
work-rate inside the deforming body equals the work-rate done by the
surface tractions for all velocity fields that are continuous and continuously
differentiable (virtual velocity fields).
Let Ofj be any stress field that is in equilibrium and vv, be any virtual
velocity field. Then the principle is expressed by

where V is the volume of the body and S is the surface. Since otj is

FIG. 4.4 Equilibrium of surface tractions.

 Plasticity and Viscoplasticity 63

symmetric, eq. (4.26a) is also written as

where e,y is the strain-rate derivable from vv, according to eiy = 2(wy + wjti).
For proof, note that

and

For the two-dimensional case, eq. (4.26b) can be written in unabridged

form as

4.5 Plastic Potential and Flow Rule

Hooke's law is well known for describing the relationships between
stresses and corresponding deformation in the elastic deformation regime.
When deformation extends to the plastic range, the stress and plastic strain
relationships are derived using the concept of plastic potential.
Plastic Potential and Flow Rule
The ratios of the components of the plastic strain-rate e/ (or infinitesimal
plastic strain de^p) are defined by

or

where g and h are scalar functions of the invariants of deviatoric stresses

and / is the yield function (if / = 0, neutral loading, and / < 0, unloading).
The function g(a,y) is called the plastic potential. Although eq. (4.27a) is
written in the rate form, the relations between stress and strain are
independent of time.
64 Metal Forming and the Finite-Element Method

Assuming a simple relation g =/, eq. (4.27) becomes

where A or dX is a positive proportionality constant, being equal to hf or

h df. Equation (4.28) is a flow rule associated with the yield function /(o,-,-).
In differentiation of /(a,y) with respect to shear stress components, in
unabridged form, note that, for example,

The plastic strain-rate can be represented in the same (al, o2, o3) space by a
free vector G(e1p, e2p, £3P), where the factor G is introduced to obtain the
dimension of stress. This vector lies in the jr-plane since k-f + s2p + £3P =
0 (no volume change). Because df/do^ are the direction cosines of the
outward normal to the yield surface /(%) = C, the plastic strain-rate
vector is normal to the yield locus at point (cr/, o2'', o3) in the n plane.
The plastic strain-rate vector associated with the yield locus of the
distortion energy criterion is shown by PQ in Fig. 4.5. The concept of
plastic potential and its associated flow rule also apply to the shear stress
criterion. It will not be discussed here, however, since the shear stress
criterion is not used in this book, as stated in Section 4.3. It should be
noted that the yield locus must be convex (concave to the origin) at all
points if the stress corresponding to a given strain-rate is to be unique.
Maximum Plastic Work Principle
The maximum plastic work principle follows from the flow rule and the
convexity of the yield locus, and is

where otj is a yield state of stress, ei/> is the associated strain-rate, and a,,*

FIG. 4.5 Normality of the strain-rate vector to the yield locus.

 Plasticity and Viscoplasticity 65

FIG. 4.6 Two states of stress and a strain-rate vector.

is any other stress state represented by a point either on or inside the yield
surrace.
Since £,yp is parallel to the outward normal to the yield locus at the point
Oi/, (0i/ ~ Oij*)£ijp is proportional to the scalar product of the outward
normal to the yield locus at the point a// with the chord joining a^*' to o^'
(see Fig. 4.6). Therefore, eq. (4.29) holds and the equality sign applies
when <T,/ = ai*' or cr,-, and ot* differ only by a uniform hydrostatic stress.
Drucker [11] arrived at eq. (4.29) from a definition of a stable plastic
material.
The Prandtl-Reuss Equation and the Levy-Mises Equation
With the yield function given by f{otj) =J2 = 2<V<V> (eq. 4.21a), eq.
(4.28) becomes, in the rate form,

since

noting that the repeated subscripts k and / indicate summation with respect
to these quantities.
Equation (4.30) can be written as

Combining the elastic strain-rate components e^ and the plastic strain-rate

components eff, according to e,y = e,/ + e^, we obtain the Prandtl-Reuss
equations for elastic-plastic solids, as
66 Metal Forming and the Finite-Element Method

where G, E, and v are shear modulus, Young's modulus, and Poisson's

ratio, respectively.
For rigid-plastic materials, the assumption is that e;; = E/, and Levy-
Mises equations are obtained by removing the superscripts/) in eq. (4.31).
They are expressed in terms of stress components oif by three equations of
the type

and three of the type

The Levy-Mises equations (4.33) will be used extensively throughout the

book. In the following sections and in the chapters to follow, the plastic
strain-rate and the infinitesimal plastic strain are indicated without
superscript p, unless otherwise specified.

4.6 Strain-Hardening, Effective Stress, and Effective Strain

When a real metal is deformed at room temperature, its resistance to
further deformation increases (strain hardening). A hypothesis that the
degree of hardening is a function of the plastic work has been mentioned
in Section 4.3. For a mathematical formulation of srain hardening, it is
assumed that the final yield locus is the same, no matter by what
strain-path a given stress state is reached (isotropic hardening).
The total plastic work per unit volume during a certain finite deforma-
tion is

where the integral is taken over the actual strain-path.

The hypothesis that the radius of the yield locus of the distortion energy
criterion is a function only of Wn may be written as

where a is written for Y or \/3 k, and known as the flow stress, the
effective stress, generalized stress, or equivalent stress.
Another hypothesis for strain-hardening relates a to a certain measure
of the total plastic deformation. A quantity de, known as the effective,
generalized, or equivalent infinitesimal plastic strain is defined according to

The effective strain e = J de, integrated over the strain-path, provides a

measure of the plastic distortion. It is assumed that strain-hardening
characteristics can be formulated by
 Plasticity and Viscoplasticity 67

where H is a certain function depending on the metal concerned.

Equation (4.37) is most frequently used for process analysis in metal
forming. Methods of experimental determination of stress-strain pro-
perties are discussed in Chap. 3 and are also described in the book by
Johnson and Mellor [7].
It should be noted that an explicit expression for the effective strain can
be obtained when the principal axes of successive strain-increments do not
rotate relative to the element, and, further, when the components of any
strain-increments bear constant ratios to one another. It should also be
noted that, although quanties J d£t can always be formed (de, =
infinitesimal principal strain), even where the principal axes rotate, they
cannot generally be evaluated explicitly, nor do they possess any geometri-
cal significance. With the measures of o and de (and therefore e, effective
strain-rate), the proportionality factor A in the Levy-Mises equations
(4.33) can be expressed by

(4.38)

The derivation of eq. (4.38) is as follows. The plastic work-rate per unit
volume, using the flow rule (4.28), is written as

Because the yield function ijf{a) is a homogeneous function of degree 2,

namely, /(to,y) = ^/(cr/,), and using Euler's theorem for homogeneous
functions,i.e.,if is a homogeneous function of degree n, then
the plastic work rate becomes

from which eq. (4.38) results for the yield function /(a,y) of eq. (4.21a),
and a denned by eq. (4.35). The expression for e in terms of strain-rate
components, corresponding to the effective stress o of eq. (4.35), can be
obtained by considering the inversion of the flow rule. Then

from which

In unabridged form, eq. (4.39a) is

68 Metal Forming and the Finite-Element Method
The numerical factors in the definition of a and e have been chosen so that
the functional relationship between a and J df. is identical with the
relationship between true stress and logarithmic strain in a tension or
compression test.

4.7 Extremum Principles

By analogy with extremum principles in the theory of elasticity, two
related extremum principles pertaining to a consistent state are proven for
a rigid-plastic material that undergoes plastic deformation under pre-
scribed surface traction over the surface SF and prescribed velocities over
the surface Su. A complete solution to this problem consists of an
equilibrium stress field and an associated velocity field satisfying the
boundary conditions, and the stress field satisfies the yield criterion where
deformation occurs and does not violate the yield criterion in the rigid
regions. The uniqueness theorem proves that the stress field of the
complete solution is uniquely determined in the deforming region. Where
the deformation mode is not uniquely defined, it must be compatible with
the stress field.
The First Extremum Principle
Let Ojj, Uj denote a complete solution to the plastic deformation problem
of a rigid-plastic solid under the prescribed boundary conditions. Let a,y*
be an equilibrium stress field satisfying the stress boundary conditions on
SF and nowhere violating the yield criterion (statically admissible). The
principle of virtual work-rate gives

Then

according to the maximum plastic work principle. Since Fj = F* on SF,

where S = SF + Su,

Equation (4.42) is the first extremum principle and it states that a lower
bound to the rate of work of the actual surface traction can be obtained
from a statically admissible stress field.
In the quasi-static flow of a rigid plastic solid, velocity discontinuity is
permissible. When the velocity field u{ contains surfaces of discontinuity,
the principle of virtual work-rate gives, instead of eq. (4.40),
 Plasticity and Viscoplasticity 69

FIG. 4.7 Conditions along a surface of velocity discontinuity.

where SD is a surface of velocity discontinuity, T* is the shear stress

component of oif*, and k is the shear yield stress along the surface of SD,
where |A«| is the amount of tangential velocity discontinuity across SD.
Equation (4.43) is obtained by applying the virtual work-rate principle
to parts (1) and (2) in Fig. 4.7 separately, and then adding with the
condition that the traction to part (1), Fj'\ and to part (2), Fp', along the
surface SD are in equilibrium. Because the second term on the left-hand
side of eq. (4.43) is always positive, eq. (4.41) and therefore eq. (4.42) also
hold for this case.
It should be mentioned that a surface of stress discontinuity in an
equilibrium state of stress is also permissible, but that its presence does not
affect the relationships described above. Further, note that a surface of
stress discontinuity cannot be the surface of velocity discontinuity in the
associated stress and velocity fields, because physical interpretations are
the limits of the elastic region and large plastic deformation regions for
stress and velocity discontinuities, respectively.
The Second Extremum Principle
In the second extremum principle, let «,•* be a velocity field that satisfies
the incompressibility and the velocity boundary conditions on Su. Such a
velocity field is said to be kinematically admissible. The principle of virtual
work-rate gives

where i is the shear stress component of the stress a{j of the complete
solution along the velocity discontinuity surface 5D* with the amount of
tangential discontinuity |Aw*| in the velocity field «,*. If u* is the actual
70 Metal Forming and the Finite-Element Method

solution, then T = k. Thus

and also

from the maximum plastic work principle, where o^* satisfies the yield
criterion and is associated with Et*. Therefore, we have

The surface S = SF + Su and on Su, ut* = uh consequently

Equation (4.46) is the second extremum principle and it shows that an

upper bound to the rate of work of the actual surface traction on Su can be
obtained from a kinematically admissible velocity field. A different
approach to the extremum principles is given in the References [12,13].

4.8 Viscoplasticity
The mathematical theory of plasticitiy adequately describes the time-
independent aspect of the behavior of materials but is inadequate for
analysis of time-dependent behavior. An approach to achieving a satisfac-
tory formulation for time-dependent behavior has been to generalize
plasticity to cases within the strain-rate-sensitive range. One such generali-
zation has been provided by the theory of viscoplasticity. The history of
this theory goes back to as early as 1922 [14]. Since then, generalizations of
early versions have been achieved and various forms of the theory of
viscoplasticity have been provided (summarized, for example, by Perzyna
[15] and by Cristescu [16]). In this section, we describe an approach to the
construction of equations for rigid viscoplastic materials. Then, the
extremum principle is presented as the basis for the finite-element
formulation of viscoplastic flow analysis.
Constitutive Equations
What follows is based on work by Perzyna. We consider a rigid viscoplastic
material and apply the infinitesimal theory for each incremental deforma-
tion. Perzyna introduced a function F(<7y) such that

where k is the static yield stress in shear. Considering F(cTy) as the function
similar to the plastic potential in the theory of plasticity, the constitutive
 Plasticity and Viscoplasticity 71

equation is expressed as

where y" is a viscosity constant of the material, and ($(F)} is a function of

F such that

Then

where a = V(3/2) {o^'o^'}1'2 so that 0 is identical to the yield stress in

uniaxial tension, and Y = V3 k is the static yield stress in tension.
Squaring both sides of eq. (4.49),

and using t = V(2/3) {%£,•,• }1/2, we obtain

From eqs. (4.49) and (4.50), the constitutive equation becomes

with a = V(3/2) H'<V>1/2 and £ = V(2/3) {e^}1/2

Equation (4.51) is formally identical to the Levy-Mises equations. The
effective stress o in eq. (4.51), however, depends on the strain-rate-
dependent function $, which is to be determined by the properties of the
material under consideration. If we choose the function $ = ((a/ Y) — l)1/m,
for example, then

from eq. (4.50). Equation (4.52) is a familiar rate-dependence law and the
exponent m is the strain-rate sensitivity index.
Extremum Principle
Hill [17] derived the (second) extremum principle associated with the
deformation process of the rigid-viscoplastic materials. Among all the
possible constitutive equations, we will restrict our attention to the case
where there exists the work function £(£</), sucn that
72 Metal Forming and the Finite-Element Method

and where E is convex. The existence of work function £(f,y) is ensured if

Otj' is a single-valued function of e,y, satisfying doti'/dekl = dok,'13^.
Also, £ is a convex, if

where et* is a strain-rate field derived from an admissible velocity field

u*. With this restriction, we have the following relation,

where e^, «,- are actual quantities and the starred quantites are kinemati-
cally admissible ones. It can be shown that the solution is uniquely
determined at points of the body where E is strictly convex, but not
necessarily in all respects elsewhere.

References
1. Hill, R., (1950), "The Mathematical Theory of Plasticity," Oxford University
Press, London.
2. Mendelson, A., (1968), "Plasticity: Theory and Application," MacMillan, New
York.
3. Szczepinski, W., (1979), "Introduction to the Mechanic of Plastic Forming of
Metals," Sijthoff & Noordhoff International Publishers, The Netherlands.
4. Yamada, Y., (1965), "Mechanics of Plastic Deformation," Nikkan Kogio
Shinbun-Sha, Tokyo.
5. Chakrabarty, J., (1987), "Theory of Plasticity," McGraw-Hill, New York.
6. Johnson, W., Sowerby, R., and Venter, R. D., (1982), "Plane Strain Slip Line
Fields for Metal Deformation Processes," Pergamon Press, Oxford, UK.
7. Johnson, W., and Mellor, P. B., (1973), "Engineering Plasticity," Van
Nostrand Reinhold, London.
8. Sokolnikoff, I. S., (1956), "Mathematical Theory of Elasticity," McGraw-Hill,
New York.
9. Naghdi, P. M., and Trapp, J. A., (1975), "The Significance of Formulating
Plasticity Theory with Reference to Loading Surfaces in Strain Space," Int. J.
Engg. Sci., Vol. 13, p. 785.
10. Green, A. E., and Naghdi, P. M., (1965), "A General Theory of an
Elastic-Plastic Continuum," Arch. Rational Mech. Anal., Vol. 18, p. 251.
11. Drucker, D. C., (1951), "A More Fundamental Approach to Plastic Stress
Strain Relations," Proc. 1st U.S. Natl. Congress, Appl. Mech. ASME, p. 487.
12. Drucker, D. C., Greenburg, H. J., and Prager, W., (1951), "The Safety
Factor of an Elastic-Plastic Body in Plane Stress," Trans. ASME., J. Appl.
Mech., Vol. 18, p. 371.
13. Drucker, D. C., Prager, W., and Greenburg, H. J., (1952), "Extended Limit
Design Theorems for Continuous Media," Q. Appl. Math., Vol. 9, p. 381.
14. Bingham, E. C., (1922), "Fluidity and Plasticity," 1st edition, McGraw Hill,
New York, p. 125.
15. Perzyna, P., (1966), "Fundamental Problems in Viscoplasticity," Adv. App.
Mech., Vol. 9, p. 243.
16. Cristescu, N., (1967), "Dynamic Plasticity," North-Holland, Amsterdam.
17. Hill, R., (1956), "New Horizons in the Mechanics of Solids," J. Mech. Phys.
Solids, Vol. 5, p. 66.
 5
METHODS OF ANALYSIS

5.1 Introduction
In Chap. 4 we described fundamentals for the analysis of metal-forming
processes and derived some useful principles. The governing equations for
the solution of the mechanics of plastic deformation of rigid-plastic and
rigid-viscoplastic materials are summarized as follows:

Equilbrium equations:

Yield critertion

Constitutive equations:

with

Compatiblity conditionms:

The unknowns for the solution of a quasi-static plastic deformation process

are six stress components and three velocity components. The governing
equations are three equilibrium equations, the yield condition, and five
strain-rate ratios derived from the flow rule. The boundary conditions are
prescribed in terms of velocity and traction. Along the tool-workpiece
interface, the velocity component is prescribed in the direction normal to
the interface and the traction is specified by the frictional stress in the
tangential direction.
Since it is difficult to obtain a complete solution that satisfies all of the
governing equations, various approximate methods have been devised,
depending upon the assumptions and approximations. Some of them have
been mentioned in Chap. 1.
The basic principles and concepts involved in the finite-element method
are the variational principle and discretization. In the course of developing
the analysis methods, the variational principle has played a significant part
in expanding analysis capabilities to prediction of phenomena of industrial
importance. These methods are specifically the upper-bound method and
Hill's general method, leading to the finite-element method. In this
73
74 Metal Forming and the Finite-Element Method

chapter, the two methods are briefly outlined with examples. The basic
formulations for the finite-element method are given in Section 5.4.
Discretization of these basic formulations and numerical solution tech-
niques are discussed in Chaps. 6 and 7.

5.2 Upper-Bound Method

The two extremum principles discussed in Chap. 4 provide an upper bound
and a lower bound to the rate of work of the unknown surface tractions on
Su. Often the velocity on Su is uniform; then the extremum principles can
be used to obtain lower and upper bounds for the forming load. For many
metal-forming operations, no exact solutions, even for the load to cause
unconstrained plastic deformation, are available. If the two bounds can be
made sufficiently close to each other, an almost exact solution can be
estimated. However, in the case of metal-forming problems, it is well
known that a good lower bound is difficult to obtain [lj. Consequently, the
upper-bound method was used in most applications. Even though the
advantage of bracketing the exact solution is usually abandoned, upper
bounds are valuable to mechanical or production engineers, since they
have to estimate a load that is required to perform a forming operation.
The second extremum principle states (see Chapter 4) that

where u* is a kinematically admissible velocity field, e,y* is the strain-rate

field derivable from «*; |Aw*| is the amount of velocity discontinuity along
the surface of discontinuity S^, and otl* satisfies the yield criterion and is
associated with £,-,*.
In plane-strain problems an admissible velocity field can consist of
rigid-block sliding along the lines of velocity discontinuity. Then, since
£,y* = 0 in V and usually Ft = 0, except along the tool-workpiece interface
where the magnitude of the frictional stress fs, acting in the opposite
direction to the relative sliding, is prescribed, the upper-bound calculation
becomes very simple according to

where \us*\ is the amount of relative sliding along the tool-workpiece

interface Sc.
Solutions with velocity fields, using rigid-block sliding, are presented for
many metal-forming problems in the books by Johnson and Mellor [2] and
by Johnson and Kudo [3]. Also, extension of the concept of velocity
discontinuities to formation of "plastic hinges" found its applications to
many plate-bending problems, where determination of plate-collapse loads
is of primary importance [2].
 Methods of Analysis 75

Perhaps the first successful use of kinematically admissible velocity fields

in predicting the mean pressure to extrude round bars is that of Johnson
and Kudo [4]. Kudo [5, 6] has contributed immensely to this field and his
work has helped to set the direction for many theoretical studies in
extrusion and forging. The books by Avitzur [7, 8] show the quantity of
effort required in arriving at some solutions and a variety of problems in
which the upper-bound solutions yield useful information. The principles
applied in all cases are basically identical. Differences in approach are only
seen in constructing velocity fields, and the calculation of the upper-bound
load follows the same procedure. In the following, construction of an
admissible velocity field and the upper-bound calculation are illustrated in
an example [9].
Compression of a Solid Cylinder Between Flat Rough Dies
In the compression of a cylinder shown in Fig. 5.la, the deformation is
symmetric about the r-axis. Hence the forming pressure is the same as that
obtained for the compression of a cylinder between the top rough die
(frictional stress equals to the shear yield stress k) and the bottom smooth
die (frictional stress is zero) for the height H and the diameter D. An
assumed velocity field is shown in Fig. 5.1b. The deforming workpiece is
divided into two regions by a velocity discontinuity curve (1-4). The

FIG. 5.1 (a) Compression of a solid cylinder between flat rough dies, (b) An admissible
velocity field.
76 Metal Forming and the Finite-Element Method
velocity components in each region are assumed to be

in Region 1
and

in Region 2

In eq. (5.3), the exponent n is a parameter with which a minimum upper

bound is sought, and it is entirely different from n used as a strain-
hardening exponent in denning the stress-strain property of a material. In
each region, the radial velocity component ur is assumed and the axial
component uz is determined so that it satisfies the incompressibility
condition given by

and and the boundary conditi

From the condition that the normal velocity component must be

continuous across the curve of discontinuity, we have dz/dr = Aw r /Aw^,
along the curve (1-4), where A denotes the difference of velocity
components between the two regions. Integration results in the equation
for the discontinuity curve as

In eqs. (5.3) and (5.4), a = 2H/D with dimensions as shown in Fig. 5.la.
It is to be noted that the line of discontinuity becomes straight for n = 0
and the velocity field becomes continuous throughout the body for n = 1.
An upper bound to the rate of total forming energy E is obtained by
calculating the energy rate of deformation of a continuously deforming
body and the energy-rates due to shearing along the surface of velocity
discontinuity, based on an assumed velocity field. The upper bound to the
forming pressure, then, can be found from the equation

where A is the projected area of the die surface on a plane perpendicular

to the motion of this surface upon which the forming pressure acts, and UD
is the die velocity.
The region 1 in Fig. 5.1b moves as a rigid body and no energy-rate
dissipation occurs along the die-workpiece interface. The energy rate for
the continuously deforming region (124) is written as
 Methods of Analysis 77

where

according to the velocity field in region 2 given by eq. (5.3).

In order to simplify the calculation, an approximation is made for the
effective strain-rate as

The energy-rate dissipated as a result of the velocity discontinuity |A«|

along the curve (1-4) is given by

Thus the upper bound solution of the forming pressure becomes

Equation (5.5) is plotted as a function of 2H/D in Fig. 5.2, with the

FIG. 5.2 Comparison of upper bounds to the average forming pressure for the compression
of a cylinder.
78 Metal Forming and the Finite-Element Method
value of n that makes pja minimum, and is compared with the pressures
for n = 0 and n = 1.
Remarks
The technique outlined above together with the increased use of compu-
ters led to the development of generalized upper-bound approaches to a
wider class of problems. Examples of such approaches are the analysis of
axisymmetric forging of complex shapes [10], estimation of lateral spread
in plate rolling [11] and in forging [12], and upper bounds to extrusion and
drawing through dies of various shapes [13-16].
In these examples, construction of admissible velocity fields becomes
more complicated and upper-bound calculations must often resort to
numerical means. A question arises, then, whether pursuing this approach
is worth the effort, because detailed flow is much more important than
forces. Kudo [17] surveyed the development and use of the upper-bound
method and concluded that the approach should continue to be useful in
teaching, in workshop trials, and in research on bulk forming processes.
Also, it will be noted in Section 5.5 that the finite-element method
emerges from the principle involved in the upper-bound method and from
the concept of discretization.

5.3 Hill's General Method

Hill [18], combining flexibility with rigorous principles, proposed a new
method of analysis applicable to any technological forming process. He
systematized the details of procedure and illustrated the method with
preliminary analyses of inhomogeneous compression, bar drawing, and
forging. Ideally, a good method of approximation must be capable of
reproducing and predicting the main phenomena, and of delivering reliable
information about loads and principal dimensional changes. It should also
be governed by well-defined general principles, but be flexible enough to
cover all normal situations, arbitrary material properties, different kinds of
friction, or any tool shape. Hill believed that the proposed method comes
close to the ideal and indicated that the prospects were encouraging, but
the final assessment awaits a longer and more diverse investigation.
Since 1963, however, with few exceptions, no further investigation on
the method has appeared in the literature. Nagamatsu et al. [19] and
Murota et al. [20] analyzed plane-strain and axisymmetric compression by
Hill's method. Lahoti and Kobayashi [21,22] carried out the analysis of
ring compression with barreling, spread in Stechel rolling, and thickness
change in tube sinking. In the following, the basic ideas of the method are
outlined and the procedure of the method is illustrated in an example.
The Method of Analysis
Suppose we have some piecewise-continuously differentiable distribution
of stress a^ in a region V, together with traction F, over its surface 5. Then
 Methods of Analysis 79

the converse of the virtual work-rate principle is stated as follows: The

considered distribution o(i is in equilibrium within a body (neglecting the
body force) and with applied surface tractions given by Fh if

for all virtual velocity fields w,- that are continuous and continuously
differentiable. Introducing an approximate stress field that may not
necessarily satisfy equilibrium in eq. (5.6a) and applying the divergence
theorem, we obtain

where CT,-,M/ is the traction on S computed from the considered stress field.
If eq. (5.6b) holds for all virtual velocity wjt then dOfj/dXi = 0 in V and
a,y«, = FJ on S. Equations (5.6a) and (5.6b) thus are equivalent. Equation
(5.6b) asserts that a sufficient condition for the considered stress distribu-
tion to satisfy the required statical conditions is that the divergences should
have zero net work-rate in the stated class of virtual motions, or the
divergences should be "orthogonal" to the virtual motions.
The interpretation of this theorem leads to a criterion of approximation,
namely, the static conditions for the approximating field o^ can be
regarded as closely satisfied overall when eq. (5.6a) is satisfied for a
sufficient subclass of virtual orthogonalizing motion w,, with the traction Ff
prescribed on SF and computed from the approximating field otj on the
remainder of the surface.
For metal-forming processes, the surface 5 of the deforming zone
consists of three distinct parts, S = Sc + SF + Sj, where Sc adjoins a tool or
container, SF is unconstrained, and S/ is the junction with the rigid zone;
on surface SF, ordinarily the traction is zero; the frictional constraint over
surface Sc is represented by a constant frictional stress mk (where k is the
shear yield stress of the deforming body) or a Coulomb coefficient of
friction (i.
The selection criterion for the approximating field a,-,- is then expressed
bv
for constant
frictional stress
for Coulomb
friction

for a sufficient subclass of virtual orthogonalizing motion w,, where T(-

denotes the surface traction computed from the considered field o^; n; is
the local unit outward normal; and /, is a unit tangent vector opposite in
sense to the relative velocity of sliding in the approximating field.
80 Metal Forming and the Finite-Element Method
The initial procedure of the method is to choose a class of velocity fields
from which the approximate stress fields can be determined. The chosen
velocity fields must satisfy all kinematic conditions. The associated stress
(given by the material constitutive law) is either determined uniquely, or at
least determined to within a hydrostatic pressure if the material is
incompressible. The associated distribution of stress in the deformation
zone of a chosen velocity field will generally not satisfy all the statical
requirements. In applying the selection criterion (5.7), the orthogonalizing
family Wj must be sufficiently wide and extensive to identify a single
approximating velocity field in the particular class constructed for satisfy-
ing the kinematic conditions. If the class is defined by equations involving
an unknown function of just one position variable, then the orthogonaliz-
ing family must also involve an arbitrary function of one variable. Clearly,
the easiest choice of orthogonalizing motions is the class of approximating
velocity fields itself; or one can use differences of pairs of these fields.
Once a family is chosen, the calculus-of-variations technique is applied
to eqs. (5.7), treating w, as a variation. We obtain a system of equilibrium
equations and boundary conditions, suited to the particular approximating
class and uniquely determining its best member.

Illustrative Example—Compression of Solid Circular Cylinder

A solid circular cylinder with diameter D and height 2H is compressed
between parallel rigid platens having a constant friction mk (see Fig. 5.la).
It is required to calculate the incipient barreling of the free surface. Take
axes (r, z) with the origin at the half height and z along the direction of
compression.
Since longitudinal nonuniformity is the feature under investigation, the
approximating class of velocity field must contain an unspecified function
of z. The simplest construct for the velocity components is

with 0(—z) = -0(z) and (j)(H) = l, where the prime implies

differentiation.
For symmetry about the mid-plane, </> is an odd function of z with
continuous first and second derivatives at least.
To match the class (5.8), we must choose a comparable family of virtual
orthogonalizing motions, also involving an arbitrary function of z. Take

with V(~ 2 ) = ~'<P(Z)-

Noting that Sj = 0, nf is a unit vector in the direction of z, {«,} =
{nr, ng, nz} = {0, 0, 1}, and /, is a unit vector in the direction of r,
 Methods of Analysis 81

{lj} = (lr, le, 4} = {-1, 0, 0}, the criterion (5.7) can be written as

Substitution of eq. (5.9) into eq. (5.10) results in

where o'z — oz — j(az + or + og) and P is the compression load given by

Integrating by parts and noting that

0, eq. (5.11) becomes

Since i/>' is completely arbitrary, it follows that

and

at any z, which integrates to

In deriving eq. (5.14), the yield condition in the form of a'2= — §Y,
where Y denotes the yield stress in uniaxial compression, is used as a first
order of approximation.
Evaluating eq. (5.14) at z=H and using eq. (5.13), we obtain the
expression for the compression load P as

By replacing mk with fiY in eq. (5.15), the familiar Siebel formula is

obtained.
82 Metal Forming and the Finite-Element Method
To obtain the differential equation for (j>, Trz is calculated from the
approximating field (5.8), using the flow rule as

and thus

Substitution of eq. (5.16) into eq. (5.14) with the help of eq. (5.15), results
in

where

Integration of eq. (5.17), with 0(0) = 0 and $(//) = !, determines the

bulge function <f>(z) as

Substituting eq. (5.18) into the approximate velocity field (5.8), the
incipient of bulge formation can be described by the radial velocity
component at r = D/2. Figure 5.3 shows the profile of the radial velocity
component along the free surface for H/D = 0.5 and several values of
friction factor m. In order to trace the development of bulge, calculation
must be repeated step by step, updating the current geometrical
configuration.
To illustrate the method, the formulation given in this example uses a
simpler approximating velocity field than that given by Hill.
Remarks
It appears that the class of problems to which the Hill method can be
applied effectively is one for which the determination of the geometrical
changes and load are the main objectives. This class of problems can be
grouped into three areas: (1) simple upsetting of rings and blocks, (2)
flat-tool forging and its allied problems, (3) steady-state processes with
unknown steady-state configurations. The method is well-based on mathe-
matically sound principles and is amply flexible in introducing various
simplifying assumptions into the analysis. Actual procedure using the
method usually involves extensive calculations and sometimes encounters
overwhelming difficulties in obtaining final solutions. However, as pointed
 Methods of Analysis 83

FIG. 5.3 Computed profile of the radial velocity component along the free surface in
compression.

out in Section 5.5, the basis of the finite-element method is the discrete
representation of the principles and ideas given in Hill's method.

5.4 The Finite-Element Method

The four approaches used for the derivation of the basic equations for the
finite-element analysis have been stated in Chap. 1. They are (1) the direct
approach, (2) the variational method, (3) the method of weighted
residuals, and (4) the energy balance approach. In the following, the
variational method and a special case of the weighted residual method are
described.
Basis for the Finite-Element Formulation
The variational approach is based on one of two variational principles. It
requires that among admissible velocities ut that satisfy the conditions of
compatibility and incompressibility, as well as the velocity boundary
conditions, the actual solution gives the following functional (function of
functions) a stationary value:

for rigid-plastic materials

and

for rigid-viscoplastic materials

84 Metal Forming and the Finite-Element Method

where o is the effective stress, e is the effective strain-rate, Ff represents

surface tractions, and E(eLj) is the work function (see eq. (4.53) in Chap.
4). The solution of the original boundary-value problem is then obtained
from the solution of the dual variational problem, where the first-order
variation of the functional vanishes, namely,

where o = O(E) and a = o(e, e) for rigid-plastic and rigid-viscoplastic

materials, respectively. The incompressibility constraint on admissible
velocity fields in eq. (5.20) may be removed by introducing a Lagrange
multiplier A [23, 24] and modifying the functional (5.19) by adding the
term J Ac,, dV, where ev = ea, is the volumetric strain-rate. Then,

Another way of removing the constraint is to use the penalized form of

the incompressibility [25] as

where K, a penalty constant, is a very large positive constant.

In eqs. (5.21) and (5.22), dut and <5A are arbitrary variations and de and
dev are the variations in strain-rate derived from dut. Equation (5.21) or
(5.22) is the basic equation for the finite-element formulation.
It should be noted that under the plane-stress condition that applies to
the analysis of sheet-metal forming (Chap. 11), the change of sheet
thickness is determined from the incompressibility condition and the
constraint is not imposed to admissible velocities. Also, the incompres-
sibility condition does not apply to the deformation of porous materials,
discussed in Chap. 13.
An alternative approach to eq. (5.20) is to begin with a weak form of the
equilibrium equations, namely,

where duf is an arbitrary variation in «,-. Equation (5.23) becomes

Note that eq. (5.24) is the expression of the virtual work-rate principle, if
dut is considered to be the virtual motion (see eq. (4.26) in Chap. 4).
Using the symmetry of the stress tensor and the divergence theorem, eq.
(5.24) results in
 Methods of Analysis 85

For the surface integral term in eq. (5.25), the boundary conditions that
diij = 0 on Su (essential boundary condition) and o^n, = Ft on SF
(suppressible boundary condition), are imposed.
Substitution of a,y = a,/ + <5,yOm, where am = \okk, the hydrostatic stress,
into eq. (5.25) gives

From the constitutive equation and the definition of e, it is seen that,

a// (5e,y = ode, and the final form of eq. (5.26) suitable for the finite-
element formulation becomes

with the incompressibility constraint given by £„ = 0 in V. The incompres-

sibility constraint is removed by modifying eq. (5.27a) as

By comparison of eqs. (5.21) and (5.27b), the Lagrangian multiplier A in

eq. (5.21) is identified as the mean stress om. Further, eq. (5.22) is
regained by the use of om = (K/2)£v in eq. (5.27b), thus providing the
interpretation of K as a constant similar to the bulk modulus.
Equation (5.21) or (5.22) is the basic equation for the finite-element
discretization. Once the solution for the velocity field that satisfies the
basic equation is obtained, then corresponding stresses can be calculated
using the flow rule and the known mean stress distribution.
Comments on Basic Equations
The duality of the boundary-value problem and the variational problem
can be seen clearly by considering the construction of the functional (5.19).
If the functional is n(u), then dn(u) = 0 breaks down into two parts, that
of the domain and that of the boundary. Suppose that the functional Jt(u)
is given by

where x is the coordinate, u is the field variable and u' is the derivative of
u with respect to x. Then

Integrating the second term by parts, we have

86 Metal Forming and the Finite-Element Method
Thus,

dn(u) = Euler Equation} du dV + {the boundary term)

The Euler equation is a differential equation taken over the domain V

that is argued to be zero because of the arbitrariness of du. Given the
differential equations, the functional n and the boundary terms can often
be constructed by manipulation. For our boundary value problem,

The essential boundary condition that <5w, = 0 on Su is imposed and the

boundary requirement on SF are suppressed in eq. (5.28), since duf is
arbitrary on SF. The manipulation of the first term of eq. (5.28) follows
eqs. (5.23), (5.24), and (5.25). Then, eq. (5.28) becomes

from which the functional (5.19) is obtained.

The construction of the functional discussed above also shows that the
variational formulation involves redundancy in manipulation, compared to
the alternative approach. Interpretation of eq. (5.23) leads to the concept
of error minimization. Galerkin's method is the obvious discretization of
eq. (5.23), and is a specific method of using weighted residuals where the
weighting functions are the same as the approximating functions.

Boundary Conditions
The traction boundary condition on SF is either zero-traction or ordinarily
at most a uniform hydrostatic pressure. However, the boundary conditions
along the die-workpiece interface are mixed. Furthermore, in general,
neither velocity nor force (including magnitude and direction) can be
prescribed completely along this interface, because the direction of the
frictional stress is opposite to the direction of the relative velocity between
the deforming workpiece and the die, and this relative velocity is not
known a priori. Situations exist, e.g., in extrusion and drawing, in which
the direction of metal flow relative to the die is known. This class of
problems can be solved if the magnitude of the frictional stress fs is given
according to the well-known Coulomb law, fs = ftp, or the friction law of
constant factor m, expressed by fs = mk (where k = Y/VJ); here, p is the
die pressure and k is the shear yield stress.
For problems such as ring compression, rolling, and forging, the
unknown direction of the relative velocity between the die-workpiece
interface makes it difficult to handle the boundary condition in a
straightforward manner. A unique feature of this type of problem is that
there exists a point (or a region) along the die-workpiece interface where
 Methods of Analysis 87

the velocity of the deforming material relative to the die becomes zero (see
Section 3.8 of Chap. 3), and the location of this point (or region) depends
on the magnitude of the frictional stress itself. In order to deal with these
situations, a velocity-dependent frictional stress is used as an approxima-
tion to the condition of constant frictional stress. At the interface Sc the
velocity boundary condition is given in the direction normal to the
interface by the die velocity, and the traction boundary condition is
expressed by

where fs is the frictional stress, € is the unit vector in the opposite direction
of relative sliding, us is the sliding velocity of a material relative to the die
velocity (relative sliding velocity), and u0 is a small positive number
compared to us. The approximate expression (5.30) for a constant
frictional stress has been used for the smooth transition of the frictional
stress in the range near the neutral point [26].
Treatment of a Rigid Region
The principles described in the preceding sections apply to a domain in
which the entire body is deforming plastically. In metal-forming processes,
however, situations do arise in which rigid zones exist, and unloading
occurs. The rigid zones are characterized by a very small value of effective
strain-rate in comparison with that in the deforming body. If these portions
are included within the control volume V (the volume included in the
statement of boundary value problems), the value of the first term of the
basic equation (5.21) or (5.22) cannot be uniquely determined because of
the undefined value of the effective stress when the effective strain-rate
approaches zero.
This difficulty is removed by assuming that the stress-strain-rate
relationship in eq. (5.1c) is approximated by

where £0 takes an assigned limiting value, say 10~3 [27]. This presumed
stress-strain-rate relationship is equivalent to the assumption of a New-
tonian fluid-like material behavior for nearly-rigid regions. For these
regions, the first term of the basic equation, Svode dV, is then replaced
by

Thus, the finite-element discretization process is based on eq. (5.21) or

(5.22), with eq. (5.30) in mind for the die-workpiece interface boundary
condition, and eq. (5.31) for the regions that are considered to be nearly
rigid.
88 Metal Forming and the Finite-Element Method

5.5. Concluding Remarks

The upper-bound method is based on one of the extremum principles. The
conventional minimization technique in the method is associated with a
weaker statement in the form of a variational principle, namely, eq. (5.20).
Equation (5.20) also serves as a basis for the finite-element discretization.
Thus both methods are devised essentially from the same principle.
However, admissible velocity fields in the upper-bound approach are
traditionally formulated in closed forms, and thus a considered class of
velocity fields is limited. In the finite-element approach, because of
discrete representation of approximating velocity fields, the class of
considered velocity fields is much wider.
In Hill's general method, Hill stated that we have considerable freedom
of choice for orthogonalizing virtual motions and suggested that the easiest
choice is the class of approximating fields itself; alternatively, one can use
differences of pairs of these fields, or infinitesimal variations within the
class. When the variations within the approximating field are chosen as
orthogonalizing virtual motions in Hill's general method, the variational
principle, eq. (5.20), is supplied. Thus, it is clear that only the concepts of
discretization and discrete representation of approximate velocity fields
distinguish the finite-element method from the others.

References
1. Kobayashi, S., and Thomson, E. G., (1965), "Upper and Lower Bound
Solutions to Axisymmetric Compression and Extrusion Problems," Int. J.
Mech. ScL, Vol. 7, p. 127.
2. Johnson, W., and Mellor, P. B., (1973), "Engineering Plasticity," Van
Nostrand Reinhold, London.
3. Johnson, W., and Kudo, H., (1962), "Mechanics of Metal Extrusion,"
Manchester University Press, Manchester, UK.
4. Johnson, W., and Kudo, H., (1960), "Use of Upper Bound Solutions. . .For
Axisymmetric Extrusion Processes," Int. J. Mech. Sci., Vol. 1, p. 175.
5. Kudo, H., (1960), "Some Analytical and Experimental Studies of Axisym-
metric Cold Forging and Extrusion—I," Int. J. Mech. Sci., Vol. 2, p. 102.
6. Kudo, H., (1961), "Some Analytical and Experimental Studies of Axisym-
metric Cold Forging—II," Int. J. Mech. Sci., Vol. 3, p. 91.
7. Avitzur, B., (1968), "Metal Forming: Processes and Analysis," McGraw-Hill,
New York.
8. Avitzur, B., (1983), "Handbook of Metal Forming Processes," Wiley, New
York.
9. Kobayashi, S., (1964), "Upper-Bound Solution to Axisymmetric Forming
Problems," Trans. ASME, J. Engr. Ind., Vol. 83, p. 326.
10. McDermott, R. P., and Bramley, A. N., (1974), "An Elemental Upper-Bound
Technique for General Use in Forging Analysis," Proc. 15th M.T.D.R. Conf.,
p. 437.
11. Oh, S. I., and Kobayashi, S., (1975), "An Approximate Method for a
Three-Dimensional Analysis of Rolling," Int. J. Mech. ScL, Vol. 17, p. 293.
12. Juneja, B. L., (1973), "Forging of Polygonal Discs with Barrelling," Int. J.
Machine Tool Des. Res., Vol. 13, p. 87.
 Methods of Analysis 89

13. Juneja, B. L., and Prakash, R., (1975), "An Analysis for Drawing and
Extrusion of Polygonal Sections," Int. J. Machine Tool Des. Res., Vol. 15, p.
1.
14. Nagpal, N., and Allan, T., (1975), "Analysis of Three-Dimensional Metal
Flow in Extrusion of Shapes with the Use of Dual Stream Functions," Proc.
3rd NAMRC, Carnegie-Mellon University, p. 26.
15. Yang, D. Y., and Lee, C. H., (1978), "Analysis of Three-Dimensional
Extrusion of Sections Through Curved Dies by Conformal Transformation,"
Int. J. Mech. ScL, Vol. 20, p. 541.
16. Kiuchi, M., (1984), "Overall Analysis of Non-Axisymmetric Extrusion and
Drawing," Proc. 12th NAMRC, Michigan Technological Univ. Houghton,
Michigan, p. 111.
17. Kudo, H., (1985), Upper Bound Approach to Metal Forming Processes—To
Date and in the Future, "Metal Forming and Impact Mechanics," edited by
S. R. Reid, Pergamon Press, Oxford, p. 19.
18. Hill, R., (1963), "A General Method of Metal-Working Processes," /. Mech.
Phys. Solids, Vol. 11, p. 305.
19. Nagamatsu, A., Murota, T., and Jimma, T., (1970), "On the Non-Uniform
Deformation of Block in Plane Strain Compression Caused by Friction," Bull.
JSME, Vol. 13, p. 1389.
20. Murota, T., Jimma, T., and Nagamatsu, T., (1966), "A Theoretical Analysis
of Circular Cylinder in Axial Compression with Friction," Proc. 16th Japan
Natl. Congr. Appl. Mech., p. 141.
21. Lahoti, G. D. and Kobayashi, S., (1974), "On Hill's General Method of
Analysis for Metal-Working Processes," Int. J. Mech. ScL, Vol. 16, p. 521.
22. Lahoti, G. D., and Kobayashi, S., (1974), "Flat-Tool Forging," Proc. 2d
North American Metal Working Res. Conf., May, University of Wisconsin-
Madison, p. 73.
23. Washizu, K., (1968), "Variational Methods in Elasticity and Plasticity,"
Pergamon Press, Oxford.
24. Lee, C. H., and Kobayashi, S., (1973), "New Solutions to Rigid-Plastic
Deformation Problems Using a Matrix Method," Trans. ASME, J. Engr. Ind.,
Vol. 95, p. 865.
25. Zienkiewicz, O. C., (1977), "The Finite Element Method," 3d Edition,
McGraw-Hill, New York.
26. Chen, C. C., and Kobayashi, S., (1978), "Rigid-Plastic Finite-Element
Analysis of Ring Compression," Application of Numerical Methods to
Forming Processes, ASME, AMD, Vol. 28, p. 163.
27. Chen, C. C., Oh, S. I., and Kobayashi, S., (1979), "Ductile Fracture in
Axisymmetric Extrusion and Drawing, Part 1: Deformation Mechanics of
Extrusion and Drawing," Trans. ASME, J. Engr. Ind., Vol. 101, p. 23.
 6
THE FINITE-ELEMENT METHOD—PART I

6.1 Introduction
The concept of the finite-element procedure may be dated back to 1943
when Courant [1] approximated the warping function linearly in each of an
assemblage of triangular elements to the St. Venant torsion problem and
proceeded to formulate the problem using the principle of minimum
potential energy. Similar ideas were used later by several investigators to
obtain the approximate solutions to certain boundary-value problems. It
was Clough [2] who first introduced the term "finite elements" in the study
of plane elasticity problems. The equivalence of this method with the
well-known Ritz method was established at a later date, which made it
possible to extend the applications to a broad spectrum of problems for
which a variational formulation is possible. Since then numerous studies
have been reported on the theory and applications of the finite-element
method.
In this and next chapters the finite-element formulations necessary for
the deformation analysis of metal-forming processes are presented. For hot
forming processes, heat transfer analysis should also be carried out as well
as deformation analysis. Discretization for temperature calculations and
coupling of heat transfer and deformation are discussed in Chap. 12. More
detailed descriptions of the method in general and the solution techniques
can be found in References [3-5], in addition to the books on the
finite-element method listed in Chap. 1.

6.2 Finite-Element Procedures

The path to the solution of a problem formulated in finite-element form is
described in Chap. 1 (Section 1.2). Discretization of a problem consists of
the following steps: (1) describing the element, (2) setting up the element
equation, and (3) assembling the element equations. Numerical analysis
techniques are then applied for obtaining the solution of the global
equations. The basis of the element equations and the assembling into
global equations is derived in Chap. 5 (eq. (5.20) and eqs. (5.21) or
(5.22)).
The solution satisfying eq. (5.20) is obtained from the admissible
90
 The Finite-Element Method—Part I 91

velocity fields that are constructed by introducing the shape function in

such a way that a continuous velocity field over each element can be
denned uniquely in terms of velocities of associated nodal points. In the
deformation process shown in Fig. 6.1, the workpiece is divided into
elements, without gaps or overlaps between elements. In order to ensure
continuity of the velocities over the whole workpiece, the shape function is
defined such that the velocities along any shared element-side are
expressed in terms of velocity values at the same shared set of nodes
(compatibility requirement). Then a continuous velocity field over the
whole workpiece can be uniquely defined in terms of velocity values at
nodal points specified globally.
We define a set of nodal point velocities in a vector form as

where the superscript T denotes transposition and N = (total number of

nodes) x (degrees of freedom per node).
An admissibility requirement for the velocity field is that the velocity
boundary condition prescribed on surface Su (essential boundary condi-
tion) must be satisfied. This condition can be imposed at nodes on Su by

FIG. 6.1 Finite-element mesh and nodal point specifications in a forming process.
92 Metal Forming and the Finite-Element Method
assigning known values to the corresponding variables. It is to be noted
that the incompressibility condition is not required for defining a velocity
field in the formulation of eq. (5.21) or (5.22).
Equations (5.20) and (5.21) or (5.22) are now expressed in terms of
nodal point velocities v and their variations 6v. From arbitrariness of 6v{,
a set of algebraic equations (stiffness equations) are obtained as

where (/) indicates the quantity at the/th element. The capital-letter suffix
signifies that it refers to the nodal point number.
Equation (6.2) is obtained by evaluating the (dn/dvj) at the elemental
level and assembling them into the global equation under appropriate
constraints.
In metal-forming, the stiffness equation (6.2) is nonlinear and the
solution is obtained iteratively by using the Newton-Raphson method.
The method consists of linearization and application of convergence
criteria to obtain the final solution. Linearization is achieved by Taylor
expansion near an assumed solution point v = v0 (initial guess), namely,

where Au/ is the first-order correction of the velocity v0. Equation (6.3)
can be written in the form

where K is called the stiffness matrix and f is the residual of the nodal point
force vector.
Once the solution of eq. (6.4) for the velocity correction term Av is
obtained, the assumed velocity v0 is updated according to vn + <x Av, where
a is a constant between 0 and 1 called the deceleration coefficient. Iteration
is continued until the velocity correction terms become negligibly small.
The Newton-Raphson iteration process is shown schematically in Fig. 6.2.
It is seen from the figure that convergence of Newton-Raphson iterations,
the initial guess velocity should be close to the actual solution. When a
deformation process is relatively simple, the initial guess velocity can be
provided, for instance, by the upper-bound method. However, if the
process is complex and obtaining a good initial guess solution is difficult,
then the use of the direct iteration method discussed in Section 7.4 of
Chap. 7 may be appropriate.
Two convergence critera may be used. One measures the error norm of
the velocities, ||Av||/||v||, where the Euclidean vector is defined as
||v|| = (vTv)1/2, and requires such an error norm to decrease from iteration
to iteration. The other criterion requires the norm of the residual
equations, ||3jr/3v||, to decrease.
In general terms, the first criterion is most useful in the early stages of
 The Finite-Element Method—Part I 93

FIG. 6.2 Schematic representation of the Newton-Raphson method: (a) convergence; (b)
divergence.

iteration, when the velocity field is still far from the solution. The second
test is most useful when slightly ill-conditioned systems reach the final
stage of iterations. The final solution is considered to be achieved when the
error norm reaches a specified small value, say 5 x 1(T5.
The finite-element method procedures outlined above are implemented
in a computer program in the following way.
1. Generate an assumed solution velocity.
2. Evaluate the elemental stiffness matrix for the velocity correction
term Av in eq. (6.4).
3. Impose velocity conditions to the elemental stiffness matrix, and
repeat step 2 over all elements defined in the workpiece.
4. Assemble elemental stiffness matrix to form a global stiffness
equation.
94 Metal Forming and the Finite-Element Method
5. Obtain the velocity correction terms by solving the global stiffness
equation.
6. Update the assumed solution velocity by adding the correctional term
to the assumed velocity. Repeat steps 2 through 6 until the velocity
solution converges.
7. When the converged velocity solution is obtained, update the
geometry of the workpiece using the velocity of nodes during a time
increment. Steps 2 through 7 are repeated until the desired degree of
deformation is achieved.
The order of the certain steps mentioned above may change depending
upon the computer programming practice.
The above procedure applies to the analysis of nonsteady-state-
processes. For steady-state processes, updating the geometry of the
workpiece is not necessary. The procedure for steady-state processes is
described in Chap. 10 for the analysis of axisymmetric extrusion and
drawing.

6.3 Elements and Shape Function

The order of a boundary-value problem is denned as that of the highest
derivative of the field variable. It is 2 in our problem, and the integrand in
eq. (5.20) is of order 1. Thus, the admissibility requirements are that the
velocity must have a continuous first-order derivative within the element
(completeness requirement) and the velocity should be continuous at the
subdomain interfaces (compatibility requirement). Symbolically, a function
of class Cr has continuous derivatives of all orders up to and including r.
Then shape functions for the velocity field within the element must be of
class Cl for completeness, and the compatibility of elements requires them
to be of class C° globally.
The geometry of an element, in general, is uniquely defined by a finite
number of nodal points (nodes). The nodes are located on the boundary of
the element or within the element, and the shape function defines an
admissible velocity field locally in terms of velocities of associated nodes.
Thus, elements are characterized by the shape and the order of shape
functions.
In the finite-element method, interpolation of a scalar function f(x, y)
defined over an element is introduced in a form

where fa is a function value associated with ath node, and qa(x, y) is the
shape function. It is, in general, a polynomial function of x and y defined
over the element in such a way that
 The Finite-Element Method—Part I 95

FIG. 6.3 Area coordinate system of the triangular element.

where (xe, y@) is the coordinates of /3th node and 6a/i is the Kronecker
delta. Owing to the property of the shape functions given by eq. (6.6), /a
in eq. (6.5) has the value of the function / at (xa, ya) and the fa are
independent of each other.
There are various types of elements, depending upon the shape of the
element and the polynomial order of shape funtions. In the following, the
elements used in this book are discussed.
Triangular Element Family
In the triangular element family, it is convenient to define shape functions
in the area coordinate system, L1; L2, L3. The area coordinates for a
triangle, as shown in Fig. 6.3, are defined by the following linear relations:

where (xa, ya) are the coordinates of a corner of the triangle. It can be
readily shown that an alternative definition of the coordinate of a point P
can be given by the ratio of the area of the shaded triangle to that of the
total triangle as

Solving eq. (6.7) for L t , L2 and L3 gives

96 Metal Forming and the Finite-Element Method
where
A = area of (123)
and

With a linear triangular element with nodes at its corners (see Fig. 6.4a),
the shape functions qa are linear and are given by

A quadratic triangular element has primary nodes at the corners and

secondary nodes at the mid-sides (see Fig. 6.4b) and the shape functions
are

The admissible velocity field within the element can be represented, for
both linear and quadratic elements, by

where u^a\ u^ are the velocity components of the <*th node.

Elements that use the same shape functions as the coordinate transfor-
mation are called isoparametric elements. Linear triangular elements,
owing to eqs. (6.7) and (6.10), are isoparametric. However, with quadratic
shape functions, isoparametric elements are curved triangular elements.

FIG. 6.4 (a) Linear triangular element; (b) quadratic triangular element.
 The Finite-Element Method—Part I 97

FIG. 6.5 (a) Natural coordinate and rectangular parent element; (b) isoparametric element
(mapped on Cartesian coordinate; quadrilateral element); (c) shape function.

In this book, the linear triangular element is used for the analysis of
plate-bending under plane-strain conditions in Chap. 8. It is also used for
the analysis of sheet-metal forming in Chap. 11.
Rectangular Element Family
The shape functions of rectangular elements are, in general, denned in a
parametric form over a domain — ! < £ < ! , — l < r j < l in a natural
coordinate system (|f, r/). The element defined in the natural coordinate
system is sometimes called the parent element. The simplest of the
rectangular elements is the 4-node linear element shown in Fig. 6.5a. The
shape functions, qa, which are bilinear in £ and r/, are defined as

where (£„., r]a) are the natural coordinates of a node at one of its corners.
98 Metal Forming and the Finite-Element Method
The value of the shape function, given by eq. (6.13), is shown schemati-
cally in Fig. 6.5c. Admissible velocity fields can be defined uniquely over
the rectangular element by the nodal velocity components as

where (u^a\ u(ya)) is the velocity at the <*th node and summation is over all
four nodes.
Coordinate transformation from the natural coordinate (£, 77) to the
global coordinate (x, y) is denned by

where (xa, ya) are the global coordinates of the o-th node. Since the
coordinate transformation (6.15) uses the same shape functions (6.14), the
linear element is isoparametric and takes quadrilateral shape in the
Cartesian map, as shown in Fig. 6.5b.
Elements with shape functions of higher-order polynomials can be
defined in a similar manner. Figure 6.6 shows a quadratic rectangular
element with 8 nodes in the natural coordinate and global coordinate
systems. The shape functions are defined by
corner nodes:

mid-side nodez;

Rectangular quadratic elements can also be defined by 9 nodes by

adding an internal node to an 8-node quadratic rectangular element. The
shape functions of the 9-node quadratic element can be derived from those
of 8-node element. It may be of interest to note that the internal node does
not affect the velocity distributions at the element boundary and the
element shape. The rectangular element with 9 nodes is said to be of the
Lagrangian family, since the shape functions are derived from Lagrange
polynomials [6].
The isoparametric element can be distorted, for instance as shown in
Fig. 6.6b, for a quadratic rectangular element by mapping according to eq.
(6.15) with the shape functions given by eq. (6.16). The distorted shapes of
 The Finite-Element Method—Part I 99

(b)
FIG. 6.6 Eight-node quadratic rectangular element: (a) natural coordinate and parent
element; (b) isoparametric element after Cartesian mapping.

the isoparametric elements are sometimes useful, since they allow more
flexibility and convenience in mesh generation. On the other hand, there
are some limitations on the element geometry. To be a valid element, the
coordinate transformation given by eq. (6.15) should be one-to-one, or the
Jacobian of the coordinate transformation (see eq. (6.29)) should be
positive at all points within the element. The validity of the element
geometry is important, since elements deform considerably in metal-
forming simulation.
Elements used for axisymmetric deformation are the same as two-
dimensional elements in terms of shape functions and coordinate transfor-
mations of the element. However, the axisymmetric element represents
the cross section of a torus (ring element), while the other two dimensional
elements represent the cross section of a straight bar. The strain-rate
definition and the volume integration procedure are therefore different.
100 Metal Forming and the Finite-Element Method
The 4-node linear isoparametric element (quadrilateral element) is used
extensively in this book for the analyses of two-dimensional and axisym-
metric deformation processes.
Three-Dimensional Brick Element
The three-dimensional brick element is a natural extension of the
two-dimensional linear element. The parent element is defined over a
domain -!<£<!, -1 < 77 < 1, — 1 < £ :£ 1, in natural coordinate system
with a node at each corner. The shape functions are defined as

In eq. (6.17), (£a, rja, £a) are the coordinates of the orth node in the
natural coordinate system. Figure 6.7 shows the brick element and its

(b)
FIG. 6.7 Three-dimensional brick element in natural and Cartesian coordinate systems.
 The Finite-Element Method—Part I 101
node-ordering convention. The velocity field can be expressed by

where (u^a\ u^, u^') is the velocity of the oth node. The coordinate
transformation is given by

where (xa, ya, za) are the coordinates of the o-th node.
The use of the brick element is shown in Chap. 14 for the analysis of
three-dimensional deformation.
Note on Notation
In expressing elemental functions, we treat each velocity component as a
scalar. Sometimes it is more convenient to express a velocity field in a
vector form as

where superscript T denotes transposition. The use of the transpose is only

for the convenience of expressing formulations in matrix forms.
The vectors u and v are defined by their components according to
OT ={«,(£,»»), «,(£»,)}
V T = {Mi1), <>, U?\ Uf>, ...} = {Vl,V2,V3,...}

respectively, for the case of two-dimensional deformation.

