.....-- .

BRIDGE STRUDL MANUAL

''::.:o:.:.--:.~;:

..... &Ai, .au;.

a-

November 1973

.a+

APPENDIX B
STIFFNESS OR DISPLACEMENT METHOD

CONTENTS
Section

Page

B.l

Introduction

B-2

B.2

Basic Displacement Approach Using

Example Problem 2.3

B-8

Direct Stiffness Approach Using

Example Problem 2.3

B-14

B.3

B-1

'

APPENDIX B
STIFFNESS OR DISPLACEMENT METHOD
B.l

Introduction:
I.

Brief Discussion of Force or Flexibility Method

Indeterminate systems comprise the large majority

of structures to be analyzed and designed and hence the
solution process must satisfy the conditions of compatibil
ity and the material stress-strain behavior. Traditional
methods of structural analysis employing the concept of
redundancies and "consistent deformations" have not proven
to be as simple and direct in application as the "stiffness"
or "displacement" approach to be treated here and also used
in the STRUDL program. The traditional method involving
redundancies has been formalized into a matrix approach and
is now referred to as the "force" method.
a.
Propped Cantilever Example: The "force method"
of structural analysis (often referred to as the "flexibility
method") is probably most familiar to us for the solution of
statically indeterminate structures. The propped cantilever
beam of Figure B.la provides a simple example of the use of
the force method.

rnA

ol

.(

t~

;;t
L

RA

1
B

Fig. B.la
A concentrated load, P, is acting at a distance aL from the
left support. This load produces the reactions RA, MA and PB
as shown. Since we have only two equations of equilibrium,
LFv= 0 and l:M =0 , and three unknown reactions, this beam
is considered to be indeterminate to the first degree. To
gain an additional equation we consider the deflections of
the structure. The traditional way to approach this problem
is to remove one of the redundant reactions, in this case ~
B-2

and determine the deflection, 8 0 at B on the statically

determinate cantilever due to the external load P, Figure
B.lb. Since our actual structure does not have a vertical
deflection at B the redundant reaction, ~, must be of such

Fig. B.lb

Fig. B.lc

a magnitude that it pushes the beam of Figure B.lb upward

with a displacement equal to Sc. If we apply a unit value
of the redundant ~ to the cantilever shown in Figure B.lc,
we will have a deflection at B upward equal to 8o . Therefore
we can write

This is our compatibility equation saying that the deflection

at B is zero. Here 891 is the vertical deflection at B due
to a unit load at B. We solve Eq. 1 for ~

=
Having

(2)

allows us to determine MA and RA by statics.

b. Four Span Beam Example: For a beam with a

larger number of redundancies we could proceed in a very
similar manner. For example, consider the four span beam of
Figure B.ld. In this case we can consider R1 , R2 , R3 and R4
as the redundants, leaving us with the cantilever beam of
Figure B.le.

30

40
Fig. B.ld

Fig. B.le
B-3

The applied loads produce deflections 8 10 ,8 20'8 30

and 8 4 o.
As before, these deflections do not represent the true state
of our structure so we must consider that the redundants push
upward just enough to eliminate these displacements.
In this
instance we shall arrive at four compatibility conditions.
For example, applying a unit load at support 1 yields deflec
tions 811 t 821 '831 and841 (see Figure B.lf). Similarly, a
unit load at point 2 yields 8 12, 8 22'8 32, 8 42.

Fig. B.lf

Fig. 8.1g

We could continue applying the unit load at each point and

determining the deflections. Our compatibility equations
become

&1o

d20

~30 +
b40 +

R1 &11
R1 S21
R1 b31
Rl 041

+
+
+
+

R2 ~2
R2 ~2

R3 ~13

R4 s14

R3 523
R3 '33
R3 s43

R4 S"24
R4 634
R4 S44

~32 +

R2
R2 S42

+
+

=0
=0
=0
=0

( 3)

