MSCE 504

Matrix Structural Analysis

Lesson Module 5
DIRECT STIFFNESS METHOD

In the previous modules, we form the stiffness matrices of the structure using the general formula

(K) = (A) x (S) x (AT)

Thus, the size of (A) and (S) are proportionate to the size of the structure which are invariably large. The
direct stiffness method considers the members as individual structures assembled or connected
together to form the total structure. Thus for member i,

(ki) = (Ai) x (Si) x (Ai)T

Once the individual member matrices (k) are formed, they are superimposed to form the Global
Stiffness Matrix:

(K) = ∑(ki) = ∑ ((Ai) x (Si) x (Ai)T)

COORDINATE SYSTEM AND SIGN CONVENTIONS:

Two sets of coordinate systems will be used. One is the LOCAL COORDINATE which is oriented along the
axis of a given member and used for the member alone. The other is the GLOBAL COORDINATE which is
arbitrarily oriented at the convenience of the problem solver to describe the positions and
displacements of all members in the structure.
THE MEMBER STIFFNESS IN LOCAL COORDINATES
As a convention, the local x axis runs parallel to the length of the member while y or z axis are the minor
and major axis of the member cross section:
FRAME ANALYSIS

Frame – an assemblage of beam elements not restricted to lying along a straight line.

A. Stairway Support Frame

1 2
P
L2

3 4

L1 L2 L1

Determine the deflection at node 2 and bending moments at nodes 1 and 2.

u1 v1 1 u 2 v2 2 u3 v3 3 u 4 v4 4
0 sym  0 sym 
 12   12 
0 3  0 3 
 L1   L1 
 6 4   6 4 
0  L 2 L1  0  L 2 L1 
K 1 2  EI  1  K 3 4  EI  1 
 0 0 0 0   0 0 0 0 
 12 6 12   12 6 12 
0  L 3 L1
2
0
L1
3  0  L 3 L1
2
0
L1
3 
 1   1 
0  6 2
0
6 4 0  6 2
0
6 4
 2
L1 2
L1   2
L1 2
L1 
 L1 L1  L1 L1
 u2 v2 2 u3 v3 3
6
sym
L2 3
6 6
L2 3 L2 3
3 2 3 2 4
L2 2 L2 2 L2
K 2-3 = EI 6 6 3 2 6
L2 3 L2 3 L2 2 L2 3
6 6 3 2 6 6
L2 3 L2 3 L2 2 L2 3 L2 3
3 2 3 2 2 3 2 3 2 4
L2 2 L2 2 L2 L2 2 L2 2 L2

Boundary conditions are as follows

u1  v1  1  u 4  v 4   4  0

Superimposing the stiffness matrices for the three individual elements and immediately imposing
boundary conditions

6
sym
L2 3
6 12 6
X2  3 +
Y  L2 3 L1 L2 3 u2
 2  3 2 6 3 2 4 4 v2
M 2  +
   EI L2 2 2
2
L1 L2 2 L1 L2
 X3  6 6 3 2 6 u3
 Y3 
  L2 3 L2 3 L2 2 L2 3 v3
 M 3  6 6 3 2 6 12 6 3
3 +
L2 3 L2 3 L2 2 L2 3 L1 L2 3
3 2 3 2 2 3 2 6 3 2 4 4
+ +
L2 2 L2 2 L2 L2 2 L1 2 L2 2 L1 L2

Impose symmetry conditions

u 2  u 3 v 2  v3  2   3
 12 EI
X2  3
u2 X 3
L2

 12 12 6 
Y2   3 u 2  3 v 2  2  2  EI
 L2 L1 L1 

 12 12 6 
Y3    3 u 2  3 v 2  2  2  EI
 L2 L1 L1 

 6 2 6  4 2  
M 2    2 u 2  2 v 2     2  EI
 L2 L1  L1 L2  

 6 2 6  4 2  
M 3    2 u 2  2 v 2     2  EI
 L2 L1  L1 L2  

X2  X3  0 therefore u2  0

 12 6 
Y2  Y3   3 v 2  2  2  EI , Y2  Y3   P
 L1 L1 

 6  4 2  
M 2   M 3   2 v 2     2  EI , M2  M3  0
 L1  L1 L2  

3
P PL1
v 2  45 
EI 4.8 EI
2
P PL1
2  9 
EI 4 EI

Having determined v2 and θ2, the internal nodal moments can be immediately calculated by using
the solution for the displacements and the individual member stiffness equation.
  6  2 
M 1   2 v 2    2  EI  4.5 P
 L1  L1  