The shape functions are also arranged in the matrix form as

for the two-dimensional case.

6.4 Element Strain-Rate Matrix

In Chap. 4, the strain-rate components in Cartesian coordinate system are
defined by
102 Metal Forming and the Finite-Element Method

It was also shown, in Section 6.3, that the admissible velocity for all type
of elements can be expressed by

Substituting eq. (6.21) into eq. (6.20), we have

It is seen from eq. (6.22) that strain rate components can be evaluated if
dqa/dXi is known.
For the Cartesian coordinate system, we denote the coordinate jc, by
(x, y, z) for three-dimensional deformation, by (r, z, 0) for axisymmetric
deformation, and by (x, y) for two-dimensional deformation.
Let Xa, Ya, and ZK be denned as

Then the strain-rate components given by eq. (6.22) are expressed by

It is convenient to arrange the strain-rate components in a vector form.

For two-dimensional elements and axially symmetric deformation, the
strain-rate components can be written as

for plane-stresss
 The Finite-Element Method—Part I 103

for plane- strain

and

for axisymmetric
deformation

Substituting eqs. (6.23) into eqs. (6.24a, 6.24b and 6.24c), the strain-rate
vectors are represented in a unified form, as

In eq. (6.25), u^ and u2 correspond to ux and uy, respectively, for

two-dimensional deformation, and Pa is zero for plane-strain and the row
of £3 is deleted for plane-stress deformation. For the axially symmetric
case, MI and u2 represent ur and uz, respectively, and Pa becomes qalr.
Equation (6.25) can be written in the matrix form as

where B is called the strain-rate matrix and written as

The number of columns of the B matrix is determined by the number of

degrees of freedom allowed to the element.
The evaluation of the strain-rate matrix, or of Xa, Ya, and Za, requires
the differentiation of shape functions with respect to the global coordinate.
104 Metal Forming and the Finite-Element Method
Since the shape functions are expressed in the natural coordinate system, it
is necessary to express the global derivatives in terms of the derivatives
with respect to the natural coordinate. Consider a coordinate transforma-
tion given by eq. (6.19), where shape functions are denned in the natural
coordinate system. Then the derivatives of the shape functions with respect
to the natural coordinate system can be expressed, using the chain rule, as

where J is the Jacobian matrix of the coordinate transformation, given by

Then the derivatives in eq. (6.23a) can be obtained as

where J"1 is the inverse of J.

It may be mentioned that in plane-strain deformation, the strain rate e3
is not necessary, since it is always zero. However, it is convenient to
include £3 in eq. (6.25) so that the strain rate matrix B of the plane strain
deformation has the same form as that of the axisymmetric deformation as
shown in eq. (6.27).
Triangular Element Family
The strain-rate matrix of the triangular family of elements can be derived
by applying eq. (6.30) to the shape functions given in eqs. (6.10) and
(6.11). Since the area coordinates are not independent of each other, we
can eliminate L3 from the expressions of qa by using L3 = 1 — Ll — L2.
Equation (6.30) can be written for triangular element as

where |J| is the determinant of the Jacobian matrix and expressed by

The strain-rate matrix of a linear triangular element, shown in Fig. 6.4a,

can be obtained in a closed form by substituting q1 = L1, q2= L2,
 The Finite-Element Method—Part I 105

q3 = 1 — L! - L2, and is written as follows:

where

Note that |J| is twice of the area of the triangle. It may also be noted that
all the strain-rate components of a linear triangular element are constant
over one element, since Xa and Ya of eq. (6.33) are not functions of the
area coordinates.
The strain-rate matrix of the quadratic triangular element can be derived
in a similar manner to that shown above. However, the expressions for Xa
and Ya are much more complex, and it is easier to evaluate these
numerically following procedures similar to those used for the linear
element. It may be also mentioned that the strain-rate is not constant and
varies within the quadratic element.
Rectangular Element Family
For the rectangular family of elements, Xa and Ya in eq. (6.25) can be
written as

where |J| is the determinant of the Jacobian matrix of eq. (6.15) and is
expressed by

For a quadrilateral element with the numbering of nodes shown in Fig.

6.5c, Xa, Ya, and |J| can be expressed in the closed form as
106 Metal Forming and the Finite-Element Method
and |J| is expressed by

where xu =x,-Xj and yu = y,- y,.

For higher-order elements, it is easier to evaluate Xa and Ya numerically
according to eq. (6.30).
Three-Dimensional Brick Element
The strain-rate matrix, or Xa, Ya, and Za for a three-dimensional brick
element, can be derived by extending the derivation for the rectangular
family elements. Rewriting eq. (6.22) in the matrix form, we have

By using procedures similar to those of rectangular element family, we can

derive the differential operator as

with

where IJI, the determinant of the Jacobian matrix, is given by

The evaluation of Xa, Ya, and Za in eq. (6.36) can be made by using eqs.
 The Finite-Element Method—Part I 107

(6.37) and (6.38). The strain rate matrix B becomes

Matrices of Effective Strain-Rate and Volume Strain-Rate

In the finite-element formulation for the analysis of metal forming, the
effective strain-rate £ and the volumetric strain-rate EV are frequently used.
Therefore, it is necessary to express the effective strain-rate and volumetric
strain-rate in terms of the strain-rate matrix.
The effective strain-rate is defined in terms of strain-rate components,
according to eq. (4.39) in Chap. 4, as

or, in the matrix form

The diagonal matrix D has f and 3 as its components; corresponding to

normal strain-rate and engineering shear strain-rate, respectively. Substi-
tution of eq. (6.26) into eq. (6.40) gives

where P = BTDB.
The matrix D in eq. (6.40) takes different forms depending upon the
expression of the effective strain-rate, in terms of strain-rate components.
For example, the effective strain-rate in plane-stress problems is expressed
in a different form from that of plane-strain problems, although the
definition of the effective strain-rate is identical in both cases. The matrix
D written for plane-stress problems is not diagonal (Chap. 11). The
expression of the effective strain-rate also depends on the yield criterion.
Thus, the matrix D is different for anisotropic materials (Chap. 11) and for
porous materials (Chap. 13), and is described in corresponding chapters.
The volumetric strain-rate £„ is given by

and expressed by

with C, = Bu + B2i + B3I, where Bu is an element of the strain-rate matrix

B.
108 Metal Forming and the Finite-Element Method

6.5 Elemental Stiffness Equation

It can easily be seen from the way in which the element was introduced
that the global integrals over the whole workpiece stem from the assembly
of integrals over the local domain of disjoint finite elements. Therefore, it
is convenient to evaluate the stiffness matrix given by eq. (6.3) at the
elemental level, and to assemble into a global stiffness matrix.
Of the two variational formulations (5.21) and (5.22) in Chap. 5, we first
discuss the penalty function method, eq. (5.22). Denote the first, second,
and third terms (including signs) of eq. (5.22) dnD, dnp, 6j[Sp,
respectively. In metal forming, the boundary conditions along the die-
workpiece interface are mixed (see Section 5.4 of Chap. 5). Therefore,
along the interface Sc the treatment of traction depends on the friction
representation. This aspect of the surface integral terms is discussed in
Chap. 7.
Using the discrete representation of the quantities involved in dn that
are developed in Sections 6.3 and 6.4, we can express the integrals of dn in
terms of nodal-point velocities. Equation (6.3) then becomes

where

It should be noted that the term (—djtSp/dvj) is the applied nodal point
force and that dnDl' dv, + djtp/dv, is the reaction nodal point force.
The second derivatives of n are expressed as

Evaluating stiffness matrices at the elemental level from eqs. (6.43) and
(6.44), assembling them for the whole workpiece, we obtain a set of
simultaneous linear equations (6.4).
When effective strain-rate e approaches 0, or becomes less than a
preassigned value e0, we have, by eq. (5.31) in Chap. 5,
 The Finite-Element Method—Part I 109

where OO/EO = constant. The derivatives of nD can be expressed by

The penalty constant K and the limiting strain rate £0 are introduced
rather arbitrarily for computational convenience. However, proper choices
of these two constants are important in successful simulation of metal-
forming processes. A large value of K is, in general, preferred to keep the
volume strain-rate close to zero. However, too large a value of K may
cause difficulties in convergence, while too small a K results in unaccept-
ably large volumetric strain. Numerical tests show that an appropriate K
value can be estimated by restricting volumetric strain rate to 0.0001—0.001
times the average effective strain-rate.
The limiting strain-rate, e0, under which the material is considered to be
rigid, has been introduced to improve the numerical behavior of the
rigid-plastic formulation [7]. Too large a value of the limiting strain-rate
results in a solution in which the strain-rate of the rigid zone becomes
unacceptably large. On the other hand, if we choose too small a value of
limiting strain-rate, then the convergence of the Newton-Raphson method
deteriorates considerably. Numerical tests show that an optimum result
can be obtained by choosing the limiting strain rate as -^ of the average
effective strain-rate.
Equation (5.21) is the basic formulation using a Lagrange multiplier in
order to remove the incompressibility constraint. It replaces dnp in eq.
(5.22) by <5jrA = J <5(Ae w ) dV. The Lagrange multiplier A is treated as an
independent variable and a function of material points denned over the
workpiece. The Lagrange multipliers are defined at points where the
volume constancy is enforced and are introduced at the reduced integra-
tion points (see Section 7.1 of Chap. 7). For example, one variable is
assigned for each linear rectangular [8] and three-dimensional brick
element, and four A are needed for a quadratic rectangular element.
Since 6A is also arbitrary in eq. (5.21), we obtain a set of simultaneous
equations given by

The first equation of (6.47) is nonlinear and eq. (6.3) applies for
linearization. The additional terms necessary for evaluation of matrices in
the stiffness equation are

Then eqs. (6.3) and (6.47) can be arranged in the system of equations
110 Metal Forming and the Finite-Element Method

Each coefficient appearing in eqs. (6.48) can be evaluated at the elemental

level and assembled into the global stiffness equation.
It is difficult to assess which method, the penalty function method or the
Lagrange multiplier method, is better in the finite-element implementa-
tion. However, it can be mentioned that the Lagrange multiplier method
introduces a larger number of independent variables. Also, the diagonal
terms of the stiffness matrix corresponding to the A always become zero
and special attention is required during the assembly of the stiffness matrix
so that an equation corresponding to any A does not become the first one
in the stiffness equations after applying the boundary conditions.

References
1. Courant, R., (1943), "Variational Methods for the Solution of Problems of
Equilibrium and Vibrations," Bull. Amer. Math. Soc., Vol. 49, p. 1.
2. Clough, R. W., (1960), "The Finite Element Method in Plane Stress Analysis,"
/. Struct. Div., ASCE, Proc. 2nd Conf. Electronic Computation, p. 345.
3. Bathe, K. J., and Wilson, E. L., (1976), "Numerical Methods in Finite-Element
Analysis," Prentice-Hall, Englewood Cliffs, NJ.
4. Oden, J. T., (1972), "Finite Element of Nonlinear Continua," McGraw-Hill,
New York.
5. Desai, C. S., and Abel, J. F., (1972), "Introduction to the Finite Element
Method," Van Nostrand Reinhold, New York.
6. Hildebrand, F. B., (1974), "Introduction to Numerical Analysis," 2d Edition,
McGraw-Hill, New York.
7. Chen, C. C., Oh, S. I., and Kobayashi, S., (1979), "Ductile Fracture in
Axisymmetric Extrusion and Drawing, Part I: Deformation Mechanics of
Extrusion and Drawing," Trans. ASME, J. Engr. Ind., Vol. 101, p. 23.
8. Lee, C. H., and Kobayashi, S., (1973), "New Solutions to Rigid-Plastic
Deformation Problems Using a Matrix Method," Trans. ASME, J. Engr. Ind.,
Vol. 95, p. 865.
 7
THE FINITE-ELEMENT METHOD—PART II

7.1 Numerical Integrations

Numerical integration is an important part of the finite-element technique.
As seen in Section 6.5 of Chap. 6, volume integrations as well as surface
integrations should be carried out in order to represent the elemental
stiffness equations in a simple matrix form. In deriving the variational
principle, it is implicitly assumed that these integrations are exact.
However, exact integrations of the terms included in the element matrices
are not always possible. In the finite-element method, further approxima-
tions are made in the procedure for integration, which is summarized in
this section.
Integration Formula
Numerical integration requires, in general, that the integrand be evaluated
at a finite number of points, called Integration points, within the integra-
tion limits. The number of integration points can be reduced, while
achieving the same degree of accuracy, by determining the locations of
integration points selectively. In evaluating integration in the stiffness
matrices, it is necessary to use an integration formula that requires the
least number of integrand evaluations. Since the Gaussian quadrature is
known to require the minimum number of integration points, we use the
Gaussian quadrature formula almost exclusively to carry out the numerical
integrations.
Consider a one-dimensional scalar function/(:c) defined in — l < x < l .
In the Gaussian quadrature, the integration of f ( x ) can be evaluated [1] by

where xt is the coordinate of an integration point, w, is a weight factor, and

the summation is carried out over n (order of integration) integration
points. Assuming f(x) as a polynomial function, the formula given by eq.
(7.1) provides the exact integration of polynomial of degree (2n — 1).
The optimum integration points and corresponding weight coefficients
are given in Table 7.1 [1]. The integration of higher-degree polynomial
functions can be found elsewhere [1,2].
Ill
112 Metal Forming and the Finite-Element Method

TABLE 7.1 Integration Points and Weight Factors of the

Gaussian Quadrature Formula [1]

n X, w,

1 0 2.000000000000000
2 ±0.577350269189626 1.000000000000000
3 ±0.774 596 669 241 483 0.555 555 555 555 556
0.000000000000000 0.888888888888889
4 ±0.861 136311594053 0.347854845137454
±0.339981043584856 0.652 145 154 862 546

n = number of integration points.

x, = coordinate of integration points.
w, = integration weight factors.

The integration of a scalar function /(£, 77) defined by the natural

coordinates (—1 < £ < 1, — 1 < 77 < 1) in the two-dimensional space can be
obtained by applying the integration formula successively, namely,

and, in the three-dimensional space (-1 < g < 1, - l < j j < l , -!<£<!);

For a function f ( x , y) defined over an isoparametric element, integration

off(x, y) can be evaluated as

where |/(£, T])\ is the determinant of the Jacobian of the coordinate

transformation matrix and given by eq. (6.34b).
For a function, f ( r , z), defined over an isoparametric axisymmetric
element, the volume integration over 1 radian is evaluated by

where /"(£/, f]j) is the radial position of the integration point.

 The Finite-Element Method—Part II 113

The integration over the triangular family elements can be evaluated as

where (Lu, L2I, L3f) is the area coordinate of an integration point [3].
The integration formula given by eq. (7.6) gives the exact integration of
a polynomial function of degree n. The integration points and weight
factors of the Gaussian quadrature formula for triangular elements are
listed in Table 7.2 [1].
Application of Integration Formulas
The use of Gaussian quadrature and isoparametric elements in FEM
formulation will introduce errors. Usually, it is true that a higher-order
integration is suggested in order to obtain accurate evaluations of
integrands. But a large portion of the computer execution time is spent in
performing this numerical integration. Therefore, a proper choice should
be the lowest possible order of integration that does not introduce much
error into the results.
The minimum order of integration that ensures convergence has been
established for different types of elements. For a linear rectangular
element, 2 x 2 integration points are sufficient to guarantee convergence.
For a three-dimensional brick element, it is necessary to use 2 x 2 x 2
integration points. The required integration order for quadratic elements is
3 x 3 . For the triangular element family, one-point integration is sufficient
for a linear element, while three-point integration is required for a
quadratic element. It is also known that increasing the order of integration
above the required minimum does not necessarily improve the solution
accuracy. In determining the order of integration, possible overconstraints
imposed by the FEM formulation must be also considered. For example,
the incompressibility requirement cannot be satisfied at all points in a

TABLE 7.2 Numerical Integration Formulas for Triangular Elements [1]

Number of
Integration Integration Coordinates Weight
Points Order (^,, L2, L3) Factors
1 Linear i i i 1
3> 3> 3

3 Quadratic U,o i
i
o,U
I n 1
3
1
2>«> 2 3
1 1 1 27
4 Cubic 3> 3' 3 48
25
0.6,0.2,0.2 48
25
0.2,0.6,0.2 48
25
0.2,0.2,0.6 48
114 Metal Forming and the Finite-Element Method
rectangular linear element except for uniform deformation [4]. In order to
relieve this overconstraint, one-point integration is used for the volumetric
strain-rate term in a linear rectangular element. This is known as the
reduced integration scheme. The reduced integration in a linear rectangu-
lar element, in effect, imposes the volume constancy averaged over that
element.
For higher-order elements, it is found that the integration order of the
dilatational contribution may be reduced by 1 to obtain the best results.
That is, the volumetric strain-rate term is integrated at 2x2 integration
points for a quadratic rectangular element and at one point for a quadratic
triangular element. For the linear triangular element, which gives constant
strain-rate over the element, the reduced integration technique cannot be
applied. This is why the linear triangular element is not used for the case in
which volume constancy is required. When the linear triangular element
with the regular integration order is used for forming simulation, the
model tries to achieve the volume constancy averaged over several
elements. It has been shown that if four linear triangular elements are
arranged to form a rectangle, then the volume constancy is satisfied over
the four elements within the rectangle [4]. It may be also mentioned that
the number of Lagrange multipliers necessary to enforce volume constancy
should be the same as the number of the reduced integration points of the
element used.
The Gaussian integration points are the natural locations to evaluate the
strain-rate and stress within the element, since all the necessary informa-
tion to calculate the strain-rate and stress are already evaluated at these
points. It is shown [5], however, that the calculated stress values at the
regular integration points deviate considerably from the actual ones when
volume constancy is enforced. On the other hand, stress and strain-rates
evaluated at the reduced integration points are known to be more
accurate. Consequently, throughout this book, reduced integration points
are used for stress and strain-rate evaluations.
The integrations of the boundary terms can also be evaluated by the
Gaussian quadrature. It is shown that a large number of integration points
are necessary when the integrand contains higher-order polynomials. The
friction term given by eq. (5.30) cannot be approximated by a low-order
polynomial over the usual range of the sliding velocity [6] and it is
necessary to use a higher-order integration formula. It may sometimes be
more convenient to use Simpson's formula to calculate the frictional
energy rate along the tool-workpiece interface [7], since this formula can
be applied to represent the local behavior of the integrand more accurately
by subdividing the integration range. Simpson's formula is given by

where a and b are the limits of the variable of each subdivision.

 The Finite-Element Method—Part II 115
7.2 Assemblage and Linear Matrix Solver
Element Assemblage
The concept of assemblage can be simply explained. When the element
equations are assembled, the global constraints are that (1) the global
nodal-point velocities are identical to the nodal-point values of the
elements surrounding that nodal point, and (2) the global nodal-point
forces are the sums of the elemental nodal-point forces. However, the
assembly procedure requires extensive bookkeeping in computer im-
plementation. In order to handle this extensive bookkeeping, it is
customary to assign sequential numbers to nodes and elements, in order to
identify them in the finite-element method model. Figure 7.1 shows an
example of node and element numbering. Here, nodes are numbered 1
through 30 and elements are numbered 1 through 19. To further identify
the nodes associated with each element, we define an array called the
element connectivity for each element. For example, the connectivity of
element 8 is defined as (16,10,11,17), where the nodes identified by
numbers define the element. It is also customary to use a predefined
ordering convention for a type of element in defining the connectivity.
Ordering conventions for different types of elements are indicated in
Section 6.3.
The velocity field that was constructed in Section 6.3 is defined in such a
way that the change of a velocity component at a node influences the
velocity field only over the elements that share the node. For example, if a
velocity component of node 10 in Fig. 7.1 changes, then the velocity field is
affected only over the elements 4, 5, 7, and 8, while it is unaffected over all
other elements. According to eq. (6.3) in Chap. 6, we have the global
stiffness equation corresponding to node 10 as

FIG. 7.1 Example of node and element numbering.

116 Metal Forming and the Finite-Element Method

Since the stiffness matrix has been evaluated at the elemental level,
evaluated elemental stiffness matrices should be assembled as shown in eq.
(7.8).
Often the assembled stiffness equations are numbered in the order of
nodal-point numbering and the elemental stiffness equations are numbered
in the order of element connectivity, so that the connectivity of each
element can be used to correlate the corresponding equations and
variables in global and elemental stiffness equations.
It is also seen from eq. (7.8) that the stiffness equation corresponding to
node 10 contains the variables of nodes associated with the elements
surrounding node 10. That is, the equation contains the velocity com-
ponents of nodes 5, 6, 7, 9, 10, 11, 15, 16, and 17. Therefore, the global
stiffness matrix is a sparse matrix because of the limited range of influence
of a node velocity on the admissible velocity field. The sparseness of the
global stiffness matrix can be utilized to reduce the computational effort in
solving the linear equations. Thus, the stiffness matrix is arranged in a
banded matrix form, as shown in Fig. 7.2, by proper numbering of nodes.
Gaussian Elimination
Completing assemblage of elements, the global stiffness equation can be
written in the form of eq. (6.4) in Chap. 6, namely,

The most common procedure of solving eq. (7.9) is Gaussian elimina-

tion. The reader may refer to any book on linear algebra or the
finite-element method for the Gaussian elimination technique [for ex-
ample, 8, 9].
In the solution process, solving linear matrix equations takes a substan-
tial amount of computer time. It is therefore necessary to program a linear
equation solver in the most efficient way. Depending upon the computer
programming technique, the linear matrix solver may take several different

FIG. 7.2 Banded stiffness matrix.

 The Finite-Element Method—Part II 117

forms. The simplest way is to store the stiffness matrix in a banded matrix
form, and to apply Gaussian elimination over the maximum band width. In
a banded matrix solver, the same number of matrix elements is stored for
each equation. In the skyline solver, K is stored column-wise, from the
diagonal matrix element to the last nonzero element in the column [8].
The number of operations necessary to solve a linear matrix can be
estimated by ^nm\, where n is the total number of equations and mk is the
half band width of the matrix. Here, one operation consists of one
multiplication (or division), which is almost always followed by an
addition. The estimated number of operations shows that the computer
time necessary to solve the matrix equation is proportional to the square of
the half band width. It is therefore a must to number the nodes in such a
way that the assembled stiffness equation has minimum band width.
The variables of the stiffness matrix can be "eliminated" when the
stiffness matrix is assembled partially. That is, when the stiffness equation
for node / is completely assembled, then the corresponding variable can be
expressed by other variables, although the assembly is not complete. The
frontal solution technique is a linear equation solver based on the
elimination of variables during assembly. The frontal solution technique
[10] is known to be advantageous when the stiffness matrix with minimum
band width is still sparse. Such cases arise for the three-dimensional
finite-element method. In the frontal method, element numbering, but not
node numbering, influences the matrix solution efficiency; therefore, new
nodes can be added without renumbering the nodes.

7.3 Boundary Conditions

Since the boundary condition along the tool-workpiece interface Sc is
mixed, it is convenient to write the boundary surface S in three distinct
parts

Imposition of the traction boundary conditions on SF is straightforward.

Recalling the boundary integral 6nSp or the first derivative of nSp with
respect to a node velocity component, the traction boundary condition is
imposed in the form of nodal-point forces. It should be mentioned that the
same nodal-point force can be obtained with different traction
distributions.
The velocity boundary conditions on Su are essential boundary condi-
tions. In the finite-element discretization, the velocity boundary condition
is enforced only at nodes on Su, and the velocity along the element-side is
determined automatically in terms of velocities of nodes and element
shape functions. For the node at which the velocity is defined, the velocity
correction Au M is zero. Consequently, the corresponding stiffness equation
should be removed. The simplest way to implement this procedure is to
replace the corresponding rows and columns by zeros and to set the
118 Metal Forming and the Finite-Element Method
diagonal term to 1, as shown below:

On the surface Sc, the traction is prescribed in the tangential direction

and the velocity is prescribed in the normal direction to the interface.
When the interface direction is inclined with respect to the global
coordinate axis, the coordinate transformation of the stiffness matrix upon
the inclined direction is necessary in order to impose mixed boundary
conditions.
Consider a velocity vector v in the global coordinate system and the
corresponding vector v' in the inclined boundary coordinates. Then the
vector is transformed from the global to the local coordinate system by

where T is the coordinate transform matrix.

Similarly, the nodal-point force vector f is transformed to f according to

In the two-dimensional coordinate system, the transformation matrix of

node / is written as

where d is measured from the x axis in the global coordinate system to the
x' axis of the local coordinate system in counterclockwise direction. In the
three-dimensional coordinate system, the transformation matrix is ex-
pressed by

where T^ is a direction cosine between the coordinate axis of xt and x'j. The
transformation matrix for all nodes on the surface Sc can be constructed as

and the stiffness equation (7.9) is transformed to

 The Finite-Element Method—Part II 119
The velocity boundary condition at the tool-workpiece interface is given
by

where UD is the tool velocity and n is the unit normal to the interface
surface.
In the direction of the relative sliding between the die and the
workpiece, the frictional stress /, is prescribed as the traction boundary
condition. The frictional stress is usually represented according to the
Coulomb law or as a constant frictional stress, as discussed in Chap. 3. The
friction representation by a constant friction factor m (eq. (3.2) in Chap. 3)
is approximated by eq. (5.30) in Chap. 5, in order to deal with
neutral-point problems in metal forming.
Equation (5.30) expresses that the magnitude of the frictional stress is
dependent on the magnitude of the relative sliding and that their directions
are opposite to each other. Then, the relationship can be written as

The approximation of the frictional stress by the arctangent function of

the relative sliding velocity eliminates the sudden change of direction of
the frictional stress mk at the neutral point. Figure 7.3 shows the
arctangent function of the relative sliding as an approximate representation
of the constant frictional stress. The figure shows that the frictional stress
approaches mk asymptotically as the relative sliding velocity us increases.
However, the frictional stress, fs approximated by eq. (7.12) deviates
considerably from the value of mk as us approaches zero. It may be noted
that the value of u0 was introduced arbitrarily for performing numerical

RELATIVE SLIDING VELOCITY U s / D I E VELOCITY

FIG. 7.3 The arctangent function for the friction representation.
120 Metal Forming and the Finite-Element Method

calculations and that the choice of u0 could have a significant influence on

the reliability of the solution. It is seen from the figure that the ratio us/u0
should be equal to or larger than 10 in order to attain the friction value
within 9% of the one originally intended. On the other hand, if we choose
the ratio too large, then the sudden change of the frictional stress near the
neutral point can cause difficulties in numerical calculation [11]. Since the
order of magnitude of us is 0.1 (with the unit die velocity), a recommended
value for «0 is 10~3 ~ 10~4-
For the discretization, consider a die and an element that is in contact
with the die, as shown in Fig. 7.4. The boundary condition normal to the
contact surface is enforced at the contact nodes. Also, the relative sliding
velocity at the nodes vs can be evaluated. It should be noted, however,
that the element-side cannot be made to conform to the die surface.
However, it may be assumed that the relative sliding velocity us can be
approximated in terms of nodal-point values vsa by using a shape function
of the elements as

where the subscript a denotes the value at aih node.

In deriving the stiffness equation, dn (see eq. (6.43) in Chap. 6) should
include the term dnSc, and the final form of the stiffness equation should
contain the terms

FIG. 7.4 An element in contact with the die.

 The Finite-Element Method—Part II 121

and

in addition to those given in eqs. (6.43) and (6.44) in Chap. 6.

The finite-element method approximation of the boundary conditions
introduces errors to the solution of the boundary-value problem. Note that
the surface integration in eqs. (7.14) and (7.15) is carried out over the
element surface rather than the actual die surface. When linear elements
are used with a curved die, the interface area represented by elements is
always smaller than the actual interface area, and the effect of friction in
the analysis is always smaller than the actual. For deformation processes
that are sensitive to friction, this type of error could be quite serious.
Also, the velocity boundary condition imposed by the finite-element
model can be considerably different from that of the actual problem. In
expressing the sliding velocity by eq. (7.13), it was assumed that the
mismatch angle between the element-side and the tangent direction of the
die at contact node (angle 6 in Fig. 7.4) is small. When this angle of
mismatch is large, the deformation mode is not modeled correctly. The
errors resulting from the boundary conditions imposed by the finite-
element method can be minimized by increasing the number of elements at
the boundary.

7.4 Direct Iteration Method

The convergence of the Newton-Raphson method, which is described in
Section 6.2 of Chap. 6, is usually very good, provided that the "initial
guess" is in the vicinity of the solution. Also, the Newton-Raphson
method, in general, takes a small number of iterations if it is convergent.
However, it may not be easy to obtain a good initial guess velocity when
the deformation process is complex.
Another technique for solving a nonlinear equation is the direct iteration
method [6,12]. In the direct iteration method, it is assumed that the
constitutive equation ((5.1c) in Chap. 5) is linear during each iteration.
Then a/e in dnD/dv, in eq. (6.43) is assumed to be constant during each
iteration. The nonlinear friction term expressed by eq. (7.14) is also
approximated by a linear relationship (viscous friction coefficient) between
the frictional stress and the relative sliding velocity. Then the stiffness
equation resulting from dn = 0 becomes linear.
The computational procedures of the direct iteration method are as
follows:
1. Assign an assumed strain-rate e for each element. If a previous
solution or iteration is not available, assign a constant average
strain-rate to each element. If a previous solution or iteration is
available, then use the strain-rate obtained previously for each
element.
122 Metal Forming and the Finite-Element Method

2. Assign an assumed sliding velocity to each element-side that is in

contact with a die. If a previous solution or iteration is not available,
assign a constant average sliding velocity to all relevant element-
sides. If it is available, use a sliding velocity that is obtained from
previous solution or iteration.
3. Calculate alt at each integration point of the element, where a is
evaluated for e assigned in step 1.
4. Calculate the viscous friction coefficient for each die contact side
from the linear relationship between frictional stress and the relative
sliding velocity.
5. Evaluate the stiffness matrix and obtain a velocity solution.
6. Calculate the strain-rate for each element by using the velocity
solution of step 5.
7. Calculate the sliding velocity for each die contact element-side.
8. Check whether solution converges, using convergence criteria.
9. If the solution does not converge, go to step 3.
Figure 7.5 shows schematically the direct iteration solution procedure. It
may be noted, from steps 1 and 2, that the direct iteration method does
not require any "initial guess velocity." For metal-forming applications,
the direct iteration method converges fast towards the solution during the
earlier stages of iteration. However, as the solution point is approached,
the convergence becomes very slow. It seems that the best computational
efficiency can be obtained by using (1) the direct iteration method for
generating the initial guess and for cases where the Newton-Raphson
method does not converge, and (2) the Newton-Raphson method for all
other cases.

NODE VELOCITY

FIG. 7.5 Schematic representation of the direct iteration method.

 The Finite-Element Method—Part II 123

7.5 Time-Increment and Geometry Updating

When the velocity solution is obtained, then the deformed geometry of the
workpiece, in the case of two dimensions, for example, can be obtained by
updating the coordinates of the nodes (Lagrangian mesh system) by

where (x,, y/) are the coordinates of node /, t0 is the time at current
configuration, and Af is the time-increment. The strain is updated in a
similar manner from the strain-rate solution. In general, the time
increment At can be determined by considering several factors, such as the
time (A£die) necessary for a next free node to contact the die surface, a
desired maximum strain-increment (Afstrain), and a maximum allowable
time-increment (A?a). The actual time-increment is determined by taking
the minimum of A£die, Afstrain, and Af a . The time necessary for a next free
node to contact the die can be determined by calculating these time
increments for all free nodes and choosing the minimum time-increment.
The time-increment required to limit the maximum strain-increment can
be readily obtained from strain-rate solutions. The maximum allowable
time-increment is given rather arbitrarily. However, consideration of the
error in the volume constancy is a factor for determining its magnitude.
During the time increment Af, elements lose volume after the geometry
is updated. Consider a two-dimensional element, as shown in Fig. 7.6,
where the element (1234) with a width of W and a height H is deformed to
the shape (12'3'4') after a time increment At. The volume constancy
requires that

FIG. 7.6 Two-dimensional uniform deformation of a rectangular element.

124 Metal Forming and the Finite-Element Method
where the dot represents the time derivative. After the time increment Af,
the volume change can be calculated by

or

where V and AF are the volume of the element and the amount of volume
change, respectively. Equation (7.18) suggests that updating by eq. (7.16)
always results in a volume loss during a time-increment and that the
volume loss rate is proportional to the square of (A///H). Assuming that
(A/////) is constant at all increments, we have, after n increments,

The total volume loss given by eq. (7.19) is plotted in Fig. 7.7 as a function
of total reduction in height and of &.H/H. It is seen from the figure that
0.7% volume is lost at 50% reduction in height if (A/f///) is 0.01, and the
volume loss reaches 2.3% at 90% reduction in height with (AH/H) of
0.01. Therefore, it is necessary to use (A/////) less than 0.01 in order to
maintain less than 2% volume loss at 90% reduction in height. It is also
seen that volume loss increases considerably with larger step size.
In case of axisymmetric deformation, the volume of an element (solid
cylinder) is represented by V — nR2H. The volume loss rate can be

FIG. 7.7 Volume loss during plane-strain FEM simulation as function of deformation for
various step sizes of A/////.
 The Finite-Element Method—Part II 125

FIG. 7.8 Volume loss during plane-strain FEM simulation as function of deformation for
various step sizes of A///W 0 .

calculated by

Equation (7.20) shows that the volume loss in axisymmetric deformation is

about 25% less than that in plane-strain deformation.
Taking the step size proportional to the current height H, a total of 225
steps are required to reach 90% reduction in height with (A///H) = 0.01.
When the step size is controlled by a constant A////f,,, where H0 is the
initial height, the volume loss is accelerated as deformation proceeds. As
shown in Fig. 7.8, the effective (A///H) becomes larger as H is reduced.
The volume loss rate, due to geometry update with finite time-increment
estimated above, is based on uniform deformation. In the simulation of
practical forming processes, the amount of volume loss will vary within the
workpiece. However, the estimated volume loss rate based on the uniform
deformation is a good reference for determining the maximum allowable
time-increment for simulation of practical forming processes.
Maintaining the volume loss within certain small percentage of the total
deforming volume is a serious consideration in the prediction of proper die
filling and defect formation, which are important in process design.
 126 Metal Forming and the Finite-Element Method

7.6 Rezoning
In practical forming processes, deformation is usually very large, and it is
not uncommon to encounter effective strain with magnitudes of 2 or more.
Moreover, the relative motion between the die surface and the deforming
material is also large. Such large deformations and displacements cause the
following computational problems during FEM simulation with a Lagran-
gian mesh:

• Difficulties in incorporating the die boundary shape into the FEM

mesh, with increasing relative displacement between die and
workpiece
• Difficulties in accommodating the considerable change of deformation
mode with one mesh system
• Formation of unacceptable element shapes with negative Jacobian due
to large local deformation
In order to overcome these difficulties, it is necessary periodically to
redefine the mesh system [13,14]. The rezoning consists of two proce-
dures. One is the assignment of a new mesh system to the workpiece and
the other is the transfer of information from the old to the new mesh
through interpolation. Generation of the new mesh is essentially the same
as the initial mesh generation performed by using a commercially available
mesh generator. The field variables, which depend on deformation history,
are effective strains and temperatures, and they must be interpolated on
the new mesh.
Temperatures are given at nodal points in the finite-element method, as
discussed in Chapter 12 for thermo-viscoplastic analysis. Thus, the
temperature distribution is expressed by using element shape functions
over the whole workpiece. Interpolation from the old mesh to the new
mesh is done simply by evaluating the temperature at the new node
locations.
Interpolation of effective strain requires an additional step, since
effective strains are given at the reduced integration point of each element.
Therefore, before interpolation it is necessary to obtain the effective strain
values at nodal points from the values given at the regular integration
points. Thus, effective strain distribution can be expressed by nodal-point
values and element shape functions.
For a linear rectangular element, one value of effective strain is given at
the center of each element. Among the various methods tested [13], it
appears that the area-weighted average method is the most convenient and
provides sufficient accuracy for remeshing in metal-forming simulations. In
the area-weighted averaging scheme, the nodal value is determined on the
basis of the average of the adjacent element values weighted by the
associated element size. Consider a node N surrounded by adjacent
elements, as shown in Fig. 7.9. The nodal value of the effective strain at
 The Finite-Element Method—Part II 127

FIG. 7.9 Node N surrounded by adjacent elements for area-weighted average.

node N can be obtained by

where EN is the effective strain value at node N and £7 is the effective strain
value at the center of element / that surrounds the node N. AjN is the area
contribution of ;th element to node N and is denned by

where qN is the element shape function of element ;' at node N. The

summation of eq. (7.21) is done over all the elements that surrounds the
node N. Once the effective strains are determined at all nodes, the strain
distribution over each element can be denned by

where qa is the element shape function. The interpolation from the old
mesh to the new mesh is done by evaluating the effective strain value at
the integration point locations of the new mesh system.
As can be expected, the area-weighted average method, in general,
provides very accurate interpolation results for internal nodes when the
trend of the field variable distribution is relatively well defined by the
values at surrounding elements. The method often fails at the workpiece
boundary when the field variable has large gradients, since the boundary
nodes do not have a sufficient number of surrounding elements. However,
for all practical purposes, the method provides sufficient accuracy for
metal-forming simulations. Other methods that improve the accuracy of
interpolation at the workpiece boundary have been suggested and are
discussed in Reference [13].
The rezoning procedure has been applied to simulations of various
 128 Metal Forming and the Finite-Element Method

Rnoiusmm
FIG. 7.10 FEM mesh assignment for rezoning.

metal-forming processes, and an example of a disk forging process with

flash [15] is shown in Fig. 7.10. The figure shows the distorted mesh of the
disk forging process (upper half) at 65% reduction in height. It shows that
the elements are highly distorted near the die corner radius between the
bore and flange joint. Such a highly distorted grid requires remeshing.
However, in this example, the remeshing procedure was necessary not
only because of the large element distortion, but also because of the lack

FIG. 7.11 Example of strain contours before and after rezoning.

 The Finite-Element Method—Part II 129

of degrees of freedom near the flash. The new mesh is also shown in Fig.
7.10 (bottom half). Note that the new mesh system allows far more
degrees of freedom near the flash. Interpolation of effective strain was
carried out by the area-weighted average method and the results are shown
in Fig. 7.11. It is seen from the figure that the strain distributions are
almost identical before and after rezoning.

7.7 Concluding Remarks

In Chaps. 6 and 7', discretization of the basic equations for the finite-
element method developed in Chap. 5 was formulated. The solution
techniques for the finite-element formulations were explained, and impor-
tant aspects of numerical procedures for metal-forming process analysis
were presented in some detail.
We did not discuss the programming of the method, because it is outside
of the scope of this book. However, since it is an important part of the
finite-element method, a simple FEM code with the program description is
given in the Appendix.
In the following chapters, in which applications of FEM to various
metal-forming processes are discussed, some finite-element formulations
are recapitulated and explained in further detail.

References
1. Zienkiewicz, O. C., (1977), "The Finite Element Method," 3d Edition,
McGraw-Hill, New York.
2. Strang, G., and Fix, G. J., (1973), "An Analysis of the Finite Element
Method," Prentice-Hall, Englewood Cliffs, NJ.
3. Cowper, G. R., (1973), "Gaussian Quadrature Formulas for Triangles," Int. J.
Num. Meth. Engr., Vol. 7, p. 405.
4. Nagtegaal, J. C., Parks, D. M., and Rice, J. C., (1974), "On Numerically
Accurate Finite Element Solutions in the Fully Plastic Range," Comput. Meth.
Appl. Mech. Eng., Vol. 4, p. 153.
5. Zienkiewicz, O. C., and Hinton, E., (1976), "Reduced Integration, Function
Smoothing and Non-Conformity in Finite Element Analysis," /. Franklin Inst.,
Vol. 302, p. 443.
6. Oh, S. I., (1982), "Finite Element Analysis of Metal Forming Problems with
Arbitrarily Shaped Dies," Int. J. Mech. Sci., Vol. 24, p. 479.
7. Lee, G. J., and Kobayashi, S., (1982), "Spread Analysis in Rolling by the
Rigid-Plastic Finite Element Method," Numerical Method in Industrial Form-
ing Processes, p. 777.
8. Bathe, K. J., (1982), "Finite Element Procedures in Engineering Analysis,"
Prentice-Hall, Englewood Cliffs, NJ.
9. Noble, B., (1969), "Applied Linear Algebra," Prentice-Hall, Englewood
Cliffs, NJ.
10. Irons, B. M., (1970), "A Frontal Solution Program for Finite Element
Analysis," Int. J. Num. Meth. Engr., Vol. 2, p. 5.
11. Chen, C. C., and Kobayashi, S., (1978), "Rigid-plastic Finite Element Method
Analysis of Ring Compression," Application of Numerical Method to Forming
Processes, ASME, AMD, Vol. 28, p. 163.
130 Metal Forming and the Finite-Element Method

12. Lyness, J. F., Owen, D. R. J., and Zienkiewicz, O. C., (1974), "Finite
Element Analysis of Steady Flow of Non-Newtonian Fluid Through Parallel
Sided Conduits," Int. Symp. on Finite Element Method in Flow Problems,
Swansea, p. 489.
13. Oh, S. I., Tang, J. P., and Badawy, A., (1984), "Finite Element Mesh
Rezoning and its Applications to Metal Forming Analysis," Proc. of 1st ICTP
Conference, Tokyo, p. 1051.
14. Badawy, A., Oh, S. I., and Allan, T., (1983), "A Remeshing Technique for
the FEM Simulation of Metal Forming Processes," Proc. Int. Computer Engr.
Conf., ASME, Chicago, IL., p. 143.
15. Ficke, J. A., Oh, S. I., and Malas, J., (1984), "FEM Simulation of Closed Die
Forging of Titanium Disk using ALPID," Proc. of NAMRC XII, Houghton,
ML, p. 166.
 8
PLANE-STRAIN PROBLEMS

8.1 Introduction
This chapter is concerned with the formulations and solutions for plane
plastic flow. In plane plastic flow, velocities of all points occur in planes
parallel to a certain plane, say the (jc, y) plane, and are independent of the
distance from that plane. The Cartesian components of the velocity vector
u are ux(x, y), uy(x, y), and uz = 0.
For analyzing the deformation of rigid-perfectly plastic and rate-
insensitive materials, a mathematically sound slip-line field theory was
established (see the books on metal forming listed in Chap. 1). The
solution techniques have been well developed, and the collection of
slip-line solutions now available is large [1]. Although these slip-line
solutions provide valuable insight into deformation modes and forming
loads, slip-line field analysis becomes unwieldy for nonsteady-state prob-
lems where the field has to be updated as deformation proceeds to account
for changes in material boundaries. Furthermore, the neglect of work-
hardening, strain-rate, and temperature effects is inappropriate for certain
types of problems. Many investigators, notably Oxley and his co-workers,
have attempted to account for some of these effects in the construction of
slip-line fields. However, by so doing, the problem becomes analytically
difficult, and recourse is made to experimental determination of velocity
fields, similarly to the visioplasticity method. Some of this work is
summarized in Reference [2]. The applications of the finite-element
method are particularly effective to the problems for which the slip-line
solutions are difficult to obtain.

8.2 Finite-Element Formulation

The finite-element formulation specific to plane flow is recapitulated here.
The variational form as a basis for finite-element discretization is given by

for the Lagrange multiplier method, where A is the Lagrange multiplier, or

131
132 Metal Forming and the Finite-Element Method

FIG. 8.1 Quadrilateral element and natural coordinate system.

for the penalty function constraints, where K is a large positive constant.

In equations (8.la) and (8.1b), EV is the volumetric strain-rate, and the
effective strain-rate t includes only nonzero strain-rate components,
namely, ex, £y, and yxy.
A quadrilateral element (linear isoparametric element) is used for
discretization and is shown in Fig. 8.1. The velocity field u is approximated
by shape functions in terms of nodal point velocity values as

where

with

according to eq. (6.13) in Chap. 6.

For convenience of combined programming of plane-strain and axisym-
metric problems (see eq. (6.24b) in Chap. 6), the strain-rate vector is
formed according to

then
 Plane-Strain Problems 133

where Xa and Ya are expressed by eq. (6.35) for the node numbering
shown in Fig. 8.1.
The volumetric strain-rate ev is expressed by

where

The effective strain-rate e, in a discrete form, is denned by

by putting P = BTDB.
The matrix D in eq. (8.5) is given by

Then the basic equation, for example, eq. (8.1b) is discretized and a set of
nonlinear simultaneous equations (stiffness equations) is obtained from the
arbitrariness of <5v as

with P, C, and N defined above. For the solution procedure, eq. (8.6) is
linearized according to eq. (6.3) in Chap. 6.

8.3 Closed-Die Forging with Flash [11]

Forging in closed dies is an important operation in shaping metals into
useful objects. The process involves the compression of a billet, usually
between two dies. The metal flow is restricted to fill the closed die cavity,
and excess material flows through the gap between the closing dies and
forms a flash. In order to completely fill the die cavity, more material than
the actual volume of the finish forged part is needed. The flash due to this
excess material is subsequently trimmed from the forging.
At the finishing stage of a closed-die forging operation, the actual plastic
deformations are mainly at the flash portion, with a minor amount of flow
of the forging material to fill the die cavity. Solutions at this stage,
assuming that the cavity is filled and that material flows into the flash, were
based on the slab method [3,4,5]. The upper-bound method was also
applied to forging problems by several researchers [6, 7, 8]. In addition to
the mathematical analysis, experimental studies can be found in several
references. Kasuga et al. [9] observed the material flow in plane-strain and
134 Metal Forming and the Finite-Element Method

FIG. 8.2 Schematic diagram of plane-strain closed-die forging.

axisymmetric rib-web type forgings. Lyapunov and Kobayashi [10] con-

ducted experiments to examine the metal flow in plane-strain closed-die
forgings.
A schematic diagram of the apparatus used in the experiment by
Lyapunov and Kobayashi is shown in Fig. 8.2. In this study, a lead
specimen 50.8mm (2.0 in.) in height and width, with a square cross-
section, is placed inside the cavity between the upper and lower dies. The
cross-section of the cavity, when two dies are closed, consists of a flange
with a tapered rectangular cross-section and a tapered shaft extended in
the axial direction.
The results obtained from this experiment, including grid patterns,
velocity fields, load-displacement curves, flash dimensions, and the height
variations of the specimen, were compared with the results of rigid-plastic
finite-element analysis [11].
The specimen used in the analyses is made of pure lead, which is
characterized by a rigid-perfectly plastic (i.e., nonwork-hardening) mate-
rial behavior with the constant flow stress Y0 assumed to be 2500 psi
(17.236 MN/m2). Two extreme cases of friction conditions were con-
sidered: the first assumed perfect lubrication at the interfaces of the die
and workpiece so that no frictional force exists; the second assumed a
sticking condition that implies that once the material touches the die
surface the material is completely adhered to the die.
The initial mesh system consists of 208 elements interconnected at 238
nodal points. A very fine mesh was used for elements located below the
fillet of the upper die and also for elements along the free surface. The
nonsteady-state forging process was analyzed in a step-by-step manner
with a die displacement at each step that was 1% of the initial height of the
specimen. It is to be noted that additional elements were provided at the
stage of flash formation.
 Plane-Strain Problems 135

The computed results were obtained in terms of grid distortion, velocity

distribution, load-displacement curve, flash geometry, and axial height
variations.
Predicted and experimental velocity distributions, for the case of sticking
friction, are given in Fig. 8.3. It is seen that the computed velocity
distributions are in good agreement with the experimental observations.
Material particles near the upper fillet move sideways in the earlier stages
and change direction downward in subsequent stages. The particles at the
core portion remain stationary until the flash is formed and then begin to
move upwards when the lower die cavity is almost filled.
When the side cavity of the upper die is essentially filled, the material
starts to flow through the gap between the upper and lower dies, and a
small portion of the flash is formed. This metal flow into the flash enhances
the filling of the lower die cavity. A comparison of grid distortions between
theory and experiment at this stage is shown in Fig. 8.4. The figure
demonstrates that the agreement between theory (sticking) and experi-
ment is good and reveals the deforming region at the stage of flash
formation.
The study of closed-die forging suggested the need for (1) handling of
complex die geometry more efficiently, and (2) establishment of a
remeshing scheme for simulations of severe deformation (such as flash
formation).

FIG. 8.3 Velocity distributions in closed-die forging: (a) FEM (sticking friction); (b)
experiment [10].
136 Metal Forming and the Finite-Element Method

FIG. 8.4 Grid patterns in closed-die forging. Upper: experiment (33rd step) [10]. Lower:
FEM (32nd step, dark portion indicates rigid elements).

A significant advance in analysis technique along these lines was

accomplished by Oh [12] who improved the rigid-plastic finite-element
method to handle arbitrarily shaped boundary conditions and to generate
automatically an initial guess for a solution. These studies resulted in the
development of a general-purpose program named ALPID.
A discussion of remeshing is presented in Section 7.6 of Chap. 7. The
scheme described in Chap. 7 was incorporated into the ALPID program by
Wu, Oh, and Altan [13]. Mori et al. [14] simulated the axisymmetric
backward extrusion stage of a forging sequence by using spatially fixed
elements as a remeshing scheme.
 Plane-Strain Problems 137

8.4 Sheet Rolling

Rolling is one of the oldest processes used in the metal-working industry.

In view of the tremendous amount and wide variety of rolled products
manufactured each year, rolling can be considered to be one of the most
important forming processes. For more than half a century, numerous
analytical and experimental investigations have been carried out on rolling.
The slab, the slip-line, and the upper-bound methods have been widely
used in theoretical analyses.
In recent years, several attempts have been made to solve the rolling
problem by using the finite-element technique. Zienkiewicz et al. [15] and
Dawson [16] have dealt with the rolling problem in terms of viscoplasticity;
Dawson, moreover, incorporated temperature effects into the computa-
tion. Shima et al. [17] analyzed several cases of rolling using the
rigid-plastic model; they did not include work-hardening behavior. All the
studies cited above used the plane-strain assumption, and no comparisons
were made between these computed results and experimental
observations.
The plane-strain rolling problem was solved by the rigid-plastic finite-
element method by Li and Kobayashi [18], on the basis of the infinitesimal
theory of plastic deformation. As a comprehensive investigation on rolling,
a series of cases with different dimensions and material properties were
numerically analyzed and the results were compared with the experiments
found in the literature. The process of rolling is shown schematically in
Fig. 8.5.
The process variables investigated were material properties, roll dia-
meter, the initial and final workpiece thicknesses, and the friction
conditions. Most of the process conditions were taken from the literature
in order to compare computed and experimental results.
The analysis was performed using both nonsteady-state and steady-state
procedures. For the nonsteady-state procedure, deformation of the work-
piece was simulated in a step-by-step manner. The solution scheme for the
steady-state procedure is described in Chap. 10.
The mesh system and the boundary conditions used for the analysis are
shown in Fig. 8.5, the frictional stress was assumed to be velocity
dependent and represented by eq. (5.30) in Chap. 5.
The computations performed for each case resulted in the evaluations of
the velocity field, the grid distortion pattern, the distributions of stresses,
strain-rates and total effective strain, the normal pressure variation along
the roll-workpiece interface, and the roll-separating force and the roll
torque.
Comparisons were made between the computed results and the ex-
perimental data on contact pressure, roll separating force, and torque.
Al-Slehi et al. [19] not only measured the roll separating force and the roll
torque, but also measured the contact pressure distribution and the
coefficient of friction with the pressure-pin technique.
138 Metal Forming and the Finite-Element Method

For steady-state approach

FIG. 8.5 (a) Geometry and external forces in rolling, (b) Mesh systems used for
nonsteady-state and steady-state computations.

The classical slab method gives the shape of the stress distribution,
known as the "friction hill," with maximum pressure located at the neutral
point, on both sides of which the pressure decreases monotonically.
However, in some circumstances the pressure distribution curves were
observed to have double peaks. A slip-line solution [20] also predicted the
pressure drop in the middle of the arc of contact. In the computed results,
double-peak pressure distribution curves, as well as friction-hill type
 Plane-Strain Problems 139

4 dec,

FIG. 8.6 Comparison of measured [19] and computed roll-workpiece contact pressure
distributions: (a) P = 1.79 and (b) P = 13.18 for aluminum; (c) P = 1.85 and (d) P = 11.57 for
copper.

distributions, were obtained. Figure 8.6 shows the pressure variations for
some typical cases, comparing the experimental and computed results.
An important parameter that determines the deformation mode is the
ratio of the roll-workpiece contact length to the initial workpiece
thickness. This ratio is equivalent to the parameter P defined by
P = R(H0-Hl)/Hl = (Reduction)(R/H0). In Fig. 8.6a, the pressure dis-
tribution curve shows maxima near entrance and exit with a pressure drop
in the middle of contact length. This is characteristic for small values of the
parameter P. For large values of the parameter P, Fig. 8.6b shows a
friction-hill type pressure distribution. There is considerable discrepancy in
magnitude between the computed values and the experimental measure-
ments. The same observations apply to the pressure distributions for a
different material given in Figs. 8.6c and d. As typified by the results
shown in Fig. 8.6, two modes of deformation are observed. In one the
contact pressure distribution shows double peaks and deformation is more
inhomogeneous, and in the other the pressure distribution is a friction-hill
type and homogeneous deformation is dominant.
For the roll separating force and the torque in rolling aluminum and
copper, the comparisons between the computed results and the ex-
perimental values [19] are shown in Fig. 8.7. The computed results for the
roll separating force are smaller than the measured values, while the
140 Metal Forming and the Finite-Element Method

Reduction

FIG. 8.7 Comparison of measured [19] and computed roll separating force and roll torque;
(a) and (b) for aluminum; (c) and (d) for copper.

computed torque values generally overestimate the actual values. Similar

comparisons of the roll separating force and the torque for steel [21] are
shown in Fig. 8.8. In these cases, the predictions are lower than the
experimental values.
Discrepancies between the computed and measured values of contact
pressure, roll separating force, and torque shown in Figs. 8.6-8.8 may be
attributed to the rigid-roll assumption in the finite-element analysis and to
the uncertainty in friction modeling. It is a known fact that roll flattening is
 Plane-Strain Problems 141
R / H 0 = 12.5 R/H 0 - 39

FIG. 8.7 (continued)

an important aspect in the analysis of rolling and contributes to an increase

in the contact pressure; hence the actual roll separating force should be
larger than that obtained with the assumption of rigid-rolls.

8.5 Plate Bending

The air bending operation is a simple but widely used process in the
sheet-metal industry. Air bending implies that the shape taken up by the
metal depends on the punch position and not on the die shape. Although
the process is simple, the bending operation presents several technical
 142 Metal Forming and the Finite-Element Method

Reduction

FIG. 8.8 Comparison of measured [19] and computed roll separating force and roll torque
for steel; (a) and (b) with R/Ha = 65; (c) and (d) with R/Hn = 130.

problems, such as the limitations in flexibility and the deviations in bend

angle, due to elastic spring-back. There are also shape inaccuracies that
arise from edge distortion and punch shape.
Several analytical solutions for sheet bending and spring-back have been
published, but they are usually based on simple beam theory and include
an assumption about displacement distribution in the thickness direction or
in the plate direction, or in both. A different approach to this problem is to
apply plate bending theory, or shell theory, by using the finite-element
method [22]. Another approach [23] is to use the finite-element method by
considering the sheet as a bulk material. Hibbit et al. [24], using the
finite-element formulation, solved a three-point bending problem with
counter-pressure at the bottom of a sheet.
 Plane-Strain Problems 143

FIG. 8.8 (continued)

The analysis of sheet bending as a bulk deformation process was

performed by the rigid-plastic finite-element method using the incremental
theory, and by the elastic-plastic finite-element method with large-strain
formulation [25]. Some of the rigid-plastic loading solutions are shown
here and compared with those of the elastic-plastic solution. In Chap. 16,
trie large-strain formulation for the elastic-plastic finite-element method is
described and more detailed comparisons of the two solutions are given.
Sheet bending, as shown in Fig. 8.9, is considered to take place under
the plane-strain condition. The punch and the die are assumed to be rigid.
The workpiece material is aluminum alloy 2024-0 and isotropic. The
normalized dimensions (taking the workpiece thickness T = 1) are: die
opening, W = 7.5, punch radius, RP = 1.8, and die corner radius, RD =
1.2. It is assumed that there is no frictional stress acting along the punch
and die surfaces.
144 Metal Forming and the Finite-Element Method

FIG. 8.9 (a) Schematic diagram of sheet bending operation used in the computation with
T = l, W = 7.5, RP = 1.8, RD = 1.2. (b) Boundary conditions in sheet bending.