Note that 8ij -is the deflection at support i due to a unit

load at support j. The solution of these four equations gives
values for the redundants R1 , R2 , R3 and R4 One thing that
we can observe is that in order to determine the redundants we
must calculate the deflections at all of the redundant points
for all positions of the unit load-.II. Brief Discussion of the Stiffness or Displacement
Methods
The previous discussion has dealt with the flexibil
ity or force method of analysis. vle can handle the same
problems by considering the stiffness or displacement method
of analysis.
In this case we take the unknown displacements
of the structure as the redundants.
a. Propped Cantilever Example:
propped cantilever beam of Figure B.lh.
B-4

Again consider the

Fig. B.lh

F.ig B.lc

In this case the only unknown displacement is the rotation,

Ba , at end B. We then say that this structure only has one
degree of freedom.
In order to eliminate this unknown dis
placement we clamp the end. The applied loads then produce
fixed end moments MA and ~ as shovm in Figure B .li. How
ever, we know that this is not the actual condition of our
structure. The redundant rotation, Be , produces a moment
of magnitude equal to
but of oppos~te direction. We can

ME

consider the effect of this rotation by considering the effect

of a unit rotation at end B (Figure B.lj).

-- ____
.....

~l

/~

Fig. B.lj
Then our compatibility condition becomes

Me

mss

eB

=~

eB

=o

(4)

ms

where

~Bis

(3)

the moment at B due to a unit rotation at B.

Having BB , we can determine all other moments and reactions.

For example, the moment at A, MA' is given by

MA

= MAFEM +, mAB

(5)

where mAB is the moment at A due to a unit rotation at B.

B-5

b. Four Span Beam Example: This may be a strange

way to look at this problem because it is normally more
difficult to calculate the reaction caused by a unit dis-.
placement than to calculate the displacement caused by a
unit reaction. Hm.,ever, we shall soon see that there is an
advantage in looking at the problem from this point of view.
Consider again the continuous beam of Figure B.lk. We see
that there are fOUr Unknown rotatiOnS, 81 I 8 2 I 8 3 1 8 4

Fig. B.lk
We begin by fixing all supports against rotation and deter
mine the FEM's (Figure B.lL).

___,
8' -'-""' ',' c S"f<, lD /i
f -,, A
,/~-,'
--..I. ---...
..!! . . . ____
M1~ M18 -M2S M~o - M~o M 30 Mfo
~

".~--

-/~

Fig. B.. lL
Now we apply a uni't rotation at each support. For example,
a unit rotation at support 1 (Figure B.lrn) produces moments
MLl at the left support, m11 and m21 at the first and second
supports respectively.

Fig. B.lm

Fig. B.ln

A unit rotation at support 2 yields (Figure B.ln) moments

m12 , m22 and m32
at supports 1, 2 and 3, respectively. We
can \.,rite the compatibility equation at support 1.

B-6

'Ne can cont:.J.nue to apply the unit rotation and get three
additional compatibility equations, for example, at joint 2

o = Me

20

M8

20

21

22

23 93

(7)

Solving this system of equations gives values for 81 82 83

and 8 4 The thing to note is that these equations 'only
involve the effects produced by members adjacent to the joint
in question.
In other -v1ords we do not have to determine
effects on the structure due to rotations at distant points.
This allows an efficient manner of storing the problem in
the computer and allows the computer time to be reduced.
Furthermore, the method is applicable to determinate and
indeterminate systems with equal ease and the degree of
indeterminancy need not even be determined.
The preceding discussion of stiffness method was
presented to give an overview of the method. We shall next
consider a more detailed application of the stiffness or dis
placement approach.

B-7

B.2

Basic Displacement Approach Using Example Problem 2.3,

Indeterminate Truss

~-~~-~P4_a

tR~t~t .

8'

Fs. Us

tR3,A3 '

. JOINT LOADS
AND DISPLACEMENTS

GEOMETRY

BAR FffiCES
AND ELQNGATIONS

fig. B.2a

Note that the external applied loads, P, have a

one-to-one correspondence with the external joint displace
ments, X, and the internal bar forces, F, have a one-to-one
correspondence with the member elongations, u. Also note
the one-to-one correspondence between unknown reaction com
ponents, R and known support displacements ~ (not necessarily
zero) .
a. Equilibrium Matrix:
Rewriting the equilibrium
matrix, shown on Page A-10, to include the additional bar gives

3 ..

0._

-4. -10.
o.. -4. =
0.