 6  4 
M 2   2 v 2    2  EI  1.5 P
 L1  L1  

B. Balcony Support Frame z,Mb,θ

y,v L
1 2

L
3
Mt, x

5 4
L P

Determine the nodal displacements (v,θ,)and nodal forces (Y,Mb,Mt) at the joints

Y3   P , EI and GJ are taken to be constant throughout. Under this loading the frame displaces out
of its original plane. This produces torsional moments as well as bending in supported members 1-2
and 4-5

The total stiffness matrix prior to imposing boundary or symmetry conditions is of order 15x15.

Imposing boundary conditions vi  i  i  0, i  1,5 reduces this to order 9x9.

Imposing symmetry conditions v 2  v 4 2  3  4  2   4  3 0 reduces this to

order 4x4

The displacements which remain are v 2 ,  2 ,  2 , v3

 The stiffness matrix for members 1-2 and 5-4 which are both subjected to combined bending and
torsion is given by

v1 1 1 v2 2 2
 12 EI 
 L3 SYM 
 6 EI 4 EI 
 2 
 L L 
 0 0
GJ 
 L 
K 1 2  
12 EI 6 EI 12 EI
 3 0 
 L L2 L3 
  6 EI 2 EI
0
6 EI 4 EI 
 L2 L L2 L 
 GJ GJ 
 0 0  0 0 
 L L 

v5 5 5 v4 4 4
 12 EI 
 L3 SYM 
 6 EI 4 EI 
 2 
 L L 
 0 0
GJ 
 L 
K 5 4  
12 EI 6 EI 12 EI
 3 0 
 L L2 L3 
  6 EI 2 EI
0
6 EI 4 EI 
 L2 L L2 L 
 GJ GJ 
 0 0  0 0 
 L L 
 v2 2 2 v3 3 3
12 EI 2 
 L3  SYM 
 6 EI 4 EI 2 GJ 2 
 2     
 L L L 
 6 EI  4 EI GJ  4 EI 2 C 2 
 2        
L  L L  L L
K 2 3   12 EI 6 EI 6 EI 12 EI

   2  
 L3 L2 L L3 
 6 EI 2 EI 2 GJ 2  2 EI GJ  6 EI 4 EI 2 GJ 2 
 2            
 L L  2
 L L L L L L 
 6 EI  2 EI GJ  2 EI 2 GJ 2 6 EI  4 EI GJ  4 EI 2 GJ 2 
 L2   2       2        
  L L  L L L  L L  L L 

v3 3 3 v4 4 4
12 EI 2 
 L3  SYM 
 6 EI 4 EI 2 GJ 2 
 2     
 L L L 
 6 EI  4 EI GJ  4 EI 2 C 2 
 2        
L  L L  L L
K 3 4   12 EI 6 EI 6 EI 12 EI

   2  
 L3 L2 L L3 
 6 EI 2 EI 2 GJ 2  2 EI GJ  6 EI 4 EI 2 GJ 2 
 2            
 L L L  L L  L2 L L 
 6 EI  2 EI GJ  2 EI 2 GJ 2 6 EI  4 EI GJ  4 EI 2 GJ 2 
 L2   2       2        
  L L  L L L  L L  L L 

Superimpose the stiffness matrices for the four members

 24 EI 
 L3 SYM 
 6 EI 4 EI  GJ 
 
 Y2   L2 L  v 2 
M   6 EI 4 EI  GJ   2 
 b2    2 0  
M t2   L L
  2 
   12 EI 0
6 EI 24 EI
  v 
 3   L3
Y L2 L3  3 
 
M b3   0 
GJ
0 0 2
GJ   3 
 Y4   L L  v 
   12 EI 24 EI  4 
 b4   0
M 0 0  0   4 
L3 L3
M    
 t4   0 0 0 0 
GJ 6 EI 4 EI  GJ   4 
 L L2 L 
 6 EI 6 EI 4 EI  GJ 
 0 0 0  0 0 
 L2 L2 L 
Impose symmetry conditions
  24 EI 6 EI 6 EI 12 EI 
 L3  
L2 L2 L3  v
 Y2  0   6 EI 4 EI  2 
 M  0  0 0   
 b2   L2 L   2 
  4 EI  GJ 6 EI   2
M t2  0   2
6 EI
0  
Y3   P   L L L2   v3 
 24 EI 12 EI 24 EI 
 3 0 
 L L2 L3 