Since the process is symmetric with respect to AB in Fig. 8.9b, the

analysis is performed for one half of the workpiece. The boundary
conditions along AB are u x=0 and Fy=Q, where ut,Fi denote velocity and
traction respectively. Along GH, where the circular punch and workpiece
are in contact, un = Up cos 9 and F/ = 0. The subscript n denotes the
direction normal to the circular arc GH, / is the tangential direction, and
UP is the punch velocity.
On the free surfaces, BC, DE, EF, FG, HA, the boundary conditions
are simply Fx = 0 and Fy=0. The boundary conditions on the surface, CD,
are written as «„ = 0 and Ft — 0.
Solutions during loading were obtained for the punch loads, the stress
 Plane-Strain Problems 145

and strain distributions, and the deformed geometries at each stage of the
process.
Figure 8.10 compares the grid distortion patterns obtained from rigid-
plastic and elastic-plastic formulations. The black area of the workpiece
under the punch represents the deforming zone in the rigid-plastic analysis
and the plastic zone in the elastic-plastic analysis. From the elastic-plastic
calculations, it can be seen that yielding begins at the outer fiber near the
bending axis and spreads gradually toward the rest of the sheet. The trend

Elasto-Plastic Rigid-Plastic
FIG. 8.10 Comparison of grid distortions by rigid-plastic and elastic-plastic (elasto-plastic)
formulations for punch displacements: (a) 0.6, (b) 1.2, (c) 1.8, (d) 2.25, (e) 2.8, (f) 3.2. The
darkened area denotes the plastically deforming (plastic) zone.
 146 Metal Forming and the Finite-Element Method
is the same as that predicted by the elementary beam theory. As the
process continues, a well-defined wedge-shaped elastic-plastic boundary
moves away from the bending axis. But, after a certain punch displace-
ment, the boundary remains nearly stationary. The rigid-plastic calculation
does not show the well-defined continuous spread of the plastic zone. It
can be seen, however, that the size of the deforming zone determined by
the rigid-plastic analysis compares well with that of the plastic zone
obtained in the elastic-plastic analysis. It can also be seen from the figure
that unloading takes place near the bending axis during the later stages of
deformation in the elastic-plastic case, and that the rigid zone appears
during the same stage in the rigid-plastic case.
Another interesting feature in the deformation mode of sheet bending is
that the bend radius of the workpiece does not follow that of the punch;
the workpiece separates from the punch and only a small portion of the
workpiece is in contact with it, resulting in shape inaccuracies in the bent
sheet. This is shown quantitatively in Fig. 8.11, where the bend angle and
the clearance between the punch pole and the workpiece are plotted
against the punch displacement for both analyses. The figure shows that
the curves for bend angle vs. punch displacement obtained from the two
formulations are in almost perfect agreement. From the figure it can be
seen that workpiece separation from the punch pole occurs when the
punch displacement reaches 1.48 according to the elastic-plastic calculation
and 1.68 according to the rigid-plastic analysis. The clearances that are
calculated from the two formulations show the same trend, but the
rigid-plastic formulation results in slightly larger clearance than that of
elastic-plastic analysis.

PUNCH DISPLACEMENT

FIG. 8.11 Relationship of bend angle to punch displacement, and of clearance between the
punch poles and the workpiece to punch displacement.
 Plane-Strain Problems 147

Elasto-Plastic Rigid-Plastic
FIG. 8.12 Comparison of distributions of normalized bending stress (as/initial yield stress)
and effective strain by rigid-plastic and elastic-plastic (elasto-plastic ) formulations at the end
of loading (punch displacement = 3.2).

Figure 8.12 shows the distributions of bending stress and the effective
strain at the end of loading, for rigid-plastic and elastic-plastic materials.
From the figure it can be seen that the bending stress increases toward the
outer surface and toward the punch axis, as expected. It can be also seen
that the neutral line (os = 0) shifts toward the punch near the punch axis
and is at about half of the thickness in other locations. The effective strain
distributions show a similar pattern. The figure also reveals that the
distributions of bending stress and effective strain for both analyses agree
with each other very well, with minor differences.
Unloading and resulting spring-back and residual stresses are discussed
in Chap. 16.
148 Metal Forming and the Finite-Element Method

8.6 Side Pressing

In side pressing, a long circular cylinder is compressed sideways (perpen-
dicular to the cylinder axis) between two flat dies. This process has been
shown to be very useful in studying flow behavior at hot working
temperatures. Experiments show that hot isothermal side pressing of
(a + /J)Ti-6242 alloy, for example, leads to stable deformation, while that
of ()8)Ti-6242 produces severe shear bands and unstable flow. This is due
to two completely different deformation modes and to the differences
betwen the flow stress behaviors of the two structures; that is, the (a + /3)
microstructure exhibits a stable strain-rate hardening, while the (/3)
microstructure shows substantial amounts of strain softening behavior.
In order to estimate the strain concentration, an isothermal rigid-
viscoplastic finite-element method (see Chap. 12 for viscoplastic analysis)
was used to simulate the side pressing of cylinders with two different flow
stress characteristics corresponding to the two microstructures [26]. The

(a)

(b)

FIG. 8.13 ^umparison of predicted strain-rate (per second) distributions and experimentally
determined flow localization in side pressing of Ti-6242-0.1Si at 913°C (1675T): (a) (a + ft)
microstructure; (b) (/J) microstructure [26],
 Plane-Strain Problems 149
diameter of the specimen was 10.2 mm (0.40 inch) and the deformation
was assumed to be plane-strain (zero strain in the axial direction). The
temperature during deformation was 913°C (1675°F). Figure 8.13 shows
the experimentally observed transverse sections of side-pressed cylinders
of (a + /?) and (/3) microstructures and the predicted effective strain-rate
distributions for both cases. It can be seen that the method effectively
predicts, for the same alloy and forming conditions, detailed variations in
metal flow that are due to differences in microstructure and flow stress
behavior.

References
1. Johnson, W., Sowerby, R., and Venter, R. D., (1982), 'Plane-Strain Slip-Line
Fields for Metal Deformation Processes," Pergamon Press, Oxford.
2. Oxley, P. L. B., and Hastings, W. F., (1976), "Minimum Work as a Possible
Criterion for Determining the Frictional Conditions at the Tool/Chip Interface
in Machining," Phil. Trans. R. Soc., London, Series A. Vol. 282, p. 565.
3. Kobayashi, S., Herzog, R., Lapsley, J. T., Jr., and Thomsen, E. G., (1959),
"Theory and Experiment of Press Forging Axisymmetric Parts of Aluminum
and Lead," Trans. ASME, J. Engr. Ind., Vol. 81, p. 228.
4. Kobayashi, S., and Thomsen, E. G., (1959), "Approximate Solutions to a
Problem of Press Forging," Trans. ASME, J. Engr. Ind., Vol. 81, p. 217.
5. Akgerman, N., and Altan, T., (1972), "Modular Analysis of Geometry and
Stresses in Closed-Die Forging: Application to a Structural Part," Trans.
ASME, J. Engr. Ind., Vol. 94, p. 1025.
6. Johnson, W., (1958), "Over-Estimates of Load for Some Two-Dimensional
Forging Operations," Proc. 3d U.S. Congr. Appl. Mech., ASME (New York),
p. 571.
7. Kudo, H., (1958), "Studies on forging and Extrusion Process, I," Koken-
shuho, Tokyo University, Vol. 1, p. 37.
8. McDermott, R. P., and Bramley, A. N., (1974), "An Elemental Upper-Bound
Technique for General Use in Forging Analysis," Proc. 15th Int. MTDR
Conference, Birmingham, England, p. 437.
9. Kasuga, Y., Tsutumi, S., and Saiki, H., (1974), "Material Flow in Sunken
Forging Dies," /. Japan Soc. Tech. Plast., Vol. 15, p. 147.
10. Lyapunov, N. L, and Kobayashi, S., (1974), "Metal Flow in Plane-Strain
Closed-Die Forging," Proc. of 5th North Amer. Metalworking Res. Conf.
NAMRC, p. 114.
11. Chen, C. C., and Kobayashi, S., (1980), Rigid-Plastic Finite Element Analysis
of Plane-Strain Closed-Die Forging, "Process Modeling," ASM, Metals Park,
Ohio, p. 167.
12. Oh, S. I., (1982), "Finite Element Analysis of Metal Forming Processes with
Arbitrary Shaped Dies," Int. J. Mech. ScL, Vol. 24, p. 479.
13. Wu, W. T., Oh, S. L, and Altan, T., (1984), "Investigation of Defect
Formation in Rib-Web Type Forging by ALPID," Proc. 12th NAMRC, p.
159.
14. Mori, K., Osakada, K., and Fukada, M., (1983), "Simulation of Severe Plastic
Deformation by Finite Element Method with Spatially Fixed Elements," Int. J.
Mech. ScL, Vol. 25, p. 775.
15. Zienkiewicz, O. C., Jain, P. C., and Onate, E., (1978), "Flow of Solids
During Forming and Extrusion: Some Aspects of Numerical Solutions," Int. J.
Solids Structures, Vol. 14, p. 15.
150 Metal Forming and the Finite-Element Method

16. Dawson, P. R., (1978), Viscoplastic Finite Element Analysis of Steady-State

Forming Processes Including Strain History and Stress Flux Dependence,
"Applications of Numerical Methods to Forming Processes," ASME, AMD,
Vol. 28, p. 55.
17. Shima, S., (1980), "Rigid-Plastic Finite Element Analysis of Strip Rolling,"
Proc. 4th Int. Conf. on Prod. Eng., p. 82.
18. Li, G.-J., and Kobayashi, S., (1982), "Rigid-Plastic Finite Element Analysis of
Plane Strain Rolling," Trans. ASME, J. Engr. Ind., Vol. 104, p. 55.
19. Al-Salehi, F. A., Firbank, T. C., and Lancaster, P. R., (1973), "An
Experimental Determination for the Roll Pressure Distributions in Cold
Rolling," Int. J. Mech. ScL, Vol. 15, p. 693.
20. Firbank, T. C., and Lancaster, P. R., (1965), "A Suggested Slip-Line Field for
Cold Rolling with Slip Friction," Int. J. Mech. ScL, Vol. 7, p. 84.
21. Shida, S., and Awazuhara, H., (1973), "Rolling Load and Torque in Cold
Rolling," /. Japan Soc. Tech. Plast., 14, p. 267.
22. Popov, E. P., Khojesteh-Bakht, M., and Yaghmai, (1967), "Analysis of
Elastic-Plastic Circular Plates,"/. Eng. Mech. Div. Proc., ASCE, p. 49.
23. Cupka, V., Miyamoto, H., Miyoshi, T., and Suzuki, K., (1974), "Analysis of
Counterpressure Bending by FEM," Proc. Int. Conf. Prod. Engr., p. 257.
24. Hibbitt, H. D., Marcal, P. V., and Rice, J. R., (1970), "A Finite Element
Formulation for Problems of Large Strain and Large displacement," Int. /.
Solids Structures, Vol. 6, p. 1069.
25. Oh, S. I., and Kobayashi, S., (1980), "Finite Element Analysis of Plane-Strain
Sheet Bending," Int. J. Mech. ScL, Vol. 22, p. 583.
26. Oh, S. I., Lahoti, G. D., and Allan, T., (1981), "ALPID—A General Purpose
FEM Program for Metal Forming," Proc. NAMRC-IX, State College,
Pennsylvania, p. 83.
 9
AXISYMMETRIC ISOTHERMAL FORGING

9.1 Introduction
According to Spies [1], the majority of forgings can be classified into three
main groups. The first group consists of compact shapes that have
approximately the same dimensions in all three directions. The second
group consists of disk shapes that have two of the three dimensions (length
and width) approximately equal and larger than the height. The third
group consists of the long shapes that have one main dimension sig-
nificantly larger than the two others. All axially symmetric forgings belong
to the second group, which includes approximately 30% of all commonly
used forgings [2]. A basic axisymmetric forging process is compression of
cylinders. It is a relatively simple operation and thus it is often used as a
property test and as a preforming operation in hot and cold forging. The
apparent simplicity, however, turns into a complex deformation when
friction is present at the die-workpiece interface. With the finite-element
method, this complex deformation mode can be examined in detail. In this
chapter, compression of cylinders and related forming operations are
discussed. Since friction at the tool-workpiece interface is an important
factor in the analysis of metal-forming processes, this aspect is also given
particular consideration. Further, applications of the FEM method for
complex-shaped dies are shown in the examples of forging and cabbaging.

9.2 Finite-Element Formulation

Finite-element discretization with a quadrilateral element is similar to that
given in Chap. 8. The cylindrical coordinate system (r, 9, z) is used instead
of the rectangular coordinate system. The element is a ring element with a
quadrilateral cross-section, as shown in Fig. 9.1.
The £ and rj of the natural coordinate system vary from -1 to 1 within
each element. An arbitrary point (r, z) inside the element can be
expressed in terms of the natural coordinate (£, rj), and the coordinate
transformation is expressed by

151
152 Metal Forming and the Finite-Element Method

FIG. 9.1 Quadrilateral element and natural coordinate system.

where ra, za (a = 1, . . . ,4) are the positions of the four surrounding nodal
points of an element in the global coordinate system. For an isoparametric
element of Fig. 9.1, discussed in Chap. 6, transformation functions qa in
eq. (9.1) are the same as shape functions defined by eq. (8.2b) in Chap. 8.
Then

and the matrix IMT is identical to eq. (8.2a) in Chap. 8.

The strain-rate vector is defined (see eq. (6.24c) in Chap. 6) by

The strain-rate matrix B in eq. (9.3) is given by

where
 Axisymmetric Isothermal Forging 153

and

and

Equations (9.4b) and (9.4c) are obtained by replacing (x, y) in eqs. (6.35a)
in Chap. 6 by (r, z).
The determinant of the Jacobian of transformation |J| is expressed by
eq. (6.35b) in Chap. 6 with (r, z) replacement. The matrix D and the
vector C in the expressions of the effective strain-rate and the volumetric
strain-rate are the same as those given by eqs. (8.4) and (8.5) in Chap. 8.
The stiffness equations based on, for example, eq. (8.la) in Chap. 8,
become

and

where the subscript j indicates the element number and M is the total
number of elements. Linearization of eq. (9.5) is made according to eq.
(6.48) in Chap. 6.

9.3 Compression of Solid Cylinders and Heading of Cylindrical

Bars
In compression of circular, solid cylinders between parallel, flat dies, the
deformation is homogeneous when there is no friction, but with friction
the distribution of the compressive stress is nonuniform, and the average
compression stress differs from the flow stress. Furthermore, the free
surface barrels, and fracture may occur at the barreled surface.
The complexity of nonuniform deformation is not only represented by
barreling of the free surface, but is further indicated by the fact that a part
of the initially free surface comes into contact with the die during
compression. This flow phenomenon is known as "folding." The mode of
deformation is also influenced by the workpiece geometry. With specimens
of small height-to-diameter ratio (say, less than 1.6), barreling results in a
single bulge, the maximum diameter being at the equatorial surface. For
154 Metal Forming and the Finite-Element Method
specimens of larger height-to-diameter ratio, however, the mode of
deformation is different and a double bulge is sometimes observed along
the free surface [3, 4].
Applications of the elastic-plastic finite element method to this problem,
based on the infinitesimal deformation theory, revealed that the analysis
did not produce the phenomenon of folding. This shortcoming was the
main motivation for developing the finite-element formulation based on
rigid-plastic material behavior [5]. The method was applied to obtain
complete rigid-plastic solutions in upsetting solid cylinders under various
interface friction conditions by using the work-hardening property of SAE
1040 steel at room temperature [6]. The detailed mechanics were investig-
ated by this method and the theoretical results predicted the observed
folding phenomenon remarkably well, as seen in Fig. 9.2.
An important objective of the analysis of simple compression is to
predict the possible occurrence of fracture at the free surface. Since the
fracture condition can be observed for cracks occurring at free surfaces,
upsetting of cylindrical specimens has been used extensively by several
investigators for the study of ductile fracture and workability of materials.
Workability of materials implies the extent to which materials are able
to deform without cracks during a mechanical working process. Kudo and
Aoi [3], in their investigation of carbon steel, directly measured the
equatorial free surface strains in upsetting of solid cylinders. The friction
condition at the interface was controlled qualitatively by using grooved
dies and conical dies, with and without lubrication. Stresses at the free
surface were derived from measurements of the strains, and the fracture
was observed for various stress and strain histories at the critical site.
With the finite-element method, strain paths of a critical site can be
computed under various friction conditions, and they are compared with
experimental results in Fig. 9.3. Combining the observed fracture condi-
tions and the computed results on deformation characteristics, it was
possible to predict workability in terms of reduction in height at fracture
for various friction conditions and initial workpiece dimensions, as shown
in Fig. 9.4.
The analysis of compression of tall cylinders required the development
of a method that could handle the treatment of rigid zones. With this
development the analysis was performed for compression of cylinders with
height-to-diameter ratio larger than unity [7]. The grid distortions at 36
and 66% reductions in height for H0/D0 = 2.5 are shown in Fig. 9.5. The
double-bulge formation on the free surface, typical of compression of tall
cylinders, is quite noticeable at a reduction of 36%.
Rigid zones are also present in the heading process, which is described
in Chap. 2. Although cold heading is a very common metal-working
operation, only a few theoretical studies have been undertaken to
investigate the details of this process [8-10]. The finite-element solutions
were obtained and some of the computed results were compared with
experiments [11].
 Axisymmetric Isothermal Forging 155

FIG. 9.2 Grid distortions in simple compression at 50% reduction in height for the two
friction conditions.

The deformation pattern during the heading operation can be observed

in detail by examining the effective strain distributions, which are plotted
on the right-hand side in Fig. 9.6 at 60% reduction in height. An important
feature of the strain distributions predicted at various stages is that the
pattern does not seem to change much as the reduction in height increases.
As seen in the figure, there is a strain concentration at the edge of the
contact surface on the top die, and on the contact surface at the bottom
FIG. 9.3 Strain paths of a point at the equatorial free surface in simple compression for
initial height-to-diameter ratio of 1.0 [6].

156
 Axisymmetric Isothermal Forging 157

FIG. 9.4 Workability of SAE 1040 (annealed) steel at room temperature in slow-speed
upsetting.

die. Otherwise, the maximum strain is near the axis and almost at
mid-height of the upset head. Since hydrostatic pressure seems to play an
important role in ductile fracture, it is instructive to examine the
distribution of am. This distribution is plotted on the left-hand side of Fig.
9.6. Although the pattern of the effective strain distribution remains
almost unchanged with increasing reduction in height, the hydrostatic
stress distribution changes a great deal. A very important observation,
from the fracture point of view, is that with increasing reduction in height
the hydrostatic pressure increases near the axis of the specimen, whereas it
decreases near the free surface. In fact, the hydrostatic stress becomes
tensile at higher reductions. This explains why, in heading operations,
 66 Percent Reduction

FIG. 9.5 Grid distortions in simple compression at 36 and 66% reductions in height for initial
height-to-diameter ratio of 2.5.

FIG. 9.6 Effective strain (e) contours and hydrostatic pressure (—a m /Y 0 ) contours in heading
at 60% reduction in height.
158
 Axisymmetric Isothermal Forging 159

FIG. 9.7 Experimental and theoretical flow lines in heading at 31% reduction, 47%
reduction, and 63% reduction (60%, theory) [11].

surface cracking is the predominant failure mode rather than the formation
of internal cracks.
A significant comparison between theory and experiment can be made in
terms of the grid distortion pattern. The flow patterns obtained ex-
perimentally by etching the formed specimens are shown in Fig. 9.7 at
approximately 31%, 47%, and 63% reductions in height. The figure also
shows the grid patterns obtained theoretically at 31%, 47%, and 60%
reductions. It can be seen that the distortion of axial lines predicted by the
theory is almost identical to the flow lines observed experimentally.

9.4 Ring compression

When a short ring specimen is compressed between two flat, parallel
platens (Fig. 9.8a) the diameter of inner surface either increases or
decreases as the height of the specimen is reduced, depending on the
friction condition at the interface. The inner diameter of the ring is
160 Metal Forming and the Finite-Element Method

(C)

FIG. 9.8 Two deformation modes in ring compression, (a) Compression geometry; (b)
situation with low interface friction; (c) situation with high interface friction.

increased if the interface friction is low (Fig. 9.8b) and it is decreased if

this friction is high (Fig. 9.8c). Because the change in internal diameter of
the compressed ring is sensitive to friction at the die-workpiece interface,
ring compression has been widely used as a test to evaluate the friction
condition in metal-forming processes.
To determine the friction condition quantitatively, however, the re-
lationship between the geometrical change of the workpiece and the
friction condition at the tool-workpiece interface must be established.
Since Kunogi [12] analyzed the process mathematically, this aspect of ring
compression has been the subject of many investigations in the past (see
Avitzur [13]). Most of the mathematical methods are based on an
approximate stress analysis or on the upper-bound analysis using relatively
simple velocity fields. While these theories provided useful information,
more accurate and realistic solutions were needed for wider and meaning-
ful application of the ring test in practice.
In implementing the finite-element method for analyzing metal-forming
problems, particular attention must be paid to boundary conditions. There
 Axisymmetric Isothermal Forging 161

is a class of problems such as ring compression, rolling, and forging, in

which there exists a "neutral" point (or a region) along the die-workpiece
interface. At this point the velocity of the deforming material relative to
the die velocity becomes zero, and the frictional stress usually changes its
direction. As discussed in Chap. 5, Chen and Kobayashi [14] used the
velocity-dependent frictional stress expressed by eq. (5.30) for analyzing
these problems. This approach was successful in analyzing plane-strain
sheet rolling (Chap. 8) and also ring compression. The finite-element
solution of ring compression was obtained in terms of geometrical changes
of the specimen, velocity distribution, grid distortions, and stress and
strain distributions under various friction conditions for several materials.
Some of the computed results are shown in Fig. 9.9 in terms of changes of
minimum internal diameters as functions of reduction in height. The figure
shows that the geometrical changes differ for the same m value, depending

REDUCTION IN HEIGHT,(H 0 -H)/H O x IOO%

FIG. 9.9 Comparison of the finite-element solution with the upper bound solutions for ring
compression. Experimental data: O, copper; •, copper; A, pure aluminum [17].
162 Metal Forming and the Finite-Element Method

upon the selection of the initial or current shear yield stress k. It is also
seen that the geometrical changes are affected by the material property, as
indicated by the curves calculated with ra = 1.0 for copper and for
annealed aluminum.
The problem of ring compression was also analyzed by the elastic-plastic
finite-element method. Hartley et al. [15,16] developed a method for
introducing friction into the finite-element analysis by the inclusion of a
layer of elements. The stiffness of these elements was modified by a
function of the interfacial shear factor and the method was applied to the
analysis of ring compression.

FIG. 9.10 Grid distortions in ring compression at 50% reduction in height for various
m-values: (a) m = 0.12 (pure aluminum); (b) m=0.25 (annealed Al 1100); (c) m = 0.6
(copper), (d) m - 1.0 (copper).
 Axisymmetric Isothermal Forging 163

The finite-element method not only provides overall quantities, such as

changes in specimen dimensions and load-displacement curves, but it also
reveals the details of the deformation characteristics. Figure 9.10 shows
grid distortions at 50% reduction in height under various interface friction
conditions. At low friction, the outward flow dominates, but with
increasing friction, the neutral flow point appears along the die-workpiece
interface, resulting in both inward and outward barreling. Furthermore,
for high friction, folding occurs at the inner surface as well as the outer
surface. The deformation characteristics shown in Fig. 9.10 are in
qualitative agreement with experimental observations reported in the
literature [17].

9.5 Evaluation of Friction at the Tool-Workpiece Interface

In analyzing the mechanics of metal-forming processes, a realistic friction
condition must be specified in order for a theory to yield a reliable
solution. Also, in practice, understanding and controlling friction often
lead to successful metal-forming operations. Consequently, considerable
effort has been devoted to determining satisfactorily the friction condition
in various metal-forming processes [18]. Yet, the mechanism of friction is
still not well known and, as Thomsen [19] has pointed out, friction remains
one of the most elusive variables in metal forming technology.
It has been demonstrated that in simple upsetting, for an assumed
frictional stress distribution, the finite-element method is capable of
accurately computing, among other details of the deformation mechanics,
the relative displacement distributions at the die-workpiece interface. This
suggests that the method also permits, in turn, the calculation of the
frictional stress distribution when the displacement distributions at the
interface are measured and used as the boundary condition for computa-
tion [20]. Such a study has been conducted. The experiments consisted of
measuring the displacement distributions at the die-workpiece interface at
various reductions in height when upsetting solid circular cylinders
between two rigid, flat, and parallel dies. The specimens of 20.32mm
(0.8 in.) diameter and 20.32mm (0.8 in.) height were prepared from an
aluminium alloy 7075-T6. A series of indentation marks with 1.016mm
(0.04 in.) spacing were made diametrically at one end of the specimen
using the Vickers Micro-hardness Tester, as shown schematically in Fig.
9.1 la. Also, short lines parallel to the end-plane were scribed on the
cylindrical surface near the end-plane in order to determine the displace-
ment distribution when this region comes into contact with the die. The
results are examined in Figs. 9.lib, and c for two types of representation
of the friction conditions. One is for the coefficient of friction /* and the
other for the friction factor m. It is seen in the figure that the variations for
(j, and m are similar, but that the curve for jit shows more uniform
distribution.
In ring compression, friction values are evaluated by comparing the
 (c)
FIG. 9.11 (a) Solid circular cylinder specimen of aluminum alloy 7075-T6 with indented
marks and scribed lines; distributions of coefficient of friction n and friction factor m; (b) with
wax as a lubricant; (c) without any lubricant [20].

164
 Axisymmetric Isothermal Forging 165
experimental data with the so-called calibration curves, as shown in Fig.
9.9. This procedure, however, raises some questions regarding the
accuracy and efficiency of this evaluation scheme, because the curves are
not unique and depend on strain, strain-rate, and the thermal characteris-
tics of the material, and also on the specimen geometry. Furthermore,
experimental data usually do not follow the trend of predicted curve for a
constant friction value. Thus, Hwang and Kobayashi [21] proposed a
method of direct evaluation of the friction value from the experimental
measurements in ring tests. The method uses the finite-element technique,
and includes fitting curves to the experimental data with an iteration
scheme for evaluating a current friction value based on measured changes
of the ring dimensions. This scheme for evaluating friction was applied to
the experimental data of DePierre and Gurney [22] for annealed aluminum
1100, for a ring geometry of 6:2:1 (outer diameter:inner diameter:thick-
ness). One set of experiments was conducted with Johnson's wax as a
lubricant and another without a lubricant.
The results for the case with a lubricant are shown in Fig. 9.12a. The
upper part shows experimental points and the fitted curve, and the lower
part shows the computed variation of the friction factor m, which only
ranges from m = 0.1 to 0.2. The results for the case without a lubricant are
shown in Fig. 9.12b. It can be seen that the variation in friction value is
large and the estimation of friction from the calibration curves generated
for constant friction values results in an erroneous interpretation of the
experimental data.

9.6 Forging and Cabbaging

Axisymmetric Spike Forging
In spike forging, a cylindrical billet is forged in an impression die
containing a central cavity. The material flows radially as well as axially
into the upper cavity, forming a spike. The height of the spike depends on
(1) the geometrical dimensions of the dies, (2) the initial workpiece, and
(3) the interface friction. Experiments show that spike height increases
with increasing friction if the other process conditions remain unchanged
[23].
Figure 9.13a shows the undeformed workpiece geometry with 16
nine-node elements (see Section 6.3 of Chap. 6), with the top and bottom
dies at their initial positions. The material used in these calculations was
(or + j8) Ti-6242-0.1Si at 954°C (1750°F). Figure 9.13b shows the effect of
friction on metal flow. The die velocity used for these simulations was
25.4 mm/s (1.0 in./s). The predictions shown in the figure were in excellent
agreement with experiments [24].

Cabbaging [25]
In the initial stage of backward extrusion, called cabbaging, a round billet
is placed in a container. A punch is used to upset and partially pierce one
 (b)
FIG. 9.12 Variation of friction value determined from experimental data [22]: (a) lubricated
with wax; (b) without any lubricant.
166
FIG. 9.13 Axisymmetric spike forging: (a) undeformed grid; (b) deformation at a die stroke
of 0.58H0 for two m values (H0 is initial billet height) [24].
167
168 Metal Forming and the Finite-Element Method
end of the billet. Thus, the outside diameter of the billet is sized, and the
pierced recess on top of the billet provides punch guidance for the
subsequent backward extrusion operation. The cabbaging operation has
been simulated using the actual production conditions with minor assump-
tions. The undeformed billet had a diameter of 73.7mm (2.94 in.) and a
height of 354mm (13.85 in.) The billet material was AISI 1046 steel, and
the operation was done at 1100°C (2012°F). The frictional shear factor
used in the analysis was m = 0.3. The punch speed used for the simulation
was 59.3 mm/s (2.3 in/s).
Figure 9.14 shows the undeformed grid lines, the calculated grid
distortions at various punch displacements, and the predicted punch load
versus displacement curve. Because of the unusually high strain concentra-
tion, the element near the punch tip underwent too much distortion, as can
be seen in Fig. 9.14. A partial "remeshing" near the punch tip was done at
a punch displacement of 91.4 mm (3.6 in.).
Compressor Disk Forging [26]
A simulation of a compressor disk forging from Ti 6242 alloy is illustrated
in Fig. 9.15. Because of symmetry, only a quarter of the disk is shown. The
cylindrical preform shape used in this analysis was 158.8mm (6.5 in.) in

PUNCH DISPLACEMENT (INCH)

FIG. 9.14 Predicted grids and punch load-displacement curve for the cabbaging process [25].
 Axisymmetric Isothermal Forging 169

(c) Predicted Effective Strain Distribution at 70 Percent Reduction

FIG. 9.15 Results of compressor disk forging simulation [26].

diameter and 63.5 mm (2.5 in.) in height. The velocity of the upper die
used in the simulation was 5.1 mm/min (0.2in./min). The bottom die was
stationary.
Forging was done isothermally at 900°C (1650°F), with an average
nominal strain-rate of about 0.175mm"1. The results show that the
finite-element method can be used effectively for simulating disk-type
forgings and for predicting strains, strain-rates, and stresses for a given
preform shape.
170 Metal Forming and the Finite-Element Method

Flashless Forging of a Gear Blank

In practical metal-forming operations, deformations are usually very large.
As a result, during simulation of a forming operation these large strains
distort the initial mesh to such a degree that a new mesh (remesh) is
necessary in order to continue the simulation. For this purpose a
"remeshing" method has been developed [27] (see Section 7.6 of Chap. 7).
To evaluate the remeshing algorithm, a simulation of a gear blank
forging was conducted, using the following process variables:
• The undeformed cylindrical billet had a height of 169.16mm (6.66 in.)
and a diameter of 69.98mm (2.755 in.) (Figure 9.16).
• The billet material was AISI 8620 steel and the forging temperature
was 1100°C (2012°F). The flow stress data for 8620 steel were obtained
by performing isothermal uniform compression tests at strain rates of
25-30s"1 and 2.5-3s~1. The flow stress a was calculated using the
expression a = Ce"em (ksi), where C, n and m were given as 10.92,

FIG. 9.16 Metal flow patterns in flashless forging of a gear blank at 0, 40, 60, and 78%
reductions in initial billet height.
 Axisymmetric Isothemal Forging 171

0.0, and 0.134, respectively.

• The velocity of the upper die used in the analysis was 127.0mm/s
(5.0in./s), and the bottom die was stationary.
• The friction factor used in the analysis was chosen as m = 0.4.
Figure 9.16 shows the metal-flow patterns after 0, 40, 60, and 78%
reductions in initial billet height. It should be noted that at the 60%
reduction the element at the lower right-hand corner was distorted to such
a degree that it was impossible to continue the simulation without
generating a new mesh. A new mesh was generated at the 61% reduction
in billet height, as shown in Fig. 9.17. The strain values of the last step
before remeshing (at 61% reduction) were then interpolated on the nodes
of the new mesh. The simulation was continued using the new mesh. Two
more remeshings were necessary to complete the simulation up to the
complete die fill and to obtain the results shown in Fig. 9.18.

FIG. 9.17 Distorted grid (right half) and "remeshed" grid (left half) at 61% reduction in
billet height for flashless forging.

FIG. 9.18 Grid distortions near completion of die filling for flashless forging.
172 Metal Forming and the Finite-Element Method

References
1. Spies, K., (1957), "The Preforms in Closed-Die Forging and Their Preparation
by Reducer Rolling," (in German) Doctoral Dissertation, Technical University
Hannover.
2. Lange, K., (1958), "Closed-Die Forging of Steel," (in German), Springer-
Verlag, Berlin.
3. Kudo, H., and Aoi, K., (1967), "Effect of Compression Test Condition upon
Fracturing of a Medium Carbon Steel—Study on Cold Forgeability Test: Part
II," (in Japanese), J. Japan Soc. Tech. Plast., Vol. 8, p. 17.
4. Nagamatsu, A., Murota, T., and Jimma, T., (1971), "On the Nonuniform
Deformation of Material in Axially Symmetric Compression Caused by
Friction," Bull. JSME, Vol. 14, p. 331.
5. Lee, C. H., and Kobayashi, S., (1971), "Analysis of Axisymmetric Upsetting
and Plane-strain Side-pressing of Solid Cylinders by the Finite-element
Method," Trans. ASME, J. Engr. Ind., Vol. 93, p. 445.
6. Kobayashi, S., and Lee, C. H., (1973), "Deformation Mechanics and
Workability in Upsetting Solid Circular Cylinders," Proc. North American
Metalworking Research Conference, Vol. 1, p. 185.
7. Shah, S. N., Lee, C. H., and Kobayashi, S., (1974), "Compression of Tall
Circular, Solid Cylinders Between Parallel Flat Dies," Proc. International
Conference on Production Engineering, Tokyo, p. 295.
8. Lecocq, A. G., (1971), "Stresses in the Shank of a Bolt During Cold
Heading," Wire (English version of Draft) Coburg, Vol. 115, p. 197.
9. Thomason, P. F., (1969-70), "The Effect of Heat Treatment on the Ductility
in a Cold Heading Process," Proc. Inst. Mech. Engr., Vol. 184, p. 875.
10. Gill, F. L., and Baldwin, W. M., (1964), "Proper Wire Drawing Improves
Cold Heading Process," Metal Progress, Vol. 85, p. 83.
11. Shah, S. N., and Kobayashi, S., (1974), "Rigid-Plastic Analysis in Cold
Heading by the Matrix Method," Proc. 15th International Machine Tool
Design Research Conf., p. 603.
12. Kunogi, M., (1954), "On Plastic Deformation of Hollow Cylinders under
Axial Compressive Loading,"/. Sci. Res. Inst., Tokyo, p. 2.
13. Avitzur, B., (1983), "Handbook of Metal Forming Processes," Wiley, New
York.
14. Chen, C. C., and Kobayashi, S., (1978), "Rigid-plastic Finite-element Analysis
of Ring Compression," Application of Numerical Methods to Forming
Processes, ASME, AMD, Vol. 28, p. 163.
15. Hartley, P., Sturgess, C. E. N., and Rowe, G. W., (1979), "Friction in
Finite-element Analysis of Metal-forming Processes," Int. J. Mech. Sci., Vol.
21, p. 301.
16. Hartley, P., Sturgess, C. E. N., and Rowe, G. W., (1979), "An Examination
of Frictional Boundary Conditions and Their Effect in an Elastic-plastic
Finite-element Solution," Proc. MTDR, p. 157.
17. Male, A. T., and DePierre, V., (1970), "The Validity of Mathematical
Solutions for Determining Friction from the Ring Compression Test,"
Lubrication Technology, p. 389.
18. Schey, J. A., Ed., (1970), "Metal Deformation Processes—Friction and
Lubrication," Marcel Dekker, New York.
19. Thomsen, E. G., (1969), "Friction in Forming Processes," Annals CIRP, Vol.
17, p. 149.
20. Oh, S. I., Thomsen, E. G., and Kobayashi, S., (1975), "Calculation of
Frictional Stress Distributions at the Die-Workpiece Interface in Simple
Upsetting," Proc. 3d North American Metalworking Research Conference, p.
159.
 Axisymmetric Isothermal Forging 173

21. Hwang, S. M., and Kobayashi, S., (1983), "A Note on Evaluation of Interface
Friction in Ring Tests," Proc. NAMRC XI, University of Wisconsin, Madison,
WI, p. 193.
22. DePierre, V., and Gurney, F., (1972), "A Method for Determination of
Constant and Varying Friction Factors During Ring Compression Tests," Air
Force Materials Laboratory, Report AFML-TR-72-37.
23. Oh, S. I., (1982), "Finite Element Analysis of Metal Forming Problems with
Arbitrarily Shaped Dies," Int. J. Mech. Sci., Vol. 17, p. 293.
24. Oh, S. I., Lahoti, G. D., and Altan, T., (1981), "ALPID—A General Purpose
FEM Program for Metal Forming," Proc. NAMRC-IX, State College, PA, p.
83.
25. Oh, S. I., Lahoti, G. D., and Altan, T., (1982), "Analysis of Backward
Extrusion Process by the Finite Element Method," Proc. NAMRC—X,
Hamilton, Canada, p. 143.
26. Oh, S. I., Park, J. J., Kobayashi, S., and Altan, T., (1983), "Application of
FEM Modeling to Simulate Metal Flow in Forging a Titanium Alloy Engine
Disk," Trans. ASME, J. Engr. Ind., Vol. 105, p. 251.
27. Badawy, A., Oh, S. I., and Altan, T., (1983), "A Remeshing Technique for
the FEM Simulation of Metal Forming Processes," Proc. Int. Comp. Eng.
Conf., ASME, Chicago, p. 143.
 10
STEADY-STATE PROCESSES
OF EXTRUSION AND DRAWING

10.1 Introduction
Except at the start and the end of the deformation, processes such as
extrusion, drawing, and rolling are kinematically steady state. Steady-state
solutions in these processes are needed for equipment design and die
design and for controlling product properties.
A variety of solutions for different conditions in extrusion and drawing
have been obtained by applying the slip-line theory and the upper-bound
theorems [1-3]. Early applications of the finite-element method to the
analysis of extrusion [4-6] have been for the loading of a workpiece that
fits the die and container, and for the extrusion of a small amount of it
rather than extruding the workpiece until a steady state is reached. An
exception is the work by Lee et al. [7] for plane-strain extrusion with
frictionless curved dies using the elastic-plastic finite-element method. In
view of the computational efficiency, various numerical procedures par-
ticularly suited for the analysis of steady-state processes have been
developed by several investigators [8-14]. Shah and Kobayashi [8]
analyzed axisymmetric extrusion through frictionless conical dies by the
rigid-plastic finite-element method. The technique involves construction of
the flow lines from velocities and integration of strain-rates numerically
along flow lines to determine the strain distributions. An improvement of
the method was made by including friction at the die-workpiece interface.
The steady-state deformation characteristics in extrusion and drawing were
obtained as functions of material property, die-workpiece interface
friction, die angle, and reduction [15].

10.2 Method of Analysis

In kinematically transient or nonsteady-state forming problems, a mesh
that requires continuous updating (Lagrangian) is used. In steady-state
problems, a mesh fixed in space (Eulerian) is appropriate, since the
process configuration does not change with time. For steady-state prob-
lems whose solutions depend on the loading history or strain history of the
material, combined Eulerian-Lagrangian approaches are necessary
[16,17]. In deformation of rigid-plastic materials under the isothermal
conditions, the solution obtained by the finite-element method is in terms
174
 Steady-State Processes of Extrusion and Drawing 175

of velocities and, hence, strain-rates. In the nonsteady-state processes, the

effective strain-rates are added incrementally for each element to deter-
mine the effective strains after a certain amount of deformation. For the
analysis of a steady-state process, a technique is needed for determining
the distributions of effective strain if the deforming material is a strain-
hardening material. The following method is developed to obtain the flow
lines, distorted grid pattern, and the effective strain distribution.
In the finite-element technique, the strain-rates for the elements are
assumed to be the values at the centers of the elements; the velocities are
known at the nodal points of the elements. The coordinates of these
element centers and nodal points are also known. First, a point is selected
along the entrance boundary of the deformation zone where the effective
strain is zero. From the known coordinates of this point, components of
the velocity and the effective strain-rate are determined by interpolation
from surrounding nodal-point values and from surrounding element-center
values, respectively. These velocities are then incrementally added to the
coordinates of the point to determine its new position. The effective
strain-rate multiplied by the time-increment is added to the effective strain
to determine the effective strain at the new location of the point, and the
procedure is repeated. The procedure of incrementally adding the velocity
and the strain-rate is continued until the point reaches the exit boundary of
the deformation zone. Beyond the exit, the point moves with a velocity
equal to the exit velocity, and the effective strain remains unchanged.
Starting with a different point at the entrance boundary, another flow line
and the values of effective strain along this flow line can be determined in a
similar manner. This procedure yields the entire network of grid distor-
tions and the effective strain distribution.
In nonsteady-state processes, the effect of material work-hardening can
be readily incorporated into the analysis by computing the incremental
strains and modifying the flow stress at each deformation step according to
the increase in the total effective strain. In the analysis of steady-state
processes, however, the flow stress distribution must be consistent with the
final effective strain distribution according to the work-hardening charac-
teristics of the material. This requirement can be achieved by using the
following computational procedure. During iterations to obtain a converg-
ing solution, the flow lines corresponding to the latest velocity field are
constructed after each iteration. The network of grid distortions and the
effective strain distributions are then determined from these flow lines.
The effective strains for all elements are then interpolated from these
values and, using a given stress-strain relationship, corresponding flow
stress distributions are determined for the elements. Using this new flow
stress distribution, the next iteration is carried out to determine the
velocity field, and the same procedure is repeated until a converged
solution is obtained. The solution depends not only on the velocity
distribution but also on the flow stress distribution. Therefore, when the
velocity solution converges, the flow stress and effective strain distributions
176 Metal Forming and the Finite-Element Method

also match each other according to the stress-strain behavior of the

material.

10.3 Bar Extrusion

Boundary Conditions and Mesh System
The boundary conditions and the mesh system used to analyze extrusion
through conical dies are shown in Fig. 10.1 [15]. The material in the
container moves axially with a uniform velocity of unit magnitude. The
container is assumed to be frictionless, and along the conical die surfaces
the tangential traction is equal to the frictional stress at the die-workpiece
interface. The extruded material moves axially with a uniform velocity
whose magnitude is determined by the area reduction and the incompres-
sibility relationship. Also, there are no tractions on the surface of the
extruded part. Along the axis of symmetry of extrusion, the conditions are
such that the shear traction and the radial velocity must vanish.
It must be noted that in performing the computations the die corners
were slightly modified by connecting with a straight line the two material
nodal points located closest to the die corners. This modification was made
in order to avoid singularities in the velocity components near the die
corners.

FIG. 10.1 Boundary conditions and mesh system for steady state axisymmetric extrusion
analysis.
 Steady-State Processes of Extrusion and Drawing 177

Results
The computations were carried out under various extrusion process
conditions for SAE 1112 steel and aluminum alloy 2024-T351. Since the
solution with nonwork-hardening material was used as the initial guess in
analyzing the extrusion of work-hardening materials, the results for
nonwork-hardening materials are also discussed. The computation was
performed for each solution until the error norm of ||Av||/||v|| = 0.000 08
was reached. The number of iterations to reach the above convergence
depends on the initial guess, but the average number of iterations required
for the final solution was about 25-30.
The predicted results were obtained for average extrusion pressure,
normal pressure distribution on the die, grid distortions, and for velocity,
stress, and strain-rate distributions. Some of these are discussed below.
Detailed differences in deformation and f}ow behavior, due to material
properties and friction at the die-workpiece interface, are clearly indicated
in calculated grid distortions in Fig. 10.2. The steady-state grid distortion
patterns are compared for nonwork-hardening and work-hardening (SAE
1112 steel) cases for two friction conditions. Figure 10.2a shows the
distortion of the grid lines that are originally perpendicular to the axis of
the workpiece. It is seen that there is a cusp or double peak on the axis in
the extruded part for both the work-hardening and nonwork-hardening
materials. However, the magnitude of the cusp is smaller with the former
material. With increasing friction at the die-workpiece interface, the

FIG. 10.2 Grid distortion patterns for the extrusion of nonwork-hardening (right halves) and
work-hardening (left halves) materials with (a) frictional stress fs = 0, and (b) fs = 0.4Y0: die
semi cone angle, 30°, R0/R,, = 2.37.
178 Metal Forming and the Finite-Element Method

double-peak or cusp type distortion disappears almost completely for both

work-hardening and nonwork-hardening materials as seen in Fig. 10.2b.
A typical effective strain distribution is shown for a nonwork-hardening
material in Fig. 10.3. The strain has the highest value near the die and the
lowest value near the extrusion axis throughout the deformation zone. The
effective strain distribution at the exit section is also shown in the figure,
which indicates that the strain has its lowest value near the extrusion axis
and increases towards the extrusion surface.
In general, the velocity distributions calculated for aluminum alloy
2024-T351 do not differ significantly from those obtained for SAE 1112
steel. Direct comparisons of the computed velocity fields with those
measured in a visioplasticity study conducted by Shabaik and Thomsen
[18] was possible, and such a comparison is shown in Fig. 10.4. The
computed results are given for the case of a frictionless die. The
comparison of the computed and measured velocity distributions in Fig.
10.4 reveals that the theoretical results, in both velocity components, are
in agreement with the experimental results in magnitude as well as in
trends.
The distribution of the mean stresses for the aluminum alloy is also
similar to the pattern obtained for SAE 1112 steel. As the reduction
decreases, the mean stress increases and becomes tensile in the zone near
the center, as shown in Fig. 10.5. The computations also reveal that with
increasing friction at the die-workpiece interface the mean stress becomes
more compressive throughout the deformation zone. The tensile stresses,
in the central zone of the extrusion, are responsible for the occurrence of
center bursting cracks.

10.4 Bar Drawing

The drawing process is characterized by smaller die angles and smaller
reductions than those used in extrusion. Initially, it appears that the
differences between extrusion and drawing are merely in geometrical
quantities and hydrostatic stress components (extrusion being essentially a
process of compression, and drawing a process of tension). However, this
is not the case and the finite-element results obtained in extrusion cannot
be extrapolated to obtain results in drawing by simply taking into account
the geometrical conditions and the concepts of pushing in extrusion and
pulling in drawing.

Computational Conditions and Procedures

The material selected for the computation was SAE 1144 cold-drawn steel
[15]. The computation was performed for semi-die angles a of 6° and 8°,
two friction conditions, and several reductions.
For computational purposes, the solution for extrusion with a = 45° and
Rn/Re = 1.25 (/?o = billet radius, Re = radius of extruded bar) was used as
an initial guess for drawing with a = 6° and R(,/Re = 1.2 by modifying the
FIG. 10.3 A typical effective strain distribution in extrusion of a nonwork-hardening material.
FIG. 10.4 Comparison of (a) computed and (b) experimental [18] (K = 0 and 20 correspond
to Z/R0 = Q and 1.0, respectively) velocity distributions in extrusion; R0/Re = 2, semi die
angle a = 45° (frictionless in computation).

180
 Steady-State Processes of Extrusion and Drawing 181

FIG. 10.5 The mean stress distribution in extrusion through a frictionless die at R0/RC =
1.25.

solution according to geometrical proportions. Thirty-four iterations were

required to achieve the converged solution. This solution was then used as
an initial guess for the computation of all the other drawing conditions.
The convergence was excellent and only 6—10 iterations were necessary to
obtain the solutions in most cases.
Results
The results obtained for drawing are examined in detail for a = 6°. The
variation of effective strain distribution in drawn bars is similar to that
found in extrusion, but the magnitudes are smaller owing to smaller die
angles and reductions in drawing. By comparing the results of drawing
with those of extrusion, it is found that the die angle is the most important
variable in controlling the nonuniformity of the strains and properties in
extruded or drawn bars.
The die pressure distribution is plotted for various reductions and for
two die friction conditions in Fig. 10.6. A higher die pressure is obtained
for a smaller reduction, as was the case for extrusion. However, contrary
to the case in extrusion, the die pressure is higher in frictionless drawing
than in drawing with friction. These findings on die pressure are in
agreement with experimental results [19] and with the results predicted by
the slip-line theory. Further observations of the results for drawing and
comparison of the results with those of extrusion revealed the following.
Although materials properties somewhat influence metal flow, as seen in
grid distortion patterns, their effects on the overall deformation and flow
behavior appear to be insignificant in extrusion and drawing processes.
182 Metal Forming and the Finite-Element Method

FIG. 10.6 The die pressure distributions in drawing for various reductions and for the two die
friction conditions (die semi-cone angle = 6°).

Among other variables, friction at the die-workpiece interface plays an

important role in determining the detailed mechanics in these processes.
With increasing friction, the degree of grid distortion becomes larger and
the size of the deformation zone expands in both processes, but the effect
of friction on nonuniformity of product properties is less significant. The
main difference between the extrusion process and the drawing process can
be found in the die pressure distribution and the distribution of hydrostatic
stress components. In both processes, the die pressure decreases with
increasing reduction in area. However, the die pressure is greater for
larger interface friction in extrusion, while the reverse is true in drawing.
 Steady-State Processes of Extrusion and Drawing 183

Similarly, in extrusion the hydrostatic stress component decreases with

increasing interface friction, but in drawing this value increases with
increasing friction.

10.5 Multipass Bar Drawing and Extrusion

In practice, many bar drawing and extrusion processes are multipass
operations. Internal fracturing (center bursting) in these processes usually
occurs after significant area reduction through several passes. In order to
predict and prevent the occurrence of such defects, it is necessary to
determine the deformation characteristics involved in multipass opera-
tions. For this purpose, commercially pure aluminum was considered for
simulation. The effect of work-hardening was examined in single-pass
operations, and the effect of flow stress distributions on the flow pattern
was revealed by analyzing two-pass drawing and extrusion processes [20].
Extrusion
In Fig. 10.7, the steady-state grid distortion patterns are compared for the
two passes. In both extrusion and drawing, deformation behavior during

FIG. 10.7 Comparison of grid distortions in extrusion of pure aluminum for the first and
second passes (/, = 0).
184 Metal Forming and the Finite-Element Method

FIG. 10.8 Stress and srain histories of a material point along the axis of (frictionless)
extrusion.

the second pass is generally closer to that of nonwork-hardening materials,

in comparison with the first-pass deformation. The critical site for the
occurrence of center bursting in extrusion is along the axis of extrusion.
The stress and strain histories are important in applying ductile fracture
criteria. Figure 10.8 shows these stress and strain histories of a point
moving along the axis of extrusion for the frictionless case. The axial stress
becomes tensile over a certain range of deformation and a larger tension
zone is obtained for nonwork-hardening materials than for work-hardening
materials. The tensile stress increases during the second pass. It was found
that the interface friction had the effect of decreasing the axial stress, thus
suppressing the possibility of producing center bursting. These results are
qualitatively in agreement with experimental observations.
Drawing
The hydrostatic component distributions in two-pass drawing are given in
Fig. 10.9. The top figure is for nonwork-hardening material and is valid for
both first and second passes. The rigid zone boundaries are also shown in
the figure. The size of the deformation zone in the first pass is larger with
work-hardening materials than that for nonwork-hardening materials.
However, the deformation zone becomes smaller in the second pass and is
almost identical to that for nonwork-hardening materials. The strain
distributions of the work-hardening materials during the second pass are
closer to those for a nonwork-hardening material, although the differences
between the strains for the first pass of work-hardening materials and of
nonwork-hardening materials are not great. The hydrostatic stress com-
ponent distributions revealed that the region under the die is in a state of
compressive hydrostatic stress, while the core and exit regions are in a
state of tensile hydrostatic stress. The mean stress becomes more tensile
FIG. 10.9 Hydrostatic pressure distributions in two-pass drawing (die semi-cone angle, 6°
R0/«e = l.l,/ s = 0.4F0).

185
186 Metal Forming and the Finite-Element Method

FIG. 10.10 Stress and strain histories of a material point along the axis of drawing.

with larger die friction. This is contrary to the results obtained in

extrusion. The stress and strain histories along the axis of drawing are
plotted in Fig. 10.10. The peak axial stress appears near the entrance; for
extrusion it is observed near the exit. The axial stress is greater for
nonwork-hardening materials than for work-hardening ones during the first
pass, but increases during the second pass. This is also true for extrusion,
as seen in Fig. 10.8, suggesting an increased possibility of center burst
formation in multipass processes.

10.6 Applications to Process Design

An important application of metal-flow simulation is determining the
process conditions necessary to prevent the formation of defects. The
predicted stress-strain histories along the axis of extrusion and drawing,
discussed above, when combined with the ductile fracture criterion, can be
utilized in selecting the preferred process conditions, such as reduction in
area, semi-die angles and die-workpiece interface friction, so that center
bursting can be avoided [21].
Another significant application concerns computer-aided design of
extrusion dies. Streamlined die design is now being •considered in extrusion
of certain structural shapes and workpiece combinations for a number of
applications in the aerospace industry. These components generally have
complex part geometries, and the workpiece material in general has poor
workability. This combination of conditions can be found in almost every
structural alloy family, including aluminum, steel, titanium, and nickel-
based alloys. Failure in these alloy systems is usually attributed to
center-burst phenomena or porosity due to the decohesion of second-phase
particles from the matrix material. The use of the finite-element method in
 Steady-State Processes of Extrusion and Drawing 187
metal-flow simulation for the optimum die design in extrusion has been
investigated by Gegel et al. [22]. They performed extrusion simulations
with a variety of die geometries, including shear, conical, parabolic, and
streamlined dies. They have developed various optimization criteria, and
with these criteria and metal-flow simulations they successfully designed
streamlined dies for extrusion. Research in this area will result in further
applications of finite-element metal-flow analysis for process design and
optimization.

