4-4 STATIC AND DYNAMIC ANALYSIS

Of greater significance, however, is the comparison of the axial stress

distribution summarized in Figure 4.2, using both the force and displacement
methods of analysis.

5
4 FORCE METHOD 
     s
     s 3
     e
     r
      t 2 DISPLACEMENT 
      S
METHOD 
1
0
0 20 40 60 80
Distance "s"

 Figure 4.2 Comparison of Stresses for Force and Displacement Method 

At the end of the tampered rod, the displacement method produces only 33
percent of the maximum stress of 5.0 ksi. Of course, if a fine mesh is used, the
results will be closer. Also, if higher order elements are used, with interior points,
the displacement method results can be improved significantly. Nevertheless, this
example clearly illustrates that  the force approach should be used to predict the
 behavior of one-dimensional elements.

4.3 TWO-DIMENSIONAL FRAME ELEMENT

A non-prismatic frame element with axial, bending and shearing deformations

will be developed to illustrate the power of the force method. The displacement
method has the ability to calculate a stiffness matrix of any element directly in
terms of all displacement degrees-of-freedom associated with the elements; and
the element automatically includes the rigid body displacement modes of the
element. The force method only allows for the development of the element
flexibility matrix in terms of displacements relative to a stable support system.
ONE DIMENSIONAL ELEMENTS 4-5

The general frame element is composed of any number of non-prismatic frame

segments. Each segment can have independent axial, shear or bending properties.
properties.
Therefore, at the ends of the element, rigid bending segments are possible, with
or without shearing and axial deformations. Hence, it is possible to approximate
the behavior of the finite connection area. A typical frame member is shown in
Figure 4.3.

θ , M 
Deformed Position

s v, V 

∆, P
S i Semi Rigid
S i +1 Segment

 L

 Figure 4.3 Arbitrary, Two-Dimensional Frame Element

The relative displacements are the axial displacement ∆ , vertical displacement

e nd rotation θ . The corresponding loads are the axial load P , vertical
v , and the end
load V  , and the end moment  M . At a typical cross-section at location s , the
force-deformation
force-deformation relationship is:

 1 
 0 0 
 ε (s)  E(s) A(s)   P(s) 
d(s) = C(s) f(s) , or γ (s) =  0 1
0

V (s)
 (4.8)
 G(s) A (s)  
ψ (s)  s  
   1   M(s)
 0 0
E(s) I (s)

All cross-sectional properties, including the effective shear area  As , can vary
within each segment of the frame element.
4-6 STATIC AND DYNAMIC ANALYSIS

The section forces within a typical segment at location s can be expressed

directly from statics in terms of the arbitrary end forces R . Or:

 P(s)  1 0 0  P 
f( s) = P( s) R , or
 V (s)  = 0 1 0  V  (4.9)
    
 M(s) 0 L − s 1  M

The 3 x 3 flexibility matrix as defined by the force method is calculated from:

L I  MAX Si + 1

∫ 
F = P(s) T C(s)P(s) ds =
0
∑ ∫ P(s)
i Si
T
C( s)P(s) ds (4.10)

It is of interest to note that because of the discontinuity of the properties of the

segments, each segment produces a separate 3 by 3 flexibility
flexibility matrix. Therefore,
Equation (4.10) can be written in the following form:

I  MAX  Si + 1

F= ∑Fi
(i)
, where F ( i ) ∫ 
= P(s) T C(s)P(s) ds
Si
(4.11)

Equation (4.11) can be termed the  direct flexibility

flexibil ity method , because the segment
flexibility terms are directly added. It should be pointed out that if any cross-
sectional stiffness properties are infinite, as defined in Equation (4.9), the
contribution
contribution to the flexibility
flexibility at the end of the element is zero.

The C and P matrices contain a significant number of zero terms. Therefore, the
element flexibility matrix for a straight member contains only four independent
terms, which are illustrated by:

 FP 0 0 
 
F =  0 FVV  FVM  , (4.12)
 0 FVM F MM 

It can easily be shown that the individual flexibility terms are given by the
following simple equations:
ONE DIMENSIONAL ELEMENTS 4-7

I  MAX SI + 1
1
FP = ∑ ∫ 
i Si
E(s) A(s)
ds (4.13a)

I  MAX SI + 1
 (L − s)2 1 
FVV  = ∑ ∫  i 
+
 E(s) I (s) G(s) A (s)  ds
s 
(4.13b)
Si

I  MAX SI + 1
FVM = ∑ ∫  E((Ls)−I s()s) ds
i Si
(4.13c)

I  MAX SI +1
1
F MM = ∑ ∫  i Si
E(s) I (s)
ds (4.13d)