-10

o.
0:

R11 R12
R21 . R22
R31

R32

0 -1 0 0
1. 0 0 0
1 0 .8
0
0 0 0 0
0 0 1 .6.

-1 0 0
=- 0 0 "()
0 0 .1

F11

F12l

F21

F22

.o

F31

F32

.8

F41

F42

Fs1

F;;2

F51

Fs2

-:-8

0
0
1

.6

0 0
-. 8 -1 0
0 0 -.6

-.6

= [A) ( F}

(8)

F11. F12
F21 F22
F31 F32
F41 F42
Fs1 Fs2
Fs1 F52

B-8

tP}

{Rj

[AR]

[F)

b. Compatibility Matrix: The equilibrium matrix

is no longer square (A 5 x 6 } and hence cannot be inverted.

The degree of indeterminacy is NF - NP = 6 - 5 = 1

Therefore, we must apply the stress-strain relationships and

the conditions of compatibility.

The stress-strain assumption for axially loaded

members is simply

cr

= Eex

or Fx
Ax

= E Ux

or

(9)

One such equation can be written for each bar, hence

tF) 6x2

::.

[5] 6x6 {u

1 6x2

EA1/L1
EA2/L2
tFl

ZEROS

EA3/L3
EA41L4

EA 5!L 5
EA6

ZEROS

/J

u11

u12

u21

u22

u31

u32

u41

u42

u51

u52

us1

u62

(10)

Applying compatibility conditions to a truss means

simply that the member elongations !u~ must be consistent

with the joint displacements [xJ and t~} .

u1 = b11x1 + b12x2 +
. u2 = b21x1

b22x2

+ b15x5 + b16~1 +
+

b25x5

b36 ~1

b46 ~1

us = b51x1

+ b56~1

b62x2

+ .. +

bssxs

b1a ~3

+ b26~1

u3 = b31xl
u4 .= b41x1
us = bslxl

b66 ~1

)(1 1)

bsa D-3

(12)
B-9

__ \..._,

--

What we are saying is that bar elongations are some linear

combination of the external joint displacements.
To obtain the coefficients of [ B] defining the
compatibility matrix, we may apply a unit displacement in
the direction of each of the external joint displacements.
For example,

xl

=1'

x2

ul

t4

= bll

u2 = b21 u3 = b31 etc

ul = bll = 0 (small deflections)

u2 = bzl = -1.0

us = bs1 = -0.8

u3

= u4 = u5 = 0

Fig. 8.2b
as another example set

1;

x1. = x2 = x3

x4

A
A
=ul
=u2

=~3

u2 = b25 = 0 (small deflections)

u3 = b35 = +1.
u4 = b45 = +O.S
Fig. 8.2c

ul = u5 = us

B-10

= o.

= 0

Each column may be determined successively to yield

-1.
0
[8]=
0
0
-.8

1.
0
0

0
0
.6

1.
0
.8

0
0

0
0
0
0

.6

1.
.8

-1.
0
0
0
0
0
[BRJ= -. 6 -. 8

0
0

0
0

-1.
0

(13)

- 1. 0
0 -.6

(we've just done columns 1 and 5)

c. Relationship of Compatibility and Equilibrium
Matrices:
It is extremely interesting and significant to
note at this time the transpose relationship between the
equilibrium and the compatibility matrices or

[8]

[8R]

= [A]T
= [ARJT

(14)

This relationship always holds for linearly elastic struc

tures and can be proved by the principle of virtual work.
d.

System Stiffness Matrix:

{P) = [A] {F)

{R} = [~] {F}

{F) = [S] {u}
{u} = [A]T {xJ

+ [ARJT {~}

In summary

Equilibrium

(15)

Stress-Strain

(16)

Compatibili_ty

(17)

Substitute (17) into (16) to obtain

(18)
B-11

Then substitute 18 into 15

(P}

= [ASAT J

{x}

{Pj - [ASA~] {a)

[ASA~]

= [K]

fa}

(19)

(xj

(PJ -[ASA~] (~J

(20)

{R 1J = [ARSAT]

fxJ + [ARSA~ J

fa)

(21)

{x}

= [K]-1

If all support displacements are zero the basic solution

process for the example is

(x}5x2

= [ASATJ5!s

(F] 6x2 = [SAT]

(Plsx2

= [KJ5!s

{PJsx2

6x5 fxlsx2

EA1

(22)

(2.3)

Ll
-EA2

EA 2

o.