References
1. Johnson, W., Sowerby, R., and Venter, R. D., (1982), "Plane-Strain Slip Line
Fields for Metal Deformation Processes," Pergamon Press, Oxford.
2. Johnson, W., and Kudo, H., (1962), "The Mechanics of Metal Extrusion,"
Manchester University Press, Manchester, UK,
3. Avitzur, B., (1983), "Handbook of Metal Forming Processes," Wiley, New
York.
4. Murota, T., Jimma, T., and Kato, K., (1970), "Analysis of Axisymmetric
Extrusion," Bull. JSME, Vol. 13, p. 1366.
5. Iwata, K., Osakada, K., and Fujino, S., (1972), "Analysis of Hydrostatic
Extrusion by the Finite Element Method," Trans. ASME, J. Engr. Ind., Vol.
94, p. 697.
6. Lee, C. H., Iwasaki, H., and Kobayashi, S., (1973), "Calculation of Residual
Stresses in Plastic Deformation Processes," Trans. ASME, J. Engr. Ind., Vol.
95, p. 283.
7. Lee, E. H., Mallet, R. L., and Yang, W. H., (1976), "Stress and Deformation
Analysis of the Metal Extrusion Process," SUDAM No. 76-2, Stanford
University, June.
8. Shah, S. N., and Kobayashi, S., (1977), "A Theory on Metal Flow in
Axisymmetric Piercing and Extrusion," J. Prod. Engr., Vol. 1, p. 73.
9. Zienkiewicz, O. C., and Godbole, P. N., (1974), "Flow of Plastic and
Viscoplastic Solids with Special Reference to Extrusion and Forming Proc-
esses," Int. J. Num. Meth. Engr., Vol. 8, p. 3.
10. Zienkiewicz, O. C., Jain, P. C., and Onate, E., (1978), "Flow of Solids
During Forming and Extrusion; Some Aspects of Numerical Solutions," Int. J.
Solids Structures, Vol. 14, p. 15.
11. Zienkiewicz, O. C., Onate, E., and Heinrich, J. C., (1978), "Plastic Flow in
Metal Forming, I. Coupled Thermal Behavior in Extrusion. II Thin Sheet
Forming," Applications of Numerical Methods to Forming Processes, ASME,
AMD, Vol. 28, p. 107.
12. Dawson, P. R., (1978), "Viscoplastic Finite Element Analysis of Steady State
Forming Processes including Strain History and Stress Flux Dependence,"
Applications of Numerical Methods to Forming Processes, ASME, AMD,
Vol. 28, p. 55.
13. Dawson, P. R., and Thompson, E. G., (1978), "Finite Element Analysis of
Steady-State Elasto-Visco-Plastic Flow by the Initial Stress-Rate Method," Int.
J. Num. Meth. Engr., Vol. 12, p. 47.
14. Derbalian, K. A., Lee, E. H., Mallet, R. L., and McMeeking, R. M., (1978),
"Finite Element Metal Forming Analysis with Spacially Fixed Mesh," Appli-
cations of Numerical Methods to Forming Processes, ASME, AMD, Vol. 28,
p. 39.
15. Chen, C. C., Oh, S. I., and Kobayashi, S., (1979), "Ductile Fracture in
188 Metal Forming and the Finite-Element Method

Axisymmetric Extrusion and Drawing—Part I, Deformation Mechanics of

Extrusion and Drawing," Trans. ASME, J. Engr. Ind., Vol. 101, p. 23.
16. Schreura, P. J. G., Veldpaus, F. E., and Brekelmans, W. A. M., (1982), An
Arbitrary-Eulerian-Lagrangian Finite Element Model for the Simulation of
Geometrical Non-Linear Hyper-Elastic and Elasto-Plastic Deformation Proc-
esses, in "Numerical Method in Industrial Forming Processes," (Edited by J.
F. T. Pittman et al.), Pineridge Press, Swansea, UK, p. 491.
17. Huetink, J., (1982), Analysis of Metal Forming Processes Based on a
Combined Eulerian-Lagrangian Finite Element Formulation, in "Numerical
Method in Industrial Forming Processes," (Edited by J. F. T. Pittman et al.),
Pineridge Press, Swansea, UK, p. 501.
18. Shabaik, A. H., and Thomsen, E. G., (1968), "Investigation of the Applica-
tion of the Visioplasticity Methods of Analysis to Metal Deformation
Processes," Final Report—Part II, Department of the Navy.
19. Wistreich, J. G., (1955), "Investigation of the Mechanics of Wire Drawing,"
Proc. Inst. Mech. Engr., Vol. 169, p. 654.
20. Chen, C. C., and Kobayashi, S., (1978), "Deformation Analysis of Multi-Pass
Bar Drawing and Extrusion," Annals CIRP, Vol. 27, No. 1, p. 151.
21. Oh, S. I., Chen, C. C., and Kobayashi, S., (1979), "Ductile Fracture in
Axisymmetric Extrusion and Drawing, Part 2: Workability in Extrusion and
Drawing," Trans. AMSE, J. Engr. Ind., Vol. 101, p. 36.
22. Gegel, H. L., Malas, J. C., Gunasekera, J. S., Morgan, J. T., and Doraivelu,
S. M., (1984), "Computer-Aided Design of Extrusion Dies by Metal-Flow
Simulation," AGARD Lecture Series, No. 137, Process Modeling Applied to
Metal Forming and Thermomechanical Processing, North Atlantic Treaty
Organization, p. 8-1.
 11
SHEET-METAL FORMING

11.1 Introduction
The stress-state is said to be plane when the direction normal to the plane
is a principal stress direction and the magnitude of the stress in this
direction is zero. This situation occurs when a sheet is loaded along its
edges in the plane of the sheet. In-plane deformation of sheet metal, such
as bore expanding and flange-drawing, is an example of plane-stress
problems. For out-of-plane deformation of sheet metals, such as punch
stretching, sheet bending, and cup drawing, a simple analytical method is
the use of membrane theory. This theory neglects stress variations in the
thickness direction of a sheet and considers the distribution of stress
components only in the plane of the sheet. Thus, the basic formulations for
the analysis of both in-plane and out-of-plane deformations contain only
the stress components acting in the plane of the sheet. However, the
analysis of out-of-plane deformation requires consideration of large
deformation, while the infinitesimal theory is applicable for in-plane
deformation analysis.
Many materials employed in engineering applications possess mechani-
cal properties that are direction-dependent. This property, termed
anisotropy, stems from the metallurgical structure of the material, which
depends on the nature of alloying elements and the conditions of
mechanical and thermal treatments. Metal sheets are usually cold-rolled
and possess different properties in the rolled and transverse directions.
Therefore, in sheet-metal forming in particular, the effect of anisotropy on
the deformation characteristics may be quite appreciable and important.
In the past the calculation of the detailed mechanics of large plastic
deformation of metal sheets has been achieved with some success by
numerical methods. However, without exception, these studies have dealt
with deformations that possess a high degree of symmetry, and were
concerned with the anisotropy existing only in the direction of sheet
thickness (normal anisotropy). Methods that are capable of solving
nonaxisymmetric problems in forming of anisotropic sheet metal are still
being sought. The finite-element method is one of those methods. It was
applied to the elastic-plastic analysis of nonaxisymmetric configurations of
sheet stretching with normal anisotropy by Mehta and Kobayashi [1].
Yamada [2] presented a stress-strain matrix for a material that is
189
190 Metal Forming and the Finite-Element Method

elastically isotropic, and that obeys Hill's anisotropic yield criterion in the
plastic range. He treated the incipient deformation of a circular blank of
anisotropic material in the flange drawing process. For the deformation
analysis of a metal sheet having three mutually perpendicular axes of
anisotropy (orthotropic material), the rigid-plastic finite-element method
has been applied to plane-stress bore expanding and flange drawing [3,4].
To analyze out-of-plane forming processes, Wang [5] proposed a method
of solution in which two spatially independent variables are required to
define the geometry. The method is based on a variational procedure and
assumes that the material of the sheet is rigid-plastic. On the same
principle, Kim et al. [6, 7] analyzed three processes, namely, the bulging of
a sheet subjected to hydrostatic pressure, the stretching of a sheet with a
hemispherical punch, and the deep drawing of a sheet with a hemispherical
punch.
Using the elastic-plastic approach, complete solutions of stretch-forming
and deep-drawing problems, taking into account the contact problem at
the blank holder, die, die profile, and punch head, were obtained by Win
[8]. On the basis of the nonlinear theory of membrane shells, Wang and
Budiansky [9] developed a procedure for calculating the deformations in
the stamping of sheet metal by arbitrarily shaped punches and dies.
Onate and Zienkiewicz [10] presented a finite-element formulation
based on an extension of the general viscoplastic flow theory for
continuum problems to deal with thin shells.
Toh and Kobayashi [11,12] analyzed sheet-metal forming processes,
axially symmetric and nonsymmetric, by the finite-element method based
on the membrane theory. The finite-element model takes into account the
rigid-plastic material characteristics and includes the normal anisotropy of
the sheet metal as well as the finite deformation that occurs during the
sheet-forming process.
In this chapter, first the yield function for anisotropic materials is
defined and in-plane deformation processes are analyzed using the
infinitesimal theory. Then the large-strain formulation using membrane
theory is described for the analysis of out-of-plane deformation processes.
The formulation is extended to sheet-metal forming of general shapes and
applied to square-cup drawing. Finally, in Section 11.8 we discuss the
nonquadratic yield function for anisotropic materials.

11.2 Plastic Anisotropy

Considering states of anisotropy that possess three mutually orthogonal
planes of symmetry at every point, the simplest yield criterion for
anisotropic materials is one that reduces to the distortion energy criterion
when the anisotropy is vanishingly small. Hill [13] assumed the yield
criterion to be quadratic in the stress components as follows:
 Sheet-Metal Forming 191

where the orthotropic axes are taken as the coordinate axes (x, y, z) and
F, G, H, L, M, N are anisotropy parameters. The parameters in eq. (11.1)
are not definite but their ratios are for defining the behavior of a given
material. Thus, eq. (11.1) is rewritten as

where /(cry) is the plastic potential, o is the effective stress, and /, g, h, I,

m, n are pure numbers. The coefficients/, g, h, I, m, n in eq. (11.2) are
subjected to conditions depending upon the selection of a as a reference
material property.
If Yx, Yy, Yz are the tensile yield stresses in the principal directions of
anisotropy, it is easily shown from eq. (11.2) that

If a in eq. (11.2) is defined as

then

Under plane-stress conditions (oz = TZX = ryz = 0), eq. (11.2) reduces to

It should be noted again that the ratios of anisotropic parameters, not

their absolute values, in eq. (11.4a) define the state of anisotropy. This is
obvious, if we use, for example, the condition of eq. (11.3b) corresponding
to the definition of a expressed by eq. (11.3a), and write eq. (11.4a) in the
form

Supposing that/(cr,y) in eq. (11.4a) is the plastic potential, the strain-rate

relations from the flow rule, referred to the principal axes of anisotropy,
are
192 Metal Forming and the Finite-Element Method
The effective strain-rate £ in eq. (11.5), corresponding to the effective
stress a denned in eq. (11.4a) can be obtained first by inverting eq. (11.5)
as

and then substituting eq. (11.6) into eq. (11.4a). We have

Frequently, the anisotropy of a material is represented by the strain ratios

known as /"-values. For a tensile specimen cut at an angle a to the rolling
direction (x direction)

where o is the tensile yield stress. The ratio of the transverse to the
through-thickness strain (or strain-rate) is

Substitution of eq. (11.5), using the stresses given by eq. (11.8), into (11.9)
results in

From eq. (11.10)

where rx, r45 and ry are the /--values for a = 0° (x direction), 45°, and 90° (y
direction), respectively. For the anisotropy to be rotationally symmetric
about the z axis, f=g and n=f + 2h (and / = m) in addition to the
condition given by eq. (11.3b). If there is complete spherical symmetry, or
isotropy,

11.3 In-plane Deformation Processes

Finite-Element Formulation
Under the plane-stress condition, the mean stress om is expressed in terms
of strain-rate components. Therefore, the Lagrange multiplier is not
 Sheet-Metal Forming 193

necessary, and the variational formulation becomes

The volume constancy is not required in the assumed velocity field

because ez is now treated as a dependent variable. For an isoparametric
quadrilateral element, shown in Fig. 8.1 (Chap. 8), the matrices necessary
for discretization of eq. (11.11) in the two-dimensional case have been
derived in Chap. 6, and some of them are recapitulated in Chap. 8.
However, the matrix D in the definition of e according to

differs for anisotropic materials from that for isotropic materials. With
reference to eq. (11.7), the matrix D for orthotropic materials, becomes

with the strain-rate vector defined by e T = {ex, ey, yxy}.

In the case of normal anisotropy (isotropy in the plane), f=g and
n =f + 2h or, in terms of r-values, rx = ry = r45 = r and h/f = h/g = r and
nig = 1 + 2r. Thus, the effective stress and the effective strain-rate from
eqs. (11.4a) and (11.7), respectively, become

Since/ + g + h = 3, eq. (11.13a), in terms of r-values, becomes

If we select a = Yx = Yy as a reference material property and note that

0£ = Wp (plastic work-rate), we have

and
194 Metal Forming and the Finite-Element Method

The expressions for the matrix D corresponding to eq. (11.13a), (11.13b),

and (11.13c), from eq. (11.12), become, respectively,

and

For isotropic materials the r-value in eqs. (11.14) is equal to unity. It is to

be noted that the strain-rate component ez is eliminated in expressing e
through the condition that ez = —(ex + ey) for the case of plane stress.
Bore Expanding and Flange Drawing
The problems considered for the analysis are bore expanding and flange
drawing; these are shown schematically in Fig. 11.1.
In the bore-expanding process, a circular blank with a concentric hole is
stretched radially. The stretching is accomplished by drawing the blank
through a die opening, using a flat-bottomed punch (Fig. 11.la). The outer
periphery of the blank is fixed by a holder. To analyze the expansion of the
hole of radius /?0 in the blank of initial thickness T0, a uniform radial
displacement 6b is applied along the boundary of radius Rb. The specimen
dimension selected for the analysis is Rb/Rn = 4.4.
For flange drawing, the process is simulated by considering a circular
sheet of initial radius Rb that is subjected to a uniform radial displacement
towards its center, along the circular boundary of constant radius R0 (Fig.
ll.lb). Similarly to bore expanding, this boundary condition is imposed by
applying the displacement 60 along the boundary of radius R0. The
drawing ratio selected for the analysis was Rb/R0 = 2.5.
The stress-strain property of the material used for the computation is
expressed by

with F0 = 23,600psi (162.7 MN/m2).

Two materials were considered for the calculation. One is a material
with planar anisotropy with r-values of rx = 1.69, r45 = 1.43, and ry = 2.24,
which correspond to those of Al-killed steel. The other material is of
normal anisotropy with rave = \(rx + 2r45 + ry) = 1.70 and has the identical
effective stress-strain curve.
 Sheet-Metal Forming 195

(a) BORE EXPANDING (b) FLANGE DRAWING

FIG. 11.1 The processes of bore expanding and flange drawing.

In Fig. 11.2, the strain distributions for normal anisotropy and those in
the direction a = 45° for planar anisotropy are compared (ah the angle
from the principal axis of anisotropy). The distributions in the directions of
a = 5° and 85° are almost identical to those obtained for normal aniso-
tropy. The formation of "ears" and "hollows" in flange drawing is shown
in Fig. 11.3a and the corresponding strain contours are given in Fig. 11.3b.
The ears appear along the axes of anisotropy and a hollow is formed at
approximately a = 47° at an early stage and changes its location slightly
near the end of the process. According to Hill [13], the principal axes of
stress and strain-increment coincide for the orientations a = 0°, a ( = 49°),
and 90° for the present planar anisotropy; the ears are formed at a = 0°
and 90°, and a hollow at a = (90 - a)° = 41°. The location of the hollow,
shown in Fig. 11.3, is not in agreement with Hill's prediction, but is closer
to that for minimum r-value (a = 51°).

11.4 Axisymmetric Out-of-plane Deformation

Variational Formulation
The variational formulation used in the in-plane deformation process given
by eq. (11.11) is inadequate for analyzing out-of-plane deformation
196 Metal Forming and the Finite-Element Method

R A D I U S / R0
FIG. 11.2 Strain distributions at various deformation stages of bore expanding.

processes such as punch-stretching and cup-drawing operations. Consider a

deformation of a circular sheet of radius R, as shown in Fig. 11.4. The
velocityt of the sheet is given by

where «0 is a constant.
It is easily shown that all the components of strain rate, denned by
£ij = 2(MI,/ + "/,;)> become zero, even though the length of the sheet would
change after a finite time interval A?. This suggests that there is at least
one deformation mode that cannot be determined by the variational
principle formulated by eq. (11.11). The zero-energy deformation mode

tFrom Section 11.4 through Section 11.7, where finite strain formulation is utilized, u, is
used as a component of displacement, ut for a velocity component, and v, for a nodal point
displacement.
FIG. 11.3 (a) Deformed grid pattern in flange drawing, (b) Effective strain contours in flange
drawing.

197
198 Metal Forming and the Finite-Element Method

o R
FIG. 11.4 Out-of-plane deformation of a ring element.

given in Fig. 11.4 is the result of incomplete description of deformation by

the membrane theory.
To obtain an adequate description utilizing the membrane theory, it is
assumed that the mode of deformation can be determined by the
variational functional integrated over the finite time-increment Af, namely,

where dn-m denotes the variational functional for infinitesimal deformation

given by eq. (11.11).
In axisymmetric deformation of a membrane, the principal stress
directions remain the same with respect to the material fiber, i.e., the
meridian and circumferential directions, during deformation. We define
the logarithmic strain-increment during the time-increment Af as

In eq. (11.16), A*0 and As are the lengths of an incremental line element
in the meridian direction before and after the time increment, respectively,
and R0 and R are the radial positions of a material point before and after
the time increment, respectively. The components of the strain-rate can be
expressed by

For axisymmetric deformation, the sheet is assumed to have normal

anisotropy, represented by the r-value, and the constitutive equation,
corresponding to eq. (11.6), can be expressed by
 Sheet-Metal Forming 199
where

and

Note that the components of strain-rate and stress are functions of time,
and that the coordinate axis with which these quantities are denned rotates
with the material fiber. Owing to the work-hardening, a in eq. (11.18b) is
also a function of time and is expressed as

where H' = do/de and is the slope of the true stress vs. true strain curve of
the material.
Substituting eqs. (11.17) and (11.19) into eq. (11.15), we can readily
obtain the variational functional integrated over a time-increment At as

where T is the thickness of the sheet, A is the surface area of the sheet,
and the effective strain increment AE is expressed as

assuming that strain-rate ratios remain constant during a time-increment

Anneq. (11.18c).
Finite-Element Formulation
In the finite-element formulation, the sheet geometry is approximated by a
series of conical frustums, as shown in Fig. 11.5, treating each conical
frustum as a line element. The displacement within the element is
expressed by

where u^a) and u^ are the radial and axial displacements at the ath node,
respectively, and the summation is performed for the two nodes. The
shape function qa is written in the natural coordinates as

where - l < f <1.

Since the element used is linear, the meridian strain-increment is
constant in each element. The lengths of the undeformed and deformed
200 Metal Forming and the Finite-Element Method

FIG. 11.5 Approximation of the sheet geometry by a series of conical frustrums.

elements, sn and s, are expressed by

The radial positions of a material point before and after the deformation,
,/?0 and R, can be expressed by

The logarithmic strain-increment can be calculated by simply substituting

eqs. (11.22) and (11.23) into eq. (11.16). The effective strain-increment
can be written in vector form as

where AET = {A.E,, A£e} and D is a matrix given by

similar to eq. (11.14c).

The solution procedure for the functional given by eq. (11.20) is similar
to that described in Chap. 6. The first derivative of the functional with
respect to nodal displacement v, can be expressed by

where
 Sheet-Metal Forming 201

The second derivative of the functional can be written as

where

Note that the derivatives of strain components are given by

Assembling the eqs. (11.24) and (11.25) for all elements in the finite-
element scheme, we obtain the linear simultaneous equations

The solution of eq. (11.26) can be obtained by the technique described in

Chap. 7. The finite-strain formulation described in this section was derived
by Kim et al. [6,7]. This formulation is valid for rate-insensitive materials.
Similar formulations that are based on rigid-viscoplastic material have
been derived by Rebelo and Kobayashi [14], and also by Park et al. [15].
In both formulations, strain-rate is defined by A£/Af, which is an average
strain rate over a time-interval A?.

11.5 Axisymmetric Punch-Stretching and Deep-Drawing Processes

Punch-stretching and deep-drawing processes with hemispherical punches
are analyzed by using the formulation described in Section 11.4 [6,7].
Boundary Conditions with Hemispherical Punch
The boundary conditions in these processes are prescribed not only by
tractions and incremental displacements but also by their ratios. Further,
the radial and axial positions of the material elements in the contact region
are not independent; they are related to each other through a mathemati-
cal expression representing the geometrical requirement that they must be
actually on the surface of the punch head. The hemispherical surface of the
punch head (Fig. 11.6) can be expressed by

where r0 and z0 are radial and axial positions of a nodal point at the
present configuration, ur and uz are the increments of radial and axial
202 Metal Forming and the Finite-Element Method
z

FIG. 11.6 Geometrical requirement for a node on the contact surface.

displacements, and c is a parameter related to the punch depth d as shown

in Fig. 11.6.
Recall that the finite-element formulation has already been linearized
and what it really solves for are the perturbation terms. Therefore, we also
linearize the boundary condition (11.27) in terms of perturbation (or
correction) of displacements ur and uz, as

where starred (*) quantities are initial guesses and Aw r , A«2 are perturba-
tions. By rearranging eq. (11.28) we have

where

and

In order to implement Coulomb friction between sheet and punch, a

tangential frictional stress is assumed and a converged solution is obtained.
Then, from the computed nodal-point forces, the friction coefficient is
calculated. If the computed friction coefficient is not what was intended,
then tangential force is modified and the solution process is repeated. It
 Sheet-Metal Forming 203
should be noted here that the correction of frictional force requires the
modification only in the F vector in eq. (11.20). The deformation step is
controlled by the punch-head increment. The optimum step size was
shown to be the one that results in the effective strain-increment of 0.04
[16].
Punch Stretching
The present rigid-plastic finite-element method has been compared else-
where [6] with the finite-difference methods by Wang and by Woo. The
agreement of the solutions from the two numerical methods was excellent.
Here, the rigid-plastic finite-element solutions are first compared with
experiments. The effect of interface friction on the detailed mechanics is
then examined by evaluating the results of the computations.
1. Comparison with experiments. For soft copper (experiment by
Kaftanoglu and Alexander [17]), the parameters are:
• Stress-strain characteristics: o = 68394(0.0122 + g)° 3789 psi
= 471.56(0.0122 + ef3789 MN/m2
• Thickness: 1.219mm (0.048 in.)
• Friction condition: PFTE film lubricant
• Radius of the sheet: 18.21 mm (0.717 in.)
• Punch radius: 16.51mm (0.65 in.)
Kaftanoglu reports that the friction condition changes with deformation
and measures three different friction coefficients: ^ = 0.2 at stage 1,
/* = 0.135 at stage 2, and // = 0.07 at stage 3. To include the variation of
the friction coefficient in the analysis, more information on the friction
history is needed. Therefore, as a representative value, we use the mean of
three values of the friction coefficient, ^ = 0.135, for our computation.
Figures 11.7 and 11.8 show the distribution of the circumferential strain

FIG. 11.7 Comparison of the circumferential strain distributions between computed and
experimental [17] results, at three punch positions.
204 Metal Forming and the Finite-Element Method

FIG. 11.8 Comparison of the computed results with experiment [17] for thickness strain
distribution, at three punch positions.

and thickness strain. The agreement between the experimental data and
the numerical solution is reasonable, considering the fact that the exact
friction condition is not known.
2. Effect of friction. The parameters used for computation are:
• Stress-strain characteristics: a = Ce° 23 (C a constant)
• Normal anisotropy: r = 1.27
• Punch radius: Rp = 1.0
• Radius of the blank: Rb = 1.0

The material that corresponds to the characteristics above is aluminum-

killed steel. With other parameters held fixed, three different friction
coefficients, n = 0.1, 0.2, and 0.3, are assumed. As shown in Fig. 11.9, in
general friction does not affect the punch load, even though punch load
slightly increases with the increase in friction at later stages of deforma-
tion. The arrows in Fig. 11.9 indicate the limiting stage when the polar
region of the sheet bounded by the ring element becomes rigid. This rigid
zone is observed numerically when the computed value of the effective
strain-increment is vanishingly small.
The effect of friction on the strain distributions is shown in Fig. 11.10.
At a given punch depth, lower friction gives more thinning over the zone
that is in contact with the punch, while less thinning is observed in the
unsupported region. Over the unsupported region, the frictional effect on
the strain distribution is not as significant as in the contact region, and at
the initial stages of deformation it is almost unnoticeable.
Deep Drawing with Hemispherical Punch [6]
In deep drawing, a circular sheet of metal is placed between the blank
holder and the die and then fully drawn into the shape of a cup by a punch
(see Fig. 2.27 in Chap. 2 for the process description). The formability is
measured by the maximum blank size that can be drawn without a failure,
 Sheet-Metal Forming 205

FIG. 11.9 Punch load vs. punch depth in deep drawing with hemispherical punch for
different coefficients of friction.

or more often, by the ratio of the blank diameter to the punch diameter.
This ratio is called the limiting drawing ratio and the test used to determine
the limiting drawing ratio is called the Swift test.
Deep drawing is not only a useful method of material testing, but also
one of the basic operations in sheet-metal stamping. In practice, various
shapes may be used for the bottom of the punch; however, most past

CURRENT RADIUS /BLANK RADIUS

FIG. 11.10 Effect of friction on thickness strain distribution at punch depth (d/R ) = 0.25
and 0.45.
206 Metal Forming and the Finite-Element Method
investigations have been on deep drawing with a flat-bottomed punch.
A difference between deep drawing with a hemispherical punch and
punch stretching with a round die corner is the presence of the flange,
which is free to slide over the die. The blank holding force is implemented
in the formulation as a tangential friction force acting on the last node
located at the rim of the sheet. The increment of deformation is controlled
by the punch movement.
Woo [18] analyzed drawing with hemispherical punch by dividing deep
drawing into two component processes of (1) pure radial drawing over the
flange and (2) punch stretching over the hemispherical punch in his
analysis. He used the finite-difference method, and also conducted
experiments. The parameters used by Woo are:
• Material: Soft copper
• Stress-strain characteristics: a = 5.4 + 27.8(e)0504 ton/in2
= (0.082 08 + 0.422 568e0 504)
xlO 9 N/m 2 (£<0.36)
= 5.4 + 24.4(e)°-375 ton/in 2
= (0.082 08+ 0.3708e0'375)
xlO 9 N/m 2 (g>0.36)
• Blank radius: 55.88mm (2.2 in.), thickness: 0.889mm (0.035 in.)
• Radius of die throat: 53.92mm (2.123 in.)
• Radius of die profile: 12.7mm (0.5 in.)
• Radius of punch head: 25.4mm (1.0 in.)
• Blank holding force: 500kg (0.5 ton.)
As shown in Figs. 11.11 and 11.12, the solution by the rigid-plastic FEM
is in excellent agreement with the experiment for the flange part: however,
over the punch head it predicts more straining than does the experiment
when the friction coefficient of 0.04, as Woo suggested, is assigned for the
contact regions over the punch head and over the die. The deviation of the
numerical solution from the experimental data becomes larger as deforma-
tion progresses, which is reflected in the punch load vs. punch depth
relationship shown in Fig. 11.13.
The lubricant used in the experiment was graphite in tallow. In the
analysis the practical difficulty always lies in the determination of a
reasonable value of friction coefficient, because in real sheet-metal forming
operations, the friction coefficient is difficult to measure and may even
change during deformation.

11.6 Sheet-Metal Forming of General Shapes

Variational Formulation
As shown in Section 11.4 for axisymmetric deformation, the solution of the
out-of-plane deformation of general shapes can be obtained through the
 Sheet-Metal Forming 207

FIG. 11.11 Distribution of thickness strains for [ip=Q.Q4, ;irf = 0.04, and comparison with
experiment [18].

functional integrated over the finite time duration At. It has been shown in
Section 11.4 that the variational functional dn is integrated over the finite
time-increment A/ can be expressed in terms of the logarithmic strain-
increment, provided that the directions of principal strains are known in
advance. For a general out-of-plane deformation, the principal stretch
direction is not known in advance owing to the shear strain component in
the plane of the sheet.

FIG. 11.12 Distribution of circumferential strains for ft p =0.04, jUrf = 0.04, and comparison
with experiment [18].
208 Metal Forming and the Finite-Element Method

Punch Depth
FIG. 11.13 Punch load vs. punch depth, comparison of predictions with experiment [18].

In order to simplify the variational functional, it is assumed that the

rigid-plastic constitutive equation given by eq. (11.6) holds between the
Piola-Kirchhoff stress (second) stj and the Lagrangian strain-rate Etj with
the reference state at t0, which is the beginning of the small but finite
time-increment Af. The Piola-Kirchhoff stress and Lagrangian strain-rate
were introduced in Chap. 4. The components of Lagrangian strain are
expressed by

where «, is a component of displacement. Note that the displacement of a

particle at t = t0 is zero. The component of the Lagrangian strain-rate is
expressed by

where the dot indicates the time derivative. The Piola-Kirchhoff stress sap
is related to the Cauchy stress a,-,- according to

where / is the determinant of the deformation gradient matrix and the

deformation gradient xit<x is expressed by

Noting that o,-, = s/,- at t = t0, the variational functional integrated over a
time-increment At is obtained, similarly to eq. (11.20), as
 Sheet-Metal Forming 209

where h' = ds/dE, and s is defined in terms of s,y in the same form for a
given by eq. (11.13c).
The slope h' is related to H'( = dolde) according to h' =H' -2o(t0).
The derivation of this relationship can be found in References [11] and
[19].
The effective strain-increment can be written in the matrix form as

where

The matrix D in eq. (11.35) is identical to eq. (11.14c). Once the solution
of the displacement increment is obtained, the geometry is updated. The
updated geometry becomes the reference state for the next time increment
Af.

Finite-Element Formulation
Let the sheet-metal domain be decomposed into an assemblage of linear
triangular elements. A local Cartesian coordinate system is assigned to
each element in such a way that the (x-y) coordinate plane coincides with
that of the element plane and the z axis is normal to the element plane.
The displacement field within the element can be described by

where qa is the shape function of the linear triangular element given in eq.
(6.10) in Chap. 6. Equation (11.36) can be rewritten in a matrix form as

where vT={41), u«\ «<'>, u™, uf\ u(?, u?\ uf\ u< 3 >}.
The component of displacement gradient can be written as

where Xa and Ya are defined in eq. (6.33a) in Chap. 6.

The components of Lagrangian strain-increment defined by eq. (11.30)
210 Metal Forming and the Finite-Element Method
are expressed in terms of displacement gradient by

The first derivative of the functional, eq. (11.34), with respect to the
nodal displacement can be written

where

Here, we have

etc.
Note that 3AE/3u7 is also a function of the nodal displacements,
because of the nonlinear terms involved in the expression of AE. The
solution of eq. (11.39) is obtained iteratively by the Newton-Raphson
method described in Chap. 7. The details of the stiffness matrix evaluation
can be found in Reference [19].
Since the element stiffness equations are evaluated based on the local
coordinate system, the elemental stiffness matrices should be transformed
into the global coordinate system before they are assembled.

11.7 Square-Cup Drawing Process [11,12]

The finite-element formulation presented in Section 11.6 has been applied
to the square-cup drawing process. Figure 11.14 shows a schematic view of
the square-cup drawing process. In the simulation of this process, the
Coulomb friction law is assumed at the sheet-tool interface. The blank-
 Sheet-Metal Forming 211

FIG. 11.14 Schematic view of the square-cup drawing process.

holding force is treated as a concentrated vertical load acting on the nodal

points that are located at the outer rim of the blank.
Two sets of computational conditions were used in the finite-element
simulations. One is cup drawing from a square blank and the other is cup
drawing from various blank shapes.
1. Cup drawing with square blanks
• Material: aluminum-killed steel
• Stress-strain characteristic: a = 107 183£° 228 psi
= 739£a228MN/m2
• r- Value: 1.6
• Blank size: 110 x 110 mm (4.33 x 4.33 in.)
• Material thickness: 0.86mm (0.034 in.)
• Punch size, a: 40 x 40 mm (1.575 X 1.575 in.)
• Punch radius, Rp: 5mm (0.197 in.)
• Punch corner radius, Rc: 3.2mm (0.126 in.)
• Die opening, b: 42.5 x 42.5 mm (1.67 x 1.67 in.)
• Die corner radius, RD: 5mm (0.197 in.)
• Coefficients of friction: up=0.2, uD=0.04 for punch and die
respectively.
• Blank holding force: 500 kg.
The FEM mesh that was used in the square-cup drawing simulation is
shown in Fig. 11.15. Figures 11.16 and 11.17 show the thickness strain
distributions across the diagonal and the transverse directions of a square
212 Forming and the Finite-Element Method

FIG. 11.15 Finite-element mesh used in the square-cup drawing simulation.

cup, respectively. The data obtained experimentally by Thomson [20] are

also included. Maximum thinning occurs around the regions of punch and
die radii, particularly in the corner radii of the forming tools. The trends of
the strain distributions predicted by FEM are consistent with the ex-
perimental data. However, different strain magnitudes are observed in the
numerical prediction. The discrepancy between the predicted and ex-
perimental results may be attributed to the larger cup height achieved in
the experiment and the coefficients of friction used in the numerical
computation. The variations of the strain distributions across the diagonal
of the drawn cup are quite drastic as compared to those in the transverse
direction. Hence, it is obvious that the potential failure site in deep
drawing of a square cup is located along the diagonal of the drawn cup and
in the vicinity of the die corner where thinning is maximum. This fact has
been shown experimentally by several investigators [21,22]. Figure 11.18
shows a square cup formed by the finite-element simulation at a punch
depth of 20.2 mm (0.795 in.).

2. Cup drawing with various blank shapes

The material flow, final flange configurations, punch load, and the punch
depth are noted and compared by using various blank shapes of identical
initial surface area. The material and process variables used in the
 initial Distance from Center of Blank,mm
FIG. 11.16 Comparison of the numerical solutions with the experimental data [20] for
thickness strain distributions across the diagonal of a drawn cup.

FIG. 11.17 Comparison of the numerical solutions with the experimental data [20] for
thickness strain distributions in the transverse direction of a drawn cup.

213
214 Metal Forming and the Finite-Element Method
1/4 OF SQUARE CUP

FIG. 11.18 The square cup formed by the FEM simulation [19].

simulations are:
• Sheet material: AISI 304 stainless steel
• Stress-strain characteristic,CT= 219.6£°'43ksi
= 1514e° 43 MN/m 2
• r-value: 1.025
• Sheet thickness: 0.76 mm (0.03 in.)
• Coefficient of friction: \ip = \ID = 0.04
• Other dimensions for the punch and die are the same as those used in
drawing with square blanks.
Figure 11.19 shows the blank shapes and corresponding finite-element
meshes used in the process simulations. The same coefficient of friction
was used between the punch head, the die surface, and the sheet metal.
The deformation is controlled by incremental punch advancement and the
blank holding force is set equal to zero. The results of the computation are
compared with experiments [12].
The punch load versus displacement curves are shown in Fig. 11.20 for
various blank shapes. The finite-element solutions and experimental data
FIG. 11.19 Blank shapes and finite-element domains used in the square-cup drawing
simulations: (a) square; (b) octagonal; (c) circular.

215
216 Metal Forming and the Finite-Element Method

Punch Depth, mm

FIG. 11.20 Load-displacement curves in square-cup drawing processes with various initial
blank shapes.

are in good agreement in the early stages of deformation. However, for

punch depth larger than 6mm (0.236 in.), the discrepancies between the
theory and experiment become increasingly significant. This may be
attributable to the fact that the lubricant is squeezed out at the corners of
forming tools as the deformation proceeds, and the friction becomes
appreciably higher than the value of fi = 0.04 assumed in the simulations.
Another source for discrepancy may be the fact that constant-gap blank
holding was used in experiments, while the blank holding force is assumed
to be zero in computations. It is noted, however, that the load-
displacement curves predicted by the finite-element method are consistent
with the experimental data, in that the circular blank requires the largest
load to deform as compared to other shapes, particularly at larger punch
penetration.
Strain distributions of the drawn cups across the transverse and diagonal
directions obtained by the finite-element method are shown in Fig. 11.21
for various blank shapes. The strain distributions across the cup walls are
almost identical, but different amounts of thickening result in flange
portions due to the blank shapes. Comparisons of the deformed flange
 Sheet-Metal Forming 217

FIG. 11.21 Thickness strain distributions in square cup. (a) Transverse direction; (b)
diagonal direction [12].

contours predicted by the finite-element method with the experimental

results are shown in Fig. 11.22 for various blank shapes. Owing to the
differences in metal flow rates, depending on the blank shape, the resulting
flange configurations differ. The earing behavior is most significant in a
square blank, whereas, in a circular blank, earing is reduced to a minimum
at the same stage of deformation. The predicted flange configurations are
in excellent agreement with the experimental results. Photographs of
completely drawn square cups with various blank shapes are shown in Fig.
11.22. It is to be noted that blanks with orientations of 0°, 45°, and 90° with
respect to the rolling direction were tested, but the results are of no
significant difference since AISI 304 sheet exhibits little planar anisotropic
behavior.

11.8 Nonquadratic Yield Criterion

For in-plane isotropy (/ = g), eq. (11.4a) reduces, in terms of principal
stresses, to
FIG. 11.22 Comparisons of the deformed flange shapes predicted by FEM (dotted line) and
experiment (solid line) [12].

218
 Sheet-Metal Forming 219

Equation (11.5) becomes

Under uniaxial tension (a1, o2) = (ou, 0), then eqs. (11.41) and (11.42)
give, respectively,

Under equibiaxial tension, (a we have and

where r is the ratio of transverse to through-thickness strain-rates under ou.

Woodthorpe and Pearce [23] found r < 1 but ab/ou>l for some
materials. This pairing of inequalities is in conflict with eq. (11.44). Thus,
Hill [24] proposed the following nonquadratic yield criterion that allows
flexibility than the quadratic:

where loading is coaxial with the orthotropy. The coefficients /, g, etc.

characterize the anisotropy, a is a scaling factor with units of stress, and
m > 1 to ensure convexity of the yield locus. Hill suggested four simple
versions of eq. (11.45) for planar isotropy and particular values of
coefficients.
Parmar and Mellor [25, 26] used one of the special cases for predicting
the limit strains occurring in frictionless in-plane stretching and for
determining the plastic stress and strain distributions in bore expanding.
The special case was obtained by taking a = b = 0 and / = g = 0 in eq.
(11.45), as

We have, instead of eq. (11.43),

Then eq. (11.46) becomes

and instead of eq. (11.44), we now have

The predicted behavior is therefore anomalous with m such that ob/ou< 1,

when r > 1 or ob/ou > 1 when r < 1.
220 Metal Forming and the Finite-Element Method

Recently, Kobayashi et al. [27] examined four special cases suggested by

Hill, using experimental data for various anisotropic sheet metals. They
concluded that all four cases give reasonable results from a practical point
of view and that there is a preference for the case given by eq. (11.46).
The strain-rate ratio associated with eq. (11.48) is readily obtained from
the flow rule as

with the ordering of o-i > o2. In eq. (11.50) the effective strain-rate e is
introduced from
oj = 0^1 +CT2e2= 1(0! + o2)(i-l + e2) + 1(0! - CT^E! - £2).
The explicit formula for e can be obtained by eliminating ol and o2 from
eqs. (11.48) and (11.50) as

Wang [28] implemented the yield criterion, eq. (11.46), into the finite-
element procedure and applied the method for hemispherical punch
stretching to assess the effect of m-value on the resulting strain
distributions.

References
1. Mehta, H. S., and Kobayashi, S., (1971), "Finite-element Analysis and
Experimental Investigation of Sheet-metal Stretching," Rep. No. MD 71-2,
University of California.
2. Yamada, Y., (1969), "Recent Japanese Developments in Matrix Displacement
Method for Elasto-Plastic Problems," Paper presented at Japan-U.S. Seminar
on Matrix Methods of Structural Analysis and Design, Tokyo, Japan.
3. Lee, C. H., and Kobayashi, S., (1973), "New Solutions to Rigid-Plastic
Deformation Problems using a Matrix Method," Trans. ASME, J. Engr. Ind.,
Vol. 95, p. 865.
4. Lee, S. H., and Kobayashi, S., (1975), "Rigid-Plastic Analysis of Bore
Expanding and Flange Drawing with Anisotropic Sheet Metals by the Matrix
Method," Proc. 15th Int. Mach. Tool Des. Res. Conf. p. 561.
5. Wang, N. M., (1970), "A Variational Method for Problems of Large Plastic
Deformation of Metal Sheets," General Motors Research Publication GMR-
1038.
6. Kim, J. H., and Kobayashi, S., (1978), Deformation Analysis of Axisymmetric
Sheet Metal Forming Processes by the Rigid-Plastic Finite Element Method,
in, "Mechanics of Sheet Metal Forming," (Edited by D. P. Koistinen, and
N. M. Wang), Plenum Press, New York, p. 341.
7. Kim, J. H., Oh, S. I., and Kobayashi, S., (1978), "Analysis of Stretching of
Sheet Metals with Hemispherical Punch," Int. J. Machine Tool Des. Res., Vol
18, p. 209.
8. Win, S. A., (1976), "An Incremental Complete Solution of the Stretch-
forming and Deep-drawing of a Circular Blank Using a Hemispherical Punch,"
Int. J. Mech. ScL, Vol. 18, p. 23.
 Sheet-Metal Forming 221

9. Wang, N. M., and Budiansky, B., (1978), "Analysis of Sheet Metal Stamping
by a Finite-element Method," General Motors Research Publication GMR-
2423.
10. Onate, E., and Zienkiewicz, O. C., (1983), "A Viscous Shell Formulation for
the Analysis of Thin Sheet Metal Forming," Int. J. Mech. ScL, Vol. 25, p. 305.
11. Toh, C. H., and Kobayashi, S., (1983), "Finite-element Process Modeling of
Sheet Metal Forming of General Shapes," Grundlagen der Umformtechnik I
Symposium, Stuttgart, p. 39.
12. Toh, C. H., and Kobayashi, S., (1985), "Deformation Analysis and Blank
Design in Square Cup Drawing," Int. J. Machine Tool Des. Res, Vol. 25, No.
1, p. 15.
13. Hill, R., (1950), "Mathematical Theory of Plasticity," Oxford University
Press, London.
14. Rebelo, N., and Kobayashi, S., (1980), "Axisymmetric Punch Stretching of
Strain Rate Sensitive Sheet Metals," Proc. 8th NAMRC, University of
Missouri, Rolla, MO., p. 235.
15. Park, J. J., Oh, S. I., and Altan, T., (1987), "Analyses of Axisymmetric Sheet
Forming Processes by Rigid-Viscoplastic Finite Element Method," Trans.
ASME, J. Engr. Ind., Vol. 109, p. 347.
16. Kim, J. H., (1977), "Analysis of Sheet Metal Forming by the Finite-Element
Method," Ph.D. Dissertation, University of California, Berkeley.
17. Kaftanoglu, B., and Alexander, J. M., (1970), "On Quasistatic Axisymmetri-
cal Stretch Forming," Int. J. Mech. Sci. Vol. 12, p. 1065.
18. Wood, D. M., (1968), "On the Complete Solutions of the Deep Drawing
Problem," Int. J. Mech. Sci. Vol. 10, p. 83.
19. Toh, C. H., (1983), "Process Modeling of Sheet Forming of General Shapes
by the Finite Element Method Based on Large Strain formulation," Ph.D.
Dissertation, University of California, Berkeley.
20. Thomson, T. R., (1975), "Influence of Material Properties in the Forming of
Square Shells," /. Australian Inst. Metals, Vol. 20, No. 2, p. 106.
21. Yoshida, K., and Miyauchi, K., (1978), Experimental Studies of Material
Behavior as Related to Sheet Metal Forming, in "Mechanics of Sheet Metal
Forming" (Edited by D. P. Koistinen, and N. M. Wang), Plenum Press, New
York, p. 19.
22. El-Walkil, D., Kamal, M. N. E., and Darwish, A. H., (1980), "Mechanics of
the Square Box Drawing Operation of Aluminum Blanks," Sheet Metal
Industry, August, p. 679.
23. Woodthorpe, J., and Pearce, R., (1970), "The Anomalous Behavior of
Aluminum Sheet under Balanced Biaxial Tension." Int. J. Mech. Sci., Vol. 12,
p. 341.
24. Hill, R., (1979), "Theoretical Plasticity of Textured Aggregates," Math. Proc.
Camb. Phil. Soc., Vol. 85, p. 179.
25. Parmar, A., and Mellor, P. B., (1978), "Predictions of Limit Strains in Sheet
Metal Using a More General Yield Criterion," Int. J. Mech. Sci., Vol. 20, p.
385.
26. Parmar, A., and Mellor, P. B., (1978), "Plastic Expansion of a Circular Hole
in Sheet Metal Subjected to Biaxial Tensile Stress," Int. J. Mech. Sci., Vol. 20,
p. 707.
27. Kobayashi, S., Caddell, R. M., and Hosford, W. F., (1985), "Examination of
Hill's Latest Yield Criterion Using Experimental Data for Various Anisotropic
Sheet Metals," Int. J. Mech. Sci., Vol. 27, p. 509.
28. Wang, N. M., (1984), A Rigid-Plastic Rate-sensitive Finite Element Method
for Modeling Sheet Metal Forming Processes, in "Numerical Analysis of
Forming Processes," (Edited by J. F. T. Pittman et al.), Wiley, New York, p.
117.
 12
THERMO-VISCOPLASTIC ANALYSIS

12.1 Introduction
The main concern here is the analysis of plastic deformation processes in
the warm and hot forming regimes. When deformation takes place at high
temperatures, material properties can vary considerably with temperature.
Heat is generated during a metal-forming process, and if dies are at a
considerably lower temperature than the workpiece, the heat loss by
conduction to the dies and by radiation and convection to the environment
can result in severe temperature gradients within the workpiece. Thus, the
consideration of temperature effects in the analysis of metal-forming
problems is very important. Furthermore, at elevated temperatures, plastic
deformation can induce phase transformations and alterations in grain
structures that, in turn, can modify the flow stress of the workpiece
material as well as other mechanical properties.
Since materials at elevated temperatures are usually rate-sensitive, a
complete analysis of hot forming requires two considerations—the effect of
the rate-sensitivity of materials and the coupling of the metal flow and heat
transfer analyses.
A material behavior that exhibits rate sensitivity is called viscoplastic. A
theory that deals with viscoplasticity was described in Chap. 4. It was
shown that the governing equations for deformation of viscoplastic
materials are formally identical to those of plastic materials, except that
the effective stress is a function of strain, strain-rate, and temperature. The
application of the finite-element method to the analysis of metal-forming
processes using rigid-plastic materials leads to a simple extension of the
method to rigid-viscoplastic materials [1].
The importance of temperature calculations during a metal-forming
process has been recognized for a long time. Until recently, the majority of
the work had been based on procedures that uncouple the problem of heat
transfer from the metal deformation problem. Several researchers have
used the following approach. They determined the flow velocity fields in
the problem either experimentally or by calculations, and they then used
these fields to calculate heat generation. Examples of this approach are the
works of Johnson and Kudo [2] on extrusion, and of Tay et al. [3] on
machining. Another approach [4] uses Bishop's numerical method in which
222
 Thermo-Viscoplastic Analysis 223

heat generation and transportation are considered to occur instantaneously

for each time-step with conduction taking place during the time-step. This
technique was used by Altan and Kobayashi [5] on extrusion; by Lahoti
and Altan [6] on compression and torsion; and by Nagpal, et al. [7] on
forging. Usually the temperature calculations are done by using finite
differences, or finite elements, and the upper-bound technique is the most
common method for determining flow patterns theoretically.
In order to handle a coupled thermo-viscoplastic deformation problem,
it is necessary to solve simultaneously the material-flow problem for a
given temperature distribution and the heat transfer equations. Numerical
solutions of such forming problems were discussed by Zienkiewicz et al. [8]
with examples of steady flow in extrusion, drawing, rolling, and sheet-
metal forming. Zienkiewicz et al. [9] made a coupled thermal analysis of
steady-state extrusion. Rebelo and Kobayashi [10,11] developed the
method for a coupled analysis of transient viscoplastic deformation and
heat transfer. They applied the method to solid cylinder compression and
ring compression.

12.2 Viscoplastic Analysis of Compression of a Solid Cylinder

There are a number of materials that exhibit viscoplastic behavior. They
include most metals at high temperature, superplastic materials, heated
glass, and polymers. When the deformation is large, most of them can be
considered to be rigid-viscoplastic.
Because of the importance of the application of viscoplastic behavior to
the metal-forming processes, the treatment of time-dependent material
behavior within the framework of the theory of viscoplasticity is the
subject of several recent studies. Cristescu [12,13] applied the theory to
the upper-bound approach in drawing. Zienkiewicz and Godbole [14, 15]
have shown the feasibility of the finite-element approach in the deforma-
tion analysis of rigid-viscoplastic materials by treating them as non-
Newtonian viscous fluids. Price and Alexander [16,17] have applied this
formulation to creep forming.
There are only a few references in the literature in which strain-rate
effects in non-uniform deformation are explicitly in evidence. Among
them, the work by Klemz and Hashmi [18] on simple upsetting of lead
cylinders at room temperatures provides good experimental results for
comparison with the finite-element solution, since lead is strain-rate
sensitive at room temperatures. A cylinder having a diameter of 25.4mm
(1.0 in.) and a height of 24.13mm (0.95 in.) was considered. The stress,
strain, and strain-rate data for pure lead at room temperature were taken
from Reference [18] and from experimental results by Loizou and Sims
[19]. The static stress—strain curve was approximated by fitting experimen-
tal data to curves of the form o = Y[l + (e/y) m ], and the obtained values
of strain-rate exponent m and ym were interpolated in the program for
intermediate strain values. The friction factor m was taken as 0.06 (m = a
224 Metal Forming and the Finite-Element Method
fraction of the static yield shear strength), in agreement with the assumed
value in Reference [18].
For the finite-element formulation of viscoplastic deformation, a par-
ticular attention should be paid to the strain-rate sensitivity of the flow
stress of the material. The linearized stiffness equations (6.4) or (6.48) in
Chap. 6 contains the second derivative of the functional given by eq.
(6.44), in which the effective stress a is now a function of the effective
strain-rate e as well as the effective strain e.
The calculations simulate the deformation of the cylinder by a drop-
hammer, with a tup mass of 35.5 Ib, hitting the specimen at a speed of
30 ft/s. The assumption of a quasistatic process is justified for these
velocities [20]. Accordingly, the calculated work done by the contact forces
between tup and workpiece in each step of deformation was subtracted
from the kinetic energy of the tup, and the process ended when the tup
came to a stop. For comparison, the calculations were repeated without
introducing strain-rate effects (making y = °o in the formulation).
This nonsteady-state deformation problem was analyzed in a step-by-
step manner by treating it quasilinearly during each incremental deforma-
tion. The reduction in height at each step was 1%. The solution of the
velocity field for uniform compression was used as an initial guess. The
solution obtained from the previous step was then used as an initial guess
for the subsequent step. Normally, seven iterations were required for the
first step to reach an accuracy of ||Av||/||v|| < 0.000 05. For subsequent
steps, only two to four iterations were necessary to reach the same
accuracy.
The calculated overall quantities are compared with experimental data
and with the numerical dynamic analysis of Klemz and Hashmi [18] in
Figs. 12.1 and 12.2. Some observations can be made from the figures. The
results for nonstrain-rate-sensitive analysis (Fig. 12.1) tend to follow those
obtained by the analysis of Klemz and Hashmi, which did not take into
account strain-rate effects but was based on a dynamic analysis. The
agreement of the two analyses indicates that the dynamic effect is

FIG. 12.1 Comparison of the finite-element rigid-plastic analysis (•) with dynamic analysis
( ) [18] aid experiment (—) [18].
 Thermo-Viscoplastic Analysis 225

FIG. 12.2 Comparison of the finite element rigid-viscoplastic analysis (•) with dynamic
analysis ( ) [18] and experiment (—) [18].

negligible. However, both solutions show poor agreement with experi-

ments. The analysis that includes strain-rate sensitivity shows good
agreement with experiments (Fig. 12.2). No "rigid" zones were formed at
such a low value of friction in either rate-sensitive or rate-insensitive cases.

12.3 Heat Transfer Analysis

Basic Equations
The basic equations and corresponding finite-element formulations for the
deformation analysis are described in previous chapters. For the heat
transfer analysis, we begin with the energy balance equation, expressed by

where kl Ttii is the heat transfer rate, k^ denotes thermal conductivity, f is

the heat generation rate, and peris the internal energy-rate. In eq. (12.1),
the notation TiU is used for T,,_,- , with the comma denoting differentiation
and repeated subscript meaning summation (Laplace differential operator
applied to temperature T). We consider that the heat generation in the
deforming body is due only to plastic deformation

where the heat generation efficiency K, represents the fraction of mechani-

cal energy transformed into heat and is usually assumed to be 0.9. The
fraction of the remainder of the plastic deformation energy (1 - K) is
expended to cause changes in dislocation density, grain boundaries, and
phases. This energy is usually recoverable by annealing.
Along the boundaries of the deforming material, either the temperature
T is prescribed or a heat flux is given. The energy balance, eq. (12.1), can
be written in the form
226 Metal Forming and the Finite-Element Method
for arbitrary variation in temperature 6T. By using the divergence
theorem, eq. (12.3) becomes

where qn is the heat flux across the boundary surface 5g, n denotes the unit
normal to the boundary surface, and

Solutions to problems of this nature require the temperature field to satisfy

the prescribed boundary temperatures and eq. (12.4) for arbitrary pertur-
bation dT.
Finite-Element Formulation
The temperature field in eq. (12.4) is approximated by

where qa is the shape function and Ta is the temperature at ath node.

With the quadrilateral element shown in Fig. 8.1,

and

where q\, . . . , q* are given by eq. (8.2b) in Chap. 8.

Putting

and substituting eq. (12.6) into eq. (12.4),

Because of the arbitrariness of 6T, the following system of equations is

obtained,
 Thermo-Viscoplastic Analysis 227

Equations (12.9) can be expressed in the form

where C is the heat capacity matrix, Kc is the heat conduction matrix, Q is

the heat flux vector, T is the vector of nodal point temperatures, and T the
vector of nodal point temperature-rates. The heat flux vector Q in eq.
(12.10) has several components and is expressed with the interpolation
function N by

The first term on the right is the heat, generated by plastic deformation
inside the deforming body. The second term defines the contribution of the
heat radiated from the environment to the element, where a is the
Stefan-Boltzman constant, e is the emissivity, and Te and Ts are environ-
ment and surface temperatures, respectively. The third term describes the
heat convected from the body surface to the environment with heat
convection coefficient h. The fourth term represents the contribution of the
heat transferred from the workpiece to the die through their interface. Td
and Tw are die and workpiece temperatures, respectively, and /ilub is the
heat transfer coefficient for the lubricant. The last term is the contribution
of the heat generated by friction along the die-workpiece interface, qf
being the surface heat generation rate due to friction.
The theory necessary to integrate (12.10) can be found in numerical
analysis books; see for instance Dahlquist and Bjorck, [21]. The conver-
gence of a scheme requires consistency and stability. Consistency is
satisfied by an approximation of the type

where /3 is a parameter varying between 0 and 1, and t denotes time.

For unconditional stability, /? should be greater than 0.5, and a value of
0.75 was chosen. Selection of a proper value of /? is an important factor in
situations where it is desirable for the time step to be as large as possible,
provided that the increments in strain are compatible with an infinitesimal
analysis.

12.4 Computational Procedures for Thermo-Yiscoplastic Analysis

We treat the workpiece and the die separately, assuming that the die
properties do not change. Thus we greatly reduce the number of equations
to be solved simultaneously, which, in turn, reduces the cost of the
solution. There is no internal heat generation in the die and therefore
deformation calculations are not necessary.
228 Metal Forming and the Finite-Element Method

The heat generated through friction, qf, is calculated as

where fs is the friction stress, and \us\ the relative velocity between die and
workpiece. The heat qf is evenly distributed between the die and
deforming material.
It should be noted that the nodes on the die do not generally coincide
with those on the deforming material along the interface, and that the
calculation of nodal point temperatues requires interpolation. When
boundary conditions such as the convection term in eq. (12.11) apply, they
are split into two parts, wherein one containing the unknown temperatures
is added to the matrix Kc. Boundary conditions such as the radiation term
in eq. (12.11) are applied using previous iteration values for body
temperatures.
The equations for the flow analysis and the temperature calculation are
strongly coupled, making a simultaneous solution of their finite-element
counterparts necessary.
Considering Tt+&t as a primary dependent variable, we have, from eq.
(12.12), with t = 0 initially,

The circumflex over T denotes

Substituting eq. (12.14) into eq. (12.10), gives

The coupling procedure makes use of eq. (12.15) through the following
sequence.
1. Assume the initial temperature field T0.
2. Calculate the initial velocity field u corresponding to the tempera-
ture field T0.
3. Calculate the initial temperature-rate field T0 from eq. (12.10) using
values from (1) and (2).
4. Calculate the quantity T.
5. Update the nodal point positions and the effective strain of elements
for the next step.
6. Use the velocity field at the previous step to calculate the first
approximate temperature T such as
 Thermo-Viscoplastic Analysis 229

7. Calculate a new velocity field with the solution of (6).

8. Use the new velocity field to calculate the second temperature field
such as

9. Repeat steps for (7) and (8) until both have converged.
10. Calculate the new temperature rate field TA,.
11. Repeat steps from (5) to (10) until the desired deformation state is
reached.
The iteration process for temperature calculations is not likely to require
much computing time, because only the heat input vector Q is changed
during iterations and, as a result triangularization of the matrix is
necessary only once. Moreover, additional iterations necessary to obtain a
velocity field after a new temperature field is obtained should be relatively
few, because the velocity field does not show much sensitivity to small
variations of the temperature field.