Solving for displacements then yields

PL3
v2  
6 EI

PL2
2 
4 EI

PL2
2 
4EI  GJ 

PL3 8 EI  5GJ 
v3  
24 EI EI  GJ 

Nodal forces and moments are

12 EI 6 EI P
Y1   3
v2  2  2 
L L 2

6 EI 2 EI PL
M b1  2
v2  2 
L L 2

GJ PL GJ
M t1   2  
L 4 EI  GJ

12 EI 6 EI P
Y2   3
v2  2  2  
L L 2

6 EI 4 EI
M b2  2
v2  2  0
L L

GJ PL GJ
M t2  2 
L 4 EI  GJ
 6 EI 2 EI 6 EI 4 EI PL2 EI  GJ 
M t3   v2   2  2 v3  3  
L 2
L L L 4EI  GJ 

C. Right-angle Frame

1 45◦ 45◦ 3
P P

The angle at node 2 is assumed to be preserved, and EI is the same for both members. Let AE be the
axial stiffness factor of the members.

Because of the rollers and loadings at nodes 1 and 3, the following conditions apply:

v1  v3  0 u1  u 3 1   3 u2   2  0

This leaves u1 , 1 , v 2 as the unknown displacements of the problem.

 6I A 6I 6I A 
 L2  2 
L2 2  u1 
 1
X   P   2L
  E  6I 6I  
 M b1  0   4I 1 
 Y  0  L  2L 2L  
 2  12 I 12 I 12 I  v 2 
 2 A  A
L 2L L2 

Inverting to find the unknown displacements gives

  12 I 2 
 2  4 AI  
u1   L   P 
  L/E  12 AI  
1       0 
v  K  2L  
 2  12 I 2  0 
 2  4 AI  
 L 

Desired displacement u1 is then

PL3  3I 
u1   1  
6 EI  AL2 

If we assume infinite axial stiffness by letting A  

PL3
u1  
6 EI

Displacement v2 is likewise found to be

PL3  3I 
v2   1  2 
6 EI  AL 

D. Circular Frame
 2

1 3

P Ro
P

Straight line segments are used to represent the frame. The frame is loaded with forces P as shown.
Also, it has a radius Ro and uniform flexural stiffness EI.

Segments 1-2 and 2-3 carry loading P/2. With symmetry existing in the frame, we can impose an
additional condition, θ1 = 0

 P  6I A 6I A 
 X 1  
2   L2  2  u 
L2 2   1 
  E 
    
 Y  0  L 12 I 12 I  
 2  A  A v 2 
   L2 L2


Solving for the displacement gives

3
2 PRo  12 I 
u1   1  
24 EI  AL2 

3
2 PRo  12 I 
v2   1  2 
24 EI  AL 

Eliminating member axial stiffness by letting A   gives

3 3
2 PRo PR
u1  v 2    0.059 o
24 EI EI

The theoretical answers for this problem may be found in texts on structural theory. They are

PRo  1  
3 3
PR
u1      0.076 o
EI   8  EI

3 3
PRo  1 1  PR
v2      0.067 o
EI   4  EI
It is seen that the crude idealization for the circular frame gives an error of -21% for u1 and -13% for
v2. If a reasonable number of straight line segments are used to represent a curved frame, the
stiffness method should provide satisfactory engineering answers.

E. Simple Frame
Example 1. BENT – Case 1

2 Lb,Ib 3
P

La,Ia y La,Ia

1 4

Find displacement and member bending moments for the bent.