EA3

L3

[SAT]=

.8EA4

L4
0

(24)

.6EA4

L4
0

EA5

Ls
-.. 8EA6

Ls

.6EA6 0 .8EA6
L6
Ls

B-12

and

c~~-6~) L2

Ls

A
Ls

-. 48 ..__

IK1

~ (E) .

As
Ls
...

A2
---.

.48-

({I-+

.36

L2

~:)

-.48~--

Ls

. 0
6

_.

.0

A4

.48-L'
. 4

.____..., ___ .....,.....

-~

.48 ~
L
6

____

___

'

These two matrices plus the load matrix {PJ are what is

required to solve for the displacements and forces in the

structural system under investigation. The method is

generally referred to as the displacement method because

displacements are the primary unknown quantities. Also note
the symmetrical condition of the stiffness matrix. This is
proved by the reciprocity theorem.

B-13

(25)

B.3 "Direct Stiffness" Approach Using Example Problem 2.3,

Indeterminate Truss
Direct stiffness simply implies that one is going
to obtain the stiffness matrix [~ without generating the [A]
[s] , and [B] matrices and then performing the matrix multi
plication operations. This method is much more efficient
computationally and requires considerably less effort in the
preparation of data inpu~.
a. Discussion of the development of the system
stiffness matrix directly from physical considerations.
To motivate the development consider the indeter
minate truss just investigated and write the basic stiffness
equations as

Pl
p2
p3

= K11X1
= K21X1
= K31X1

p4 = K41X1
p = K51X1
5

K12X2

K13X3

K14X4

K1sXs

K22X2 .K23X3

K24X4

K2sXs

K32X2

K33X3

K34X4

K35X5

K42X2

K43X3

K44X4

+ K45X~

K52X2

K53X3

Ks4X4

(26)

KssXs

Again these coefficients may be determined by

defining a set of values for the independent variables (x] ,
in order to isolate one column of the matrix. For example,
if x1 = 1~ x2 = x3 = x4
x5 = o, then P 1 = K11 , P 2 = K21 ,
P 3 = K31 , P 4 = K41 , and P 5 = K51 .
Physically this means that

for a given state of displacement, what are the req~ired

applied loads to produce this state? Therefore, the stiffness
coefficient K.. is defined to be the load at coordinate i given
l.J
a unit displacement at coordinate j, all other displacements

equal to zero.

For this state of

displacements the
member elongations
are:

ul

= 0;

u2 = -1; u3

u4. = 0; us

Fig. 8.3a
B-14

= 0;

= 0;
u6 =-o.a

Hence the associated bar forces are

= -0.8 EAs
Ls

Then consider the equilibrium of the joints

Fig. B.3b
EA2

Kn+(-L

EA

)+0.8(-.8--2)=0

Ls

EA

K21 - o-.o.s <-0.8~> = o.

EA

K31 -(-L 2 ) -0.8 (0)

2

=0

( 28)

K41 -0 -0.8 (-0.8 EA2) = 0

L2

K51 -o -o.s <O> = o

These coefficients are the same as those obtained in the

first column of the [K] matrix when the triple matrix multi
plication was employed. If a similar operation is employed
for each of the external displacement coordinates the
remaining four columns of the stiffness matrix could be
obtained and would agree with those obtained previously.
----~

.... -----....

Using this concept to develop the stiffness matrix

indicates the composition of the individual terms and also
clearly identifies which members of the system will contribute
to the individual coefficients. In particular, any given
member will only contribute to those coefficients associated
with the external coordinates of the ends of the member. The
coefficients Kii will consist of contributions from each member
framing into the joint associated with coordinate i.
In more
general terms each member contributes to the stiffness of the
joints into which they frame. This suggests that the stiff
ness matrix could be generated from the stiffness properties
of the component parts or as the summation of the element
stiffness matrices.

B-15

27

First take an individual truss bar, subjected to

an axial force, F.