12.5 Applications
Applications of the thermo-viscoplastic analysis include compression of a
solid steel cylinder [11,22], hot nosing of a steel shells [23], and forging of
titanium alloys [24, 25].
Compression of Steel Cylinder [11]
Pohl [26] conducted temperature measurements in order to test his
uncoupled analysis, in which approximate stream functions were used for
the deformation, and finite differences for the heat balance. A solid
cylinder of a carbon steel AISI 1015 was compressed between flat dies at
room temperature. Thermocouples were inserted in the cylinder at
different locations. Upon deformation, their readings indicated the tem-
perature increases due to the heat generation. Figure 12.3 shows the
dimensions and locations of measuring points. The conditions used in
computations with the finite-element method [11] were as follows. The
deformation took place at room temperature, until a reduction of 33% in
height was achieved. The finite-element grid was composed of 132
four-node quadrilateral elements in the workpiece, and 119 in the die.
Because of symmetry, only one-quarter of the cylinder needed to be
analyzed. The friction factor m was taken as 0.65.
The die velocity was changed at each time-step to simulate a mechanical
press. Each step corresponded to 1% reduction in height, which was
equivalent to time-steps of up to 0.03 s. The flow stress was considered to
be independent of strain-rate and temperature and its values were given by
Pohl. The heat transfer characteristics, other than the thermal conductivity
and the heat capacity of the AISI 1015, which were given by Pohl, were
taken from handbooks.
FIG. 12.3 Compression of a steel cylinder: specimen geometry and thermocouple locations, and comparison of temperature distribution between theory (—)
and experiments [26].
 Thermo-Viscoplastic Analysis 231

The dimensions of the die were such that at the outside boundaries a
constant temperature was imposed. The temperature values measured by
Pohl are compared with the calculated values in Fig. 12.3. The agreements
are excellent for the internal points. For the three points near the billet
surface, however, the computed results indicate that the temperature
difference for these points is small, while experiments show larger
differences. This discrepancy may be attributed to inaccuracies in the
material constants used in computations or to inaccuracies in experimental
measurements.
Hot Compression of Steel Cylinder
Wu and Oh [22] developed an FEM-based computer program (ALPIDT)
that is capable of simulating nonisothermal forming operations with
arbitrarily shaped workpieces and dies. This program consists of two
independent FEM programs, ALPID for viscoplastic deformation analysis
[27] and the program for the heat transfer analysis. They are coupled in an
efficient manner in the program ALPIDT for simulation of thermo-
viscoplastic deformation. To demonstrate the capability of ALPIDT, a hot
compression process was simulated. The temperature changes during the
initial resting and dwelling periods were included in the simulation. The
simulation also accounted for the changes in the heat transfer coefficients
between the workpiece and the die during the process.
A cylindrical billet of AISI 1020 steel was compressed between two flat
dies. The initial billet temperature was 1232°C (2250°F) and the initial die
temperature was 204°C (400°F). In order to estimate the workpiece
temperature accurately, simulation was performed in the process consisting
of three periods as follows:
1. The heated billet is placed on the lower die for 6 seconds without
deformation (free resting period).
2. The workpiece is compressed to 67% in height (deformation period),
1.5 seconds).
3. The deformed workpiece stays on the lower die for 3 seconds after
the upper die is retracted (dwelling period).
Detailed computational conditions are given in Reference [22].
In Fig. 12.4 photographs of grid-distortions and temperature distribu-
tions are shown at various stages of the process. The temperature scale
(°F) is also shown in the figure.
The predicted temperatures at the end of the resting period are shown in
Fig. 12.4a. Owing to relatively small loss of heat to the environment, the
temperatures of the upper die and of the upper portion of the workpiece
remained almost unchanged. However, heat loss from the bottom portion
of the billet to the lower die was considerable. The temperature at the
bottom of the workpiece dropped by 280°C (536°F) during the free resting
period of 6 s. At the same time, the surface of the lower die was heated to
nearly 600°C (1112°F). It is to be noted that the flow stress of the
232 Metal Forming and the Finite-Element Method
workpiece material was doubled when the temperature was reduced from
1230° to 950°C (2246° to 1742°F).
The predicted temperature distributions and grid distortions during
the deformation period are shown in Figs. 12.4b to e. At the beginning of
the deformation period, the temperature of the workpiece at the upper die
interface was higher than that at the lower die interface. However, the
temperature of the top surface dropped quickly after the upper die
contacted the workpiece. The bottom and top surface temperatures of the
workpiece become almost the same, reaching 700-800°C (1292-1472°F) at
0.375 s after deformation started. However, it is noted that the tempera-
ture gradient at the top was higher than at the bottom of the workpiece.
These differences between the temperature distributions in the upper and
lower regions of the workpiece became less as the deformation proceeded
further. The temperature distributions shown in Figs 12.4b to e suggest
that the flow stress of workpiece near the die workpiece interface could
have been three times that in the mid-height region.
The predicted grid distortions reflect the effect of temperature on the
flow stress of the workpiece, influencing metal flow. During the early stage
of deformation (see Figs. 12.4b and c), the barreling near the top surface
was more pronounced than near the bottom surface. It is also noticed that
the upper surface moved radially more than the bottom surface, suggesting
that the average flow stress was higher near the bottom surface. It can be
also seen that, throughout the deformation period, the deformation
occurred mainly in the mid-height region, while the chilled regions near
the die workpiece interfaces remained almost rigid (Figs. 12.4d and e).
Figure 12.4f shows the temperature distributions at 3 s after the upper
die was retracted from the workpiece at the end of deformation. It is seen
from the figure that the top surface of the workpiece has warmed up again,
while the average temperature of the workpiece had dropped. This was
due to the redistribution of heat within the workpiece, without die chilling
at the top surface of the workpiece. The predictions shown in Fig. 12.4
agree with general observations made in nonisothermal forging of cylindri-
cal billets.
Forging of Titanium Alloy 776242
The preform considered is a cylindrical composite material with a central
core of (a + ft} phase and an outer ring of (/^-transformed phase, the two
diffusion-bonded together. Its dimensions are given in Fig. 12.5a. The
preform was forged isothermally (dies and workpiece at the same initial
temperature) at 1227 K (1750°F) at a constant ram speed of 5.08mm/min.
(0.2in./min). The total reduction in height was 60%. At these high
temperatures, a glass-type lubricant is very effective; a friction factor of
m = 0.2 was used in computer simulations.
In isothermal forging, very slow speeds are usually employed in order to
(1) avoid increasing the flow stress of the material, which is strain-rate
dependent, and (2) allow the heat generated during deformation to spread
FIG. 12.4 Predicted temperature distributions and grid distortions at various stages of hot
compression process [22], (a) At the end of free resting (elapsed time t ~ 6s); (b) 16.67%
reduction in height (/ - 6,375s); (c) 33,34% reduction in height (/ - 6.750s); (d) 50.00%
reduction in height (i = 7.125s); (e) 66.67% reduction in height (/ = 7.500s); (f) at the end
of 3s free resting after deformation (/ - 10.5s).
This page intentionally left blank
 Thermo-Viscoplastic Analysis 235

FIG. 12.5 Die and workpiece temperatures in forging a titanium 6242 alloy composite billet
at 60% reduction in height with different ram speeds: (a) preform dimensions; (b) ram
speed = 0.2 in./min.; (c) ram speed = 2.0 in./min.

uniformly throughout the workpiece, assuring uniformity in temperature

and deformation. If higher speeds can be used without adverse effects, the
increase in productivity is obvious. To investigate this aspect, the whole
process was simulated at 5.08 mm/min. (0.2 in./min.) and at 50.8 mm/min.
(2.0 in./min).
The difference in speed of deformation did not produce any noticeable
differences in the overall deformation process, or even in the local
deformation histories. The only changes found were in the temperature
fields, also shown in Figs. 12.5b and c. An average calculated temperature
increase of 6.5 K in the slow-speed deformation increased to 12.5 K for the
faster deformation. However, the temperature gradients were such that
the greatest difference between two points in the workpiece was roughly
5 K at the end of deformation. This indicates that the speeds of
deformation can be increased (at least to 2.0 in./min) without loss of
uniformity of properties, provided the average temperatures reached are
not critical.
236 Metal Forming and the Finite-Element Method

The resulting microstructures at various locations after forging are

shown in Fig. 12.6. The strain, strain-rate, and temperature variations
during forging were found to be almost the same at all locations, and
typical variations for 5.08mm (0.2in./min.) ram speed are shown for the
elements near the outer periphery and near the center in Fig. 12.6.

FIG. 12.6 Local strain, strain-rate, and temperature variations during (a + f})/(i composite
forging and corresponding microstructures after forging. (Microstructures courtesy of C. C.
Chen, Chen Tech. Industries.)
 Thermo-Viscoplastic Analysis 237

The temperature histories indicate that no appreciable changes took

place. Therefore, any microstructural modifications must be due to
maintaining the workpiece at an elevated temperature for some time under
pressure and also to the amount of deformation imposed.
The information given in Fig. 12.6 allows further qualitative interpreta-
tion of the relationships between local strain, strain-rate, and temperature
histories and corresponding microstructures.
Isothermal and Hot-Die Disk Forging
Using the program developed by Oh [27], coupled with the heat transfer
finite-element calculation, the compressor disk forging of a (/J)-phase Ti
6242 alloy was analyzed under two sets of conditions, isothermal forging
and hot-die forging [25].
In isothermal forging, the initial temperatures of both die and preform
were chosen to be 1171.9 K (1650°F) and the die velocity was
5.08mm/min. (0.2in./min). Again, a glass-type lubricant was used at the
die-preform interface.
In addition to isothermal forging, hot-die forging was simulated for the
same die and preform geometries in order to determine the effects of
temperature and strain-rate on the details of deformation behavior. The
initial temperature of the preform was 1171.9 K (1650°F), but the die and
environmental temperatures were assumed to be 644 K (700°F) and 293 K
(68°F), respectively. The die speed was 76.2mm/min. (3.0in./min).
The grid distortions at 70% reduction under the two forging conditions
are compared in Fig. 12.7a. The difference in metal flow, perhaps mainly
due to temperature effects, can be seen qualitatively. Distortion is more
severe in hot-die forging than in isothermal forging. Nonuniformity of
deformation due to temperature gradients within the workpiece in hot-die
forging can be seen from the shape of the grid distortions. The contact
area at the rim region is different, being larger in isothermal forging. It is
evident that the bulge at the outer surface is greater with hot-die forging.
Although the observation is qualitative, it is clear that the effect of
temperature, possibly combined with the strain-rate effect, causes the
metal flow to differ under the two forging conditions.
Figure 12.7b shows a comparison of temperature distributions under the
two forging conditions. The temperature variation within the workpiece
for isothermal forging is only a few degrees, while a severe temperature
gradient can be seen near the contact region between the die and the
workpiece in hot-die forging.
The strain distributions, on the other hand, are almost the same for the
two forging conditions, as shown in Fig. 12.7c. It appears that the strains
are predominantly determined by the preform geometry and the die
configuration.
Hot Nosing [23]
The nosing process was described in Chap. 2. To study the nature of the
nosing process of shells, Carlson [28] conducted hot nosing on a small
FIG. 12.7 (a) Grid-distortions; (b) temperature distributions; and (c) strain distributions in
isothermal and hot-die forging. Material: Ti 6242 alloy; /J-phase. Isothermal, die and
workpiece temperature 1227 K; speed 0.2in./min., hot-die, die temperature 624 K, speed
3.0in./min.

238
 Thermo-Viscoplastic Analysis 239

FIG. 12.7 (cont'd)

model of a 105-mm shell. Commercial open-hearth, cold-rolled steel was

used in making the specimens. Geometric details of the specimens are
given in Fig. 12.8.
The die that was employed throughout the experiments had a curvature
radius of 300mm (11.85 in.) for an ogive profile. To simulate the amount
of nosing, 36.8mm (1.45 in.) of the specimen was made to enter the die,
giving a maximum reduction in the mean diameter of about 33%.
In preparation for the test, the die was heated for about 1 hour to bring
it to a temperature of 811 K (1000°F). The shell temperature varied from
about 1144 K (1600°F) to 1311K (1900°F) at the tip, depending upon

FIG. 12.8 Hot nosing specimen [28].

240 Metal Forming and the Finite-Element Method

FIG. 12.9 Temperature distributions obtained by induction heating for a shell [28].

heating conditions. To measure the temperature distribution, six chromel-

alumel thermocouples were placed along the shell specimen. The
measured temperature distributions are shown in Fig. 12.9.
The heating of the shell was done in about 1 min by an induction coil.
The length of time for nosing was in the order of 1 s. For the finite-element
simulations, the flow stress expressions were obtained from experimental
data of Altan and Boulger [29], and the heat transfer characteristics were
taken from standard handbooks.
The predicted load-displacement curves for hot nosing at various
friction conditions, with initial tip temperature at 1269 K (1825°F) and die
speed of 41 mm/s (1.62 in./s), are given in Fig. 12.10a. It can be seen that
the simulation results, obtained with a friction coefficient of ju = 0.1, are
very close to the experimental results. In Fig. 12.10b, the load-
displacement curves are computed for initial tip temperatures of 1158K
(1625°F), 1227 K (1750°F), and 1269 K (1825°F), respectively, at a nosing
speed of 41 mm/s and for /j = 0.1. All of the finite-element simulation
results showed good agreement with the experimental data.

12.6 Concluding Remarks

Using the capabilities of a coupled thermo-viscoplastic analysis, attempts
were made to predict metal flow and forming loads, and to correlate the
metallurgical changes with the information obtained through simulation.
Gegel et al. [30] provided a detailed interpretation of the microstructure
 Thermo-Viscoplastic Analysis 241

FIG. 12.10 Load-displacement curves for (a) various friction conditions and (b) different
initial tip temperatures, and comparison with experiment [28].

that develops during hot forging. Furthermore, Gegel et al. [31] proposed
a new method of modeling the dynamic material behavior that explicitly
described the dynamic metallurgical processes occurring during hot
deformation.
Recently, Dawson [32] developed the numerical solution formulation for
242 Metal Forming and the Finite-Element Method
simulation of hot or warm metal forming under steady-state conditions. Of
particular importance was the incorporation of a methodology for using
particle stress-temperature trajectories in conjunction with deformation
mechanism maps. Thus, the assumptions made regarding the assumed
constitutive equations could be evaluated. The analysis of slab rolling of
aluminum was given as an example of possible applications. A significant
advance, made by Dewhurst and Dawson [33], is the development of a
finite-element program that models steady-state viscoplastic flow and heat
transfer in three dimensions.

References
1. Oh, S. I., Rebelo, N. M., and Kobayashi, S., (1978), "Finite-Element
Formulation for the Analysis of Plastic Deformation of Rate-Sensitive Mate-
rials in Metal Forming," IUTAM Symposium, Tutzing/Germany, p. 273.
2. Johnson, W., and Kudo, H., (1960), "The Use of Upper-Bound Solutions for
the Determination of Temperature Distributions in Fast Hot Rolling and
Axisymmetric Extrusion Processes," Int. J. Mech. Sci., Vol. 1, p. 175.
3. Tay, A. O., Stevenson, M. G., and Davis, G. V., (1974), "Using the
Finite-Element Method to Determine Temperature Distributions in Orthogo-
nal Machining," Proc. Inst. Mech. Engr. Vol. 188, p. 627.
4. Bishop, J. F. W., (1956), "An Approximate Method for Determining the
Temperature Reached in Steady-State Motion Problems of Plane Plastic
Strain," Q. J. Mech. Appl. Math., Vol. 9, p. 236.
5. Altan, T., and Kobayashi, S., (1968), "A Numerical Method for Estimating
the Temperature Distributions in Extrusion Through Conical Dies," Trans.
ASME, J. Engr. Ind., Vol. 90, p. 107.
6. Lahoti, G., and Altan, T., (1975), "Prediction of Temperature Distributions in
Axisymmetric Compression and Torsion," Trans. ASME, J. Engr. Materials
Technology, Vol. 97, p. 113.
7. Nagpal, V., Lahoti, G. D., and Altan, T., (1978), "A Numerical Method for
Simultaneous Prediction of Metal Flow and Temperatures in Upset Forging of
Rings," Trans. ASME, J. Engr. Ind. Vol. 109, p. 413.
8. Zienkiewicz, O. C., Jain, P. C., and Onate, E., (1978), "Flow of Solids During
Forming and Extrusion: Some aspects of numerical solutions," Int. J. Solids
Structures Vol. 14, p. 15.
9. Zienkiewicz, O. C., Onate, E., and Heinrich, J. C., (1978), "Plastic Flow in
Metal Forming—I. Coupled Thermal Behavior in Extrusion—II. Thin Sheet
Forming." Applications of Numerical Methods to Forming Processes, ASME,
AMD, Vol. 28, p. 107.
10. Rebelo, N., and Kobayashi, S., (1980), "A Coupled Analysis of Viscoplastic
Deformation and Heat Transfer—I. Theoretical Considerations," Int. J. Mech.
Sci., Vol. 22, p. 699.
11. Rebelo, N., and Kobayashi, S., (1980), "A Coupled Analysis of Viscoplastic
Deformation and Heat Transfer—II. Applications." Int. J. Mech. Sci., Vol.
22, p. 707.
12. Cristescu, N., (1975), "Plastic Flow Through Conical Converging Dies, Using
a Viscoplastic Constitutive Equation," Int. J. Mech. Sci., Vol. 17, p. 425.
13. Cristescu, N., (1976), "Drawing Through Conical Dies—An Analysis Com-
pared with Experiments," Int. J. Mech. Sci., Vol. 18, p. 45.
14. Zienkiewicz, O. C., and Godbole, P. N., (1975), Viscous, Incompressible
Flow with Special Reference to Non-Newtonian (plastic) Fluids, Chap. 2 in
"Finite Elements in Fluids" (Edited by R. H. Gallagher et al), Wiley, New
York.
 Thermo-Viscoplastic Analysis 243

15. Zienkiewicz, O. C., and Godbole, P. N., (1974), "Flow of Plastic and
Viscoplastic Solids with Special Reference to Extrusion and Forming Proc-
esses," Int. J. Num. Meth. Eng., Vol. 8, p. 3.
16. Price, J. W. H., and Alexander, J. M., (1976), "A Study of Isothermal
Forming or Creep Forming of a Titanium Alloy," Proc. of the 4th NAMRC,
Columbus, Ohio, p. 46.
17. Price, J. W. H., and Alexander, J. M., (1976), "The Finite-Element Analysis
of Two High-Temperature Metal Deformation Processes," 2d Int. Symposium
on FEM in Flow Problems, p. 717.
18. Klemz, F. B., and Hashmi, S. J., (1977), "Simple Upsetting of Cylindrical
Billets: Experimental Investigation and Theoretical Prediction," 18th MTDR
Conference, London, p. 323.
19. Loizou, N., and Sims, R. B., (1953), "The Yield Stress of Pure Lead in
Compression," J. Mech. Phys. Solids, Vol. 1, p. 234.
20. Lippman, H., (1966), "On the Dynamics of Forging," Proc. 7th Int. MTDR
Conference, Birmingham, England, p. 53.
21. Dahlquist, G., and Bjorck, A., (1974), "Numerical Methods," Prentice-Hall,
Englewood Cliffs, NJ.
22. Wu, W. T., and Oh, S. I., (1985), "ALPIDT: A General Purpose FEM Code
for Simulation of Non-Isothermal Forming Processes," Proc. NAMRI—XIII,
Berkeley, California, p. 449.
23. Tang, M.-C., and Kobayashi, S., (1982), "An Investigation of the Shell Nosing
Process by the Finite Element Method. Part 2: Nosing at Elevated Tempera-
tures," Trans. ASME, J. Engr. Ind., Vol. 104, p. 312.
24. Rebelo, N., and Kobayashi, S., (1981), "Thermo-Viscoplastic Analysis of
Titanium Alloy Forging," ASME Publications FED—Vol. 3, Manufacturing
Solutions Based on Engineering Sciences, p. 151.
25. Oh, S. I., Park, J. J., Kobayashi, S., and Allan, T., (1983), "Application of
FEM Modeling to Simulate Metal Flow in Forging a Titanium Alloy Engine
Disk," Trans. ASME, J. Engr. Ind., Vol. 105, p. 251.
26. Pohl, W., (1972), "A Method for Approximate Calculation of Heat Genera-
tion and Transfer in Cold Upsetting of Metals," Doctoral Dissertation,
University of Stuttgart.
27. Oh, S. I., (1982), "Finite Element Analysis of Metal Forming Process with
Arbitrarily Shaped Dies," Int. J. Mech. ScL, Vol. 24, p. 479.
28. Carlson, R. K., (1943), "An Experimental Investigation of the Nosing of
Shells," Forging of Steel Shells, presented at the Winter Annual Meeting of
ASME, New York.
29. Altan, T., and Boulger, F. W., (1973), "Flow Stress of Metals and its
Applications in Metal Forming Analysis," Trans. ASME J. Engr. Ind., Vol.
95, No. 4, p. 1009.
30. Gegel, H. L., Nadiv, S., Malas, J. C., and Morgan, J. T., (1980), "Application
of Process Modeling to Analysis of Microstructural Changes During the Hot
Working of a Two-phase Titanium Alloy," Appendix K, AFWAL-TR-80-
4162, p. 403.
31. Gegel, H. L., Prasad, Y. V. R. K., Malas, J. C., Morgan, J. T., and Lark, K.
A., (1984), "Computer Simulations for Controlling Microstructure During Hot
Working of Ti-6242," ASME, PVP, vol. 87, p. 101.
32. Dawson, P. R., (1984), "A Model for the Hot or Warm Forming of Metals
with Special Use of Deformation Mechanism Maps," Int. J. Mech. Sci., Vol.
26, p. 227.
33. Dewhurst, T. B., and Dawson, P. R., (1984), "Analysis of Large Plastic
Deformation at Elevated Temperatures Using State Variable Models for
Viscoplastic Flow," Proc. Symp. Constitutive Equations: Micro, Macro, and
Computational Aspects, ASME., p. 149.
 13
COMPACTION AND FORGING
OF POROUS METALS

13.1 Introduction
Powder forming, once considered a laboratory curiosity, has evolved into a
manufacturing technique for producing high-performance components
economically in the metal-working industry because of its low manufactur-
ing cost compared with conventional metal-forming processes [1,2].
Generally, the powder-forming process consists of three steps: (1) com-
pacting a precise weight of metal powder into a "green" preform with
10-30% porosity (defined by the ratio of void volume to total volume of
the preform); (2) sintering the preform to reduce the metal oxides and
form strong metallurgical structures; (3) forming the preform by repressing
or upsetting in a closed die to less than 1% residual porosity.
Powder forming has disadvantages in that the preform exhibits porosity.
Because of this porosity, the ductility of the sintered preform is low in
comparison with wrought materials [3]. In forging compacted and sintered
powdered-metal (P/M) preforms, where large amount of deformation and
shear is involved, pores collapse and align in the direction perpendicular to
that of forging and result in anisotropy. However, repressing-type defor-
mation, where very little deformation and shear are present, does not lead
to marked anisotropy [4]. A low-density preform will result in more local
flow and a higher degree of anisotropy than will a preform of high initial
density [5]. These anisotropic structures can lead to nonuniform impact
resistances of the forged P/M parts. Also, in forming of sintered preforms,
materials are more susceptible to fracture than in forming of solid
materials, and the analysis is of particular importance in producing
defect-free components by determining the effect of various parameters
(preform and die geometries, sintering conditions, and the friction
conditions) on the detailed metal flow. In this chapter, the plasticity theory
for solid materials is extended to porous materials, applicable to the
deformation analysis of sintered powdered-metal preforms.
In characterizing the mechanical response of porous materials, a
phenomeno'ogical approach (introducing a homogeneous continuum mo-
del) is employed. For the finite-element formulations of the equilibrium
and energy equations based on the infinitesimal theory, the following
assumptions are made: the elastic portion of deformation is neglected
244
 Compaction and Forging of Porous Metals 245
because the practical forming process involves very large amounts of
plastic deformation; the normality of the plastic strain-rates to the yield
surface holds; anisotropy that occurs during deformation is negligible; and
thermal properties of the porous materials are independent of the
temperatures.

13.2 Yield Criterion and Flow Rules

For porous metals, a number of plasticity theories have been proposed
with a yield function /(a/,-) of the following form [6,7]:
(13.1)
where 7X is the linear invariant of stress tensor, J2 is the quadratic invariant
of deviatoric stress tensor, and A and B are functions of void ratio or
relative density. The invariants 1^ and J2 are denned as
Il = ax + ay + oz
and

Starting with eq. (13.1), Oyane and his colleagues [8,9,10] derived the
plasticity equations. On the basis of this theory, they derived the slip-line
field equations and the upper-bound theorem applicable to porous metals.
The yield surface can be defined by

where YR is the apparent yield stress of porous materials determined by

uniaxial tension or compression. It can then be shown that B = 1 — (A/3).
The yield function f(oti) is now expressed by

With this yield function, flow rules are expressed by

where o,, and e,y are apparent stresses and strain-rates, respectively,
considering a porous metal to be a continuum. The proportionality factor A
in the flow rule, eq. (13.3), is given by

where the apparent strain-rate eR is defined according to the equivalence

246 Metal Forming and the Finite-Element Method

of the work-rate, namely, atje,j = YReK [11,12] and expressed by

where y,-,- is the engineering shear strain rate and ev is the volumetric strain
rate.
The apparent yield stress YR depends on the property of the base metal
and the relative density R (ratio of the volume of base metal to the total
volume of porous metal) [13] according to

where Yb is the yield stress of the base metal and 17 is a function of relative
density. The effects of strain, strain-rate, and temperature on the yield
stress are included in Yb = Yb(eb, eb, Tb), where eb, eb, and Tb are strain,
strain-rate, and temperature of the base metal. The relationship between
the apparent strain and strain-rate and those of the base metal are given by

and

In order to complete the constitutive equations, A in eq. (13.2) and r\ in

eq. (13.6) must be determined as functions of relative density by
experiment. Among the proposed constitutive equations, those proposed
by Doraivelu et al. [13] and by Shima et al. [8], appear to agree with
experimental measurements quite well. The expressions for A and rj by
Doraivelu et al. are

It is to be noted that for R = 0.707, r\ = 0, which implies that the apparent

yield stress is zero according to eq. (13.6). For the analysis in this chapter,
the constitutive equation (13.8) is used. Further discussion on the validity
of the constitutive equations for porous materials can be found in
References [14] and [15].

13.3 Finite-Element Modeling and Numerical Procedures

The variational form of the equilibrium equation, as a basis for discretiza-
tion, is derived as

where 8 denotes variations.

 Compaction and Forging of Porous Metals 247

Discretization
The element used for discretization is an isoparametric quadrilateral
element with bilinear shape function (see Fig. 8.1 and Fig. 9.1 for the
two-dimensional and the axisymmetric deformations, respectively). The
elemental velocity field is approximated by

where v is the velocity vector of nodal-point value and N is the shape

function matrix. Applying differentiation to eq. (13.10),

The matrices N and B have been defined for a quadrilateral element in the
previous chapters. Substitution of eq. (13.11) into eq. (13.5) leads to

and the variation 6eR becomes, because of symmetry of the matrix P,

Substituting eqs. (13.10) and (13.13) into eq. (13.9) at the elemental level
and assembling the element equations with global constraints, we obtain

where (;') indicates the y'th element. Because the variation <5v is arbitrary,
eq. (13.14) results in the stiffness equations. The matrix P in eq. (13.12) is
obtained as follows. The inversion of the flow rules (13.3) can be expressed
by

where, for the axisymmetric case,

with the stress and strain-rate vectors are defined according to

and
248 Metal Forming and the Finite-Element Method

respectively. Then, from the requirement that O T E= YRER,

Updating of Relative Density

The volumetric strain-rate is related to the density-rate according to

where the relative density R is defined by

with Vb being volume of the base metal and Vv as the volume of void.
Integrating eq. (13.16), we have

In eq. (13.17), R0 is the current relative density and Aey is the change of
volumetric strain in one deformation step. The average relative density Ra
is defined by

where Rt and Vt are the relative density and the volume of an element,
respectively.
Fully Dense Materials
For fully dense materials, R = 1.0 and A = 3. Then, the matrix D of eq.
(13.15) becomes infinity. Consequently, convergence behavior for the
solution becomes erratic when the relative density in the element
approaches unity.
A constraint was incorporated in the numerical procedures such that an
element with R = 0.9990 was considered as a fully dense element. This
constraint was helpful in obtaining well-behaved convergence for the
solution. Osakada et al. [16] and Mori et al. [17], also using this theory,
developed the finite-element method and applied it to the rigid-plastic
deformation analysis of fully dense materials.
Volume Integration
The program has been tested by analyzing compression of cylinders and
rings of porous materials [18]. During the test it was found that the
reduced integration must be applied to the terms involving the volumetric
strain-rate. Such a reduced integration strategy is straightforward in the
formulation of fully dense material, since the term of the volumetric strain
 Compaction and Forging of Porous Metals 249

energy-rate is fully uncoupled from that of the distortional energy-rate. In

the present formulation, the terms with volumetric strain-rate cannot be
decoupled from those with distortional strain-rate, as seen in eq. (13.14).
Thus, the matrix P is decomposed into two components, Pj and P2. Here,
P! contains the terms with distortional strain-rate and P2 involves the terms
with volumetric strain-rate. The matrices P1 and P2 are obtained by
decomposing the matrix D into the two components D1 and D2 as
D = D! + D2
where

and

Then, P = P! + P2 = BTDjB + BTD2B. For evaluating the matrix P, the

matrix P2 is evaluated at the reduced integration point, while Px is
evaluated at the regular integration points. The evaluation of the matrix P
is then obtained at the regular integration points. It should be noted again
that for solid materials A = 3 in eq. (13.19), which gives

and l/[3(3 — A)] is replaced by K, the penalty constant.

13.4 Simple Compression [14]

In simple compression, a cylindrical sintered P/M preform is compressed
between two flat dies. The preform was 50.4 mm (2 in.) in diameter and
50.4mm (2 in.) in height. It was assumed that the preform has uniform
relative density of 0.8. The simulations were carried out with two different
friction factors, m = 0.2 and 0.5. The flow stress of the matrix material was
assumed to be expressed by o = 1.0 + O.Ole and the constitutive eq. (13.8)
was used.
Figure 13.1 shows the predicted grid distortions at 20, 50 and 70%
reduction in height for the friction factor m = 0.5. It can be seen from the
250 Metal Forming and the Finite-Element Method

FIG. 13.1 Simple compression of a cylindrical sintered P/M preform. The predicted grid
distortions during simple compression at 20, 50, and 70% reductions in height. The dashed
lines are the predicted boundaries with fully dense initial preform. Initial relative density =
0.8; friction factor = 0.5.

figure that the predicted grid distortions resemble those expected in the
compression of a fully dense preform. The predicted boundaries of the
deformed workpiece for a fully dense initial preform are shown in dashed
lines for comparison. It is seen that the porous preform changes its volume
during the compression process. The average density change as a function
of height strain is shown in Fig. 13.2. The height strain is defined by
\n(H/H0) where H0 is the height of the undeformed preform and H is the
deformed height. It is interesting to note that the average density varies
very little with differerent friction conditions.
 Compaction and Forging of Porous Metals 251

FIG. 13.2 The predicted average relative density changes as a function of height strain in
simple compression.

Although the average density changes are not affected by different

frictions, the differences in the local density distributions are considerable.
Figures 13.3a and b show the predicted relative density distributions at 20,
50, and 70% reductions in height for the friction factors m = 0.2 and 0.5,
respectively. It is seen from Fig. 13.3a that at 20% reduction the density is
lowest near the center of the die contact surface where the deformation is
restricted by friction. The highest density is observed at the center of the
workpiece and also near the outside radius of the die contact surface. At
50% reduction in height, the density is lowest at the equator of the side
surface, and at this stage, the side surface barrels considerably. The higher
mean stress at the equator results in the low density at this point. It is also
noted that at 50% reduction in height, densification near the center of the
die contact surface has been accelerated. This can be explained by the fact
that the pressure near the z axis increases as radius-to-thickness ratio
increases. At 70% reduction in height, a large portion of the workpiece
near the z axis becomes fully dense while the material near the side surface
remains porous. In fact, the density of the workpiece near the free surface
decreases as the deformation progresses from 50% to 70% in height.
Comparing Figs. 13.3a and b, it is seen that the overall densifications of
the workpiece, with the two different interface frictions, are almost the
same. However, it is noted that at 20% reduction, the gradient of the
density distribution is larger with the higher friction. This fact supports the
above argument that the density near the center of the die contact surface
increases because of the increase in mean pressure near the axis. It is well
known that higher pressure near the axis can be achieved with higher
friction when a thin disk is compressed between two flat dies. It is also
 Radius
FIG. 13.3 The predicted relative density distributions in simple compression at 20, 50, and
70% reductions in height. Initial density = 0.8; (a) friction factor = 0.2 and (b) friction
factor = 0.5.
252
 Compaction and Forging of Porous Metals 253

observed that the density near the free surface is lower with higher
friction, owing to the larger barrelling. At 70% reduction in height, it is
noted that the size of the fully dense region near the axis is larger with
higher friction, while the density near the free surface decreases con-
siderably as the workpiece undergoes deformation from 50 to 70%
reduction in height. In practice, fracturing is often observed at the
equatorial surface because of this density reduction.
Figure 13.4 shows the predicted load-displacement curves for two
different friction conditions. For comparison, the predicted load-
displacement curve for the fully dense preform with friction factor m = 0.5
is also shown. The load is lower with the porous preform than that for the
fully dense preform. It can be seen from the figure that the difference
between the forming loads with different frictions is very small during the
early stage of deformation. However, during the later stage, the load
begins to increase faster in the case of higher friction.

13.5 Axisymmetric Forging of Flange-Hub Shapes [18]

In a closed-die forging, the stress state favors complete densification,
because of compressive mean stress. Experimental studies on closed-die

FIG. 13.4 The predicted load versus die stroke curves in simple compression. Ra indicates
the initial relative density and m indicates the friction factor.
254 Metal Forming and the Finite-Element Method

forging of metal powders were presented by Downey and Kuhn [19].

These studies allowed qualitative determination of the preform shapes for
forging to full density with a sound metallurgical structure and without
fracture. One of the forgings investigated had a flange-hub shape. The
forging dies and two preform specimens, with the initial mesh system used
for the analysis, are shown in Fig. 13.5. Since the specimens are
axisymmetric, only a quarter of the workpiece is used for calculations.
Equation (13.8) was used with the stress-strain relationship given by

for the base metal (OFHC copper). The initial relative density was
R0 = 0.800. Y0 is the initial yield stress of the base metal and was given as
42061 psi (290MN/m 2 ). The frictional condition is given by mky (ky
represents the apparent yield stress in shear) and the friction factor m was
assumed to be 0.1.
The results, illustrating the deformation zone and the extent of
densification, are given for two preforms in Fig. 13.6, where the
distributions of apparent effective strain and relative density are shown at
24% reduction in height. In this figure, the pattern of equistrain and
density contours are remarkably similar. This is because the effective strain
contains not only the term of distortion but also the volume change that is
associated with the relative density. The effective strains are largest at the
center of the forging and at the edge of the die-specimen interface, as are
the relative densities.
For preform I, the central region under the die shows no deformation
and no change in relative density from the initial value (R0 = 0.800) for

FIG. 13.5 Preform shapes and mesh systems in flange-hub forging.

 Compaction and Forging of Porous Metals 255

FIG. 13.6 Effective strain (right) and relative density (left) distributions at 24% reduction in
height in forging (a) preform I and (b) preform II.

small reduction. Another observation of importance is that the relative

density near the bulged free surface increases at first, then begins to
decrease as reduction increases. Thus, the equatorial free surface is a
possible fracture site and this was indeed observed in an experiment [19].
For preform II, the workpiece is deforming radially as well as axially
relative to the die motion to fill the cavity. Characteristics of strong
singularity, where the gradients of strain and density are large, are seen
along the die-workpiece contact surface. This results in a distinctly
different deformation pattern from that obtained for preform I. It was
noted for preform II that friction at the die-workpiece interface is
sensitive to the metal flow. It was also found that at a certain stage of
deformation the severe distortions are localized around the die corners and
remeshing was needed for continuing forging. A need for remeshing
occurred at 37% reduction in height for the case of preform II.
The technique for remeshing has been discussed in detail in Chap. 7.
The area-weighted average method is adopted for the problem because of
its simplicity and computational efficiency. In this method, the nodal point
value of relative density or apparent strain is determined from the average
of the adjacent element values surrounding that nodal point weighted by
the associated element areas. Further details of the method can be found
in Reference [20].
Figure 13.7a shows the mesh system at 37% reduction in height before
and after remeshing. The comparisons of relative density and effective
strain fields before and after remeshing are shown in Fig. 13.7b. For
remeshing, no change was made in the number of elements and nodal
points, and therefore the element connectivity was not changed.
In Fig. 13.7b close similarity between the relative density contours
before and after remeshing is found. Other comparisons of the apparent
average relative densities before and after remeshing confirm the credibi-
lity of the program as follows: At 37% reduction in height, Ra — 0.9378
256 Metal Forming and the Finite-Element Method

FIG. 13.7 (a) Mesh systems and (b) relative density contours before (left) and after (right)
remeshing in flange-hub forging.

and 0.9373 before and after remeshing, respectively; and at complete

filling, at 46% reduction in height, Ra = 0.9933 in both cases. Relative
density and effective strain contours at complete filling are shown in Fig.
13.8, which shows that the relative density reaches full density over the
entire forging for preform II.

13.6 Axisymmetric Forging of Pulley Blank

The finite-element program has been applied to the simulation of
closed-die forging of a pulley blank for the two preforms shown in Fig.
13.9 [20]. The sintered preforms used in the simulations were made of
FIG. 13.8 Distributions of relative density (left) and apparent strain (right) at the completion
of forging the flange-hub shape.

FIG. 13.9 Preform shapes and mesh systems in axisymmetric closed-die forging of a pulley
blank.
257
258 Metal Forming and the Finite-Element Method
aluminum powders. The yield stress of the preform was determined from
that of the base metal of commercially pure aluminum and the constitutive
eq. (13.8). The rate-sensitivity of the material property and the tempera-
ture effect on deformation were not included in this simulation. The initial
relative density was assumed to be uniform and to be 0.780 for both
preforms. Two values of m were assumed for friction, m = 0-l and
m = 0-5.
The nonsteady-state forging process was simulated by a step-by-step
method with increments of 2% of the initial height for preform A and 5%
of the initial height for preform B. In order to complete the calculation
successfully, four and ten remeshings were required for preforms A and B,
respectively.
Preform A. The calculation was stopped at 38% and 43% reductions in
height for m = 0.1 and m = 0.5, respectively, in forging preform A. The
program is designed to automatically discontinue the calculation if the
relative density of any element falls below the limiting value of the relative
density for which the apparent yield stress becomes zero, because this
indicates the initiation of fracture.
In Figs. 13.10a and b, the distributions of relative density and hydro-
static stress at the final stage of forging are shown for two friction
conditions. The patterns of densification and hydrostatic stress are similar
to each other. For the two friction conditions, the relative density
distributions differ somewhat in the rim area, as do the hydrostatic stress
distributions. It was found that the effect of friction on the densification of
the materials was not evident at an early stage of forging. As deformation
continues, however, the densification and the compressive hydrostatic
stress are greater with higher friction. It can also be seen that the locations
of the large and small values of relative density are the same for both
friction conditions and lie at the flange part around the die corners and
near the free surfaces, respectively.
Preform B. Contrary to preform A, complete forging was possible with
preform B. This was due to the differences in material flow in the two
preforms.
The distributions of relative density and hydrostatic stress with low
friction were compared with those with high friction at the final stage of
forging in Fig. 13.11. The distribution patterns are similar to each other for
the two friction conditions at the same stage of forging except that for
higher friction, the densification and the compressive hydrostatic stress
were greater. Because of this effect of friction on densification, the final
forged part in preform B with high friction did not completely fill the
cavity at 60% reduction in height due to the volume change, while
complete filling was achieved for low friction at the same reduction in
height (see Figs. 13.Ha and b).
As in forging of preform A, densification and the compressive hydro-
static stress are concentrated at the flange part near the die corners. The
figure also shows that the weakest mechanical properties will be near the
 Compaction and Forging of Porous Metals 259

FIG. 13.10 Relative density (left) and hydrostatic stress (3am, units of ksi) (right)
distributions at the final stage for the two friction conditions in forging preform A (darkened
areas indicate the possible fracture sites): (a)m = 0.1; (b)m=0.5.

tips of the rim section. However, the relative density in most of the forged
hub section reached full density, R = 0.999, in the final stage.
The experiments [19] also showed that preform B produced defect-free
pulley blanks. This agreement allows the prediction of the probable
locations of fracture during forging using the existing finite-element
program.

13.7 Heat Transfer in Porous Materials

Heat transfer analysis for solid metals was presented in Chap. 12. For
porous materials the energy balance equation, corresponding to eq. (12.1)
260 Metal Forming and the Finite-Element Method

FIG. 13.11 Relative density (left) and hydrostatic stress, (3am units of ksi) (right)
distributions at the final stage for the two friction conditions in forging preform B: (a)
m=Q.\, (b) m=0.5.

in Chap. 12, is expressed by

where the subscript R denotes apparent (or equivalent) quantities,

representing a porous material as an equivalent continuum. Thus, using
the apparent thermal properties, kR and CR derived from the base-metal
properties, the heat transfer analysis follows exactly the same procedure
described in Chap. 12.
In porous materials, heat transfer takes place in three ways: conduction
through base metal, convection, and radiation through the pores. When
the size of the pores is sufficiently small, convection is negligible.
 Compaction and Forging of Porous Metals 261

Radiation through pores can be neglected when the temperature is low.

Therefore, conduction plays a crucial role in heat transfer in porous
materials.
Change in thermal conductivity, as determined by porosity, has been
studied extensively and numerous attempts have been made to correlate
the complex effects of the pores with experimental results in a simplified
form [21]. Among published proposals, the most generally employed are
the expressions by Russel [22] and by Eucken [23]. Both expressions were
derived from Maxwell's relation for conductors and resistors. Although the
expressions differ from each other in form, they lead to similar results.
They assume that the pores are discontinuous, spherical, and evenly
distributed throughout a continuous base-metal matrix.
Im and Kobayashi [24] derived a simple linear relationship between the
thermal conductivities of the powdered metal and the base metal. Under
the assumptions that the heat flow is unidirectional and that the volume
pore fraction is equal to the linear pore fraction and to the cross-sectional
pore fraction, as well as that the thermal properties are homogeneous, the
heat balance equation yields

where kR is the apparent thermal conductivity, kb is the thermal conduc-

tivity of the base metal, and kv is the thermal conductivity of the voids.
According to eq. (13.21), kR depends on both the volume pore fraction
and the ratio between the thermal conductivity of the base metal and air.
In addition to the conduction of air, at high temperatures the radiation
across the pores contributes to heat transfer. In this case, the radiation
portion of heat transfer can be added to the thermal conductivity of air and
lowers the ratio between the thermal conductivity of the base metal and air
in eq. (13.21). By neglecting kv/kb terms compared to 1, eq. (13.21)
becomes

Since the effect of temperature on the yield stress is determined by the

temperature of the base metal, the relationship between the temperatures
of the base metal and the porous material should be determined. By
introducing the apparent density pR and the specific heat, CR of porous
materials, the total change in internal energy can be expressed by

where the subscripts R, b, and v denote the quantity related to the total
porous materials, base metals, and voids, respectively, t is the time, and p,
c, and V are the density, the specific heat, and the volume, respectively.
The temperature-rate of voids is at most of the same order as the
262 Metal Forming and the Finite-Element Method

temperature-rate of the base metal, since the heat generation due to plastic
deformation is limited to that of the base metal. The change in the internal
energy of voids can also be neglected, since the thermal capacity of the
voids is much smaller than that of the base metal. Therefore, eq. (13.23)
can be reduced to

where the apparent specific heat of the porous material, CR, can be
determined by

Combining this equation with eq. (13.24), noting that pRVR = pbVb, gives

Integration of eq. (13.26) leads to the expression of the temperature of the

base metal as,

13.8 Hot Pressing Under the Plane-Strain Condition

Much experimental work has been performed on powdered-metal hot
forging. Preform ductility and transient cracking in forging were analyzed
by Fischmeister et al. [25], who showed that large amounts of plastic
deformation are beneficial in obtaining good impact and fatigue properties.
Malik [26] has published work on the producibility of titanium powdered-
metal shapes for aerospace structural applications. Ferguson et al. [27]
have analyzed the hot isostatic process, which involves a consolidation of
loose powders or sintered preforms.
Im and Kobayashi [24] applied the finite-element technique to the
analysis of plane-strain compression at elevated temperatures. The work-
pieces, made of sintered iron powders, were assumed to have uniform
initial relative densities of 0.802 and 0.743. Computed results were
compared in terms of macroscopic densification and forging pressure with
experimental values published by Fischmeister et al. [28]. In upsetting long
bars between flat dies, (see Fig. 13.12), the plane-strain condition (ex =0,
txy —Txz = 0) was assumed at the central cross-section, owing to the
restraint of longitudinal material flow.
The conditions used in the computation were as follows. The dimensions
of the workpieces were 10x10x100 mm. (0.39 x 0.39 x 3.94 in.) Two
initial relative densities, R0 = 0.743 and R0 = 0.802, were selected from
four experimental cases reported in Reference [28]. Since no lubricant was
used between the workpiece and dies in the experiments, two friction
factors, 1.0 and 0.5 were assumed for both cases. Initially the temperatures
of the dies and workpieces were assumed to be 293 K (68°F), and 1433 K
 Compaction and Forging of Porous Metals 263

FIG. 13.12 Schematic of plane-strain compression: (a) geometry of the die and workpiece;
(b) boundary conditions. (Superscripts c, r, f on the heat flow qn refer to convection,
radiation, and friction, respectively.)

(120°F), respectively. The material property of the base metal was

assumed to be rigid-plastic, and the yield stress was taken as 28717 psi
(198 MN/m 2 ) from the handbook. Again, the consitutive eq. (13.8) was
used.
The thermal properties necessary for computation were taken from the
handbook as follows:
kb = 55 N/(s K), kv = 0.045 N/(s K), kd = 19 N/(s K),
2
C f c p 6 =3.6N/(mm K), capu =0.4N/(mm K), cdpd = 3.77N/(mm 2 K),
2

CT£ = 3.6xlON/(mm 2 sK 4 ),
/zc = 5.5N/(mmsK), /z00 = 0.01 N/(mms K),
K = 0.85
264 Metal Forming and the Finite-Element Method

Because of geometrical symmetry, only a quarter-section of the central

cross-section of the bar was analyzed for deformation and temperature.
The boundary conditions are also shown in Fig. 13.12. Deformation was
analyzed by the step-by-step method with increments of 0.4% of the initial
height of the workpiece. Total reduction in height was 70% for R0 = 0.743
with low friction. For R0 = 0.802 with low friction, the calculation stopped
at 67% owing to the severe distortion in the corner element of the
deformed free surface in contact with the die. For both /?„ = 0.743 and
R0 = 0.802 with high friction, the calculations stopped at 58% and 64%,
respectively, because the relative density of an element fell below the
limiting value.
The computed and the experimental results are compared in terms of
the macroscopic deformation behavior. The changes of lateral flow during
the plane-strain compression are given in Fig. 13.13. The average width Wa
of the deformed free surface is used to calculate the true strain in the y
direction, ey = ln(Wa/W0), where W0 is the original width. According to the
figure, the results show good agreement between computation and
experiment. The straight line in the figure is for the fully dense material
(R = 1.0), where ez = — ey because of incompressibility. In a porous
material, volume change occurs during deformation and the increase in
width becomes less than the decrease in height.
The macroscopic level of densification as function of axial compression is
given in Fig. 13.14. As seen in Fig. 13.14 the predictions are excellent for
all cases. It is also seen that the densification occurs mainly at an early
stage of upsetting. As deformation proceeds, the densification becomes
saturated and lateral flow is enhanced.
Figure 13.15a reproduces the local hardness distribution in the homoge-

True S t r a i n in y - d i r e c t i o n
FIG. 13.13 Comparison of variations in lateral flow between theory and experiment [28].
FIG. 13.14 Variations of average relative density during compression, and comparison with
experiment [28].

(b)
FIG. 13.15 (a) Experimental local hardness distribution in the center cross-section of a bar
[28]. (b) Relative density distributions at 40% reduction in height for Ra = 0.743 with two
friction conditions: left (m = 0.5); right (m = 1.0).
265
266 Metal Forming and the Finite-Element Method

FIG. 13.16 Relative density distributions at the final stages for Ra = 0.743 with two friction
conditions: (left) 70% reduction and m = 0.5; (right) 58% reduction and m = 1.0.

neous center region, obtained from the experimental results given in

Reference [28]. This information can be correlated with the computed
relative density distributions, such as shown in Fig. 13.15b. These figures
show that densification is effective near the edge (owing to folding during
compression) and in the central region of the deformed workpieces.
Excellent agreement of relative density and hardness distributions between
experiment and computation can be seen from these figures.
Figure 13.16 shows the relative density distributions at the final stages of
compression for R0 = 0.743 with two friction conditions. It can be seen that
most of the central region is fully dense and that less densification occurs
near the free surface. In some elements at the free surface, the relative
density reaches the limiting value and then tends to decrease. This is
interpreted as possible fracturing and the critical elements are shown as
darkened areas in the figures.
The temperature profiles in the workpiece and dies are presented in
Figs. 13.17a and b for R0 = 0.743. Figure 13.17a shows that the effect of
friction on the temperature distribution is negligible. It is also seen that the
conductive heat flow from the workpiece to the die is dominant when
compared to heat generation due to plastic deformation at an early stage.
Consequently, the isothermal lines in the workpiece are almost linear. As
deformation increases, the isothermal lines encircle the center, because
heat generation due to plastic deformation increases and is maximal at the
center, as seen in the left-hand side of Fig. 13.17b. A transition between
these two types of temperature distributions can be observed at 58%
reduction in height, shown in the right-hand side of Fig. 13.17b.

13.9 Compaction
In the process of compaction, a P/M preform is placed in a cylindrical
container and pressed in a double-action press, as shown in Fig. 13.18. For
the simulation [14], the container was 50.8 mm (2.0 in.) in diameter and
152.4 mm (6.0 in.) in height. Because of symmetry, the analysis was
performed for the half-height. The flow stress behavior of the matrix
material was assumed to be Yb =0.1. The constant-friction factor law was
FIG. 13.17 (a) Temperature distributions for R0 = 0.743 at 40% reduction in height with two
friction conditions: (left) m=0.5; (right) m = 1.0. (b) Temperature distributions for
R0 = 0.743 at the final stages with two friction conditions: (left) m = 0.5; (right) m = 1.0.
267
268 Metal Forming and the Finite-Element Method

FIG. 13.18 Schematic diagram of simple die compaction.

assumed, and the friction factors used in simulation were m = 0.1, 0.2, and
0.5. The punch velocities were 76.2 mm/s (3.0 in./s) for both the upper
and lower punches. The initial relative density in the simulation was taken
to be 0.8. It was also assumed that the air and the lubricant trapped in the
pores did not affect the compaction process.
Figures 13.19a and b show the predicted relative density distributions for
the friction facors m =0.2 and 0.5, respectively. It is seen that the density
is greatest near the outside radius of the moving punch and decreases in
regions remote from the punch, particularly near the container die wall.
As the deformation progresses towards full compaction, the density
distribution becomes more uniform; that is, the difference between the
highest and the lowest density becomes smaller. It can be seen that the
trend of density distributions for both friction conditions are similar, but
the density is more uniform with lower friction.
From the predicted density distribution (which is a function of radial as
well as height location), the average density was calculated as a function of
height. The average density was obtained by integrating the relative
density distribution over the cross-sectional area and dividing by that
cross-sectional area at a given height.
Figures 13.20a and b show the predicted density variation as a function
of height for different friction factors, m = 0.2 and m = 0.5, respectively. It
can be seen from the figures that the density is lowest at the mid-height,
and that it increases toward the punch. It is also seen that the distribution
becomes more uniform with increasing deformation and lower friction.
 Compaction and Forging of Porous Metals 269

FIG. 13.19 (a) Predicted relative density distribution during die pressing; initial relative
density = 0.8; friction factor = 0.2. (b) Predicted relative density distribution during die
pressing; initial relative density = 0.8; friction factor = 0.5.

In order to validate the prediction of the current model, experimental

data for pressing of a powder preform were sought from the literature.
None that were directly comparable could be found. However, density
measurements for powder compaction in a closed die were found in a
German doctoral dissertation [29]. Figure 13.21 shows measured density
distributions of P/M preforms that were prepared by die compaction and
isostatic compaction. From this figure, it is apparent that the density
270 Metal Forming and the Finite-Element Method
D E N S I T Y D I S . M=0.2

D E N S I T Y D I S M=0.5

FIG. 13.20 (a) Predicted average density as function of height; initial relative density = 0.8;
friction factor = 0.2. (b) Predicted average density as function of height; initial relative
density = 0.8; friction factor = 0.5.

distributions of the preform obtained by die compaction are similar to

those predicted.
Figure 13.22 shows the predicted punch load vs. punch stroke curves for
friction factors m = 0.1, 0.2 and 0.5. The figure illustrates that the pressing
load is higher for higher friction. The load is important in a die pressing
operation because the maximum allowable load determines the level of
achievable average density.
 Billet Height

FIG. 13.21 Measured average relative densities obtained in compaction of powder in closed
dies and by hydrostatic pressing [29].

271
272 Metal Forming and the Finite-Element Method

Punch Stroke (in.)

FIG. 13.22 The predicted punch load during die pressing.

References
1. Wisker, J. W., and Jones, P. K., (1974), The Economics of Powder Forging
Relative to Competing Processes—Present and Future, in "Modern Develop-
ments in Powder Metallurgy," (Edited by H. H. Hausner, and W. E. Smith),
Vol. 7, American Powder Metallurgy Institute, Princeton, NJ, p. 33.
2. Jones, P. K., (1973), "New Perspectives in Powder Metallurgy," Vol. 6,
Plenum Press, New York.
3. Kaufman, S. M., and Mocarski, S., (1971), "Effect of Small Amount of
Residual Porosity on the Mechanical Properties of P/M Forgings," Int. J.
Powder Metal., Vol. 7, p. 19.
4. Dieter, G. E., (1976), "Mechanical Metallurgy," 2d Ed., McGraw Hill, New
York.
5. Moyer, K. H., (1974), A Comparison of Deformed Iron-Carbon Alloy Powder
Preforms with Commercial Iron-Carbon Alloys, in "Modern Developments in
 Compaction and Forging of Porous Metals 273

Powder Metallurgy," (Edited by H. H. Hausner and W. E. Smith), Vol. 7,

American Powder Metallurgy Institute, Princeton, NJ, p. 235.
6. Kuhn, H. A., and Downey, C. L. Jr., (1971), "Deformation Characteristics
and Plasticity Theory of Sintered Powdered Materials," Int. J. Powder Metal.,
Vol. 7, p. 15.
7. Green, R. J., (1972), "A Plasticity Theory for Porous Metals," Int. J. Mech.
Sci., Vol. 14, p. 215.
8. Shima, S., and Oyane, M., (1976), "Plasticity Theory for Porous Metals," Int.
J. Mech. Sci., Vol. 18, p. 285.
9. Tabata, T., and Masaki, S., (1975), "Plane-Strain Extrusion of Porous
Materials," Memoirs of the Osaka Institute of Technology, Series B., Vol. 19,
No. 2.
10. Tabata, T., and Oyane, M., (1975), "The Slip-Line Field Theory for a Porous
Material," Memoirs of the Osaka Institute of Technology, Series B., Vol. 18,
No. 3.
11. Hill, R., (1950), "The Mathematical Theory of Plasticity." Oxford University
Press, London.
12. Johnson, W., and Miller, P. B., (1980), "Engineering Plasticity." Van
Nostrand Reinhold, London.
13. Doraivelu, S. M., Gegel, H. L., Gunasekera, J. S., Malas, J. C., and Morgan,
J. T., (1984), "A New Yield Function for Compressible P/M Materials," Int.
J. Mech. Sci., Vol. 26, p. 527.
14. Oh, S. L, and Gegel, H. L., (1986), "ALPIDP—Modeling of P/M Forming by
the Finite Element Method," Proc. NAMRC XIV, Minneapolis, MN, p. 284.
15. Oh, S. L, Wu, W. T., and Park, J. J., (1987), "Application of the Finite
Element Method to P/M Forming Processes," Proc. 2nd ICPT, Stuttgart, West
Germany, p. 961.
16. Osakada, K., Nakano, J., and Mori, K., (1982), "Finite Element Method for
Rigid-Plastic Analysis of Metal Forming—Formulation for Finite Deforma-
tion," Int. J. Mech. Sci., Vol. 24, p. 459.
17. Mori, K., Shima, S., and Osakada, K., (1980), "Finite Element Method for
the Analysis of Plastic Deformation of Porous Metals," Bull. JSME, Vol. 23,
No. 178.
18. Im, Y. T., and Kobayashi, S., (1985), Finite Element Analysis of Plastic
Deformation of Porous Materials, in "Metal Forming and Impact Mechanics,"
(Edited by S. R. Reid), Pergamon Press, Oxford, p. 103.
19. Downey, C. L. Jr., and Kuhn, H. A., (1975), "Application of a Forming Limit
Concept to the Design of Powder Preforms for Forging," /. Engr. Mat. Tech,
Vol. 97, p. 121.
20. Im, Y. T., and Kobayashi, S., (1986), "Analysis of Axisymmetric Forging of
Porous Materials by the Finite Element Method," Advanced Manufacturing
Processes, Vol. 1, p. 473.
21. Austin, J. B., (1939), Factors Influencing Thermal Conductivity of Non-
metallic Materials, in "Symposium on Thermal Insulating Materials," Ameri-
can Society for Testing Materials, Philadelphia, p. 3.
22. Russel, H. W., (1935), "Principles of Heat Flow in Porous Insulators," /. Am.
Ceram. Soc., Vol. 18, p. 1.
23. Eucken, A., (1932), VDI, Forschungsheft, No. 353 (in German), Forschung
auf dem Gebiete des Ingenieurwesens Ausgabe B, Band 3, March-April.
24. Im, Y. T., and Kobayashi, S. (1986), "Coupled Thermo-Viscoplastic Analysis
in Plane-Strain Compression of Porous Materials," Advanced Manufacturing
Processes, Vol. 1, p. 269.
25. Fischmeister, H., Sjoberg, G., Elfstrom, B. O., Hamberg, K., and Mironov,
V., (1977), Preforms Ductility and Transient Cracking in Powder Forging, in
"Modern Developments in Powder Metallurgy," (Edited by H. H. Hausner
274 Metal Forming and the Finite-Element Method

and P. V. Taubenblat), Vol. 9, American Powder Metallurgy Institute,

Princeton, NJ. p. 437.
26. Malik, R. K., (1974), "Hot Pressing of Titanium Aerospace Components,"
Int. J. Powder Metal, and Powder Tech., Vol. 10, No. 2.
27. Ferguson, B., Kuhn, A., Smith, O. D., and Hofstatter, F., (1984), "Hot
Consolidation of Porous Preforms Using Soft Tooling," Int. J. Powder Metal,
and Powder Tech., Vol. 20, No. 2, p. 131.
28. Fischmeister, H. F., Aren, B., and Eastering, K. E., (1971), "Deformation
and Densification of Porous Preforms in Hot Forging," Powder Metal., Vol.
14, No. 27, p. 144.
29. Schacher, H. D., (1978), Kaltmassivumformen von Sintermetall, Ph.D.
Dissertation, Institute fur Umformtechnik, Universitat Stuttgart.
 14
THREE-DIMENSIONAL PROBLEMS

14.1 Introduction
A majority of the finished products made by metal forming are geometri-
cally complex and the metal flow involved is of a three-dimensional nature.
Thus, any analysis technique will become more useful in industrial
applications if it is capable of solving three-dimensional metal-flow
problems.
Nagpal and Altan [1,2] introduced dual-stream functions for describing
metal flow in three dimensions. This work showed that the proper
selection of a flow function makes the incompressibility requirement
automatically satisfied and provides general kinematically admissible
velocity fields. Yang and Lee [3] utilized the conformal transformation of a
unit circle onto a cross-section in the analysis of curved die extrusion. They
derived the stream-line equation from which a kinematically admissible
velocity field was determined. The upper-bound method was then applied
to determine the extrusion pressure for a rigid-perfectly plastic material.
An important aspect of three-dimensional plastic deformation is the
analysis of spread in metal-forming operations, such as spread in rolling or
in flat tool forging, and spread in compression of noncircular disks.
Solutions to such problems have been obtained by the use of Hill's general
method [4] and the upper-bound method [5-7].
The extension of the finite-element method to solve three-dimensional
problems is natural and not new, particularly in the area of elasticity [8].
However, the simulation of three-dimensional forming operations by the
finite-element method is relatively recent. Park and Kobayashi [9] de-
scribed the formulation for the three-dimensional rigid-plastic finite-
element method and the implementation of the boundary conditions. They
applied this technique to the analysis of block compression between two
parallel flat platens. For certain forming problems, such as those involving
lateral spread, the use of a simplified three-dimensional element is efficient
and some examples can be found for analysis of spread in rolling and flat
tool forging [10, 11].