For frame segments with constant or linear variation of element properties, those
equations can be evaluated in closed form. For the case of more complex
segment properties, numerical integration may be required. For a prismatic
element without rigid end offsets, those flexibility constants are well-known and
reduce to:

L
FP = (4.14a)
EA

L3 L
FVV  = + (4.14b)
3EI  GAs

L2
FVM = (4.14c)
2EI 

L
F MM = (4.14d)
EI 

5
For rectangular cross-sections, the shear area is  As =  A .
6

One can easily consider loading within the segment by calculating the additional
relative displacements at the end of the element using simple virtual work 
4-8 STATIC AND DYNAMIC ANALYSIS

methods. For this more general case, the total relative displacement will be of the
following form:

∆   FP 0 0   P  ∆ L 
v  =  0 F F   V  +  v  or, v = FR + v L (4.15)
   VV  VM    L 
 θ   0 FVM F MM   M  θ L 

The displacements caused by span loading are designated by v L . Equation (4.15)

can be rewritten in terms of the element stiffness as:

r = Kv - Kv L = Kv - rL (4.16)

The element stiffness is the inverse of the element flexibility, K = F -1 , and the
fixed-end forces caused by span loading are rL = Kv L . Within a computer
program, those equations are evaluated numerically for each element; therefore,
it is not necessary to develop the element stiffness in closed form.

4.4 THREE-DIMENSIONAL
THREE-DIMENSIONAL FRAME ELEMENT

The development of the three-dimensional frame element stiffness is a simple

extension of the equations presented for the two-dimensional element. Bending
and shearing deformations can be included in the normal direction using the same
equations. In addition, it is apparent that the uncoupled torsional flexibility is
given by:

I  MAX SI + 1
1
FT  = ∑ ∫ 
i G(s) J (s)
ds (4.17)
Si

The torsional stiffness term, G(s) J ( s) , can be difficult to calculate for many
cross-sections. The use of a finite element mesh may be necessary for complex
sections.

An arbitrary, three-dimensional frame element is shown in Figure 4.4. Note that

only the six forces at the J end are shown. The six relative displacements at node
J have the same positive sign convention as the forces at node J.
ONE DIMENSIONAL ELEMENTS 4-9

T
z M2

P
V2

J 1
2

V3

3
M3 LOCAL
SYSTEM 
I
y

 Figure 4.4 Member Forces in Local Reference Systems

The 6 by 6 stiffness matrix is formed in the local 1-2-3 coordinate system, as

shown in Figure 4.4. The order of the forces and relative deformations are given
by:

 P  k 11 0 0 0 0 0  ∆ 
 V    0 k 0 0 0 k 26   v 2 
  
2 22  
 V 3   0 0 k 33 0 k 35 0 v3 
  0 =   or, f J = k J d J (4.18)
 T    0 0 k 44 0 0  φ T  
 M 2   0 0 k 53 0 k 55 0 θ 2 
    
 3  
 M 0 k 62 0 0 0 k   θ 3 
66 

The bold terms indicate the shear and bending contributions in the 1-2 plane. For
a curved member in three dimensions, the 6 by 6 k matrix may be full without
the existence of any zero terms. Note that the 6 by 6 stiffness matrix formed in
the local system does not have the six rigid body modes.

The forces acting at node I are not independent and can be expressed in terms of 
the forces acting at node J by the application of the basic equations of statics.
Therefore:
4-10 STATIC AND DYNAMIC ANALYSIS

−1 0 0 0 0 0  P
 P   1  
 V    0 − 1 0 0 0   V  
  
2 L
  2
 V 3   0 0 − 1 0 1 0   V 3 
 T   =  L   T   or, fI = bIT J f J (4.19)
   0 0 0 − 1 0 0  
 M 2     M2 
   0 0 L 0 − 1 0  

 3  I  0 L 0 0 0 − 1  M 3  J
 M

The twelve forces at both ends of the beam can now be expressed in terms of the
six forces at the J end of the beam by the following submatrix equations:

fI   bIT J 
f  =  f J or, fI J = b T f J (4.20)
  J   I 

Also, from the relationship between the equations of statics and compatibility,
the following displacement transformation equation exists:

dI = b dI J (4.21)

Therefore, the 12 by 12 frame element stiffness, k I J , with respect to the local
1-2-3 reference system, is:

k I J = b T k J b (4.22)

Hence, the force-displacement equations in the local 1-2-3 system can be written
as:

fI J = k I J uI J (4.23)

To use the direct stiffness formulation, it is necessary to transform the local

element stiffness into the global x-y-z reference system. The global 12 by 12
stiffness matrix must be formed with respect to the node forces shown in Figure
4.5. All twelve node forces R and twelve node displacements u have the same
sign convention.