GEOMETRY
Fig. B.3c.
EQUILIBRIUM

pl
p2
p3
p4

= -Fi
= -F
= +Fl.
= +Fl.
.~

STRESS-STRAIN

COMPATIBILITY

cos o<:i
sino< i
Fl.

cos~

l.

EAl.
= L
l.

sin ex i

Now the element stiffness matrix [EK] can be deter

mined as the product of the element equilibrium, stress
strain, and compatibility matrices, or

-cos o<i
. [EK]

-sine< i
cosoe i

!-EA~l

LLi

J [-cos oci -sin o<i

since:: i

B-16

coso<:i sin~i] (29)

- . ..

cos 2cc:. i

,.

l
cos oc::. sin oc:.

cos:. ; sin oci

l.
2

-cos oc.
l.

-cosoe. sin
l.

oe:..l.

-cos oc. sinoc. -sin oc:.

l.

cos2o<

l.

l.

l.

l.

oC

l.

cos

cos oc. since.

l.

l.

oC.

l.

',

--~r--~.

. :

Ele~ent'\

-cos ex:. sinoc.

l.

-cos oc: sin ex:.

2

2
-cos oc::

, '-.";,

- sin2ex:

-sin

l.

l.

Matrix

l.

In the sketch shown, take direction from the initial

to the terminal end of the member according to the arrow direc
tion and calculate the element stiffness mat~ices.

-. -----

.-..

~1

x1

x2

R2

(EA1) R1
Ll p1
..,.,2

~EA2)
L2

p2 o
p3

1 0

Ps 0

Ps Xs

CP

3 .x3

(!)
0
0

t:: ~3

P4,x4

Fig B.3d
~2

p4

(EA3)" R3

-1

p3

Ps

-1

l3

.,~)
l4

R2
R1
p3
Ps

~1

48

-.64
-.48 .36
-.64 -.48
-.48 .- .36

x3
-.64

Xs
-.48

-.48 \ -.36
-.64

.48

.48

.36

..}:l--.
'''

c. Summation of the element stiffness matrices

to form the system stiffness matrix.

~2

..,

Stiffness

since.

sin~ i

P2,x2
lli:._xl

~--.

The system stiffness matrix is now obtained by

summing the element stiffness matrices. The addition is
done by summing the coefficients having identical row and
column identifications.

(~
L2

. 5-i ~'

t;)

-.41~

Le

1&36~)

Le

I\Ll

-~

1c~
L2

L2

-.64~

-48~

La

Ls

, ranspose

.64~)
L4

.48~

.0

L4

-.48~

_Al

Ll

-.64~
L4

-48~ -.36~

La

-.36 A4

-.48~

T4

L4

Ls

r-

(~ .64~)

Ls

.48~
L6

Ls

-~
L3

B-18

- __..).

c~
L
3

'Al

~Ll

.3ss
L4 1

+.36~)

L4

. 48 -A4

L4

(64 A4 +As)
L4
Ls

of

L5

------+------

.36
(~
L3

As)

1:6

~-

The matrix equations may be written as

1 = [K]

(P

1+

(X

[ KER ]

{A}

( R} = [KER] T ( x J + [ KRR ]
Since the

{A J

{X} matrix contains the only ~nknowns

Up to this point we have attempted to develop the

fundamental basis upon which the displacement method of
structural analysis is based. The development was evolved
in the following order.
1. Matrices and elementary matrix operations.
2. Simple examples illustrating the use of matrices
in the solution of statically determinant prob
lems from equilibrium concepts.
3. The basic displacement method approach involving
the determination of equilibrium, stress-strain,
and compatibility matrices and the combination
of these to determine the system stiffness matrix.
4. Development of the system stiffness matrix
directly from physical considerations and using
this concept to help rationalize the develop
ment of the stiffness matrix as the su~~ation
of the element stiffnesses of the component parts.
Thus, the most important feature in developing any
computer program utilizing the displacement method is in the
proper formulation of element stiffness matrices. lve have
developed the element stiffness matrix for an axial force
member. STRUDL develops additional element stiffness ma
trices for members comprising plane frame, plane grid, space
truss and space frame systems.

B-19