275
276 Metal Forming and the Finite-Element Method
14.2 Finite-Element Formulation
Element Equations
The matrices for evaluation of elemental stiffness equations are defined for
a three-dimensional brick element in Chap. 6 and some of them are
recapitulated in this section. A three-dimensional brick element used for
the analysis is an eight-node hexahedral isoparametric element. As shown
in Fig. 14.1 the global coordinate system (x, y, z) is transformed into the
natural coordinate system (£, rj, £). The natural coordinate system is
defined such that £, r\, and £ vary from —1 to 1 within each element. The
velocity field within an element is expressed as

where

with

For eqs. (14.1), (14.2), and (14.3) above, reference can be made to eqs.
(6.17), (6.18), and (6.19) in Chap. 6.

FIG. 14.1 Three-dimensional hexahedral element.

 Three-Dimensional Problems 277

The strain-rate vector E T = {ex, ey, ez, jxy, yyz, YZ*} is expressed by

by taking the space derivatives of velocities, where the matrix B is given by

where

and

The determinant of the Jacobian matrix of the coordinate transformation

that is necessary for integration of elemental stiffness equation is given by
eq. (6.38) in Chap. 6.
Traction Prescribed Boundary
The boundary conditions along the interface between workpiece and die
are mixed, because the velocity is given in the direction normal to the
interface surface and the traction is prescribed by friction along the contact
surface in the direction opposite to the relative sliding of the workpiece
with respect to the die. The prescribed velocity is treated as an essential
boundary condition, and the prescribed traction, which is a suppressible
boundary condition, is imposed during the discretization process.
Suppose that the surface (1-2-3-4) of the element shown in Fig. 14.1 is
in contact with the die (x, y plane). On that surface, the natural coordinate £
is equal to unity and the shape functions become qt = \(l + ££,)(! + r/Jj,).
Equation (14.2) reduces to

and the nodal point velocity vector is given by

he determinant of the Jacobian transformation matrix is given (see eq.

(6.35b) in Chap. 6) by
278 Metal Forming and the Finite-Element Method

where
and

The frictional stress is represented as velocity dependent (see eq. (5.30)

in Chap. 5). For the Newton-Raphson procedure, the complete formula-
tion requires the detailed manipulations for the derivatives of the
functional and can be found elsewhere [12].

14.3 Block Compressions [9]

Compression of Rectangular Blocks
The compression of rectangular blocks was investigated by Kanacri et al.
[13] using annealed Al-1100. The same conditions of workpiece geometry,
material property, and interface friction were used for simulations.
The workpiece dimensions were 1/1/0.5 and 2/1/0.5 (width W/breadth
B/thickness H). The friction factors m — 0.1 and 0.2 were chosen for the
lubricated conditions for the two geometries, while m = 0.5 was used for
the dry (nonlubricated) condition. The flow stress of annealed Al-1100 is
approximately represented by

with r0 = 9102 psi (62.74 MN/m 2 ).

The mesh systems used for computation consist of a total of 75 elements
interconnected at 144 nodal points for W0/B0 = 1, and 72 elements with 104
nodal points for W0/B0 = 2. The computation was carried out up to 50%
reduction in height with 2% height reduction as a step size. The
convergence was rapid and steady. Only four iterations were necessary to
reach the accuracy of the velocity error norm less than 0.00001 with a
deceleration coefficient of 0.5. Each simulation needed about 600 CP
seconds execution time on a CDC 7600 computer.
The results are summarized in terms of contact pressure distributions,
compression loads, spread contours, and strain distributions. Comparisons
of the computed loads with the experimental results are shown in Fig.
14.2. It is of interest to note that the computed results are closer to the
interrupted loadings for both friction conditions. Also, it is seen that
assuming m = 0.1 underestimates the actual friction for the lubricated case
for WQ/B0 = 1. However, m = 0.2 and m = 0.5 are good choices for overall
estimations of the friction conditions for W0/B0 = 2.
Geometrical changes of the equatorial planes under both friction
conditions are compared in Fig. 14.3. Because the height of the workpiece
is relatively short compared to the other two dimensions, geometrical
changes along the height direction are negligibly small. As the deformation
continues the workpiece spreads out and the compression load increases.
A discrepancy in the spread contour at the corner is shown for W0/B0 = 1
(lubricated), which suggests that m = 0.1 used in simulation is less than the
 FIG. 14.2 Comparisons of the computed compression loads (broken curves) with the experimental results (solid curves) [13] in rectangular block
KJ compression.
£
o

FIG. 14.3 Comparison of the computed (broken curves) and experimental (solid curves) [13] spread contours at the equatorial plane in rectangular block
compression.
 Three-Dimensional Problems 281
actual friction value that is present in the experiment. On the other hand,
in the simulation for WJB0 = I, m = 0.5 is a proper value to assume for
the dry condition. The simulation, for W0/B0 = 2 with m = 0.2 for the
lubricated condition, gives an excellent agreement with the experimental
result. Some discrepancy around the corner in simulation for the dry
condition may indicate that friction is not uniform over the entire
die-workpiece interface.
Compression of Wedge-Shaped Blocks
Compression of a wedge-shaped block (Fig. 14.4) between two flat parallel
dies is used in practice for workability and microstructural studies in
forging. The boundary conditions for the analysis are more complex in
comparison with those for rectangular block compression owing to the
lower degree of symmetry in the geometry. The determination of the
neutral regions at the contact surfaces, with both upper and lower dies,
represents an important problem. In particular, along the die-workpiece
contact surface, at the lower die the extent of the actual contact is not
known and varies during deformation. The finite-element code for this
problem includes additional schemes in order to take into account these
boundary conditions, as well as folding, that occur in rectangular block
compression.
The workpiece dimensions used in the simulations are shown in Fig.
14.4. The workpiece geometry and deformation are symmetrical with
respect to the (x, z) plane, so that one-half of the workpiece is taken as the
control volume for simulation. The flow stress of the material, Al-2024
(annealed), is represented by a = 225 (1 + e/1.6147) (MN/m2). Two
simulations were performed under different friction conditions: dry (m =
0.4) and lubricated (m = 0.1). A step size of 2% in height reduction was

FIG. 14.4 Dimensions of the workpiece in compression of wedge-shaped block.

282 Metal Forming and the Finite-Element Method
used up to 30% reduction in height, and a step size of 1%, thereafter.
Sixty elements (126 nodes) were used for the lubricated condition, while 48
elements (105 nodes) were used for the dry condition. The computational
time (CP time) for the total deformation of 60% reduction in height was
2300 s with 60 elements and 1700 s with 48 elements on a CDC 7600
computer.
Experiments were performed, and measurement such as load-
displacement curves and geometrical changes were compared with some of
the computed results. Details of the experiments and comparison with the
theory are reported in Reference [14].
Grid distortions at three stages of height reduction are shown in Fig.
14.5. Several observations about the mode of deformation can be made
from this figure. The side surface of the block portion bulges outward,
while the side surface of the wedge portion becomes concave. This mode
of deformation is more pronounced with the dry interface condition. The
contours of the die-workpiece contact area show more severe distortion in
the width direction for higher interface friction.
Effective strain distributions in the mid-plane (plane of symmetry) are
shown in Fig. 14.6. Locations of the neutral zones are indicated by arrows.
The results show that higher concentration of strain occurs around the

FIG. 14.5 Grid distortion of the wedge-shaped workpiece at several height reductions (20,
40, and 60%).
NJ FIG. 14.6 Effective strain distributions and location of neutral zones (indicated by arrows) in the plane of symmetry at several height reductions of
w compression of a wedge-shaped block.
284 Metal Forming and the Finite-Element Method

lower center of the block portion and that the strain decreases monotoni-
cally toward the end of the wedge. The strains are, in general, uniform in
the height direction, particularly in the wedge portion and for low friction.
Locations of the neutral zones are symmetrical and approximately posi-
tioned at the mid-point of the contact surface at larger reductions in
height. However, at 20% reduction, the interfacial friction condition
appears to have a significant influence on the location of the neutral zones.
In using the wedge compression for a workability test, fracture is usually
observed on the side surface. The present analysis provides some
information concerning the occurrence of fracture, such as the strain path
and the corresponding stress variations at critical sites.

14.4 Square-Ring Compression [15]

In square-ring compression, a cube shaped billet with a coaxial square hole
is pressed between flat dies. A schematic diagram of the undeformed
square-ring is shown in Fig. 14.7. It is expected, in this process, that
different modes of deformation may result depending upon the aspect ratio
of workpiece and friction. The three-dimensional finite-element method
was used to simulate the process [15]. The undeformed preform was a cube
(50.8 x 50.8 x 50.8 mm) with co-axial square hole (25.4 x 25.4 x 50.8 mm).
The friction between the workpiece and the die was assumed to follow
the constant shear factor friction law with m=0.5. The workpiece
material was annealed Al-1100 and its flow stress was represented by o =
62.74(1 + £/0.052)a3 (MN/m2). Owing to the symmetry, it was sufficient to
include one-sixteenth of the workpiece in the analysis, as shown by thick

FIG. 14.7 Schematic diagram of the square ring used in simulation. The thick lines indicate
the volume included in the analysis.
 Three-Dimensional Problems 285

lines in Fig. 14.7. A total of 250 elements with 396 nodes were used
for the analysis. Simulation was done in a step-by-step manner and 2%
of the undeformed height was used as the step size.
The results show that the buckling-type deformation mode takes place at
the beginning of deformation. That is, the mid-height of the side wall
moves outward while the die contact surface moves very little in the
horizontal direction. As a result, nodes at the die contact surface near the
inner edge separate from the die at the initial stage of the deformation.
The separated nodes touch the die shortly afterwards. Figure 14.8 shows
the perspective view of one-quarter of the workpiece with the predicted
mesh distortions of the cross section on the YZ plane at steps of 20% in
height. It is seen in the figure that at 20% reduction in height, the
mid-plane moves outward while the top and bottom surfaces change their
positions very little. This deformation mode causes buckling. At 40%
reduction in height, the outside wall forms a considerable bulge, while the
inner wall forms a dimple. The dimple at the inner wall becomes sharper
as deformation progresses and, finally, it collapses, forming a fold at 54%
reduction in height. This inner surface folding takes place first at the YZ
cross section and propagates towards the corner of the hole. The folding of
the inner surface can be seen clearly from the predicted grid distortion at
60% reduction in height.
In Fig. 14.8, locations of the neutral points are indicated by arrows. It is
seen that the spatial position of the neutral point on the YZ cross section
changes very little with deformation. However, its location relative to a
material point changes considerably.
Figure 14.9 shows the top view of the distorted mesh at steps of 20%
reduction in height. In the figure, the neutral lines are indicated by broken
lines. It is seen from the figure that the material points at the mid-side of
the inner surface move inward, while points at the corner of the inner
surface move outward until 40% reduction in height is reached; the corner
points then start to move inward. Because of the different movements of
the material points on the die contact surface, the inner surface, looking
from the top, becomes inwardly convex and the angle between the two
inner surfaces becomes less than 90°.
Aku et al. [16] carried out square-ring compression tests by using
plasticine as a model material. Though the detailed measurements of the
process have not been reported, they include the sketch of the top view of
the deformed workpiece at 30, 50 and 77% reductions in height. By
comparing their sketch with Fig. 14.8, it is found that the present
predictions agree qualitatively very well with the experimental observa-
tions. Seen from the top, the inner hole (which is inwardly convex), the
bulge at the inner surface (which is not noticeable), and the large amount
of bulge at the outside surface all indicate reasonable agreement between
the predictions and experiment. These observations are only qualitative
and further critical comparison may be necessary to fully verify the results
of the simulations.
FIG. 14.8 Perspective views of one-quarter of the workpiece for square-ring compression,
with the predicted mesh distortions on the YZ cross section at 0, 20, 40 and 60% reductions
in height (units of mm). (Arrows indicate the position of the neutral surface.) [15].

286
 Three-Dimensional Problems 287

FIG. 14.9 Top views of grid distortions predicted by the analysis at 0, 20, 40, and 60%
reductions in height during square-ring compression. The broken lines indicate the neutral
lines (units of mm) [15].

A total of 44 hours of CPU time was required for 412 iterations to

complete the simulation on a VAX 750. About 70% of the total CPU time
was used after 40% reduction in height, owing to a large amount of free
surface folding.

14.5 Simplified Three-Dimensional Elements

The use of simplified three-dimensional elements can be reasonable in
analyzing some aspects of three-dimensional deformation. One such aspect
is the analysis of lateral spread. Simplified hexahedral elements can be
derived by taking one layer of elements in the direction normal to the
plane of symmetry and assuming that the element-sides, which are initially
normal to the plane, remain normal during deformation.
In the element shown in Fig. 14.1, suppose that the plane of symmetry is
(XOY) and that the nodes 5, 6, 7, and 8 will remain on that plane
288 Metal Forming and the Finite-Element Method
throughout the process. Then the assumption states that the velocity
components ux and uy are independent of the z coordinate. We then have

and

We also have

and

This assumption leads to a linear distribution of uz in the z direction,

and the coordinate transformation and shape functions become

With this simplification, a three-dimensional eight-node element can be

described by the coordinates and the velocities of only four lateral nodes
instead of eight nodes.
For the simplified 8-node hexahedral element defined by eqs. (14.11),
(14.12) and (14.13), the strain-rate matrix B and the Jacobian matrix of
transformation simplify accordingly.
In the three-dimensional treatment of friction, where the contact surface
is inclined with respect to the global coordinates, the surface integration
must be performed numerically. Therefore, special care should be taken in
the evaluation of the second-order derivative in the numerical integration
scheme. With reference to the element arrangement shown in Fig. 14.1, let
(3-4-8-7) be the surface of an element on which the frictional stress is
acting, in the example of the rolling process (Fig. 14.10). The relative
velocity components with respect to the die velocity in the tangential and
the transverse directions are u, and uz, respectively. According to the
velocity distribution in the simplified element, we have

In eq. (14.14), the directions of «,<3) and w,(4) are tangential to the contact
surface at points 4 and 3, respectively. Thus, the distribution of u, given in
eq. (14.14) is the result of an approximation usually involved in
discretization.
 Three-Dimensional Problems 289

FIG. 14.10 Roll-workpiece contact surface of an element.

For surface integration of these expressions, the Gaussian quadrature

was used in the z direction and the Simpson rule was used in the tangential
direction. Further, the Gaussian quadrature at eight integration points was
used for volume integration over the element.
14.6 Analysis of Spread in Rolling and Flat-Tool Forging
Rolling [10]
In rolling wide strips or plates, where the ratio of width-to-length of the
deformation zone is large (say, more than 10), spread in width is usually
negligible and such deformation problems can be considered to be in
plane-strain. Thus, most of the early theories of rolling were concerned
with predicting the load and torque under the assumption of plane-strain
deformation. However, in rolling flat sections for which the width-to-
length ratio is small (say, less than 6), spread is appreciable and cannot be
neglected.
Since there was no theory of rolling to deal with three-dimensional
deformation, with few exceptions, studies on deformation in rolling have
been largely experimental. Chitkara and Johnson [17] and Helmi and
Alexander [18] conducted experiments of rolling of lead bars and
mild-steel bars, respectively, and compared the experimentally determined
spreads with those predicted by empirical formulas.
Few attempts have been made to predict the spread theoretically.
Gokyu et al. [19] derived the expression for the width spread, applying the
minimum-work hypothesis to the rolling problem. Lahoti and Kobayashi
290 Metal Forming and the Finite-Element Method
[20] performed the computation for spread neglecting friction between roll
and strip, and using the general method proposed by Hill (see Chap. 5).
Kummerling and Lippmann [21] analyzed the spread by using a simplified
two-parameter theory of plasticity. Oh and Kobayashi [5] analyzed
three-dimensional deformation in rolling using an extremum principle for
rigid, perfectly plastic materials.
Several researchers have tried to solve the rolling problem by the use of
the finite-element technique [22-27]. All the studies, however, used the
plane-strain assumption, and the only approach taking into account the
lateral spread in rolling has been the analysis by Kanazawa and Marcal
[28], using line elements. At the international conference on numerical
methods in industrial forming processes, held at the University College of
Swansea in 1982, two papers [10, 29] were presented on the analysis of
spread in rolling. Both used the rigid-plastic finite-element method with
simplified three-dimensional elements.
For the analysis of spread in rolling, two arrangements of elements are
possible. The first considers the plane of symmetry (XOY) and the second
refers to the plane (XOZ) as a plane of symmetry, as shown in Figs.
14.11a and b, respectively. The element arrangement of Fig. 14.11a
provides the bulge profile of the front and side surfaces of the rolled stock,
in addition to the amount of spread, while the arrangement of Fig. 14.lib
results in the spread contour of the front edge, but does not give bulging of
the spread surfaces. Therefore, the choice of the element arrangement
depends on the information required and possibly on the rolling condition,
such as the width-to-thickness ratio. Mori and Osakada [29] applied the
first type of element arrangement for the analysis of edge rolling
(WQ/H0«1), and the second type for the spread analysis in plate rolling
(WQ/HQ a 1). The rolling problem can be analyzed either as a nonsteady-
state deformation or as a steady-state deformation. Mori and Osakada
adopted the steady-state approach. For this case, a solution is obtained in
terms of velocities and strain-rates. Then the flow lines are constructed and
the effective strain distribution, and thus the flow stress distribution, is
computed by integrating the effective strain-rates along the flow lines. The
calculation is repeated until the solution converges (see Chap. 10).
For the nonsteady-state procedure, deformation of the workpiece is
simulated in a step-by-step manner from the beginning of the bite,
updating the coordinates of material points and the material property after
each step. The steady state is assumed to be reached when the spread
contour becomes stationary in space and the roll torque has reached a
steady value.
The analysis of lateral spread was performed using the nonsteady-state
approach with the simplified eight-node element and the element arrange-
ment shown in Fig. 14.11a. The conditions under which simulation were
performed were as follows: roll radius R = 203 mm (8 in.); initial thickness
of the plate 2H0-2.5mm (0.1 in.); width-to-thickness ratios WQ/H0 = l
and 3; reduction in thickness, 5, 12.5, and 20%; material, annealed AISI
steel 1018; friction factor m = 0.5.
FIG. 14.11 Flat section rolling: configuration and two arrangements of elements.
291
292 Metal Forming and the Finite-Element Method

FIG. 14.12 The profiles of a rolled square bar for 20% reduction: W0/HQ= I.

These computational conditions and the experimental data were taken

from Lahoti et al. [30]. Figure 14.12 shows the spread of the rolled stock
for 20% reduction with a width-to-thickness ratio of unity. The average
lateral spreads of the rolled bars, computed and measured, are compared
in Fig. 14.13. Also, the reduction in cross-sectional area is plotted as a
function of reduction in height, as an indirect measure of the amount of
spread. The figure shows that the agreement between the theoretical
predictions and the experimental results is excellent for both W0/H0 = 1
and 3. It is to be noted that in the rolling process, as the material is
compressed, its cross-sectional area also is reduced. In the case of
plane-strain, the reduction in height is equal to that in the cross-sectional
area; however, for smaller W0/H0 ratios, the reduction in cross-sectional
areas is less than that in height. Figure 14.13 also shows the relationship
between these two reductions, i.e., in cross-sectional area and in height.
 Three-Dimensional Problems 293

FIG. 14.13. (a) Comparison between computed and measured [30] lateral spread for
room-temperature rolling of mild-steel plates, (b) Reduction in cross-sectional area vs.
reduction in height for room-temperature rolling of mild-steel plates.

Flat-Tool Forging [11]

In flat-tool forging the initial efforts to predict the lateral spread were
based on experiment [31-33]. Theoretical predictions of spread can be
made by the use of simplified three-dimensional elements. During conven-
tional bar forging, the bar is moved longitudinally with respect to the die
several times between forging strokes in one pass (Fig. 14.14). Some
results of the analysis of a three-bite flat-tool forging operation in which
the thickness of a bar of initial length L0 is reduced from 2H0 to 2H^ by
three strokes, are shown in Figs. 14.14 and 14.15. An example of the strain
distributions after each bite is shown in Fig. 14.14 for B/2W0 = 0.5. It can
be seen that interaction of the deformation zones occurs near to the die
edges where strain concentrations are observed.
Experimental results of spread and elongation are compared with the
computed results in Fig. 14.15. Comparisons of the changes in width and
length show that the agreement between theory and experiment is good in
terms of trends but not in terms of magnitudes. Several reasons can be
given for this discrepancy. In experiments the specimen dimensions were
not exact and the reductions in height were not exactly controlled.
Furthermore, the measurements of the final specimen dimensions included
some errors. In the theoretical calculations, besides the use of simplified
elements, some uncertainties are obviously associated with the friction
conditions. It was shown, however, that the use of the simplified
three-dimensional elements in the finite-element analysis provides good
predictions of spread in rolling and flat-tool forging.
 Stroke 3
FIG. 14.14 Flat-bar forging and strain distributions in three-bite flat-bar forging.

294
 Three-Dimensional Problems 295

Stroke

FIG. 14.15 Comparison between theory and experiment for spread (AWmax/W0) and
elongation (AL/L0) in three-bite flat-bar forging (Bar and tool geometry are shown in Fig.
14.14). [11]

14.7 Concluding Remarks

It is concluded that for geometrically simple block compression, the
finite-element method provides reasonably good solutions for the detail of
three-dimensional plastic deformation with relatively few elements. How-
ever, it is evident that the economic constraint becomes more severe for
three-dimensional metal-flow analysis in comparison with two-dimensional
problems. Therefore, in the development of analysis techniques for the
three-dimensional metal flow, special consideration must be given to
achieving a balance between computational efficiency and solution ac-
curacy. In this regard, the use of simplified three-dimensional elements can
be helpful. Although the problems for which simplified elements can be
used are limited, the approach can be effectively applied to a class of
296 Metal Forming and the Finite-Element Method
problems in which prediction of spread is important. Nevertheless, the
development of the three-dimensional analysis technique will continue for
more complex problems. This is indeed indicated by recent publications.
These developments must deal with not only geometrical complexities, but
also with three-dimensional heat transfer and its coupling with flow
analysis. Also, complex numerical techniques, such as rezoning, must be
developed. In addition, the analysis program must be interfaced with the
capabilities of a computer-graphics and data-display system in order to
evaluate the results quickly and effectively. Eventually, three-dimensional
programs will become useful for process design; for example, for design of
roll-pass schedules, design of operational sequences in flat-tool forging,
and design of preforms in hot and cold forging.

References
1. Nagpal, V., (1977), "On the Solution of Three-Dimensional Metal Forming
Processes," Trans. ASME, J. Engr. Ind., Vol. 99, p. 624.
2. Nagpal, V., and Allan, T., (1975), "Analysis of the Three-Dimensional Metal
Flow in Extrusion of Shapes with the Use of Dual Stream Functions," Proc. 3d
NAMRC., Pittsburgh, PA, p. 26.
3. Yang, D. Y., and Lee, C. H., (1978), "Analysis of Three-Dimensional
Extrusion of Sections Through Curved Dies by Conformal Transformation,"
Int. J. Mech. ScL, Vol. 20, p. 541.
4. Lahoti, G. D., and Kobayashi, S., (1974), "Flat-tool Forging," Proc. 2d
NAMRC, University of Wisconsin, Madison, p. 73.
5. Oh, S. I., and Kobayashi, S., (1975), "An Approximate Method for
Three-Dimensional Analysis of Rolling," Int. J. Mech. ScL, Vol. 17, p. 293.
6. Sagar, R., and Juneja, B. L., (1979), "An Upper Bound Solution for Flat Tool
Forging Taking into Account the Bulging of Sides," Int. J. Machine Tool. Des.
Res., Vol. 19, p. 253.
7. Braun-Angott, P., and Berger, B., (1982), An Upper Bound Approximation
for Spread and Pressure in Flat Tool Forging, in "Numerical Method in
Industrial Forming Processes," (edited by J. F. T. Pittman, R. D. Wood, J. M.
Alexander and O. C. Zienkiewicz) Pineridge Press, Swansea UK, p. 165.
8. Zienkiewicz, O. C., (1977), "The Finite Element Method," 3d Edn. McGraw-
Hill, New York.
9. Park, J. J., and Kobayashi, S., (1984), "Three-Dimensional Finite-Element
Analysis of Block Compression," Int. J. Mech. Sci., Vol. 26, p. 165.
10. Li, G.-J., and Kobayashi, S., (1982), Spread Analysis in Rolling by the
Rigid-Plastic Finite Element Method, in "Numerical Method in Industrial
Forming Processes," (edited by J. F. T. Pittman, R. D. Wood, J. M.
Alexander and O. C. Zienkiewicz) Pineridge Press, Swansea UK, p. 777.
11. Sun, Jie-Xian, Li, G.-J., and Kobayashi, S., (1983), "Analysis of Spread in
Flat-Tool Forging by the Finite Element Method," Proc. llth North American
Manufacturing Research Conf., May 1983, Madison, Wisconsin, p. 224.
12. Park, J. J., (1982), "Applications of the Finite Element Method to Metal
Forming Problems," Ph.D. Dissertation, Department of Mechanical Engi-
neering, University of California, Berkeley.
13. Kanacri. F., Lee, C. H., Beck, L. R., and Kobayashi, S., (1972), "Plastic
Compression of Rectangular Blocks Between Two Parallel Platens," Proc.
13th. Int. Mach. Tool Des. Res. Conference, Birmingham, p. 481.
 Three-Dimensional Problems 297

14. Sun, Jie-Xian, and Kobayashi, S., (1984), "Analysis of Block Compression
with Simplified Three-Dimensional Element," Proc. 1st Int. Conf. on Tech-
nology of Plasticity, Tokyo, p. 3.
15. Park, J. J., and Oh, S. I., (1987), "Application of Three-Dimensional Finite
Element Analysis to Metal Forming Processes," Proc. NAMRC XV, Beth-
lehem, PA, p. 296.
16. Aku, S. T., Slater, A. C., and Johnson, W., (1967), "The Use of Plasticine to
Simulate the Dynamic Compression of Prismatic Blocks of Hot Metal," Int. J.
Mech. ScL, Vol. 9, p. 495.
17. Chitkara, N. R., and Johnson, W., (1966), "Some Experimental Results
Concerning Spread in the Rolling of Lead," Trans. ASME, J. Basic Engr.,
Vol. 88, p. 489.
18. Helmi, A., and Alexander, J. M., (1968), "Geometric Factors Affecting
Spread in Hot Flat Rolling of Steel," J. Iron Steel Inst., Vol. 206, p. 1110.
19. Gokyu, I., Kiharai, J., and Mae, Y., (1970), "Study on the Width Spread in
Flat Rolling," /. Japan Soc. Tech. Plast, Vol. 11, p. 11.
20. Lahoti, G. D., and Kobayashi, S., (1974), "On Hill's General Method of
Analysis for Metal-Working Processes," Int. J. Mech. ScL, Vol. 16, p. 521.
21. Kummerling, R., and Lippmann, H., (1975), "On Spread in Rolling," Mech.
Res. Commun., Vol. 2, p. 113.
22. Tamano, T., (1973), "Finite Element Analysis of Steady Flow in Metal
Processing," J. Japan Soc. Tech. Plast., Vol. 14, p. 766.
23. Zienkiewicz, O. C., Jain, P. C., and Onate, E. (1978), "Flow of Solids During
Forming and Extrusion: Some Aspects of Numerical Solutions," Int. J. Solids
Structures, Vol. 14, p. 15.
24. Zienkiewicz, O. C., Onate, E., and Heinrich, J. C., (1981), "A General
Formulation for Coupled Thermal Flow of Metals using Finite Elements," Int.
J. Num. Meth. Eng., Vol. 17, p. 1497.
25. Dawson, P. R., (1978), "Viscoplastic Finite Element Analysis of Steady-state
Forming Processes Including Strain History and Stress Flux Dependence,"
Applications of Numerical Methods to Forming Processes, ASME, AMD,
Vol. 28, p. 55.
26. Shima, S., Mori, K., Oda, T., and Osakada, K., (1980), "Rigid-Plastic Finite
Element Analysis of Strip Rolling," Proc. 4th Int. Conf. on Prod. Eng.,
Tokyo, Japan, p. 82.
27. Li, G.-J., and Kobayashi, S., (1982), "Rigid-Plastic Finite Element Analysis of
Plane Strain Rolling," ASME Trans., J. Engr. Ind., Vol. 104, p. 55.
28. Kanazawa, K., and Marcal, P. V., (1982), "Finite Element Analysis of Plane
Strain Rolling," ASME Trans., J. Engr. Ind., Vol. 104, p. 55.
29. Mori, K., and Osakada, K., (1982), "Simulation of Three-Dimensional Rolling
by the Rigid-Plastic Finite Element Method," Proc. Num. Methods Ind.
Forming Processes, Swansea, p. 747.
30. Lahoti, G. D., Akgerman, N., Oh, S. I., and Altan, T., "Computer-aided
Analysis of Metal Flow and Stresses in Plate Rolling," /. Mech. Work. Tech.,
Vol. 4, p. 105.
31. Tomlinson, A., and Stringer, J. D., (1959), "SPread and Elongation in
Flat-tool Forging," /. Iron Steel Inst., Vol. 193, p. 157.
32. Baraya, G. L., and Johnson, W., (1964), "Flat-bar Forging," Proc. 5th Int.
Conf. MTDR, p. 449.
33. Kudo, H., and Nagahama, T., (1969), "Experimental Results for Upsetting
Pressure and Material Spread—Study in Transverse Upsetting Process of
Circular Rods, 1st Report," /. Japan Soc. Tech. Plast., Vol. 10, No. 106, p.
827.
 15
PREFORM DESIGN IN METAL FORMING

15.1 Introduction
Preform design in metal forming refers to the design of an initial shape of
the workpiece that, when it has undergone an associated forming process,
forms the required product shape with desired property successfully
without formation of defects and without excessive waste of materials. A
carefully selected preform can contribute significantly to the reduction of
the production costs. Preform design problems are encountered in various
metal-forming processes, such as closed-die forging, shell nosing, rolling,
and sheet-metal forming. Design of an optimal preform shape requires
simultaneous determination of optimal process conditions. However, we
are here concerned with the determination of the best preform shape
under a given set of process conditions.
In this chapter, a new method of "backward tracing" is introduced as an
alternative approach to the solution of preform design [1], and the
applications of this method to some specific processes are discussed.

15.2 Method for Design

Similarly to the forward simulation technique, the backward tracing
method uses the finite-element method. The forward simulation technique
has been discussed in the previous chapters.
Backward tracing refers to the prediction of the part configuration at any
stage in a deformation process, when the final part geometry and process
conditions are given. The concept is illustrated in Fig. 15.1. At time t = ta,
the geometrical configuration x0 of a deforming body is represented by a
point Q. The point Q is arrived at from the point P, whose configuration is
given as XQ.J at t = f 0 -i, through the displacement field during a time-step
At, namely, XQ = XQ_I + u 0 _! Af, where u 0 _i is the velocity field at t = f 0 -i.
Therefore, the problem is to determine UG-I, based on the information (XQ)
at point Q. The solution scheme is as follows: taking a loading solution u()
(forward) at Q, a first estimate of P can be made according to
Pm = XQ — u0 Af. Then, the loading solution u^, can be calculated from
the configuration of P(1), with which the configuration x() at Q can be
compared with P(1) + u^ t Af = Q(l\ If Q and Q(1) are not sufficiently close
to each other, then P(2) can be estimated by P(2) = XQ- u ( V_i Af. The
298
 Preform Design in Metal Forming 299

FIG. 15.1 Concept of the backward tracing method (geometrical configuration only).

solution for loading at P(2) is then u,^, and the second estimate of the
configuration Q(2) = F (2) + u,(,221 Af can be made. The iteration is carried
out until Q(n) = P(n> + v$2i Af becomes sufficiently close to Q.
The formulation of the concept described above will further clarify the
backward tracing technique. After finite-element discretization, the solu-
tion of the boundary value problem associated with isothermal viscoplastic
deformation satisfies a system of nonlinear, coupled algebraic equations,
expressed in the form

with the boundary conditions in terms of velocity and traction, where K is

the stiffness matrix; f is the force vector; x is the nodal point coordinates; v
is the nodal point velocities; e is the integration point effective strains; e is
the integration point strain rates; and T is the given temperature
(constant).
In eq. (15.1), x and E at time t are supposed to be known, and e is a
function of v. In forward simulation (t & t0), the solution v() of eq. (15.1) at
t = tn is used for the solution at the next step t = t(>+ { = t(} + Af, according to
the explicit scheme as
and
300 Metal Forming and the Finite-Element Method
LOADING P R O G R A M BACKWARD TRACING PROGRAM

FIG. 15.2 Flow chart for the backward tracing procedure implemented into the finite-
element flow analysis.

In backward tracing (t<t0), at time t = t0_1 = t0 — At, XQ_I, e0-i should

satisfy the following conditions, in addition to eq. (15.1),

For a given time-increment, the substitution of eq. (15.3) into eq. (15.1) at
time t0-i results in
 Preform Design in Metal Forming 301

where XQ, §o are known and E O _I is a function of v0_!. It is to be noted that

CQ is needed only for work-hardening materials. Equations (15.4) are
solvable for v 0 _i if the boundary conditions can be specified at time t0_i.
Repeated calculations of the solutions for times £0_2, t0_3, . . . will lead to
backward tracing of the entire deformation process. A flow chart for the
backward tracing procedure implemented into the finite-element deforma-
tion analysis is shown in Fig. 15.2. It is to be noted that the application of
forward simulation (loading) and backward tracing becomes straightfor-
ward if the change of the boundary conditions during a process is known.
In many cases the change of the boundary conditions during loading
depends on the shape of the preform itself. Then it becomes an important
issue in the design methodology to develop appropriate design criteria for
controlling the change of the boundary conditions during deformation.
This aspect of design, together with some others, such as consideration of
work-hardening and temperature, are discussed in the following specific
applications of the method.

15.3 Shell Nosing at Room Temperature

Shell nosing is a process of forming an ogive nose at an open end of
cylindrical shell by forcing the shell into a contoured die, as described in
Chap. 2. In all modern methods of shell manufacturing, the cavity is
formed to a finish shape and machining is restricted to the outer surface of
the nosed shell. Thus, in this process it is essential to determine a preform
shape that forms precisely the required final configuration with a specified
wall thickness distribution after nosing. The metal flow in nosing is
complex. Some large shells are manufactured at elevated temperatures
owing to the severe deformation required at the open end of the shell.
Until recently, there was very little quantitative information available
and, in most shell manufacturing plants, extensive experience and expen-
sive trial-and-error techniques were necessary in order to design the nosing
process. Pioneering work in this field is attributed to Nadai [2], who
outlined an approach for the strain analysis in the shell nose. Using
Nadai's approach, Carlson [3] suggested a method for determining the
original shell profile, knowing the final shape. Based on Carlson's method,
Lahoti et al. [4] developed a computer program for the preform design
procedure. Kobayashi [5] derived approximate solutions for determining
the shell profile from the tangential velocity distributions along the die
contour. The approaches mentioned above, however, give only approxim-
ate answers.
Park et al. [1] applied the backward tracing method to the determination
of preforms that result in a nosed shell with uniform wall thickness. For die
geometry and nosed shell configurations shown in Fig. 15.3, two preform
configurations were determined. The wall thickness distributions required
in the preform can be accommodated by varying either the inner or the
outer diameters of the preform. This criterion was imposed for controlling
the boundary condition during the backward tracing process.
302 Metal Forming and the Finite-Element Method

FIG. 15.3 Die dimensions and nosed shell configurations. L0 = 47.26, Lf = 48.84, R =
21.526, Rt = 14.648, Bf = 14.47, Hf = 6.88 (uniform), a0 = 237, b0 = 18.75 (units of mm).

For the preform shape with constant outer diameter (Type-0 preform),
RO = R0 for all n, where RQ is the radius of a nodal point located along the
outer surface of the shell and R0 is the radius at the die entrance. The
boundary condition for this type of preform is controlled in such a way that
nodes that are in contact with the die (7?o < RO) become force free when
the condition that RQ = R0 is reached during backward tracing.
For the preform shape with constant inner diameter (Type-I preform),
the criterion for controlling the boundary condition is that as soon as
R? = Rf for the nodes Rpt < Rh the condition of the corresponding outside
node is changed from that of die-contact to the force free condition, where
Rpt is the radius of a nodal point along the inner surface of the shell.
Backward tracing was carried out computationally by taking 2 mm
penetration (in loading) as one step. Thus, a total of 22 steps was
necessary to complete the calculation. The solution convergence needed
about six to seven iterations. For completing one step of backward tracing,
it was necessary to have one to two iterations for the case where the
boundary condition does not change, and two to three iterations were
required for the case where a change in the boundary condition occurs.
Convergence for backward tracing was determined by the closeness of the
solution (between Q and <2(M) described in Section 15.2) within a limit.
The two types of preforms determined by this method are compared with
the final nose configuration in Fig. 15.4. In the figure, the lengths £° and
£o, for Tpye-0 and Type-I preforms, respectively, indicate the portion of
the preform that becomes the nosed shell length Lf after nosing. Because
 Preform Design in Metal Forming 303

FIG. 15.4 Nosed shell configuration and two types of preform shapes for nonwork-hardening
materials [1].

the shell elongates during nosing, the preforms are shorter than the nosed
shell. The shell material was assumed to be rigid-perfectly plastic
(nonwork-hardening).
In order to include the work-hardening effect in preform design, it is
necessary to know the strain distributions in the nosed shell, in addition to
the geometrical configurations. However, the strain distributions in the
nosed shell depend on the preform shapes and are not known (or cannot
be specified). In Fig. 15.5 the procedure used to take into account the
work-hardening effect is illustrated.
By applying the backward tracing scheme to nonwork-hardening mate-
rials, the two types of preforms can be determined. With these preforms
the solutions for loading are obtained for a work-hardening material. The
results provide strain (effective strain) distributions and the nosed shell
configurations that are very close to but not identical to the specified
geometries. These configurations are then corrected exactly according to
the specifications, maintaining the same effective strain distributions. With
FIG. 15.5 Schematic illustration of the procedure to include the work-hardening effect in prefr»rn design.
 Preform Design in Metal Forming 305

these strain distributions, the application of the backward tracing scheme

results in the preform shapes for a work-hardening material. At each step,
the strain and corresponding stress are also traced backward following the
stress-strain curve of the material. It should be mentioned that the first
computational procedure given in Fig. 15.5 may be eliminated by using
approximate solutions for preform shapes of nonwork-hardening materials.

15.4 Plane-Strain Rolling [7]

In rolling a flat-ended ingot, shapes with defective ends may be formed.
These shapes have an overhang at the front and rear ends, and form
overlaps after several rolling passes. In addition, the so-called "fishtail" is
formed at the rear end if edge-rolling is applied. These defective end
regions, shown in Fig. 15.6, are cut off and lost as scrap. Johnson [6] has
remarked that the end shape changes found in rolled slabs will be difficult
to predict, since they occur in the context of nonsteady-state conditions
and are of a complicated three-dimensional character. However, theoreti-
cal predictions of the defective shapes and end shape designs to eliminate
them are possible if the plane-strain condition is assumed. Hwang and
Kobayashi [7] tested an approximate design of preformed end shapes for
nonwork-hardening materials [8] by simulating plate rolling by the
finite-element method.
Figure 15.7 shows the contours of end shapes used in rolling nonwork-
hardening materials. Figures 15.7a and b show the changes in the front and
the back end shapes, respectively. After rolling, the end shapes are not
perfectly flat, but they are fairly flat. This demonstrates that the design
based on approximate solutions is very good. When the required end
flatness is of the degree shown in the figure, the approximate approach will
give various preform end shapes under various process variables, such as
the roll diameter-to-plate-thickness ratio and reduction in thickness.
The backward tracing scheme was used to determine the preform end
shapes that are perfectly flat after rolling. For the front end, the shape
obtained with an approximate solution was modified to a flat end and the
loading pass that would result in this desired end shape was traced
backward. The front end shapes obtained at several locations from the exit

FIG. 15.6 Defective end shapes in flat and ingot rolling.

FIG. 15.7 (a) Front and (b) back end contours at several stages of rolling of a plate with a
preform design based on approximate solution (reduction in thickness = 0.2; roll diameter =
4.0; thickness of rolled plate = 1.0; material, nonwork-hardening; friction, sticking).

306
FIG. 15.8 Backward tracing applied to a preform design of (a) the front end and (b) the back
end in rolling (conditions are the same as those of Fig. 15.7).

307
308 Metal Forming and the Finite-Element Method
are shown in Fig. 15.8a. The preform front end obtained by this method
has a circular convex contour. The figure also shows that during rolling the
curvature of the front end contour is reduced and becomes slightly concave
before the front end becomes flat at the exit. Similarly, in order to apply
the backward tracing scheme to the determination of back end preform
shapes, the back end contour obtained by the approximate solution was
modified to a flat end, and the rolling process was then traced backward.
Variations of the back end contour at several stages are shown in Fig.
15.8b. The resulting preform shape is a circular arc, similar to the front
end preform, but with a smaller curvature.
For work-hardening materials, the procedure illustrated in Fig. 15.5 was
used. With preform shapes of a plate obtained for nonwork-hardening
materials (Fig. 15.8), the rolling simulation is performed for a work-
hardening material. The results provide the strain distributions and the end
contours. The end shapes are not exactly flat. The ends are then corrected
geometrically to become flat ends, maintaining the same strain distribu-
tions. With these strain distributions, the application of the backward
tracing scheme results in preform shapes for a work-hardening material.
The results for an aluminum alloy are shown in Fig. 15.9. Comparing the
preform shapes for a work-hardening material with those for nonwork-
hardening materials, it is seen in Fig. 15.9 that at both front and back ends

FIG. 15.9 Comparison of preform end shapes for nonwork-hardening and work-hardening
materials in rolling.
 Preform Design in Metal Forming 309

the deviations of the contours from the flat ends are smaller for a
work-hardening material. The end shapes obtained by the backward
tracing scheme for a work-hardening material are more complex than
those for nonwork-hardening materials. This indicates that during
nonsteady-state rolling, the deformation is influenced by the material
property.

15.5 Axially Symmetric Forging [14]

In forging, preform design involves the determination of the number of
preforms and the design of the shapes and dimensions of each preform.
According to Biswas and Knight [9], the bulk of the work conducted to
determine the preform shapes for complex forging components has been
carried out in Germany and the Soviet Union, as reported by Spies
[10]. Chamouard [11] in France and Akgerman et al. [12] in the U.S. dealt
with some practical problems of preform design for certain forging
cross-sections. Yu and Dean [13] recently reviewed existing guidelines for
the design of dies and discussed previous computer-aided approaches.
Examples of preform design for steel forgings of various H-shapes are
shown in Fig. 15.10. The present practice in handling this complex
problem is to computerize the design calculations required for the
designer's decisions on basis of qualitative guidelines that have been
derived mainly from experience or experimental studies. A preform is
required to provide complete filling of the die cavity during finish-forging.
In addition the preform must be such that it assists in obtaining desired
product properties. An example of such a design problem is encountered
in disk forging [14].

FIG. 15.10 Preforms for steel finish forgings of H cross sections.

310 Metal Forming and the Finite-Element Method
Problem Statement
A simple compression of a cylindrical bar stock between two flat parallel
dies produces a flat disk. If friction at the die-workpiece interface were
negligible, the forged disk would have uniform strain throughout the
deforming body. The problem, then, is how to produce a uniformly
deformed disk under the presence of friction at the die-workpiece
interface. In the present case we assume that friction at the die-workpiece
interface is equal to the shear strength of the deforming material (frictional
stress = mk, where k is the shear strength, and m = 1.0), and that the
material is nonwork-hardening. The process conditions are: (i) the stock is
of cylindrical shape and has a radius of 1.25 and a height of 1.0 and (ii) the
stock is to be deformed to 50% reduction in height. Under these
conditions, it is necessary to determine the number of forming operations
and the preform shapes for producing a uniformly deformed disk. It should
be noted, however, that we have some flexibility in selecting the process
conditions and that the design requirements are not very rigid. For
example, the amount of final strain in the forged disk, and the stock
geometry, are not definitely prescribed. Note also that the requirement of
uniform straining of the disk throughout the body cannot be satisfied in a
strict sense. All that we can hope for is to achieve strain uniformity as
much as possible.

Preliminaries
For developing design procedures, it would be helpful if we could obtain
some knowledge of the metal flow involved in forging. Thus, two
simulations by the finite-element method were performed as part of a
preliminary investigation.
One was the simulation of the simple compression of a cylinder between
two parallel flat dies (with m = 1.0). The grid patterns before and after
compression are shown in Fig. 15.11. Also shown in the figure are the
effective strain distributions after 50% reduction.
From the results of Fig. 15.11, it is seen that in the core region, a nearly
dead-metal zone is formed beneath the die, resulting in a large strain
gradient in the thickness direction. In the peripheral region, strains are less
than the average strain value within the main body, but they are greater
than average near the contact surface, because of distortion due to friction.
The strain gradient in the core region may be reduced if the size of the
dead zone can be reduced. One may achieve this by applying indentation
at the core region, followed by flat die compression. A symmetric
indentation was simulated by a hemispherical punch with a radius of 1.5 to
the depth of z = 0.35 (the amount of punch displacement = 0.15). Simula-
tion of flat-die compression was then performed until the final disk
configuration was reached. The grid distortions in the workpiece after
indentation, and at the final stage, are shown in Fig. 15.12 along with
resulting strain distributions. It is seen that the strain distribution in the
 Preform Design in Metal Forming 311

FIG. 15.11 The grid patterns before and after deformation and the effective strain
distributions after 50% reduction in height in disk forging (black zone: e > l . l ; white zone:
1.1 >e>0.8; gray zone: e<0.8).

core region was altered and some improvement toward uniform strains, in
general, was obtained by indentation and compression.
Design Procedures
In the present problem, and in preform design in forging in general, the
boundary conditions and their variations during the process depend on the
preform shape itself. During the actual forming operation, a change in the
boundary conditions occurs when a nodal point on the free surface comes
into contact with the die. In backward tracing, nodal points along the
die-workpiece interface in the final configuration should be detached from
the die at some stage of deformation and all the nodal points must become
free at the completion of backward tracing.
In developing design procedures for these cases, it is necessary to
consider criteria for controlling when and where and in what sequence a
boundary node should be detached from the die. Two criteria are
conceivable at this point—one for controlling the workpiece geometry and
the other for controlling deformation.
For the first backward tracing step, the boundary conditions were
312 Metal Forming and the Finite-Element Method

FIG. 15.12 The grid patterns after identation and in the finish-forged disk, and the effective
strain distributions in the finished disk (black zone: £>!.!, white zone: 1.1>E>0.8; gray
zone: e<0.8).

primarily controlled by geometrical considerations of the preform shape.

The preliminary investigations show that the preform shape with a dimple
at the center improves the strain distributions in the core region.
Therefore, the boundary condition was regulated in such a way that the
preform shape was smooth and convex, similar to the shape obtained by
indentation. This was done by allowing a node to be detached from the die
at each time-step, sequentially from both the center and the periphery of
the interface.
During the second tracing step, the residual strain distributions were
examined after each time-increment and used for controlling the boundary
conditions. When the residual strains in the thickness direction reach the
value within a given limit (say, ±0.1) in the main body (excluding a narrow
region near the interface), a corresponding boundary node was detached
from the die. In the peripheral region, however, the residual strains never
reach the desirable values (values close to zero), and separation of the
boundary nodes was required mostly for the physical reason that the nodal
force must be compressive in order for a nodal point to remain in contact
with the die.
 Preform Design in Metal Forming 313

Final Design
Following the design procedures discussed above, the shapes of the
preform and the stock and the resulting strain distributions in the finished
disk were obtained, and are shown in Fig. 15.13a. In Fig. 15.13a it is seen
that improvement has been achieved for the strain distributions in the core
region, but not in the peripheral region. Further improvement in design
appears to be possible only by increasing the amount of distortion in the
peripheral region with an additional preforming operation. A radial
forging operation with circular-contoured dies is considered as a process
for this purpose. In forward simulation as well as in backward tracing, the
radial forging process is replaced by a fictitious process in which the
diameter of a ring-die decreases or increases radially. During backward
tracing of this process, the boundary conditions were regulated in such a
way that the side-surface of the stock became as close to a circular cylinder
as possible. The resulting shape of the stock and the corresponding strain
distributions in the finished disk are shown in Fig. 15.13b. It can be seen

FIG. 15.13 Shapes of preforms and stocks and the effective strain distributions in the finished
disk: (a) design I (black zone: e > l . l ; white zone: 1.1>£S0.8; gray zone: esO.8); (b)
design II (black zone: e a 1.3, white zone: 1.3 > £ > 1.0; gray zone: e < 1.0).
314 Metal Forming and the Finite-Element Method
that improvement in uniformity of strain distributions in the peripheral
region was achieved.
In the design procedures discussed above, no particular attention was
paid to irregular shapes of designed stocks and preforms. The reason is
that the shapes and contours can always be smoothed out or modified to
obtain realistic and simple configurations. The final design can then be
evaluated by simulating the operation sequence with modified stock and
preform shapes. The modified stock shape and the radial forging processes
are shown in Fig. 15.14a. The initial stock is a cylindrical solid with a
hemispherical dimple at the center. Each die for radial forging has a
circular contour, and moves radially inward.
The second preforming operation is forging with a die that has the

FIG. 15.14 (a) Simulation of first preforming operation—radial forging, (b) Simulation of
second preforming operation—forging, (c) Simulation of finishing operation—flat-tool
compression.
 Preform Design in Metal Forming 315

FIG. 15.14 (continued)

smoothed contour of the preform 2 shown in Fig. 15.13b. The grid

distortions before and after the operation are shown in Fig. 5.14b. The
simulation of the finishing operation and the results are shown in Fig.
15.14c. The strain distributions in the finished disk obtained for the final
design show that geometrical modification of stock and preforms did not
change the results appreciably.

15.6 Hot Forming [15]

For preform design problems in thermo-viscoplastic deformation, back-
ward tracing of the process should also include temperature variations. In
addition to the velocity equations given by eq. (15.1), the following
temperature equations, along with the boundary conditions, must be
satisfied:
at time t
316 Metal Forming and the Finite-Element Method
Kc is the heat conduction matrix; C is the heat capacity matrix; Q is the
heat flux vector; T is the vector of nodal point temperatures; T is the
vector of nodal-point temperature rates.
For backward tracing (t < t0), at time t = t0_i = t0 — Af, x0_!, E0-i should
satisfy the conditions given by eq. (15.3) and T0^ should satisfy the
conditions that

(see eq. (12.12) in Chap. 12). Substitution of eqs. (15.3) into eq. (15.5) at
time t0-i, gives

In eq. (15.7), x0, £ 0 > VQ-I, EO-I are supposed to be known or obtained from
the solution of the velocity equations. The temperature T
is a function of T 0 _i according to eq. (15.6). Since T0 and T0 are known,
eq. (15.7) can be solved for T0-i by assuming the boundary conditions at
time t0_i. Once T0_! is obtained, then T0_! can be calculated and
compared with the assumed boundary values. Iterative procedures will
result in backward tracing of temperature calculations. Details of the
calculation procedure involved in backward tracing of a thermo-
viscoplastic deformation process is given in Reference [15].
To illustrate the design method, a preform design in shell nosing at
elevated temperatures is considered. The specifications described in the
study by Lahoti et al. [16] and by Tang and Kobayashi [17] were used.
They were: dimensions of the nosed shell and die; interface friction
(frictionless); material (AISI 1045 steel); initial temperature distributions;
and nosing speed (270 mm/s). The preform design procedure consists of
the following three steps.
Step (1)—Loading simulation. Backward tracing begins with the strain and
the temperature fields in the specified configuration and in the die. In
order to know these fields, loading is simulated with a trial or "guessed"
preform, using a given initial temperature distribution. A trial preform
may be designed by considering that the material volume of the preform
must be the same as that of the shell to be formed, and by assuming that
the inner profile of the preform is a straight line. The geometry of the trial
preform is then determined by the requirement of straight outer profile.
The trial preform shape and the initial temperature distributions in the
preform are shown in Fig. 15.15.
Step (2)—Preparation for backward tracing. The final deformed shape,
obtained from the simulation in Step (1) is not the same as that specified.
The nodal point positions are adjusted so that the specified shell
configuration results and strain and temperature distributions in this
configuration are obtained. Figure 15.16 shows these distributions.
Step (3)—Backward tracing and control of boundary conditions. Starting
from the specified shell configuration and corresponding strain and
temperature fields, backward tracing is performed. In order to obtain a
 Preform Design in Metal Forming 317

FIG. 15.15 (a) The trial preform shape and finite-element layout for loading simulation of
nosing, (b) The initial temperature distributions in the preform.

preform with straight outer wall surface, we control the boundary

conditions in such a way that each node that is in contact with the die
leaves the die at the die entrance and then moves axially. The final
preform shape is shown in Fig. 15.17a. Figure 15.17b gives the initial
temperature distribution, obtained from backward tracing of temperature
changes. The temperature distribution of Fig. 15.17b is very close to the
318 Metal Forming and the Finite-Element Method

FIG. 15.16 (a) The strain distributions and (b) the temperature distributions in the specified
shell configuration and in the die.

given initial temperature distribution shown in Fig. 15.15b, indicating that

there is no need for iteration.

15.7 Concluding Remarks

A new technique, the backward tracing scheme, was devised for preform
design. This is a rather unique application of the finite-element method to
solving problems in metal forming. It was demonstrated that the technique
 Preform Design in Metal Forming 319

FIG. 15.17 (a) The preform shape and (b) the initial temperature distributions obtained at
the completion of backward tracing for shell nosing.

can be applied to preform design in shell nosing and rolling at room

temperature. It was found that one of the aspects critical for further
development of the technique is the treatment of the boundary conditions.
In the disk-forging problem, where the boundary conditions change during
the process depending on the preform shape, criteria were suggested for
controlling the boundary conditions during the process. Further develop-
ment of the method was presented for backward tracing of a thermo-
viscoplastic deformation process, and the method was applied to preform
320 Metal Forming and the Finite-Element Method
design in shell nosing at elevated temperature. In conclusion, it is
important to note that in order to obtain the desired (optimal) preform of
a forming process, the path for backward tracing must be known. Thus,
future research should concentrate on developing criteria for the selection
of optimal preforms and corresponding deformation path in general
forming processes.

References
1. Park, J. J., Rebelo, N., and Kobayashi, S., (1983), "A New Approach to
Preform Design in Metal Forming with the Finite Element Method," Int. J.
Machine Tool Des. Res., Vol. 23, p. 71.
2. Nadai, A., (1944), "Plastic State of Stress in Curved Shells: The Forces
Required for Forging of the Nose of High-Explosive Shells," Forging of Steel
Shells, ASME Trans., p. 31.
3. Carlson, R. K., (1944), "An Experimental Investigation of the Nosing of
Shells," Forging of Steel Shells, ASME Trans., p. 45.
4. Lahoti, G. D., Subramanian, T. L., and Allan, T., (1978), "Development of a
Computerized Mathematical Model for the Hot/Cold Nosing of Shells,"
Report ARSCD-CR-78019 to U.S. Army Research and Development
Command.
5. Kobayashi, S., (1983), "Approximate Solutions for Preform Design in Shell
Nosing," Int. J. Machine Tool Des. Res., Vol. 23, p. 111.
6. Johnson, W., (1980), The Mechanics of Some Industrial Pressing, Rolling, and
Forging Processes, in "Mechanics of Solids," (Edited by H. G. Hopkins and
M. J. Sewell), Pergamon Press, Oxford, p. 303.
7. Hwang, S. M., and Kobayashi, S., (1984), "Preform Design in Plane-Strain
Rolling by the Finite Element Method." Int. J. Machine Tool Des. Res., Vol.
24, p. 253.
8. Kobayashi, S., (1984), "Approximate Solutions for Preform Design in Roll-
ing," Int. J. Machine Tool Des. Res., Vol. 24, p. 215.
9. Biswas, S. K., and Knight, W. A., (1974), "Computer-Aided Design of
Axisymmetric Hot Forging Dies," Proc. 15th Int. MTDR Conf., p. 135.
10. Spies, K., (1957), "The Preforms in Closed-Die Forging and Their Preparation
by Reducer Rolling" (in German), Doctoral Dissertation Technical University,
Hannover.
11. Chamouard, A., (1964), "Estampage et Forge." Dunod, Paris.
12. Akgerman, N., Becker, J. R., and Allan, T., (1973), "Preform Design in
Closed Die Forging," Metal Forming, Vol. 40, p. 135.
13. Yu, G. B., and Dean, T. A., (1985), "A Practical Computer-Aided Approach
to Mould Design for Axisymmetric Forging Die Cavities," Int. J. Machine
Tool Des. Res., Vol. 25, p. 1.
14. Hwang, S. M., and Kobayashi, S., (1986), "Preform Design in Disk Forging,"
Int. J. Machine Tool Des. Res., Vol. 26, p. 231.
15. Hwang, S. M., and Kobayashi, S., (1987), "Preform Design in Shell Nosing al
Elevaled Temperatures," Int. J. Machine Tool Manufacture, Vol. 27, p. 1.
16. Lahoti, G. D., Oh, S. I., and Allan, T., (1981), "Development and
Confirmation of a Series of Mathematical Models for the Blocking, Cabbaging,
Piercing, and Nosing Operations Involved in Shell Manufacluring," Report
ARSCD-CR-81010, to U.S. Army Research and Development Command.
17. Tang, M. C., and Kobayashi, S., (1982), "An Investigation of the Shell Nosing
Process by the Finite Element Method, Part 2. Nosing at Elevated Tempera-
tures (Hot Nosing)," ASME Trans., J. Engr. Ind., Vol. 104, p. 312.
 16
SOLID FORMULATION, COMPARISON
OF TWO FORMULATIONS, AND
CONCLUDING REMARKS

16.1 Introduction
In the previous chapters we have discussed only the applications of flow
formulation to the analysis of metal-forming processes. Lately, elastic-
plastic (solid) formulations have evolved to produce techniques suitable for
metal-forming analysis. This evolution is the result of developments
achieved in large-strain formulation, beginning from the infinitesimal
approach based on the Prandtl-Reuss equation.
A question always arises as to the selection of the approach—"flow"
approach or "solid" approach. A significant contribution to the solution of
this question was made through a project in 1978, coordinated by Kudo
[1], in which an attempt was made to examine the comparative merits of
various numerical methods. The results were compiled for upsetting of
circular solid cylinders under specific conditions, and revealed the impor-
tance of certain parameters used in computation, such as mesh systems and
the size of an increment in displacement. This project also showed that the
solid formulation needed improvement, particularly in terms of predicting
the phenomenon of folding. For elastic-plastic materials, the constitutive
equations relate strain-rate to stress—rates, instead of to stresses. Conse-
quently, it is convenient to write the field equation in the boundary-value
problem for elastic-plastic materials in terms of the equilibrium of stress
rates.
In this chapter, the basic equations for the finite-element discretization
involved in solid formulations are outlined both for the infinitesimal
approach and for large-strain theory. Further, the solutions obtained by
the solid formulation are compared with those obtained by the flow
formulation for the problems of plate bending and ring compression. A
discussion is also given concerning the selection of the approach for the
analysis.
In conclusion, significant recent developments in the role of the
finite-element method in metal-forming technology are summarized.