The structure is symmetric (about the y axis) but the loading is not. However, symmetry of
loading may be established, without changing the problem by putting

P
X2  X3 
2

Boundary conditions: u1  v1  1  u 4  v 4   4  0

Symmetry conditions: u 2  u 3 2  3

Deformations are due solely to bending; that is, axial stiffness and axial displacements are
irrelevant to this problem. Therefore v 2  v3  0

u2 v2 2 u3 v3 3
 12 I a 
 3 SYM 
 La 
 0 12 I b 
 Lb
3 
 
 6 I a 6I
 b2
4I a 4I b
 
 L 2 Lb La Lb 
K  E a 
12 I a
 0 0 0 
 La
3

 12 I b 6I b 12 I b 
 0  3 2
0 3

 Lb Lb Lb 
 6I b 2I b 6I a 6I b 4I a 4I b 
 0  2
 2 2
 
 Lb Lb La Lb La Lb 

Imposing the symmetry conditions reduces the K as follows

u2 v2 2 v3
 12 I a 6I a 
 3 0  2
0 
 La La 
 12 I b 12 I b 12 I b 
 0 3
 2
 3 
 Lb Lb Lb 
K  E  6 I a 6I
 b2
4I a 6I b

6I b 
 L 2 Lb La Lb Lb
2 
 a 
 
 0 
12 I b 12 I b 12 I b 
 Lb
3
Lb
2
Lb
3 
 
 
Imposing v2 = v3 = 0 reduces K to order 2x2 as shown in the following form of the total structure
stiffness equation:

 12 I a 6I a 
 P  3   u 
X 2  
2

 2   La La  2
4 I a 6 I b   2 
E
 M 2  0   6 I a 
 L 2 La Lb 
 a

Solving for the displacements

 4I a 6I b 6I a 
L  L 2  P
u 2  1  a La   
  12 I a   2 
b


 2 12 I  I I   6 I
 0 
a
E  3a  a  6 b  
La 
2 3
 La  La Lb   La

As a special case we let I a  2 I b  2 I , La  Lb  L then

7 PL3
u2 
192 EI

PL2
2 
32 EI

Member bending moments may now be found from the nodal displacements and the member
stiffness equation

 6 2  5
M 1  EI a   2 u 2   2    PL
 L L  16

M 2 12  EI a  
6 4  3
u 2   2    PL
 L L 
2
16

M 2 23  EI b 
4 2  3
   3   PL
2 2
L L  16
Example 2. BENT – Case 2

3
2 2’

Lb,Ib

La,Ia y La,Ia

1 1’

Determine nodal displacements and member bending moments.

Both symmetry of structure and loading are initially present. The following are the symmetry
conditions on the displacements:

u3   3  0

u 2  u 2 '

v 2  v 2 '

 2   2 '

Axial stiffness must be retained, since the vertical members are required to carry axial loading.
Hence we cannot put v2 = v2’ = 0. However, the deformed structure is such that, to be consistent
with conventional theory, we should require

u2  0

The unknown displacements which now remain in the final stiffness equation are v2,θ2, and v3
 u2 v2 2 u3 v3 3
 12 I a 
 3 SYM 
 La 
 0 A 12 I b
 3 
 La Lb 
 
 6 I a 6I b
 2
4I a 4I b
 
K 1 23  E  La 2 Lb La Lb 
 0 0 0 0 
 
 0 12 I b 6I b 12 I b 
 3 0
 Lb Lb
2
Lb
3 
 6I b 2I b 6I b 4I b 
 0  2 0 
 Lb Lb Lb
2
Lb 

u3 v3 3 u2 ' v2 ' 2 '

0 SYM 
 12 I b 
0 3 
 Lb 
 6I b 4I b 
0  2 Lb 
 Lb 
K 3 2 '1'  E 0 0 0
12 I a 
 La
3 
 
0  12 I b 6I b
0
A 12 I b
 3 
 Lb
3
Lb
2
La Lb 
 6I 2I b 6I a 6I b 4I a 4I b 
0  b   
 Lb
2
Lb La
2
Lb
2
La Lb 

The two stiffness matrices may now be superimposed. Applying symmetry conditions dives the
final form of the stiffness equation

 A 12 I b 6I b 12 I b 
L  3  2
 3 
 a Lb Lb Lb 
 Y2  0  v 2 
   6I b I I  6I b  
 M 2  0   E  2 4 a  b 
 
2  2 
Y   P   Lb  La Lb  Lb   
 3    24 I b 12 I b 24 I b   3 
v
 3 2
Lb 
3
 Lb Lb

In order to simplify algebraic calculations, the following special case is now selected.