16.2 Small-Strain Solid Formulation

The field equation for the boundary-value problem associated with the
deformation of elastic-plastic materials is the equilibrium equation of stress
321
322 Metal Forming and the Finite-Element Method
rates. As stated in Chap. 1 (Section 1.3), the internal distribution of stress,
in addition to the current states of the body, is supposed to be known, and
the boundary conditions are prescribed in terms of velocity and traction-
rate.
The derivation of the basic equations for the finite-element discretization
follows a procedure similar to that for rigid-plastic materials, (see eqs.
(5.23-5.25)), using the stress-rate equilibrium equations.
The basic equation is

where the traction-rate Ft is prescribed on SP as <7,y«; = Fit where «,• is a unit

normal to the surface SF.
The independent variables in eq. (16.1) are velocity components, and
the Prandtl-Reuss equations provide the relationship between the stress-
rate and velocity components. The Prandtl-Reuss equations are given by
eq. (4.32). Substituting A given by eq. (4.38) and noting that a,y =
dij - difidkk and 2G = £7(1 + v), eq. (4.32) becomes

where dtj is the Kronecker delta. The first term of eq. (16.2) can be
rewritten, using the strain-hardening relationship in the form of eq. (4.37),
as

where H' is the slope of the stress-plastic strain curve H' — da/de. In
deriving eq. (16.3), note that
oa = \ o'mnd'mn = | o'mn(amn - &mn\akk)
where o'mn5mn = 0. Then, eq. (16.2) can be written in the form

For discretization of eq. (16.1), we need the inversion of eq. (16.4). The
inversion was obtained by Yamada et al. [2] as

For the last term in eq. (16.5), the coefficient a is added such that a = 0
for elastic elements and a = 1 for plastic elements. After an incremental
 Solid Formulation, Comparison of Two Formulations 323

deformation of time-step Af, the stress-state is updated by adding

ACT,-,- = btj A? to the current stress <7,y.
Early applications of the finite-element method have been based on this
infinitesimal theory. The method was useful for the analysis of problems in
which plastic strains are small and the effect of material rotation on the
formulation is negligible.

16.3 Large Deformation: Rate Form

Various forms of the finite-element method dealing with large strain have
been used by several investigators [3-6]. Among them, the most elegant
forms are those given by Needleman [5] and by McMeeking and Rice [6].
Needleman employed the Lagrangian formulation using convected coordi-
nate systems, while McMeeking and Rice adopted the updated-Lagrangian
formulation. Although their approaches to the problem are different, their
formulations are equivalent to and are based on that given by Hill [7, 8]. In
this section the updated-Lagrangian formulation is outlined.
Consider a body with volume V0 and surface S0 at time t = t0 as a
reference state. After a certain increment of time Af, the body occupies
the new position V and S. At the reference state, each particle of the body
is labeled by a set of coordinates (Xt) that is embedded in the material.
Another coordinate system (#,•) that is fixed in space and not moving with
the body will be chosen, as shown in Fig. 16.1. Then, at any time t > tn, we
have the following relations between the two coordinate systems.

The coordinate system xa is chosen to be rectangular Cartesian and the

Lagrangian description uses Xf and t as independent variables. Since
Xj =Xj at t = t0, the strain-rate £,-, as a measure of deformation is given by

REFERENCE STATE

FIG. 16.1 Reference and current states, and coordinate systems.

324 Metal Forming and the Finite-Element Method
and the rotation rate is

The three stress measures were introduced in Chap. 4 and discussed in

some detail in Chap. 11 in conjunction with the considerations of large
deformation in the flow formulation. They are the Cauchy stress a,-,-,
nominal stress (or first Piola—Kirchhoff stress) pijt and the second
Piola-Kirchhoff stress s^. They are related to each other according to

where / = det \dxil9Xa\. It can be noted that the second Piola-Kirchhoff

stress sap is symmetric but that paj is not.
As a stress measure we introduce another stress Tfi, the Kirchhoff stress,
which is defined by

where p0 and p are the material densities at the reference and the
deformed states, respectively. The time derivative of the stress measure is
the Jauman derivative (DTij/Dt), where time differentiation is carried out
with the coordinate system that rotates but does not deform with the
material. It is expressed by

Because r,-, = a,-,- and Xt = xf at t = t0, and noting that t,y =

Pij: + (dUj/dxk)okj, from eqs. (16.8) and (16.9), eq. (16.10) becomes

For the finite-element discretization, the variational form of the equi-

librium equation of stress-rate is expressed as

Then

Equation (16.12) is the virtual work-rate principle in terms of stress-rate.

Substituting pn of eq. (16.11) into eq. (16.12) and using the boundary
condition for the traction-rate F/ on 5n (and further imposing the restriction
 Solid Formulation, Comparison of Two Formulations 325
that dUj — 0 on the part of S0 where the velocity is prescribed),

Using the symmetry of Dr^lDf, e,7 and a^, eq. (16.13) can be rewritten in
the form

Assuming that the Prandtl-Reuss equation holds for the Jauman deriva-
tive of the Kirchhoff stress and strain rates, the constitutive relation is
given by

Equations (16.14) and (16.15) are the bases for the finite-element
discretization.
In the updated Lagrangian description, the reference state is assumed to
coincide with the current state at time t. The procedures used to integrate
the constitutive equations for a time-step Af are based on the mean normal
method of Rice and Tracey [9] and the radial return method of Krieg and
Krieg [10]. These two methods are further explained in Reference [28].
The formulations described above have been used for the analyses of
various forming processes. A complete stress and deformation analysis has
been given for plane-strain and axisymmetric extrusion until the steady
state is reached [11,12]. In applying the finite-element method to
can-extrusion the main problem was to simulate the metal flow around a
sharp corner [13]. The use of a special element to overcome this problem
was discussed.
Complete solutions of stretch-forming and deep drawing problems,
taking into account the contact problem at blank holder, die, die profile,
and punch head, were obtained by Wifi [14]. On the basis of the nonlinear
theory of membrane shells, Wang and Budiansky [15] developed the
procedure for calculating the deformations in the stamping of sheet metal
by arbitrarily shaped punches and dies.
Plane-strain rolling was analyzed by Rao and Kumar [16]. Key et al. [17]
obtained the solutions for rolling, extrusion, and sheet stretching. Hartley
et al. [18] developed a method for introducing friction into the finite-
element analysis [19].
Hartley et al. [20] also included the effects of strain-rate and tempera-
ture variations within the billet in the analysis of axial compression of a
solid cylinder. The fixed-mesh updated-Lagrangian technique was pro-
326 Metal Forming and the Finite-Element Method

posed for the study of steady-state metal-forming processes by Derbailan

et al. [21]. Analytical formulations of computational sheet-metal forming
were given by Wang [22]. Plane-strain plate bending was analyzed by Oh
and Kobayashi [23] and the results were compared with those based on the
flow approach. Some of the comparison of the two solutions are shown in
Chap. 8 and a more detailed comparison is discussed in Section 16.5 of this
chapter.

16.4 Large Deformation: Incremental Form

In Section 16.3, the updated-Lagrangian approach was described in terms
of rate equations. An alternative formulation is the derivation of the
governing equations directly in incremental form. This formulation was
developed by Nagtegaal and De Jong [24]. Implementation of the
formulation into numerical models and applications of the finite-element
code to metal-forming analyses are discussed in a series of publications by
Nagtegaal and his group [25-28]. The outline of the formulation is given in
the following.
At a generic time t, the virtual work-rate principle is expressed in terms
of the first Piola-Kirchhoff stress p,y in the reference state as

where wy is the displacementt and Fj = Pijnoi is the traction prescribed on

SQ. Using the relationship given in eq. (16.8), eq. (16.16a) is written in
terms of the second Piola-Kirchhoff stress s,y as

where a comma denotes a spatial derivative

In the updated-Lagrangian description, the reference state is the current
state at the beginning of the increment, and various stress measures are
momentarily equal to each other. Therefore, eq. (16.16b) becomes

At the end of the increment At, the spatial position of the body x, is
described by
(16.18)
with Aw, being the incremental displacement. Then eq. (16.16b) implies
that the virtual work-rate principle, applicable at the end of the increment,

t Note that Uj has been used for velocity throughout this book but is used for displacement
in Sections 11.4 through 11.7 of Chap. 11 and in this section.
 Solid Formulation, Comparison of Two Formulations 327
is written as

Subtracting eq. (16.17) from eq. (16.19), we obtain the incremental

virtual-work equation as

The incremental constitutive equations are expressed in the form

where &Ekl is the increment of Lagrangian strain and <£ikjl are the
constitutive moduli. The increment of the Lagrangian strain A£w is related
to the displacement increment according to

The moduli J£ijk, are not the small-strain moduli Lijkl given by eq. (16.5),
and are given by

The derivation of eq. (16.23) is given in Reference [24].

The presence of the last term in eq. (16.23) causes the constitutive
equations to be nonsymmetric. However, if the deformations are nearly
incompressible, the last term can be neglected and the constitutive
equations remain symmetric.
At the end of the increment, the state variables need to be updated.
Stresses are updated, following the relationships given by eq. (16.8), by

where / is the Jacobian of the deformation increment, and will be equal to

unity if the material is incompressible.
The applications of this formulation to metal-forming processes include
sample problems, such as tension, compression, extrusion, and forging of
complex shapes, for the purpose of demonstrating the capability of the
formulations. A study undertaken by Rebelo et al. [29] is noteworthy from
the application point of view. It is a comparative study of several
algorithms implemented according to elastic-plastic (solid) and rigid-plastic
(flow) formulations. The results of the investigation are illustrated in the
simulation of ring compression and are given in the next section.

16.5 Comparison with Rigid-Plastic (Flow) Solutions

Plate Bending [23]
The analysis of plane-strain plate bending using the rigid-plastic in-
finitesimal deformation theory was presented in chap. 8 (Section 8.5). The
same process conditions are used for the elastic-plastic analysis.
328 Metal Forming and the Finite-Element Method

FIG. 16.2 Finite-element mesh system of the undeformed workpiece for (a) rigid-plastic and
(b) elastic-plastic analyses in plate bending.

With reference to Fig. 8.9 in Chap. 8, the boundary conditions for

traction in the flow formulation are changed to traction-rate. In Fig. 16.2
the undeformed grid systems are shown for both the rigid-plastic and
elastic-plastic analyses. These two mesh systems are exactly the same
except that, for the elastic-plastic analysis, the rectangular mesh is further
divided into four triangular elements where plastic strain is expected to be
large, and into two triangular elements where only elastic strain is
expected.
For elastic-plastic calculations, each incremental step solution is deter-
mined by solving a system of linear equations. This linear system assumes
that the nonlinearity of material behavior and of geometry can be
approximated by first-order linear equations. Therefore, the step size must
be kept small to obtain accurate solutions. Also, the assumption that the
boundary conditions remain unchanged during an increment limits the step
size. Changes in boundary conditions include touching of a free nodal
point to the punch, or separation of nodal points from the punch. The step
size is also restricted by limiting the element rotation to within 1° for each
increment.
In the computation the maximum punch displacement per step was set
to 0.05-0.06 (sheet thickness is unity). However, the actual step size was
limited to 0.01-0.012 after considerable yielding took place, because of the
restrictions described above. The total punch stroke to achieve a 90°
bending angle is about 3.2.
Some of the results obtained by elastic-plastic analysis were compared
with those obtained by rigid-plastic analysis in Chap. 8. Comparisons of
grid distortions, bend angle-punch displacement relationship, and stress
and strain distributions revealed that the results for both analyses agree
with each other very well with only minor differences.
 Solid Formulation, Comparison of Two Formulations 329

In order to compare the stress and strain distributions quantitatively,

bending stresses and effective strains at two stages of deformation are
plotted against undeformed thickness positions at several cross-sections of
the workpiece in Figs. 16.3 and 16.4. For identification, the numbers of the
cross-sections correspond to those given in Fig. 16.2 (shown as element
number).
Figure 16.3 shows the excellent agreement between the bending stresses
that are computed from both formulations. The agreement is almost
perfect at the outer fiber of the workpiece. However, there appear to be
some discrepancies near the mid-surface of the plate thickness.
The effective strain distributions in Fig. 16.4 show a nearly uniform
gradient across the cross-section. The figure also shows that, for both
stages of deformation, the rigid-plastic analysis predicts higher effective

FIG. 16.3 Variations of normalized bending stress along the thickness directions at various
cross sections (as indicated in Fig. 16.2), for punch displacements of (a) 2.25 and (b) 3.2.
330 Metal Forming and the Finite-Element Method

ELASTO-PLASTIC RIGID-PLASTIC

FIG. 16.4 Variation of effective strain along the thickness direction at various cross sections
(as indicated in Fig. 16.2), for punch displacements of (a) 2.25 and (b) 3.2.

strains near the outer fiber and lower values near mid-surface, than the
corresponding values predicted by the elastic-plastic analysis.
One of the main disadvantages of the rigid-plastic formulation is that it
cannot predict unloading behavior. However, in the case where the
unloading conditions can be well-defined, such as in sheet bending, it is
possible to calculate springback and the residual stress distribution by
performing the elastic-plastic unloading calculations using the results of the
rigid-plastic loading solution.
The springback angle calculated with the elastic-plastic solution is 3.03°,
and 2.94° when obtained from the other solution. Figure 16.5 shows the
distributions of the bending component of residual stress along the
thickness of several selected cross-sectional areas for both cases. The
results show that the residual stresses obtained from the elastic-plastic
solution are somewhat higher along the cross sections near the bending
axis. Comparison of the two solutions shows excellent agreement between
them, not only in overall quantities but also in detail.
The amount of springback was obtained by performing unloading
calculations. Two methods were used, namely, elastic-plastic loading and
unloading, and rigid-plastic loading and elastic-plastic unloading. The
results show that the springback angle, as well as the residual stress
distributions, calculated by both methods are in excellent agreement.
 Solid Formulation, Comparison of Two Formulations 331

FIG. 16.5 Normalized residual stress distributions in plate bending: (a) elastic-plastic loading
and unloading; (b) rigid-plastic loading and elastic-plastic unloading.

From an economic point of view, the elastic-plastic finite-element

method requires 430 steps to reach a 90° bend angle, while the rigid-plastic
program needs a total of 498 iterations for the same deformation. The total
computing time was 350 s by the rigid-plastic program and about 1000 s by
the elastic-plastic program, using the CDC 7600 computer.
Ring compression
Rebelo et al. [29] presented a comparative study of several algorithms used
in the simulation of metal-forming processes. The same forming problem
was solved with all algorithms and the results were compared. The results
of the simulation of ring compression are given here.
The same finite-element mesh was used in all tests, consisting of 100
bilinear quadrilateral elements, representing half of a ring of height
20 mm, outer diameter 40 mm, and inner diameter 20 mm. The upper die
moves at a velocity of 1 mm/s. The material being deformed was assumed
to be commercially pure aluminum. The primary set of test runs
compressed the workpiece to a 50% reduction in height in 100 increments.
In all cases, a constant friction factor m = 0.7 along the interface was used.
In the following, the test results obtained by the solid formulation
(elastic-plastic; mean normal method of integrating plasticity equations;
full Newton-Raphson algorithm; 0.5% step-size) and by the flow formula-
tion (rigid-plastic; direct substitution algorithm; 0.5% step-size) are
compared. The relative execution times on a PRIME 9955 computer were
1 and 0.26, respectively. In Fig. 16.6 the deformed meshes of the two
solutions at 50% reduction in height appear to be almost identical,
although some differences in bulge contours may be noticed. In a more
quantitative way, a comparison is shown in Fig. 16.7, displaying total
plastic strain contours at the same reduction in height as that given in Fig.
332 Metal Forming and the Finite-Element Method

FIG. 16.6 Deformed mesh in ring compression (shown for upper half of the ring
cross-section) after 50% reduction in height [29]: (a) elastic-plastic analysis; (b) rigid-plastic
analysis.

16.6. It is clear that the two solutions give almost identical results in detail.
Figure 16.8 shows plots of nodal-point loads at the die-workpiece
interface. The closeness of the two distributions confirms that both
formulations are physically consistent.
Comments on Comparison
The comparison shown above indicates that both the flow and solid
formulations are capable of predicting certain physical phenomena in
metal forming. A variety of formulations and algorithms are now available
for simulation of forming processes. The following remarks may be helpful
for the selection of an approach to a specific forming problem.
When we consider simulation of forming processes, it is important for an
engineer to be aware of the objective of simulation. Specific information
will be sought by simulation, depending on the objective. Each problem
has its peculiarities and specific requirements, and the selected method
must have the capability of delivering the required information. The

FIG. 16.7 Equivalent plastic strain contours in ring compression at 50% reduction in height
[29]: (a) elastic-plastic analysis; (b) rigid-plastic analysis.
 Solid Formulation, Comparison of Two Formulations 333

FIG. 16.8 Axial nodal force distributions along the die-workpiece contact surface in ring
compression [29]: (a) elastic-plastic analysis: (b) rigid-plastic analysis.

method should be computationally efficient. A measure of efficiency is

execution time, but true efficiency of the method must be considered in the
context of the complete system of simulation. The solution accuracy
depends on the mesh system and the step size, and influences efficiency.
The schemes for data preparation, remeshing, and postprocessing are
important factors that must be considered for total efficiency. Ease of
interfacing a simulation method with those peripheral activities greatly
enhances the overall efficiency of simulation.
Finally, solution reliability is a major concern. Reliability of the
computed solutions can only be determined by comparing predictions with
experimental observations. Since experimental observations are limited to
certain aspects of deformation, such as geometry and hardness distribu-
tion, the relative reliability of computed solutions is difficult to determine.
334 Metal Forming and the Finite-Element Method
The determination of reliability depends on how closely a boundary value
problem is defined for a physical problem. In this regard, formulations
concerning materials behavior and friction at the die-workpiece interface
must include parameters that reflect the real physical conditions. In
particular, the influence of the change of the constitutive equation
parameters upon deformation behavior was investigated for rigid-plastic
materials by Tomita [30]. Eggert and Dawson [31] used internal variable
constitutive equations for characterizing the changing state of a material
during a deformation process.

16.6 Concluding Remarks

The applications of the finite-element method in metal-forming are being
continuously improved and expanded. Many improvements are presented
in the book "NUMIFORM 86", Numerical Methods in Industrial Forming
Processes [32]. In concluding, some of the studies discussed in this book
are described here. One improvement is to expand the analysis capabilities
of the method to three-dimensional deformation. Using the flow formula-
tion, Chenot and his co-workers have developed a three-dimensional
viscoplastic finite-element code and analyzed shape rolling [33]. Further,
they used explicit and implicit Euler schemes for time integration in
connection with three-dimensional forging analysis [34].
Three-dimensional analysis capability has also been developed with the
solid formulation. Pillinger et al. [35] discussed various applications to
forging, rolling, and ring compression. In particular, the analysis of ring
compression shows the capability of the method, including predictions of
temperature and strain-rate distributions.
Deformation analyses of porous materials were discussed in Chap. 13 in
the framework of thermo-viscoplasticity. Constitutive equations of elasto-
viscoplastic type were used for the analysis of hot isostatic pressing of
metal powders [36]. A major concern is for improvement in computational
efficiency. Zienkiewicz et al. [37] suggested some new directions toward
this objective. These are mixed formulations in the iterative procedure,
and mesh regeneration—in particular, automatic mesh regeneration.
Thompson offered an approach for transient analyses of metal-forming
operations—the method of pseudoconcentrations [38]. He concluded that
the method is easy to implement and requires a minimum of user
interventions during an analysis. It was mentioned in the previous section
that improvements in solution reliability are mainly influenced by realistic
formulations of constitutive equations and friction boundary conditions. A
construction of frictional constitutive equations was presented for finite-
strain elastic-plastic material behavior by Baaijens et al. [39]. The
usefulness of the formulation, however, requires verification by ex-
perimental investigations. With respect to constitutive equations, a general
formulation of elastic-plastic theory that is valid for finite deformations,
involving strain-induced anisotropy, was developed by Lee and Agah-
 Solid Formulation, Comparison of Two Formulations 335

Tehran! [40]. Lush and Anand [41] proposed implicit time integration
procedures for internal variables of constitutive equations for large-
deformation elasto-viscoplasticity.
Pecherski [42] presented the phenomenological model for large plastic
strains and the onset of deformation instability. For calculation of the
flatness defect of rolled strip, Yukawa et al. [43] solved the eigenvalue
problem for the buckling limit and determined the shape after buckling by
the large-displacement finite-element method. Mori et al. [44], in the
coupled analysis of a forming process with elastic tools, treated the plastic
deformation of the workpiece by the rigid-plastic finite-element method
and the elastic deflection by the boundary-element method.
These investigations provide some indication for ever-increasing ap-
plications of the finite element method in metal forming.

References
1. Kudo, H., and Matsubara, S., (1979), Joint Examination Project of Validity of
Various Numerical Methods for the Analysis of Metal Forming Processes, in
"Metal Forming Plasticity," (Edited by H. Lippmann), Springer-Verlag,
Berlin, p. 378.
2. Yamada, Y., Yoshimura, H., and Sakurai, T., (1968), "Plastic Stress-Strain
Matrix and its Application for the Solution of the Elastic-Plastic Problem by
the Finite Element Method," Int. J. Mech. Sci., Vol. 10, p. 343.
3. Yamada, Y., and Takatsuka, K,., (1973), "Finite-Element Analysis of
Nonlinear Problems," J. Japan Soc. Tech. Plast., Vol. 14, p. 758.
4. Larsen, P. K., (1971), "Large Displacement Analysis of Shells of Revolution,
Including Creep, Plasticity and Viscoplasticity," Ph.D. Thesis, University of
California, Berkeley.
5. Needleman, A., (1970), "Void Growth in an Elastic-Plastic Medium," Ph.D.
Thesis, Harvard University.
6. McMeeking, R. M., and Rice, J. R., (1975), "Finite-Element Formulation for
Problems of Large Elastic-Plastic Deformation," Int. J. Solids Structures, Vol.
11, p. 601.
7. Hill, R., (1958), "A General Theory of Uniqueness and Stability in Elastic-
Plastic Solids," /. Mech. Phys. Solids, Vol. 6, p. 231.
8. Hill, R., (1959), "Some Basic Principles in the Mechanics of Solids Without a
Natural Time," /. Mech. Phys. Solids, Vol. 7, p. 209.
9. Rice, J. R., and Tracy, D. M., (1983), "Computational Fracture Mechanics,"
Proc. Symp. on Numerical and Computer Methods in Structure Mechanics,
Urbana IL, Academic Press, Orlando, p. 585.
10. Krieg, R. D., and Krieg, D. B., (1977), "Accuracies of Numerical Solution
Methods for the Elastic-Perfectly Plastic Model," /. Pressure Vessel Tech.,
ASME, Vol. 90, p. 510.
11. Lee, E. H., Mallett, R. L., and Yang, W. H., (1977), "Stress and
Deformation Analysis of the Metal Extrusion Process," Computer Methods
Appl. Mech. Engr., Vol. 10, p. 339.
12. Lee, E. H., Mallett, R. L., and McMeeking, R. M., (1977), "Stress and
Deformation Analysis of Metal Forming Processes," Numerical Modeling of
Manufacturing Processes, ASME Special Publication PVP-PB-025, p. 19.
13. Hartley, P., Sturgess, C. E. N., and Rowe, G. W., (1978), "A Finite Element
Analysis of Extrusion-Forging," Proc. VI North American Metalworking
Research Conference, SME, p. 212.
336 Metal Forming and the Finite-Element Method

14. Wifi, A. S., (1976), "An Incremental Complete Solution of the Stretch-
Forming and Deep-Drawing of a Circular Blank Using a Hemispherical
Punch," Int. J. Mech. ScL, Vol. 18, p. 23.
15. Wang, N. M., and Budiansky, B., (1978), "Analysis of Sheet Metal Stamping
by a Finite Element Method," ASME Trans. J. Appl. Mech., vol. 45, p. 73.
16. Rao, S. S., and Kumar, A., (1977), "Finite Element Analysis of Cold Strip
Rolling," Int. J. Machine Tool Des. Res., Vol. 17, p. 159.
17. Key, S. W., Krieg, R. D., and Bathe, K. J., (1979), "On the Application of
the Finite Element Method to Metal Forming Processes—Part 1," Computer
Methods Appl. Mech. Engr., Vol. 17/18, p. 597.
18. Hartley, P., Sturgess, C. E. N., and Rowe, G. W., (1979), "Friction in Finite
Element Analysis of Metal Forming Processes," Int. J. Mech. ScL, Vol. 21, p.
301.
19. Hartley, P., Sturgess, C. E. N., and Rowe, G. W., (1979), "A Prediction of
the Influence of Friction in the Ring Test by the Finite-Element Method,"
Proc. 7th NAMRC, Ann Arbor, Michigan, p. 1.
20. Hartley, P., Sturgess, C. E. N., and Rowe, G. W., (1980), "Finite Element
Predictions of the Influence of Strain-rate and Temperature Variations on the
Properties of Forged Products," Proc. 8th NAMRC, University of Missouri,
Rolla, MO., p. 121.
21. Derbalian, K. A., Lee, E. H., Mallett, R. L., and McMeeking, R. M., (1978),
"Finite Element Metal Forming Analysis with Spatially Fixed Mesh," Appli-
cations of Numerical Methods to Forming Processes, ASME, AMD, Vol. 28,
p. 39.
22. Wang, N. M., (1980), "On Analytical Formulations of Computational Sheet
Metal Forming," Proc. 4th International Conference on Production Engineer-
ing, Tokyo, p. 28.
23. Oh, S. I., and Kobayashi, S., (1980), "Finite Element Analysis of Plane-Strain
Sheet Bending," Int. J. Mech. ScL, Vol. 22, p. 583.
24. Nagtegaal, J. C., and DeJong, J. E., (1981), "Some Computational Aspects of
Elastic-Plastic Large Strain Analysis," Int. J. Num. Methods Engr., Vol. 17, p.
15.
25. Nagtegaal, J. C., (1982), "On the Implementation of Inelastic Constitutive
Equations with Special Reference to Large Deformation Problems," Computer
Methods Appl. Mech. Engr., Vol. 33, p. 469.
26. Nagtegaal, J. C., and Veldpaus, F. E., (1984), On the Implementation of
Finite Strain Plasticity Equations in a Numerical Model, in "Numerical
Analysis of Forming Processes", (Edited by J. F. T. Pittman, O. C.
Zienkiewicz, R. D. Wood and J. M. Alexander), Wiley, New York, p. 351.
27. Rebelo, N., and Wertheimer, T. B., (1986), "General Purpose Procedures for
Elastic-Plastic Analysis of Metal Forming Processes," Proc. 14th NAMRC,
Minneapolis, MN., p. 414.
28. Nagtegaal, J. C., and Rebelo, N., (1986), On the Development of a General
Purpose Finite Element Program for Analysis of Forming Processes, in
"NUMIFORM 86," Balkema, Rotterdam, p. 41 [see Ref. 32].
29. Rebelo, N., Nakazawa, S., Wertheimer, T. B., and Nagtegaal, J. C., (1986),
"A Comparative Study of Algorithms Applied in Finite Element Analysis of
Metal Forming Problems," Special Publication ASME, PED-Vol. 20, p. 17.
30. Tomita, Y., (1982), "A Rigid Plastic Finite Element Method for the Effective
Prediction of the Influence of the Change in Parameters in the Constitutive
Equation upon Deformation Behavior," Int. J. Mech. ScL, Vol. 24, p. 711.
31. Eggert, G. M., and Dawson, P. R., (1987), "On the Use of Internal Variable
Constitutive Equations in Transient Forming Processes," Int. J. Mech. Sci.,
Vol. 29, p. 95.
32. NUMIFORM 86 (1986), "Numerical Methods in Industrial Forming Proc-
 Solid Formulation, Comparison of Two Formulations 337

esses," (Edited by K. Mattiasson, A. Samuelson, R. D. Wood and O. C.

Zienkiewicz), A. A. Balkema, Rotterdam.
33. David, C., Bertrand, C., Chenot, L. J., and Buessler, P., (1986), A Transient
Three Dimensional Finite Element Analysis of Hot Rolling of Thick Slabs, in
"NUMIFORM 86," p. 219 [see Ref. 32].
34. Surdon, G., and Chenot, J. L., (1986), Finite Element Calculation of Three
Dimensional Hot Forging, in "NUMIFORM 86," p. 287 [see Ref. 32].
35. Pillinger, I., Hartley, P., Sturgess, C. E. N., and Rowe, G. W., (1986), Finite
Element Modeling of Metal Flow in Three-Dimensional and Temperature-
Dependent Forming, in "NUMIFORM 86," p. 151 [see Ref. 32].
36. Abouaf, M., Chenot, J. L., Raisson, G., and Baudin, P., (1986), Finite
Element Simulation of Hot Isostatic Pressing of Metal Powders, in
"NUMIFORM 86," p. 79 [see Ref. 32].
37. Zienkiewicz, O. C., Liu, Y. C., Zhee, J. Z., and Toyoshima, S., (1986), Flow
formulation for Numerical Solution of Forming Processes II, in "NUMIFORM
86," p. 3 [see Ref. 32].
38. Thompson, E., (1986), Transient Analysis of Metal Forming operations Using
Pseudo-Concentrations, in "NUMIFORM 86," p. 65 [see Ref. 32].
39. Baaijens, F. P. T., Brekelmans, W. A. M., Veldpaus, F. E., and Starmans, F.
J. M., (1986), A Constitutive Equation for Frictional Phenomena Including
History Dependency, in "NUMIFORM 86," p. 91 [see Ref. 32].
40. Lee, E. H., and Agah-Tehrani, A., (1986), The Structure of Constitutive
Equations for Finite Deformation of Elastic-Plastic Materials with Strain-
Induced Anisotropy, in "NUMIFORM 86," p. 29 [see Ref. 32].
41. Lush, A., and Anand, L., (1986), Implicit Time-Integration Procedures for a
Set of Internal Variable Constitutive Equations for Hot-working, in
"NUMIFORM 86," p. 131 [see Ref. 32].
42. Pecherski, R. B., (1986), The Disturbed Spin Concept and its Consequences in
Plastic Instability, in "NUMIFORM 86," p. 145 [see Ref. 32].
43. Yukawa, N., Ishikawa, T., and Tozawa, T., (1986), Numerical Analysis of the
Shape of Rolled Strip, in "NUMIFORM 86," p. 249 [see Ref. 32].
44. Mori, K., Osakada, K., Nakadori, K., and Fukuda, M., (1986), Coupled
Analysis of Steady State Forming Processes with Elastic Tools, in
"NUMIFORM 86," p. 237 [see Ref. 32].
 Appendix
THE FEM CODE, SPID
(SIMPLE PLASTIC INCREMENTAL
DEFORMATION)

A.I Introduction
The development and testing of the computer code is an important part of
the finite-element method. In programming a practical finite-element
method code, one should consider various factors, such as generality,
computational efficiency, and pre- and postprocessing. However, such
considerations are beyond the scope of this book. Instead, a simple
two-dimensional FEM code, called SPID, written for metal-forming
simulation is presented as an example and discussed in this appendix. The
main purpose of the code is to illustrate the programming of key steps
explained in this book.
The program SPID, Simple Plastic /ncremental Deformation, is based
on rigid viscoplastic finite-element formulation. SPID is capable of
handling only simple forming processes, such as simple compression and
ring compression. Some important features and limitations of SPID are
summarized below.
• SPID is valid for rigid plastic as well as rigid viscoplastic materials.
• A constant shear factor friction law is used.
• No heat transfer simulation is included.
• It can handle one flat die only.
• It cannot handle free surface folding to the die.
• The initial guess is generated automatically.
• A banded matrix solver is used.
• SPID can handle a finite-element model with up to 100 nodes.
SPID is written in FORTRAN 77 standard, with special consideration of
portability. The program was tested on two computer systems. One was
VAX 750 with VMS version 4.4 operating system. The other was IBM
PC/AT with a math coprocessor, where the professional FORTRAN compiler
of Ryan-McFarland Corp. and DOS 3.10 operating system of Micro Soft
Inc. were used. The required minimum memory size for the IBM PC to
run SPID is 256 K Bytes.

338
 Appendix The FEM Code, SPID 339
A.2 Program Structure
The program structure of SPID in terms of subroutine calling sequence is
shown schematically in Fig. A.I. Brief descriptions of each subroutine are
given as comment lines in the program listing in Section A.6. The
functional procedure of SPID is described below. Subroutine names
corresponding to each logical step are also given in parenthesis.
1. Program starts. (SPID)
2. Read input from input file SPID.DAT. (INPRED)
3. Print input information. (PRTINP)
4. Determine the maximum half-bandwidth and number of stiffness
equations. (BAND)
5. If initial guess is required, then set up for the direct iteration
procedure. If not, then use Newton-Raphson iteration method.
(NONLIN)
6. Evaluate elemental strain-rate matrix. (STRMTX)
7. Evaluate elemental stiffness matrix. (ELSHLF, VSPLON,
VSPLST)

FIG. A.I Subroutine calling sequence of SPID. The names enclosed in boxes are the
subroutine names and the double-line box indicates the continuation of the diagram.
340 Metal Forming and the Finite-Element Method

8. Calculate nodal force. (NFORCE)

9. Apply ffictional boundary conditions. (FRCBDY, FRCINT)
10. Assemble global stiffness matrix. (ADDBAN)
11. Apply displacement boundary condition. (DISBDY)
12. Solve stiffness equations. (BANSOL)
13. Adjust the velocity solution based on the result of step 12.
(NONLIN)
14. Determine the error norm. (NORM)
15. If converged, go to Step 16. Otherwise, go to Step 5. (NONLIN)
16. Update the workpiece geometry. (POTSOL)
17. Stress and strain evaluation. (POTSOL)
18. Write the results on file SPID.OUT. (PRTSOL)
19. Generate restart file, SPID.RST. (RSTFIL)
20. If desired step reached, then go to Step 21. Otherwise, go to Step 5.
(SPID)
21. Terminate the program. (SPID)

A.3 Input and Output Files

SPID requires one input file, SPID.DAT, to run a simulation and it
generates three output files, SPID.MSG, SPID.OUT, and SPID.RST at
the end of each run. These files are described below.
SPID.DAT Name of the input file for initial run and continuing run.
FORTRAN file units are indicated by INPT = 6
SPID.MSG SPID.MSG is one of SPID output files and it contains
messages generated by SPID. This message file clearly
contains
(a) information on iterations for each step solution
(b) error messages from SPID
SPID.OUT SPID.OUT is output file of SPID and it contains
(a) initial input summary
(b) simulation results for each step solution
SPID.RST SPID.RST is one of SPID output files to be used as input
file for a continuing run. SPID.RST has the same format
as that of SPID.DAT and it is ready to be used as an input
file to SPID for a continuing run. However, the user
should rename this to SPID.DAT and change data input
for NSEND (refer data input description of Line 2 in A.4)
before continuing the run.

A.4 Input Preparations

The initial input file, SPID.DAT, to be prepared by the user, is described
below. Input is read by FORTRAN list directed FORMAT.
 Appendix The FEM Code, SPID 341

1. Master Control Data

Line 1. TITLE
TITLE Heading up to 70 characters
Line 2. MINI, NSEND
NINI Starting step number
NSEND Ending step number
Line 3. ALPH, DIAT
ALPH Limiting strain-rate
DIAT Penalty constant to enforce volume constancy
Line 4. IPLNAX
IPLNAX If = 1, deformation is axisymmetric
If = 2, deformation is plane-strain
Line 5. FRCFAC
FRCFAC Constant shear friction factor
2. Node Coordinates Data
Line 6. NUMNP
NUMNP Total number of nodal points
Line 7. N, (RZ(I,N), I = 1,2)*
N Node number
RZ(1,N) x or r coordinate of node N
RZ(2,N) y or z coordinate of node N
* This line should be repeated from node 1 to node NUMNP.
3. Element Data
Line 8. NUMEL
NUMEL Total number of elements
Line 9. N, (NOD(I,N), I = 1,4)*
N Element number
NOD(I,N) Element connectivity of element N
* Element connectivity should be given counterclockwise.
4. Boundary Condition Code Data
Line 10. N, (NBCD(I,N), 1 = 1,2)*
N Node number
NBCD(1,N) Boundary condition code in * or r direction
= 0, nodal force is specified.
= 1, nodal velocity is specified.
* If the boundary condition codes for both directions are zero, then the
corresponding line can be skipped. However, data corresponding the last
node (NUMNP) cannot be skipped.
342 Metal Forming and the Finite-Element Method
NBCD(2, N) Boundary condition code in y or z direction
= 0, nodal force is specified.
= 1, nodal velocity is specified.
= 3, node is in contact with the die.
5. Node Velocity Data
Line 11. N, (URZ(I,N), 1 = 1, 2)*
N Node number
URZ(1,N) x or r velocity of node TV
URZ(2,N) y or z velocity of node TV
* This program assumes that both velocity components are zero for the
skipped node. Last node velocity data cannot be skipped.
6. Flow Stress of Workpiece Material
The flow stress of the workpiece material is given by the user supplied
subroutine, FLWSTS. The program listing in Section A.6 shows an
example for subroutine FLWSTS.

A.5 Description of the Major Variables

Most of the major variables used in the program SPID are included in
labeled COMMONS. The descriptions of these variables are given below.
COMMON/INOT/INPT, MSSG, IUNIT, IUNI2
INPT Input file unit number, INPUT = 5
MSSG Message file unit number, MSSG = 6
IUNIT Output file unit number, IUNIT = 3
IUNI2 Restart file unit number, IUNI2 = 4
COMMON/MSTR/NUMNP,NUMEL,IPLNAX
NUMNP Total number of nodal points
NUMEL Total number of elements
IPLNAX Deformation mode indicator
If = 1, axisymmetric mode
If = 2, plane-strain mode
COMMON/CNEQ/NEQ,MBAND
NEQ Total number of stiffness equations
MBAND Half-bandwidth
COMMON/RIGD/RTOL,ALPH,DIAT
RTOL Error limit to check the convergence
ALPH Limiting strain-rate
DIAT Penalty constant for volume constancy
COMMON/TSTP/NINI,NCUR,NSEND,NITR,DTMAX
NINI Starting step number
NCUR Current step number
 Appendix The FEM Code, SPID 343
NSEND Ending step number
NITR Current iteration number
DTMAX Step size in time unit
COMMON/DIES/FRCFAC
FRCFAC Constant shear friction factor
COMMON/ITRC/ITYP,ICONV
ITYP Type of iteration process
If = 1, Newton-Raphson iteration
If = 2, direction iteration
ICONV Convergence indicator
If = 1, solution is converged
If = 2, solution is not converged
COMMON/RVA1/RZ(2,100),URZ(2,100),FRZ(2,100)
RZ(I,N) Coordinates of node N
URZ(I,N) Velocity of node N
FRZ(I,N) Nodal force of node N
COMMON/RAV2/EPS(5,100) ,STS(5,100) ,TEPS(100)
EPS(I,N) Strain rate components of element N
1 = 1, ex or er
= 2, ey or ez
= 3, e2 or ee
= 4, jxy or jrz
= 5, effective strain rate e
STS(I,N) Stress components of element N
1=1, ax or or
= 1, oy or az
= 3, crz or aH
= 4, Txy or xrz
= 5, effective stress a
TEPS(N) Total effective strain of element TV
COMMON/INVR/NOD(4,100),NBCD(2,100)
NOD(I,N) Element connectivity of node TV
NBCD(I,N) Boundary condition code of node N
COMMON A(5000),B(200)*
A Array for stiffness matrix
B Array for load vector
* Variables are set up to handle up to 100-node FEM model.

A.6 Program Listing

A FORTRAN program listing of SPID is given here with the main program,
SPID, listed at the beginning and all other subroutines listed in alphabeti-
cal order by the subroutine names. Chapter and Equation numbers given
in some subroutines refer to those in this book.
344 Metal Forming and the Finite-Element Method
PROGRAM SPID
C
c*************************************************

C
C METAL FORMING AND THE FINITE ELEMENT METHOD
C
C BY SHIRO KOBAYASHI, SOO-IK OH AND TAYLAN ALTAN
C
C OXFORD UNIVERSITY PRESS, 1988
C
C THIS IS MAIN PROGRAM OF RIGID VISCOPLASTIC FINITE
C ELEMENT METHOD FOR SIMPLE FORMING PROCESSES.
C
IMPLICIT INTEGERS (I-N), REAL*8 (A-H, 0-Z)
CHARACTER TITLE*70
COMMON /TITL/ TITLE
COMMON /RVA2/ EPS(5,100), STS(5,100), TEPS(IOO)
COMMON /INOT/ INPT,MSSG,IUNIT,IUNI2
COMMON /MSTR/ NUMNP.NUMEL.IPLNAX
COMMON /INVR/ NOD(4,100),NBCD(2,100)
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /ITRC/ ITYP.ICONV
INPT - 5
MSSG - 6
IUNIT = 3
IUNI2 = 4
C
C READ INPUT
C
CALL INPRED
OPEN(IUNIT,FILE='SPID.OUT',STATUS='UNKNOWN',FORM='FORMATTED')
OPEN(MSSG, FILE='SPID.MSG',STATUS='UNKNOWN',FORM='FORMATTED')
WRITE(MSSG,1020) TITLE
CALL PRTINP
CALL BAND(NOD,NUMEL,NUMNP)
C
C STEP SOLUTIONS
C
NINI = NINI + 1
C
DO 300 N = NINI, NSEND
NCUR - N
C
WRITE(MSSG,1050) N
IF(N .NE. NINI) GO TO 80
ICOUNT = 0
50 ITYP = 2
CALL NONLIN
ICOUNT - ICOUNT + 1
80 ITYP - 1
CALL NONLIN
IF(ICONV .EQ. 2 .AND. ICOUNT .GT. 3) GO TO 900
 Appendix The FEM Code, SPID 345

IF(ICONV .EQ. 2) GO TO 50
C
CALL POTSOL
CALL PRTSOL
CALL RSTFIL
300 CONTINUE
C
CLOSE(IUNIT)
CLOSE(MSSG)
STOP
C
900 CONTINUE
WRITE(MSSG,1070)
C
STOP
1020 FORMAT(1H1,//,5X,'OUTPUT OF S P I D',//,
+ 5X,' MESSAGE FILE FOR '/,5X,A,//)
1050 FORMAT(///,' ITERATION PROCESS FOR STEP ',I5,//)
1070 FORMAT{/,' STOP BECAUSE SOLUTION DOES NOT CONVERGE. ')
END
c
SUBROUTINE ADDBAN(B,A,NQ,LM,Qq,PP)
C
C CHAPTER 7.2,
C EQUATION (7.8)
C
C ASSEMBLE GLOBAL STIFFNESS MATRIX FROM ELEMENTAL STIFFNESS MATRIX
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGERM (I-N)
DIMENSION B(l), A(NQ,1), QQ(1), PP(8,8), LM(1)
C
DO 100 I = 1, 8
II = LM(I)
DO 50 J - 1, 8
JJ = LM(J) - LM(I) + 1
IF(JJ .LE. 0) GO TO 50
A(II.JJ) = A(II,JJ) + PP(I.J)
50 CONTINUE
B(II) = B(II) + QQ(I)
100 CONTINUE
RETURN
END
c
SUBROUTINE BAND(NOD,NUMEL,NUMNP)
C
C CHAPTER 7.2
C DETERMINE MAXIMUM HALF BANDWIDTH, MBAND AND
C TOTAL NUMBER OF EQUATIONS, NEQ
C
IMPLICIT REAL*8 (A-H,0-Z), INTEGERM (I-N)
COMMON /CNEQ/ NEQ,MBAND
DIMENSION NOD(4,1)
C
MBAND = 0
346 Metal Forming and the Finite-Element Method
DO 100 N - 1, NUMEL
NMIN - NOD(1,N)
NHAX - NOD(l.N)
DO 50 I - 2,4
IF(NMIN .GT. NOD(I.N)) NMIN - NOD(I.N)
IF(NHAX .LT. NOD(I,N)) NMAX = NOD(I.N)
50 CONTINUE
MB - (NMAX - NMIN + 1) * 2
IF(MBAND .LT. MB) MBAND = MB
100 CONTINUE
NEQ - NUHNP * 2
RETURN
END
c
SUBROUTINE BANSOL(B,A,NQ,MM)
C
C CHAPTER 7.2
C THIS SUBROUTINE SOLVES THE BANDED SYMMETRIC MATRIX EQUATIONS BY
C THE GAUSSIAN ELIMINATION.
C B LOAD VECTOR
C A SYMMETRIC MATRIX IN BANDED FORM
C MM HALF BANDWIDTH
C NQ NUMBER OF EQUATIONS
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGERS (I-N)
COMMON /INOT/ INOT,MSSG,IUNIT,IUNI2
DIMENSION B(l), A(NQ,1)
C
DO 200 N = 1, NQ
IF(A(N,1) .LE. 0.) GO TO 800
DO 150 L = 2, MM
IF(A(N,L) .EQ. 0.) GO TO 150
C = A(N,L) / A(N,1)
I = N + L -1
J - 0
DO 100 K = L, MM
J - J +1
100 A(I,J) - A(I,J) - C*A(N,K)
A(N,L) = C
150 CONTINUE
200 CONTINUE
C
C LOAD VECTOR REDUCTION
C
DO 300 N - 1, NQ
DO 250 L = 2, MM
I = N + L - 1
IF(I .GT. NQ) GO TO 250
B(I) = B(I) - A(N,L) * B(N)
250 CONTINUE
B(N) = B(N) / A(N,1)
300 CONTINUE
C
C BACK SUBSTITUTION
 Appendix The FEM Code, SPID 347
r
DO 400 M - 1,NQ
N = NQ + 1 - M
DO 350 K - 2, MM
L - N + K -1
350 B(N) - B(N) - A(N,K)*B(L)
400 CONTINUE
RETURN
C
800 CONTINUE
WRITE(MSSG,1020) N
STOP
1020 FORMAT(/,' NEGATIVE PIVOT AT EQUATION NO. ',15)
END
c
SUBROUTINE DISBDY(URZ,NBCD,B,A,NEQ,MBAND,ITYP)
C
C CHAPTER 7.3
C APPLY DISPLACEMENT BOUNDARY CONDITION
C
IMPLICIT REAL*8 (A-H.O-Z), INTEGERM (I-N)
DIMENSION B(1),A(NEQ,1),NBCD(1),URZ(1)
C
IF(ITYP .EQ. 2) GO TO 120
DO 100 N > 1, NEQ
IF(NBCD(N) .EQ. 0) GO TO 100
DO 70 I = 2, MBAND
II - N - I + 1
IF(II .LE. 0) GO TO 50
A(II,I) = 0.
50 CONTINUE
II - N + I - 1
IF(II .GT. NEQ) GO TO 70
A(N,I) - 0.
70 CONTINUE
B(N) - 0.
A(N,1) = 1.
100 CONTINUE
RETURN
C
120 CONTINUE
DO 200 N = 1, NEQ
IF(NBCD(N) .EQ. 0) GO TO 200
DO 170 I = 2, MBAND
II - N - I + 1
IF(II .LE. 0) GO TO 150
B(II) = B(II) - A(II,I) * URZ(N)
A(II,I) - 0.
150 CONTINUE
II - N + I - 1
IF(II .GT. NEQ) GO TO 170
B(II) = B(II) - A(N,I) * URZ(N)
A(N,I) = 0.
170 CONTINUE
348 Metal Forming and the Finite-Element Method

B(N) * URZ(N)
A(N,1) - 1.
200 CONTINUE
END
c
SUBROUTINE ELSHLFfPP.QQ.RZ.URZ.EPS.TEPS.IPLNAX.IDREC.NEL)
C
C CHAPTER 6.5
C EQUATIONS (6.43) AND (6.44)
C EVALUATION OF ELEMENTAL STIFFNESS MATRIX
C
C IDREC : IF = 1, NEWTON-RAPHSON ITERATION
C = 2, DIRECT ITERATION
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGER** (I-N)
DIMENSION RZ(2,1), URZ(2,1), B(4,8), EPS(l), TEPS(l)
DIMENSION QQ(1),PP(8,8),S2(2),W2(2)
DATA S2/-0.57735026918963DO,0.57735026918963DO/,
+ W2/2*1.0DO/
C
DO 10 I - 1, 8
QQ(I) - 0.
DO 10 J = 1, 8
PP(I,J) = 0.
10 CONTINUE
C
C CARRY OUT ONE POINT INTEGRATION
C
S = 0.
T = 0.
CALL STRMTXfRZ.B.WDXJ.SJ.IPLNAX.NEL)
WDXJ = WDXJ * 4
CALL VSPLON(QQ,PP,B,URZ,EPS,WDXJ,IDREC)
C
C REGULAR INTEGRATION
C
DO 100 I - 1, 2
S - S2(I)
DO 50 J - 1, 2
T - S2(J)
CALL STRMTXfRZ.B.WDXJ.S.T.IPLNAX.NEL)
WDXJ - WDXJ * W2(I)*W2(J)
CALL VSPLST(QQ,PP,B,URZ,TEPS,WDXJ,IDREC)
50 CONTINUE
100 CONTINUE
RETURN
END
C
SUBROUTINE FLWSTS(YS,FIP,STRAN,STRRT)
C
C USER SUPPLIED SUBROUTINE TO DESCRIBE THE MATERIAL
C FLOW STRESS.
C THIS SUBROUTINE SHOWS THE EXAMPLE OF
C YS - 10. * (STRAIN RATE) ** 0.1
 Appendix The FEM Code, SPID 349

C
IMPLICIT INTEGER*4 (I.J.K.L.M.N), REAL*8 (A-H.O-Z)
COMMON /RIGD/ RTOL, ALPH, DIAT
C
IF(STRRT .LT. ALPH) GO TO 100
YS = 10. * STRRT ** 0.1
FIP = STRRT ** (-0.9)
RETURN
C
100 YO = 10. * ALPH ** 0.1
FIP - YO / ALPH
YS = FIP * STRRT
RETURN
END
c
SUBROUTINE FRCBDY(RZ,URZ,NBCD,TEPS,EPS,QQ,PP,IPLNAX)
C
C APPLY FRICTION BOUNDARY CONDITION
C
IMPLICIT INTEGERM (I-N), REAL*8 (A-H, 0-Z)
COMMON /DIES/ FRCFAC
COMMON /INOT/ INPT, MSSG, IUNIT, IUNI2
DIMENSION RZ(2,1),URZ(2,1),NBCD(2,1),EPS(5),QQ(1},PP(8,1),
+ ER(2,2),FR(2),XY(2,2),VXY(2,2)
C
DO 100 N = 1,4
11 = N + 1
12 = N
IF(N .EQ. 4) II = 1
IF(NBCD(2,I1) .NE. 3 .OR. NBCD(2,I2) .NE. 3) GO TO 100
CALL FLWSTS(FLOW,DUM,TEPS,EPS(5))
XY(1,1) - RZ(1,I1)
XY(2,1) = RZ(2,I1)
XY(1,2) - RZ(1,I2)
XY(2,2) = RZ(2,I2)
VXY(l.l) - URZ(1,I1)
VXY(2,1) = URZ(2,I1)
VXY{1,2) - URZ(1,I2)
VXY(2,2) - URZ(2,I2)
CALL FRCINT(XY,VXY,FLOW,FR.ER,FRCFAC,IPLNAX)
Jl = II * 2 - 1
J2 - 12 * 2 - 1
QQ(J1) - QQ(J1) + FR(1)
QQ(J2) = QQ(J2) + FR(2)
PP(J1,J1) = PP(J1,J1) + ER(1,1)
PP(J2,J2) - PP(J2,J2) + ER(2,2)
PP(J1,J2) - PP(J1,J2) + ER(1,2)
PP(J2,J1) = PP(J2,J1) + ER(2,1)
100 CONTINUE
RETURN
END
c
SUBROUTINE FRCINT(RZ,URZ,FLOW,FR.ER,FRCFAC,IPLNAX)
C
350 Metal Forming and the Finite-Element Method

C CHAPTER 7.3
C EQUATIONS (7.14) AND (7.15)
C INTEGRATION METHOD : SIMPSON'S FORMULA
C THIS ROUTINE CALCULATES THE FRICTION MATRIX
C USED FOR BOTH TYPES OF ITERATION SCHEME
C
IMPLICIT INTEGERM (I-N), REAL*8 (A-H, 0-Z)
COMMON /INOT/ INPT,MSSG,IUNIT,IUNI2
COMMON /ITRC/ ITYP.ICONV
DIMENSION RZ(2,1),URZ(1),QQ(1),PP(8,1)
DIMENSION SLIV(2),ER(2,2),FR(2)
DATA PI/3.1415926535898DO/
DATA UA/0.0005DO/
C
C INITIALIZE FR AND ER ARRAY
C
DO 10 I « 1, 2
FR(I) = 0.
DO 10 0 = 1, 2
ER(I,J) - 0.
10 CONTINUE
MINT = 5
FAC = DSQRT((RZ(1,2)-RZ(1,1))**2 + (RZ(2,2
FK FLOW * FRCFAC / SQRT(3.)
DH 2. / (NINT - 1)
S -1. - DH
CON 2. / PI * FK
WD DH / 3. * FAC * 0.5 * CON
C
DO 300 N - 1, NINT
S = S + DH
HI = 0.5 * (1. - S)
H2 = 0.5 * (I. + S)
WDXJ = WD
IF(IPLNAX .NE. 1) GO TO 90
RR = H1*RZ(1,1) + H2*RZ(1,2)
WDXJ - RR * WDXJ
90 CONTINUE
C
IF(N .EQ. 1 .OR. N .EQ. NINT) GO TO 100
NMOD = N - N/2*2
IF(NMOD .EQ, 0) WDXJ = WDXJ * 4
IF(NMOD .EQ. 1) WDXJ = WDXJ * 2
100 CONTINUE
C
US * H1*URZ(1) + H2*URZ(3)
AT = DATAN(US/UA)
IF(ITYP .EQ. 2) GO TO 200
US2 = US * US
USA - US2 + UA*UA
CT1 = AT*WDXJ
CT2 - UA/USA*WDXJ
GO TO 250
C
 Appendix The FEM Code, SPID 351

C FOR D-ITERATION CASE

C
200 CONTINUE
IF(DABS(US) .LE. l.OD-5) SLOP - UA/(UA*UA+US*US)
IF(DABS(US) .GT. l.OD-5) SLOP = AT / US
CT1 - 0.
CT2 = SLOP*WDXJ
C
C CALCULATE CONTRIBUTION TO STIFFNESS
C
250 CONTINUE
FR(1) - FR(1) - H1*CT1
FR(2) - FR(2) - H2*CT1
ER(1,1) = ER(1,1) + H1*H1*CT2
ER(1,2) = ER(1,2) + H1*H2*CT2
ER(2,2) = ER(2,2) + H2*H2*CT2
ER(2,1) - ER(1,2)
300 CONTINUE
RETURN
END
SUBROUTINE INPRED
C
C READ INPUT FROM INPUT FILE
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGERM (I-N)
CHARACTER TITLE*70
COMMON /TITL/ TITLE
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /RVA1/ RZ(2,100), URZ(2,100), FRZ(2,100)
COMMON /RVA2/ EPS(5,100), STS(5,100), TEPS(IOO)
COMMON /INVR/ NOD(4,100), NBCO(2,100)
COMMON /DIES/ FRCFAC
COMMON /RIGD/ RTOL, ALPH, DIAT
COMMON /MSTR/ NUMNP, NUMEL, IPLNAX
COMMON /INOT/ INPT, MSSG, IUNIT, IUNI2
C
C READ MASTER CONTROL DATA
C
OPEN(INPT,FILE='SPID.DAT',FORM='FORMATTED',STATUS-'OLD')
READ (INPT,1000) TITLE
READ (INPT,*) NINI.NSEND.DTMAX
READ (INPT,*) ALPH, DIAT
READ (INPT,*) IPLNAX
C
C READ DIE DATA
C
READ (INPT,*) FRCFAC
C
C READ FEM NODE INFORMATION
C
READ (INPT,*) NUMNP
IF(NUMNP .GT. 300) GO TO 500
DO 20 I = 1, NUMNP
352 Metal Forming and the Finite-Element Method
READ (INPT,*) N, (RZ{J,N), J - 1, 2)
20 CONTINUE
C
C READ ELEMENT INFORMATION
C
READ (INPT,*) NUMEL
IF(NUMEL .GT. 300) GO TO 500
DO 40 I - 1, NUMEL
READ (INPT,*) N, (NOD(J,N), J-1,4)
40 CONTINUE
C
C READ BOUNDARY CONDITION DATA
C
DO 60 N = 1, NUMNP
DO 60 I = 1, 2
NBCD(I,N) = 0
60 CONTINUE
C
DO 80 N - 1, NUMNP
READ (INPT,*) M, (NBCD(I,M), I = 1,2)
IF (M .GE. NUMNP) GO TO 100
80 CONTINUE
100 CONTINUE
C
C READ NODE VELOCITY DATA
C
DO 120 N = 1, NUMNP
DO 120 1 = 1,2
URZ(I.N) = 0.
120 CONTINUE
DO 140 N = 1, NUMNP
READ (INPT,*) M, (URZ(I.M), 1=1,2)
IF(M. GE. NUMNP) GO TO 160
140 CONTINUE
160 CONTINUE
220 CONTINUE
C
C READ STRAIN DATA
C
IF(NINI .EQ. 0) GO TO 300
DO 240 N = 1, NUMEL
READ (INPT,*) M, TEPS(M)
240 CONTINUE
300 CLOSE(INPT)
RETURN
C
500 CONTINUE
WRITE(MSSG,1010)
STOP
C
1000 FORMAT(A)
1010 FORMAT(/,' SORRY, THIS PROGRAM CANNOT HANDLE MORE THAN 100'
+ ' NODES OR ELEMENTS ')
END
 Appendix The FEM Code, SPID 353

c
SUBROUTINE NFORCE(QQ,FRZ,LM)
C
C ADD NODAL POINT FORCE
C
IMPLICIT REAL*8 (A-H.O-Z), INTEGERM (I-N)
DIMENSION QQ{1),FRZ(1),LM(1)
C
DO 100 I = 1, 8
N - LM(I)
FRZ(N) - FRZ(N) - QQ(I)
100 CONTINUE
RETURN
END
c
SUBROUTINE NONLIN
C
C THIS ROUTINE CONTROLS THE ITERATIONS
C
IMPLICIT INTEGERM (I-N), REAL*8 (A-H, 0-Z)
COMMON /INOT/ INPT.MSSG,IUNIT.IUNI2
COMMON /MSTR/ NUMNP.NUMEL.IPLNAX
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /ITRC/ ITYP, ICONV
COMMON /CNEQ/ NEQ,MBAND
COMMON /RVA1/ RZ(2,100), URZ(2,100), FRZ(2,100)
DIMENSION UNORM(2),ENORM(2),FNORM(2)
COMMON A(5000),B(200)
C
RTOL •= 0.00001
IF(ITYP .EQ. 2) RTOL - 0.0005
ACOEF - 1.
NSTEL - NEQ * MBAND
IF(NSTEL.LE.5000 .AND. NEQ.LE.200) GO TO 10
HRITE(MSSG,1010)
STOP
C
10 CONTINUE
DO 30 N - 1 , 2
UNORM(N) = 0.
ENORM(N) - 0.
FNORM(N) - 0.
30 CONTINUE
C
ITRMAX - 20
IF(ITYP .EQ. 2) ITRMAX - 200
DO 200 N = 1, ITRMAX
NITR = N
CALL STIFF(B,A,NEQ,MBAND,ITYP)
IDREC = 1
CALL NORM(FRZ,B,FDUM,DFN,NEQ,IDREC)
IF(ITYP .EQ. 2) DFN = 0.
CALL BANSOL(B,A,NEQ,MBAND)
IDREC - ITYP
354 Metal Forming and the Finite-Element Method
CALL NORM(URZ,B,UC,EC,NEQ,IDREC)
IF(ITYP .EQ. 1) WRITE(MSSG,1030) N
IF(ITYP .EQ. 2) WRITE(MSSG,1050) N
WRITE(HSSG,1070) UC.EC.DFN
IF(N .EQ. 1) GO TO 130
IF(EC .LT. RTOL .AND. DFN .LT. RTOL) GO TO 300
IF(ITYP .EQ. 2) GO TO 130
IF(EC .LT. ENORH(2)) GO TO 100
C
C ADJUST THE ACOEF
C
ACOEF = ACOEF * 0.7
GO TO 130
100 CONTINUE
IF(ENORM(1) .GT. ENORM(2) .AND. ENORM(Z) .GT. EC)
+ ACOEF - ACOEF * 1.3
IF(ACOEF .GT. 1.) ACOEF = 1.0
C
C VELOCITY UPDATE
C
130 CONTINUE
NB - 0
DO 150 I - 1, NUMNP
DO 150 J - 1,2
NB - NB + 1
IF(ITYP .EQ. 1) URZ(J.I) - URZ(J.I) + ACOEF * B(NB)
IF(ITYP .EQ. 2) URZ(J,I) = B(NB)
150 CONTINUE
C
170 CONTINUE
UNORM(l) UNORM(2)
ENORM(l) ENORH(2)
FNORH(l) FNORM(2)
UNORM(2) UC
ENORM(2) EC
FNORM(2) DFN
200 CONTINUE
C
C SET FLAG
C
ICONV - 2
RETURN
C
300 CONTINUE
C CONVERGED CASE
C SET FLAG
C
ICONV - 1
C
RETURN
1010 FORHAT(/,' YOU NEED MORE SPACE IN THE BLANK COMMON ')
1030 FORMAT(/,' N-R ITERATION NO. ',I5,/)
1050 FORMAT(/,' DRT ITERATION NO. ', 15,/)
1070 FORMAT( ' VELOCITY NORM = '.F15.7,/,
 Appendix The FEM Code, SPID 355

+ ' REL. ERROR NORM - '.F15.7,/,

+ ' REL. FORCE ERROR NORM = '.F15.7,/)
END
c
SUBROUTINE NORM(URZ,V,UC,EROR,NEQ,ITYP)
C
C CALCULATE THE ERROR NORM FOR LINEAR AND NONLINEAR CASE
C
IMPLICIT INTEGERM (I-N), REAL*8 (A-H.O-Z)
DIMENSION URZ(1),V(1)
C
UC = 0.
EROR = 0.
DO 100 N = 1, NEQ
UC = UC + URZ(N) * URZ(N)
IF(ITYP .EQ. 1) EROR = EROR + V(N) * V(N)
IF(ITYP .EQ. 2) EROR - EROR + (URZ(N)-V(N))**2
100 CONTINUE
C
UC - DSQRT(UC)
EROR - DSQRT(EROR)
IF(UC .NE. 0.) EROR - EROR / UC
RETURN
END
SUBROUTINE POTSOL
C
C THIS SUBROUTINE HANDLES THE POST SOLUTION PROCEDURES, IE,
C GEOMETRY UPDATES
C STRESS EVALUATION
C TOTAL STRAIN EVALUATION
C
IMPLICIT REAL*8 (A-H.O-Z), INTEGER*4 (I-N)
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /MSTR/ NUMNP.NUMEL.IPLNAX
COMMON /RIGD/ RTOL.ALPH.DIAT
COMMON /RVA1/ RZ(2,100), URZ(2,100), FRZ(2,100)
COMMON /RVA2/ EPS(5,100), STS(5,100),TEPS(100)
COMMON /DIES/ FRCFAC
C
C GEOMETRY UPDATES
C
DO 100 N = 1, NUMNP
RZ(1,N) - RZ(1,N) + DTMAX * URZ(l.N)
RZ(2,N) - RZ(2,N) + DTMAX * URZ(2,N)
100 CONTINUE
C
C STRESS EVALUATION
C
DO 200 N - 1, NUMEL
AL - EPS(5,N)
IF(AL .LT. ALPH) AL = ALPH
CALL FLWSTS(EFSTS, STRT, TEPS(N), AL)
EM = (EPS(l.N) + EPS(2,N) + EPS(3,N)) / 3.
356 Metal Forming and the Finite-Element Method

DO 150 I = 1,3
STS(I.N) = 2./3. * EFSTS * (EPS(I.N)-EM) / AL + DIAT * EM * 3.
150 CONTINUE
STS(4,N) = EFSTS * EPS(4,N) / AL / 3.
STS(5,N) = EFSTS
200 CONTINUE
C
C UPDATE TOTAL EFFECTIVE STRAIN
C
DO 300 N = 1, NUMEL
TEPS(N) = TEPS(N) + EPS(5,N) * DTMAX
300 CONTINUE
RETURN
END
c
SUBROUTINE PRTINP
C
C THIS SUBROUTINE PRINTS THE INPUT DATA
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGERM (I-N)
CHARACTER TITLE*70
COMMON /TITL/ TITLE
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /RVA1/ RZ(2,100), URZ(2,100), FRZ(2,100)
COMMON /RVA2/ EPS(5,100), STS(5.100), TEPS(IOO)
COMMON /INVR/ NOD(4,100), NBCD(2,100)
COMMON /DIES/ FRCFAC
COMMON /RIGD/ RTOL, ALPH, DIAT
COMMON /MSTR/ NUMNP, NUMEL, IPLNAX
COMMON /INOT/ INPT, MSSG, IUNIT, IUNI2
C
C INPUT SUMMARY
C
WRITE(IUNIT,1010) TITLE
WRITE(IUNIT,1020)
WRITE(IUNIT,1030) NINI.NSEND,DTMAX
WRITE(IUNIT,1050) ALPH,DIAT
WRITE(IUNIT,1070) IPLNAX
WRITE(IUNIT,1110) FRCFAC
WRITE(IUNIT,1130) NUMNP
WRITE(IUNIT,1150)
WRITE(IUNIT,1180) (N,(RZ(I,N), 1=1,2), N = 1, NUMNP)
C
C PRINT NODE VELOCITY
C
WRITE(IUNIT,1220)
WRITE(IUNIT,1180) (N,(URZ(I,N), 1=1,2), N = 1, NUMNP)
C
C ELEMENT INFORMATION
C
WRITE(IUNIT,1270) NUMEL
WRITE(IUNIT,1330)
WRITE(IUNIT,1350) (N, (NOD(I,N),1=1,4), N=l,NUMEL)
c
 Appendix The FEM Code, SPID 357
C BOUNDARY CONDITION
C
WRITE(IUNIT,1400)
WRITE(IUNIT,1430) (N, (NBCD(I,N),1-1,2), N=1,NUMNP)
C
C WRITE STRAIN DISTRIBUTION AT INPUT STAGE
C
WRITE(IUNIT,1500)
WRITE(IUNIT,1550) (N,TEPS(N), N=1,NUMEL)
RETURN
C
1010 FORHAT(1H1,///,5X,'OUTPUT OF S P I D ',//,
+ 5X.A,///)
1020 FORMAT(5X,'INITIAL INPUT SUMMARY'///)
1030 FORMATC INITIAL STEP NUMBER 15,/,
+ ' FINAL STEP NUMBER 15,/,
+ ' STEP SIZE IN TIME UNIT F10.5)
1050 FORMATC LIMITING STRAIN RATE F15.7,/,
+ ' PENALTY CONSTANT F15.7)
1070 FORMATC DEFORMATION CODE 15,/,
+ ' IF - 1, AXISYMMETRIC ',/
+ ' = 2, PLANE STRAIN ')
1110 FORMATC FRICTION FACTOR = ', F15.7,/)
1130 FORMAT{///' NUMBER OF NODAL POINTS = ', 15,/)
1150 FORMAT(//,' NODE COORDINATES ',//,
+ ' NODE NO X-COORD Y-COORD',/)
1180 FORMAT(5X,I5,5X,2F15.7)
1220 FORMAT(///,' NODE VELOCITY ',//,
+ ' NODE NO X-VELOCITY Y-VELOCITY'/)
1270 FORMAT(//, NUMBER OF ELEMENTS = ', 15,/)
1330 FORMAT(//, ELEMENT CONNECTIVITY ',//,
+ ELE NO. I J K L ',/)
1350 FORMAT(5I7
1400 FORMAT(//, BOUNDARY CONDITION CODE ',//,
+ NODE NO Xl-CODE X2-CODE ',/)
1430 FORMAT(3I7)
1500 FORMAT(///,' STRAIN DISTRIBUTION AT INPUT STAGE ',//,
+ ' NODE NO. STRAIN ',/)
1550 FORMAT(I5,5X,F15.7)
END
c
SUBROUTINE PRTSOL
C
C THIS SUBROUTINE PRINT THE SOLUTION RESULTS
C
IMPLICIT REAL*8 (A-H.O-Z), INTEGERM (I-N)
CHARACTER TITLE*70
COMMON /TITL/ TITLE
COMMON /INOT/ INPT,MSSG,IUNIT,IUNI2
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /MSTR/ NUMNP,NUMEL,IPLNAX
COMMON /RVA1/ RZ(Z.IOO), URZ(2,100), FRZ(2,100)
COMMON /RVA2/ EPS(5,100), STS(5,100), TEPS(IOO)
COMMON /INVR/ NOD(4,100), NBCD(2,100)
358 Metal Forming and the Finite-Element Method
C
C PRINT NODE COORDINATES
C
WRITE(IUNIT,1010) TITLE, NCUR
WRITE(IUNIT,1020)
WRITE(IUNIT,1040) (N,(RZ(I,N), 1=1,2), N -1, NUHNP)
C
C PRINT NODE VELOCITY, NODAL FORCE
C
WRITE(IUNIT,1080)
WRITE(IUNIT,1100) (N,(URZ(I,N), 1-1,2),
cc + (FRZ(I,N), 1=1,2), N = 1, NUMNP)
C STRAIN RATE, STRESS, TOTAL EFFECTIVE STRAIN
C
WRITE(IUNIT,1130)
WRITE(IUNIT,1180) (N,(EPS(I,N),1-1,5), N - 1, NUMEL)
WRITE(IUNIT,1230)
WRITE(IUNIT,1180) (N,(STS(I,N),1=1,5), N = 1, NUMEL)
WRITE(IUNIT,1330)
WRITE(IUNIT,1360) (N, TEPS(N), N = 1, NUMEL)
RETURN
C
1010 FORMAT(1H1,///,5X,'OUTPUT OF S P I D',//,5X,A,//,
+ 10X, SOLUTION AT STEP NUMBER - ',I5,///)
1020 FORMAT( /, NODE COORDINATES',//,
+ NODE NO X-COORD Y-COORD',/)
1040 FORMAT(5X,I5,5X,2F15.7)
1080 FORMAT(///, NODAL VELOCITY AND FORCE',//,
+ NODE NO X-VELOCITY Y-VELOCITY',
+ ' X-FORCE Y-FORCE ',//)
1100 FORMAT(3X,I5,3X,4F15.7)
1130 FORMAT(///,' STRAIN RATE COMPONENTS ',//,
+ ELE. NO. Ell E22 E33'
+ E12 EBAR',//)
1180 FORMAT(I5,5F15.7)
1230 FORMAT(//, STRESS COMPONENTS',//,
+ ELE. NO. Sll S22 S33'
+ S12 SBAR',//)
1330 FORMAT(///,' TOTAL EFFECTIVE STRAIN ',//,
+ ' ELE. NO. EFFECTIVE STRAIN '//)
1360 FORMAT(5X,I5,5X,F15.7)
END
c
SUBROUTINE RSTFIL
C
C GENERATE RESTART FILE
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGERM (I-N)
CHARACTER TITLE*70
COMMON /TITL/ TITLE
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /RVA1/ RZ(2,100), URZ(2,100), FRZ(2,100)
COMMON /RVA2/ EPS(5,100), STS(5,100), TEPS(IOO)
 Appendix The FEM Code, SPID 359
COMMON /INVR/ NOD(4,100), NBCD(2,100)
COMMON /DIES/ FRCFAC
COMMON /RIGD/ RTOL, ALPH, DIAT
COMMON /MSTR/ NUMNP, NUMEL, IPLNAX
COMMON /INOT/ INPT, MSSG, IUNIT, IUNI2
C
NN - NCUR + 1
OPEN(IUNI2,FILE='SPID.RSr,STATUS*'UNKNOWN',FORM='FORMATTED')
WRITE(IUNI2,1010) TITLE
WRITE(IUNI2,1040) NCUR.NN.DTMAX
WRITE(IUNI2,1060) ALPH.DIAT
HRITE(IUNI2,1080) IPLNAX
WRITE(IUNI2,1060) FRCFAC
WRITE(IUNI2,1080) NUMNP
WRITE(IUNI2,1120) (N,(RZ(I,N),1=1,2),N=l,NUMNP)
WRITE(IUNI2,1080) NUMEL
WRITE(IUNI2,1080) (N,(NOD(I.N),1-1,4), N=l,NUMEL)
WRITE(IUN12,1160) (N,(NBCD(I,N),1-1,2),N=1,NUMNP)
WRITE(IUNI2,1120) (N,(URZ(I,N),I-=1,2),N-1,NUMNP)
WRITE(IUNI2,1200) (N,TEPS(N), N-l,NUMEL)
CLOSE(IUNI2)
RETURN
C
1010 FORMAT(1X,A)
1040 FORMAT(2I10,F20.7)
1060 FORMAT(3F20.10)
1080 FORMAT(517)
1120 FORMAT(I5,2F20.10)
1160 FORMAT(3I7)
1200 FORMAT(I7,F20.10)
END
c
SUBROUTINE STIFF(B,A,NEQ,MBAND,ITYP)
C
C STIFFNESS MATRIX GENERATION
C ITYP - 1, NEWTON-RAPHSON ITERATION
C ITYP - 2, DIRECT ITERATION
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGER*4 (I-N)
COMMON /INOT/ INPT,MSSG,IUNIT.IUNI2
COMMON /RVA1/ RZ(2,100), URZ(2,100), FRZ(2,100)
COMMON /RVA2/ EPS(5,100), STS(5,100), TEPS(IOO)
COMMON /INVR/ NOD(4,100), NBCD(2,100)
COMMON /DIES/ FRCFAC
COMMON /MSTR/ NUMNP, NUMEL, IPLNAX
DIMENSION A(NEQ,1),B(1)
DIMENSION RZE(2,4), URZE(2,4), NBCDE(2,4),PP(8,8), QQ(8),
+ LM(8)
C
C INITIALIZE LOAD VECTOR, STIFFNESS MATRIX, AND
C NODAL POINT FORCE ARRAY
C
DO 20 N = 1, NEQ
B(N) = 0.
360 Metal Forming and the Finite-Element Method

DO 20 I = 1, MBAND
A(N,I) = 0.
20 CONTINUE
DO 50 N - 1, NUHNP
DO 50 I - 1, 2
50 FRZ(I.N) - 0.
C
DO 200 N - 1, NUMEL
C
C CHANGE RZ, URZ, AND NBCD FROM GLOBAL ARRANGEMENT TO ELEMENTAL
C ARRANGEMENT
C
DO 100 I - 1, 4
12 - 1 * 2
II = 1 2 - 1
NE = NOD(I,N)
RZE(1,I) - RZ(1,NE)
RZE(2,I) - RZ(2,NE)
URZE(l.I) - URZ(l.NE)
URZE(2,I) - URZ(2,NE)
NBCDE(l.I) = NBCD(l.NE)
NBCDE(2,I) - NBCD(2,NE)
LM(I2) - NOD(I,N)*2
LH(I1) - LM(I2) - 1
100 CONTINUE
C
CALL ELSHLF(PP,QQ,RZE,URZE,EPS{1,N),TEPS(N),IPLNAX,ITYP,N)
IF(ITYP .EQ. 1) CALL NFORCE(QQ,FRZ,LM)
IF(FRCFAC.NE. 0.)
+ CALL FRCBDY(RZE,URZE,NBCDE,TEPS(N),EPS(1,N),QQ,PP,IPLNAX)
CALL ADDBAN(B,A,NEQ,LM,QQ,PP)
200 CONTINUE
C
C APPLY DISPLACEMENT BOUNDARY CONDITION
C
CALL DISBDY(URZ,NBCD,B,A,NEQ,MBAND,ITYP)
RETURN
END
c

SUBROUTINE STRMTX(RZ,B,WDXJ,S,T,IPLNAX,NEL)
C
C CHAPTER 6.4
C EQUATIONS (6.25), (6.27) AND (6.35)
C EVALUATE STRAIN RATE MATRIX OF QUADRILATERAL ELEMENT
C
C B(4,8) : STRAIN RATE MATRIX
C RZ(2,4) : NODE COORDINATES
C (S,T) : NATURAL COORDINATE
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGERM (I-N)
COMMON /INOT/ INPT,MSSG,IUNIT,IUNI2
DIMENSION RZ(2,1), B(4,l)
C
R12 = RZ(1,1) - RZ(1,2)
 Appendix The FEM Code, SPID 361

R13 RZ(1,1) - RZ(1,3)

R14 RZ(1,1) - RZ(1,4)
R23 RZ(1,2) - RZ(1,3)
R24 RZ(1,2) - RZ(1,4)
R34 RZ(1,3) - RZ(1,4)
C
Z12 RZ(2,1) - RZ(2,2)
Z13 RZ(2,1) - RZ(2,3)
Z14 RZ(2,1) - RZ(2,4)
Z23 RZ(2,2) - RZ(2,3)
Z24 RZ(2,2) - RZ(2,4)
Z34 RZ(2,3) - RZ(2,4)
C
C REFER EQUATION (6.34)
C
DX08 = ( (R13*Z24 - R24*Z13) + (R34*Z12 - R12*Z34)*S +
+ (R23*Z14 - R14*Z23)*T )
DXJ = DXJ8 / 8.
IF(DXO .GT. 0.) GO TO 10
WRITE(MSSG,1010) NEL
WRITE(HSSG,1030) DXJ.SJ
STOP
10 CONTINUE
C
XI ( Z24 - Z34*S - Z23H) / DXJ8
X2 (-Z13 + Z34*S + Z14*T) / DXJ8
X3 (-Z24 + Z12*S - Z14*T) / DXJ8
X4 ( Z13 - Z12*S + Z23*T) / DXJ8
C
Yl (-R24 + R34*S + R23*T) / DX08
Y2 ( R13 - R34*S - R14*T) / DXJ8
Y3 ( R24 - R12*S + R14*T) / DXJ8
Y4 (-R13 + R12*S - R23*T) / DXJ8
C
DO 20 I = 1, 4
DO 20 J = 1,8
B(I,J) - 0.
20 CONTINUE
C
B(l,l) XI
8(1,3) X2
B(l,5) X3
B(l,7) X4
B(2,2) Yl
B(2,4) Y2
B(2,6) Y3
B(2,8) Y4
C
WDXJ = DXJ
IF(IPLNAX .NE. 1) GO TO 40
Ql = (l.-S) * (l.-T) * 0.25
Q2 - (l.+S) * (l.-T) * 0.25
Q3 - (l.+S) * (l.+T) * 0.25
Q4 = (l.-S) * (l.+T) * 0.25
362 Metal Forming and the Finite-Element Method

C
R - Q1*RZ(1,1) + Q2*RZ(1,2) + Q3*RZ(1,3) + Q4*RZ(1,4)
B(3,l) - Ql / R
B(3,3) - QZ / R
8(3,5} - Q3 / R
8(3,7) - Q4 / R
WDXJ - WDXJ * R
C
40 CONTINUE
B(4,l) - Yl
B(4,3) •= Y2
8(4,5) = Y3
B(4,7) - Y4
B(4,2) - XI
B(4,4) - X2
B(4,6) = X3
B(4,8) - X4
RETURN
C
1010 FORMAT(/,' SORRY, NEGATIVE JACOBIAN DETECTED AT ELEMENT NO.
+ 15)
1030 FORMAT(' DXJ.S.T = \3F15.7)
END
c

SUBROUTINE VSPLON(QQ,PP, B,URZ,EPS,WDXJ,IDREC)

C
C CHAPTERS 6.5 AND 7.1
C EQUATIONS (6.43) AND (6.44)
C REDUCED INTEGRATION OF VOLUME STRAIN RATE
C
C PP = ELEMENTAL STIFFNESS MATRIX
C QQ = ELEMENTAL LOAD VECTOR
C 8 = STRAIN RATE MATRIX
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGER*4 (I-N)
COMMON /RIGD/ RTOL.ALPH.DIAT
DIMENSION PP(8,8), QQ(8), B(4,8), URZ(l), EPS(l)
DIMENSION D(6),XX(8)
DATA 0/3*0.666666666666666700, 3*0.3333333333333333DO/
C
C GENERATE DILATATIONAL STRAIN RATE MATRIX
C
DO 20 I = 1, 8
XX(I) = B(1,I) + 8(2,1) + 8(3,1)
20 CONTINUE
C
C CALCULATE STRAIN RATE COMPONENTS
C
DO 40 I = 1, 5
EPS(I) = 0.
40 CONTINUE
XVOL = 0.
DO 60 J = 1, 8
XVOL = XVOL + XX(J) * URZ(J)
 Appendix The FEM Code, SPID 363

DO 60 I = 1, 4
EPS(I) - EPS(I) + B(I,J) * URZ(J)
60 CONTINUE
EB2 = (EPS(1)**2 + EPS(2)**2 + EPS(3)**2) * D(l) +
+ EPS(4)**2 * 0(4)
EPS(5) - DSQRT(EB2)
C
C EVALUATE VOLUMETRIC CONTRIBUTION Of STIFFNESS MATRIX
C
DO 80 I - 1, 8
IF(IDREC .EQ. 1)
+ QQ(I) = QQ(I) - DIAT * WDXJ * XVOL * XX(I)
TEN - DIAT * WDXJ * XX(I)
DO 80 J = I, 8
PP(I,J) = PP(I,J) + TEM * XX(J)
PP(J,I) = PP(I,J)
80 CONTINUE
RETURN
END
c
SUBROUTINE VSPLSTtQQ.PP.B.URZ.TEPS.WDXJ.IDREC)
C
C CHAPTERS 6.5
C EQUATIONS (6.43), (6.44) AND (6.46)
C FOUR POINTS INTEGRATION OF VOLUME STRAIN RATE
C
C PP = ELEMENTAL STIFFNESS MATRIX
C QQ = ELEMENTAL LOAD VECTOR
C B = STRAIN RATE MATRIX
C
IMPLICIT REAL*8 (A-H, 0-Z), INTEGER*4 (I-N)
COMMON /TSTP/ NINI,NCUR,NSEND,NITR,DTMAX
COMMON /RIGD/ RTOL,ALPH,DIAT
DIMENSION PP(8,8), QQ(8), 8(4,8), URZ(l)
DIMENSION D(6),FDV(8),E(4),XX(8)
DATA 0/3*0.6666666666666700, 3*0.333333333333333DO/
C
C ELIMINATE DIALATATIONAL COMPONENT FROM STRAIN RATE MATRIX
C
DO 20 I = 1, 8
XX(I) = (B(1,I) + B(2,I) + B(3,I)) / 3.
20 CONTINUE
DO 40 I = 1, 8
DO 40 J = 1,3
B(J,I) = B(J,I) - XX(I)
40 CONTINUE
C
C CALCULATE STRAIN RATE
C
DO 60 J = 1, 4
E(J) = 0.
DO 60 I = 1, 8
E(J) = E(J) + B(J,I) * URZ(I)
60 CONTINUE
364 Metal Forming and the Finite-Element Method
EFSR2 - D(1)*E(1)*E(1) + D(2)*E(2)*E(2) + D(3)*E(3)*E(3) +
+ D(4)*E(4)*E(4)
IF(NITR.EQ.l .AND. NCUR.EQ.NINI .AND. IDREC.EQ.2)
+ EFSR2 - (ALPH*100.) ** 2
ALPH2 - ALPH **2
IF(EFSR2 .LT. ALPH2) EFSR2 - ALPH2
EFSR - DSQRT(EFSR2)
CALL FLWSTS(EFSTS, STRAT, TEPS, EFSR)
C
C CALCULATE FIRST DERIVATE OF EFSR **2
C
DO 80 I - 1, 8
FDV(I) - 0.
DO 80 J - 1 , 4
FDV(I) - FDV(I) + D(J)*E(J)*B(J,I)
80 CONTINUE
C
C ADD POINT CONTRIBUTION TO STIFNESS MATRIX
C
Fl - EFSTS / EFSR * WDXJ
IF(IDREC .EQ. 2) GO TO 200
F2 - STRAT / EFSR2 * WDXJ - Fl / EFSR2
DO 120 I - 1 , 8
QQ(I) - QQ(I) - FDV(I) * Fl
DO 110 0 - 1 , 8
TEM - 0.
DO 100 K - 1, 4
TEM - TEM + D(K)*B(K,I)*B(K,J)
100 CONTINUE
PP(I,J) - PP(I,J) + TEM*F1
IF(EFSR2 .LT. ALPH2) GO TO 105
PP(I,J) - PP(I,J) + FDV(I)*FDV(J)*F2
105 PP(J,I) - PP(I,J)
110 CONTINUE
120 CONTINUE
RETURN
C
200 CONTINUE
DO 300 1 - 1 , 8
DO 280 J - 1,8
TEM - 0.
DO 250 K = 1,4
TEM - TEM + D(K)*B(K,I)*B(K,J)
250 CONTINUE
PP(I,J) = PP(I.J) + TEM*F1
PP(J,I) - PP(I,J)
280 CONTINUE
300 CONTINUE
RETURN
END

A.7 Example Solution

1. Simulation Conditions
A simple compression of circular cylinder was simulated by using SPID.
The computational conditions used for the simulation are as follows.
Initial billet dimension 1.0 (in.) diameter
1.0 (in.) height
Die velocity -1.0(in./s)
Friction factor 0.5
Flow stress 10.0(£)° ' (Ksi)
Total reduction in height 40%
Incremental step size 2% of initial height
 Appendix The FEM Code, SPID 365

Total number of nodes 25

Total number of elements 16
It may be mentioned that the degrees of freedom, 25 nodes, used in the
simulation are hardly enough to obtain accurate results. However, a small
number of nodes were used for demonstration purposes.
2. Input File
The input file, SPID.DAT, for the simple compression is given in the
following.

SIMPLE UPSETTING, M - 0.5

0 20 0.02
0.01 10000.
1 AXISYMMETRIC DEFORMATION
0.5 FRICTION FACTOR
25 NUMBER OF NODES
1 0.00 1.00 NODE COORDINATES
2 0.25 1.00
3 0.50 1.00
4 0.75 1.00
5 1.00 1.00
6 0.00 0.75
7 0.25 0.75
8 0.50 0.75
9 0.75 0.75
10 1.00 0.75
11 0.00 0.50
12 0.25 0.50
13 0.50 0.50
14 0.75 0.50
15 1.00 0.50
16 0.00 0.25
17 0.25 0.25
18 0.50 0.25
19 0.75 0.25
20 1.00 0.25
21 0.00 0.00
22 0.25 0.00
23 0.50 0.00
24 0.75 0.00
25 1.00 0.00
16 NUMBER OF ELEMENTS
1 1 6 7 2 ELEMENT CONNECTIVITY
2 2 7 8 3
3 3 8 9 4
4 4 9 10 5
5 6 11 12 7
6 7 12 13 8
7 8 13 14 9
8 9 14 15 10
9 11 16 17 12
10 12 17 18 13
11 13 18 19 14
12 14 19 20 15
13 16 21 22 17
14 17 22 23 18
15 18 23 24 19
16 19 24 25 20
1 1 3 BOUNDARY CONDITION CODE
2 0 3
3 0 3
4 0 3
5 0 3
6 1 0
366 Metal Forming and the Finite-Element Method
11 1 0
16 1 0
21 1 1
22 0 1
23 0 1
24 0 1
25 0 1
1 0. -1. VELOCITY BOUNDARY CONDITONS
2 0. -1.
3 0. -1.
4 0. -1.
5 0. -1.
25 0. 0.

3. Output File
The output file, SPID.OUT, contains the results of 20 solution steps. The
20th step (40% reduction in height) solution is printed here as an example
of the output file.

OUTPUT OF S P I D
SIMPLE UPSETTING, M = 0.5
SOLUTION AT STEP NUMBER 20

NODE COORDINATES
NODE NO X-COORD Y-COORD
1 0.0000000 0.6000000
2 0.2539333 0.6000000
3 0.5183240 0.6000000
4 0.7739461 0.6000000
5 1.1417572 0.6000000
6 0.0000000 0.3924766
7 0.2869400 0.3802143
8 0.5691376 0.4023207
9 0.8917992 0.4077725
10 1.2311768 0.5210101
11 0.0000000 0.2249196
12 0.3335209 0.2248486
13 0.6637572 0.2373209
14 1.0026870 0.2919173
15 1.2807303 0.3639948
16 0.0000000 0.1097141
17 0.3752439 0.0951748
18 0.7403535 0.1229732
19 1.0537857 0.1484054
20 1.3246827 0.1858872
21 0.0000000 0.0000000
22 0.3950873 0.0000000
23 0.7622312 0.0000000
24 1.0712774 0.0000000
25 1.3397564 0.0000000
 Appendix The FEM Code, SPID 367
NODAL VELOCITY AND FORCE
NODE NO X-VELOCITY Y-VELOCITY X-FORCE Y-FORCE

1 0.0000000 -1.0000000 -0.0206875 -0.1677586

2 0.0437979 -1.0000000 -0.1886367 -0.8241910
3 0.1519569 -1.0000000 -0.3876530 -1.3597889
4 0.1729627 -1.0000000 -0.7469869 -2.9630664
5 0.5416425 -1.0000000 -0.5531770 -3.5269540
6 0.0000000 -0.7072642 0.0223128 0.0000000
7 0.2229644 -0.8703353 0.0000004 -0.0000005
8 0.3261911 -0.7340251 0.0000003 0.0000004
9 0.5932088 -0.9273866 -0.0000002 0.0000012
10 0.7950078 -0.4385331 -0.0000004 0.0000004
11 0.0000000 -0.4869862 0.0064280 -0.0000001
12 0.3307826 -0.4963833 -0.0000002 0.0000000
13 0.5851140 -0.6170831 -0.0000001 0.0000002
14 0.9385745 -0.4774093 0.0000001 0.0000002
15 0.9288204 -0.2414312 0.0000000 0.0000000
16 0.0000000 -0.1617441 0.0011846 0.0000000
17 0.3893606 -0.3021957 -0.0000003 0.0000000
18 0.8405510 -0.2709927 0.0000000 0.0000001
19 1.0506590 -0.2204407 0.0000002 0.0000000
20 1.0489823 -0.1039532 0.0000000 0.0000000
21 0.0000000 0.0000000 0.0176562 0.4882912
22 0.4347056 0.0000000 -0.0000001 2.0701090
23 0.9117962 0.0000000 0.0000000 2.9386526
24 1.0834072 0.0000000 0.0000001 2.4278699
25 1.0878362 0.0000000 0.0000000 0.9161880

STRAIN RATE COMPONENTS

ELE. NO. Ell E22 E33 E12 EBAR

1 0.4916774 -0.9905694 0.4980216 -0.6989617 1.0693530

2 0.4167293 -0.8789797 0.4616160 -0.4731238 0.9206089
3 0.5239917 -0.9806039 0.4559897 -1.6138913 1.3531200
4 1.0648786 -1.5922324 0.5260508 -0.7080532 1.6726366
5 0.9058019 -1.8161495 0.9087389 -0.5110853 1.8394359
6 0.6389859 -1.4431107 0.8032143 -0.7441022 1.5083918
7 1.0831157 -1.8779439 0.7936188 -1.2569937 2.0198449
8 0.6799137 -1.4301980 0.7499446 0.2619431 1.4386273
9 1.0360787 -2.0749139 1.0371629 -0.3153663 2.0823323
10 1.0824354 -2.1205662 1.0366781 -0.7567893 2.1647993
11 0.9886168 -1.9961498 1.0066985 -0.3867207 2.0083485
12 0.1053780 -0.9711997 0.8657071 0.1472329 1.0691494
13 1.0913810 -2.1862521 1.0931731 -0.3355728 2.1942566
14 1.3115202 -2.4730094 1.1598611 -0.1449076 2.4754310
15 0.6322019 -1.7276470 1.0948233 -0.1351249 1.7497068
16 0.0186890 -0.9266512 0.9079304 0.0679924 1.0600883

STRESS COMPONENTS
ELE. NO. Sll S22 S33 S12 SBAR

1 -5.1711015 -14.4740431 -5.1312835 -2.1934270 10.0672791

2 -3.0252520 -12.3309426 -2.7028793 -1.6989706 9.9176212
3 -3.2440441 -10.8846159 -3.5893688 -4.0977962 10.3070322
4 -7.8939030 -19.0434449 -10.1548860 -1.4855362 10.5278614
368 Metal Forming and the Finite-Element Method

5 -11.7723892 -22.2574977 -11.7610756 -0.9843645 10.6284141

6 -5.6950626 -15.2834789 -4.9387623 -1.7133599 10.4196089
7 -7.6387212 -18.1237511 -8.6638214 -2.2254898 10.7283221
8 0.0451580 -10.0953735 0.3817044 0.6294081 10.3703838
9 -12.2963810 -23.0143553 -12.2926460 -0.5432490 10.7610588
10 -10.1809774 -20.8368751 -10.3332052 -1.2588613 10.8029350
11 -4.3976851 -15.0210956 -4.3333286 -0.6882100 10.7222001
12 -0.4260495 -7.1840704 4.3467787 0.4621138 10.0670875
13 -12.5221537 -23.2945029 -12.5162635 -0.5514508 10.8175458
14 -11.5784259 -22.7376738 -12.0256153 -0.2136407 10.9487637
15 -3.3509895 -12.8597561 -1.4869042 -0.2722360 10.5753933
16 -0.1836357 -6.1634682 5.4413395 0.2150461 10.0585228

TOTAL EFFECTIVE STRAIN

ELE. NO. EFFECTIVE STRAIN

1 0.1662418
2 0.1901851
3 0.3206123
4 0.5611510
5 0.4400065
6 0.4483966
7 0.5676523
8 0.4973681
9 0.7078742
10 0.6981489
11 0.6199824
12 0.4176614
13 0.8754011
14 0.8086128
15 0.6054953
16 0.4199022
~Z
 Appendix The FEM Code, SPID 369

FIG. A.2 FEM grid distortions of simple compression predicted by SPID: (a) undeformed;
(b) 40% reduction in height.
/
Figure A.2 shows (a) the initial FEM mesh and (b) the deformed FEM
mesh at 40% reduction in he/ght predicted by SPID.
The computations were performed on an IBM PC/AT with Professional
FORTRAN compiler of Ryarj/McFarland Corp. and also on VAX 750 with
VMS operating system. Tar>le A.I summarizes the CPU time requirements
to run this simulation on/VAX-750 and IBM PC/AT.

TABLE A/1 CPU Time Summary for Simple Compression

VMS version 4.4 operating system was used.

* IBM PC/AT equipped with a hard disk and math co-processor. Professional FORTRAN compiler by
Ryan-McFarland Corp. and DOS operating system, Version 3.10, were used.
This page intentionally left blank
 INDEX

Accuracy of simulation, 333 compressor disk forging, 168-169

Air bending, 21, 141-147 finite-element formulation, 151-153
Air rounding, 21 flashless forging, 170, 171, 172
ALPID program, 136 friction, evaluation of, 163-165, 166
Analysis in metal forming, 26-52 heading of cylindrical bars, 153-159
closed-die forging, 34, 35-36, 37 ring compression, 159-163, 163-165
cold extrusion, 39-41 spike forging, 163, 165
cold forging, 39-41
drawing of rod, wire, shapes, and Backward extrusion forging, 14
tubes, 45-47 Backward tracing, 298-301, 301-305
equations for, 73 in hot forming, 316-317
flow stress of metals, 28-30 in plane-strain rolling, 305-308
friction in metal forming, 30-33 in shell nosing, 301-305
hot extrusion of rods and shapes, in symmetric forging, 311-312
36-39 Bar drawing, 178-183
hot forming, 28-29, 34-35 multipass, 183-186
impression, 35-36, 37 Bar extrusion, 176-178
methods of. See Methods of multipass, 183-184
analysis Beverage cans, 46
objectives of, 26 Block compression, 278-284
rolling of strip, plate, and shapes, rectangular blocks, 278-281
41-45 wedge-shaped blocks, 281-284
sheet-metal forming, 47-52 Blocker operation, 35
temperatures, 28, 29, 33-35 Bore-expanding, 194-195, 196
variables in, 26-28 Boundary conditions, 86-87, 117-121
Anisotropy, 189 with hemispheric punch, 201-203
plastic, 190-191 Boundary lubrication, 31
Area-weighted averaging, 126-127, Brake-bending, 21
129, 255
Assemblage, element, 115-116 Cabbaging, 165-168
Automobile wheel center, 51-52 CAD. See Computer-aided design
Axisymmetric forging Can ironing press, 46-47
of flange-hub shapes, 253-256, 257 Cauchy stress, 56, 57, 58, 208, 324
in preform design, 309-315 Closed-die forging
of pulley blank, 256-259 analysis, 35-36, 37
See also Axisymmetric isothermal load vs displacement curves in, 34
forging metal forming system using, 9
Axisymmetric isothermal forging, with flash, 14, 133-136
151-172 without flash, 12
cabbaging, 165-168 Coining operation, 13
compression of cylinders, 153-159 Cold extrusion, 39-41
371
372 Index
Cold forging, 39-41 square-cup, 210-217
Compaction, 266-270 of tubes, 45-47
Compatibility conditions, 73 of wire, 45-47
Compression Dry conditions, 30
block, 278-284
of cylinder, 75-78, 80-83, 153-159, Effective strain, 66-68
223-225, 229-234, 364-369 interpolation of, 126-127
flat-tool, 314 Effective strain-rate matrix, 107
of porous metals, 249-253 Effective stress, 66-68
ring, 159-163, 163-165, 331-332, Efficiency of simulation, 333
333 Element assemblage, 115-116
square-ring, 284-287 Element connectivity, 115
See also Hot compression; Plane- Element strain-rate matrix,
strain compression 101-107
Compression test, 29 Equilibrium, with tractions, 62
Compressor disk forging, 168-169 Equilibrium equations, 61-62, 73
Computer-aided design (CAD), 35-36, Equipment, 10
93-94 Equivalent stress, 66
Computer-aided techniques, 1, 186. Euler equation, 86
See also Extrusion dies, Extremum principles, 68-70, 71-72,
computer-aided design of 74
Connectivity, element, 115 Extrusion, 19
Constitutive equations, 73 bar, 176-178
Cup drawing, 48 bar, multipass, 183-184
square, 210-217 cold, 39-41
Cylinders direct and indirect, 19
compression of, 75-78, 80-83, hot, 20, 36-39
153-159, 223-225, 229-234, of rods, 36-39
364-369 of shapes, 36-39
steel, compression of, 229-234 Extrusion dies, computer-aided design
Cylindrical bars, heading of, 153-159 of, 186

Deep-drawing processes, 201-206 FEM. See Finite-element method

Deformation mechanics, 10. See also FEM code. See SPID
Plasticity Finite deformation, 56
Deformation speed, 26-28 Finite-element method (FEM), 83-87,
Die bending, 21 88, 90-110, 111-129
Die rounding, 21 admissibility requirement for
Direct iteration method, 121-122 velocity field, 91-92
Disk forging, 311, 312 advantages of, 4
Distortion energy criterion, 59 alternative approach, 84
Drawbead, 50, 51 assemblage, element, 115-116
Drawing, 20 axisymmetric isothermal forging,
bar, 178-183 151-153
bar, multipass, 183-186 axisymmetric out-of-plane
cup, defects in, 48 deformation, 199-201
deep, 23, 201-206 basic concept of, 3
flange, 194-195, 197 basic equations, comments on,
of rod, 45-47 85-86
of shapes, 45-47 basic equations, derivation of, 83
 Index 373

basic principles and concepts in, Flat-tool forging, 293, 294, 295
73-74 Flow formulation, 4-5
basis of, 4 Flow rule, 63-66, 245-246
boundary conditions, 86-87, plasticity and, 63-66
117-121 porous metals, 245-246
computer-aided, 93-94 Flow stress, 28-30, 66
construction of model, 3-4 Folding, 153
direct iteration method, 121-122 Forging. See also Axisymmetric
discretization of problem, 90 forging; Closed-die forging
element strain-rate matrix, 101-107 classification of, 151
elemental stiffness equation, disk, 311, 312
108-110 flashless, 170, 171, 172
geometry updating, 123-125 flat-tool, 293, 294, 295
history of, 90 forward extrusion, 13
in-plane deformation processes, hot-die disk, 237, 238, 239
192-194 isothermal, 237, 238, 239. See also
introduction to, 3-4 Axisymmetric isothermal forging
Lagrange multiplier method, 110 open-die, 15
linear matrix solver, 115-117 orbital, 16
linearization, 92 process design objective, 35
metal forming and, 5-6 process design steps, 35, 36
nodal point velocities, 91 radial, 314
notation, 101 titanium alloy Ti6242, 234-237
numerical integrations, 111-114 Forward extrusion forging, 13
penalty function method, 110 Friction, 10, 30-33.
plane-strain problems, 131-133 at tool-workpiece interface,
porous metals, 246-249 163-165, 166
procedures, 90-94 Friction coefficient, 206
rectangular element family, 97-100, Friction hill, 138-139
105-106 Friction shear stress, 32
rezoning, 126-129 Frictional stress, 119-120
rigid region, treatment of, 87
ring compression, 160-163 Gaussian elimination, 116-117
sheet-metal forming of general Gaussian quadrature formula,
shapes, 209-210 111-113
stiffness equations, 92 Gear blank forging, 172
three-dimensional brick element, Generalized stress, 66
100-101, 106-107 Geometry updating, 123-125
three-dimensional problems,
276-278 H cross sections, preforms for, 309
time-increment, 123-125 Hasek method, 50
triangular element family, 95-97, Heading of cylindrical bars, 153-159
104-105, 113 Heat transfer
variational approach, 83-84, 86 in porous metals, 259-262
Flange drawing, 194-195, 197 in thermo-viscoplastic analysis,
Flange-hub shapes, axisymmetric 225-227
forging of, 253-256, 257 Hexahedral element, 276
Flashless forging, 170, 171, 172 Hill's general method, 3, 78-83, 88
Flat-face die, 36, 38 Hobbing, 14
Flat-tool compression, 314 Hooke's law, 63
374 Index
Hot compression, of steel cylinder, Metal forming and finite-element
231-234 method, 5-6
Hot-die disk forging, 237, 238, 239 Metal-forming processes, 8-24. See
Hot extrusion, 20 also specific processes
of rods and shapes, 36-39 backward extrusion forging, 14
Hot forming, 28-29, 34-35 brake-bending, 21
preform design, 315-318 classification of, 11-12
Hot nosing, 237-240 closed-die forging, 9, 12, 13
Hot pressing, under plane-strain closed-die forging, without flash, 12
compression, 262-266 coining operation, 13
Huber-Mises criterion, 59 deep drawing, 23
Hydrodynamic conditions, 31 deformation mechanics, 10. See
also Plasticity
Impression, 35-36, 37 description of, 11-12
Infinitesimal deformation theory, drawing, 20
55-56 extrusion process, 19, 20
Infinitesimal plastic strain, 63-64 forward extrusion forging, 13
In-plane deformation processes, friction, 10
192-195 hobbing, 14
Integration points ,111 ironing, 21
Ironing, 21 material variables, 9-10
Ironing press, 46-47 nosing, 15
Isoparametric elements, 96, 98, 99 open-die forging, 15
Isothermal forging, 237, 238, 239. See orbital forging, 16
also Axisymmetric isothermal product properties, 10
forging radial forging of shaft, 16
ring rolling, 18
Jauman derivative, 324 roll bending, 22
roll forming, 22
rolling, 17
Kirchhoff stress, definition of, 324
rotary tube piercing, 18
rubber-diaphragm hydroforming, 24
Lagrange multiplier method, 110 rubber-pad forming, 24
Lagrangian description, 55-57 shear forming from plate, 19
Lagrangian family, 98 sheet-metal, 11, 12
Lagrangian strain, 208 spinning, 23
Levy-Mises equations, 66 systems approach in, 8-10
Limiting drawing ratio, 205 tooling and equipment, 10
Lubrication, basic types of, 30-31 upsetting with flat-heading tool, 17
Lubricity, 32-33 Metal powders. See Porous metals
Methods of analysis, 73-88
Maximum plastic work principle, equations for, 73
64-65 finite-element method, 83-87, 88.
Maxwell-Heuber-Mises criterion, 59 See also Finite-element method
Mechanics of deformation, 10. See Hill's general method, 3, 78-83, 88
also Plasticity for non-steady-state processes,
Metal flow, 27 175-176
in non-steady-state upset forging, 27 for steady-state processes, 174-176
in rolling, 42 upper-bound method, 3, 74-78, 88
in steady-state extrusion, 27 von Mises criterion, 59
 Index 375
Modeling, process, 1-3 stress, 54-58
Multipass bar drawing and extrusion, virtual work-rate principle, 62-63
183-186 viscoplasticity, 70-72
yield criterion, 58-61
Nakajima method, 50 Plastometers, 29
Necking, 47 Plate
Newton-Raphson method, 92, 93, 121 bending of, 141-147, 327-331
Nominal stress, 324 rolling of, 41-45
Nonquadratic yield criterion, 217-220 Porous metals, 244-270
Non-steady-state flow, 27 axisymmetric forging of flange-hub
Nosing, 15 shapes, 253-256, 257
hot, 237-240 axisymmetric forging of pulley
shell, 301-305 blank, 256-259
compaction, 266-270
Open-die forging, 15 discretization, 247-248
Orbital forging, 16 finite-element modeling, 246-249
flow rules, 245-246
Parent element, 97 fully dense materials, 248
Penalty function method, 110 heat transfer in, 259-262
Piola-Kirchhoff stress, 208, 324 hot pressing under plane-strain
Piola-Kirchhoff stress tensor, 57 condition, 262-266
Plane plastic flow, 131. See also numerical procedures, 246-249
Plane-strain problems simple compression, 249-253
Plane-strain compression, hot pressing updating relative density, 248
under, 262-266 volume integration, 248-249
Plane-strain problems, 131-149 yield criterion, 245-246
closed-die forging with flash, Powder compaction, 266-270
133-136 Powder forming, 244. See also Porous
finite-element formulation, 131-133 metals
plate bending, 141-147 Prandtl-Reuss equations, 65-66
sheet rolling, 137-141 Preform design, 298-320
side pressing, 148-149 axially symmetric forging, 309-315
Plane-strain rolling, 305-309 backward tracing, 298-301,
Plastic anisotropy, 190-191 301-305, 305-308, 311-312,
Plastic strain-rate, 63-64 316-317, 318-320. See also
Plasticity, 54-70 Backward tracing
effective strain, 66-68 definition of, 298
effective stress, 66-68 H cross sections, 309
equilibrium equations, 61-62 hot forming, 315-318
extremum principles, 68-70, 71-72 method for design, 298-301
flow rule, 63-66 plane-strain rolling, 305-309
infinitesimal deformation theory, shell nosing, 301-305
55-56 Pressing, hot, 262-266
Lagrangian description, 55—57 Process modeling, 1-3
maximum plastic work principle, Product properties, 10
64-65 Pseudoconcentrations, 334
plastic potential, 63-66 Pulley blank, axisymmetric forging of,
strain, 54-58 256-259
strain-hardening, 66-68 Punch loads/displacements, 144-145,
strain-rate, 54-58 146
376 Index
Punch-stretching, 201-206 Shear stress criterion, 59
Sheet rolling, 137-141
Radial forging, 314 Sheet-metal forming, 47-52, 189-220
of shaft, 16 axisymmetric out-of-plane
Rectangular block compression, deformation, 195-201
278-281 axisymmetric punch-stretching,
Rectangular element family, 97-100 201-206
strain-rate matrix, 105-106 bore expanding, 194-195, 196
Reliability of simulation, 333-334 classification of processes, 11, 12
Rezoning, 126-129 deep-drawing processes, 201-206
Rigid-plastic formulation, 327-334 flange drawing, 194-195, 197
disadvantages of, 330 general shapes, 206-210
solid formulation vs., 327-334 in-plane deformation processes,
Ring compression, 159-165, 331-332, 192-195
333 nonquadratic yield criterion,
axisymmetric isothermal forging, 217-220
159-163 plastic anisotropy, 190-191
friction and, 163-165 square-cup drawing, 210-217
solid formulation, 331-332, 333 Shell nosing, 301-305
Ring rolling, 18 Side pressing, 148-149
Ring test, 33 Simple compression, 249-253,
Rods, 36-39, 45-47 364-369
drawing of, 45-47 of cylinder, 364-369
hot extrusion of, 36-39 of porous metals, 249-253
Roll bending, 22 Simple plastic incremental
Roll forming, 22 deformation. See SPID
Rolling, 17 Simpson's formula, 114
analysis of spread in, 289-292 Simulation of forming processes,
metal flow in, 42 332-334
plane-strain, 305-309 Slab method, 1
plate, 17, 43-44 Slip-line field method, 1-3, 138
ring, 18 Small-strain solid formulation,
of shapes, 44-45 321-323
sheet, 17, 137-141 Smooth entry dies, 38
strip, 42-43, 45 Solid formulation, 4-5, 321-335
Rotary tube piercing, 18 large deformation: incremental
Rubber-diaphragm hydroforming, 24 form, 326-327
Rubber-pad forming, 24 large deformation: rate form,
323-326
Shaft, radial forging of, 16 plate bending, 327-334
Shapes rigid-plastic (flow) solutions vs.,
drawing of, 45-47 327-334
flange-hub, axisymmetric forging of, ring compression, 331-332, 333
253-256, 257 small-strain, 321-323
hot extrusion of, 36-39 three-dimensional analysis with, 334
rolling of, 44-45 SPID (simple plastic incremental
sheet-metal forming of, 206-210 deformation), 338-369
Shear forming from plate, 19 description of major variables,
Shear strength of deforming material, 342-343
32 example solution, 364-369
 Index 377

input and output files, 340 heat transfer analysis, 225-227

input preparations, 340-342 hot nosing, 237-240
program listings, 343-364 hot-die disk forging, 237, 238, 239
program structure, 339-340 isothermal forging, 237, 238, 239
simulation conditions, 364-369 Three-dimensional brick element,
Spike forging, 163, 165 100-101
Spinning, 23 strain-rate matrix, 106-107
Springback angle, 330 Three-dimensional problems, 275-296
Square-cup drawing, 210-217 block compressions, 278-284
Square-ring compression, 284-287 brick element, 100-101, 106-107
Steady-state flow, 27 finite-element formulation, 276-278
Steady-state processes, 174-187 flat-tool forging, 293, 294, 295
applications to process design, hexahedral element, 276
186-187 rolling, analysis of spread in,
bar drawing, 178-183 289-292
bar drawing, multipass, 183-186 simplified elements, 287-289
bar extrusion, 176-178 solid formulation, 334
bar extrusion, multipass, 183-184 square-ring compression, 284-287
method of analysis, 174-176 Time increment, 123-125
Steel cylinder, compression of, Titanium alloy Ti6242, 234-237
229-234 Tooling, 10
Stiffness equations, 120 Tresca criterion (shear stress
assembled, 116-117 criterion), 59
elemental, 108-110 Triangular elements, 95-97, 104-105,
Strain, 54-58 113
effective, 66-68 strain-rate matrix of, 104-105
Strain-hardening, 66-68 True stress, 56, 57, 58
Strain-rate, 54-58 Tubes, drawing of, 45-47
in Cartesian coordinate system, 101
Strain-rate matrix, element, 101-107 Upper-bound method, 3, 74-78, 88
Strain-rate vector, 103, 153 Upsetting, with flat-heading tool, 17
Stress, 54-58
effective, 66-68
Virtual work-rate principle, 62-63
Stress measures, 324
Viscoplasticity, 70-72. See also
Strip, rolling of, 42-43, 45
Thermo-viscoplastic analysis
Swift test, 205 definition of, 222
Visioplasticity method, 3
Temperature in metal forming, 28, 29,
Volume strain-rate matrix, 107
33-35. See also Thermo-
von Mises criterion, 59
viscoplastic analysis
Thermal conductivity. See Heat
transfer Wedge-shaped blocks, 281-284
Thermo-viscoplastic analysis, 222-242 Wheels, automobile, 51-52
applications of, 229-240 Wire, drawing of, 45-47
compression of cylinder, 223-225, Wrinkling, 47, 50
229-234
computational procedures for, Yield criterion
227-229 nonquadratic, 217-220
forging titanium alloy Ti6242, plasticity, 58-61
234-237 porous metals, 245-246