I b  2 I a  2 I , La  Lb  L
 12 X 24 I 2 12 X 12 I 2 
 0 
v 2  L4  0 
4
 L
  L5
 48 AI 12 AI   0 
 2   0  3
 v  12 X 24 AEI  
2
L4 L
2  P 
 3 12 X 24 I 2 24 AI 12 AI 12 X 12 I   
  3  
 L4 L L2 L4 

The displacements are then found to be

PL PL2 PL3 PL
v2   2  v3   
2 AE 24 EI 24 EI 2 AE

Again, moments are obtained from the individual member stiffness equations and the actual
nodal displacements

 
M 1 12  E  2 I a  2   PL
 La  12

 
M 2 12  E  4 I a  2   PL
 La  6

 
M 2 23  E   6 I b2 v2  4 I b  2  6 I b2 v3    PL
 Lb Lb Lb  6

 A  P
Y1 12  E   v 2  
 L a  2

Example 3. TWO- STORY FRAME

 3 3’
Q

La,Ia

2 Lb,Ib 2’
P

1 1’

Find Nodal displacements and bending moments

Vertical members are identical (Ia,La), as are the two horizontal members (Ib,Lb).

P
X 2  X 2 '
2

Q
X 3  X 3'
2

v 2  v 2 '  v3  v3 '  0

u2  u2 '

2  2 '

u3  u3 '

3  3 '
 Imposing the above conditions,

 24 I a 
 3
SYM 
 La 
 0 8I a 4 I b 

 La Lb 
 12 I 6I a 12 I a 
 3
a
 2 3

 La La La 
 6I 2I a 6I I I  
 a
 a2 4 a  b  
 La 
2
La La  La Lb 
K  E 24 I a 
 0 0 0 0 3 
 La 
 2I b 8I a 4 I b 
 0 0 0 0  
 Lb La Lb 
 0 0 0 0 
12 I a
 2
6I a 12 I a 
 La
3
La La
3 
 
 0 2I b 6I a 2I a 6I  I a I b 
0 0  a2 4  
 
 Lb La La La  La Lb 
 u2 v2 2 v3
24 I a 12 I a 6I a
0
La 3 La 3 La 2
P 8I a 6 I b 6I a 2I a
X2 = 0 + u2
2 La Lb La 2 La
M2 = 0 12 I a 6I a 12 I a 6I a 2
Q =E La 3 La 2 La 3 La 2 u3
X3 =
2 3
M3 = 0 6I a 2I a 6I a 4I a 6I b
+
La 2 La La 2 La Lb

If we put

EI a
1 La  15
La

EI b
 2.4 Lb  25
Lb

2 0 1 7 .5 
0 420  7.5 37.5 
12 
K
225   1  7.5 1  7 .5 
 
7.5 37.5  7.5 345 

P
2 2 1 0 7.5  u 2 

 Q  12  1
 1  7.5  7.5 u 3 
   
 02  225  0  7.5 420 37.5   2 
   
7.5  7.5 37.5 345   3 
 0 

Expanding by matrix multiplication

0 12   0  7.5 u 2   420 37.5  2  

   
 7.5  7.5 u   37.5 345    
 
0 225   3    3 
P
 2  12   2  1 u 2   0 7.5  u 2  
Q          

  225   1 1  u 3   7.5  7.5 u 3  
2

Solving the first equation of the above pair for the θ’s and substituting into the second equation
yields

P
 2  12   2  1  0 7.5   420 37.5  0  7.5 u 2 
1

Q          
  225   1 1   7.5  7.5 37.5 345  7.5  7.5  v3 
2

P
 2  12  1.835  0.850 u 2 
Q     
  225  0.850 0.730   v3 
2

Solving this expression for u2 and u3

P
u 2  0.730 0.850  2 
   30.25  
u 3  0.850 1.835   Q 
2

Knowing u2 and u3, we now find θ2 and θ3

1
P
 2   28 2.5  0  7.5 0.730 0.850  2 
   15 X 30.25     
 3  2.5 23  7.5  7.5 0.850 1.835   Q 
2

P
 2   0.455 0.941  2 
   
 3  0.0294 0.545  Q 
2

Calculating member nodal bending moments

 
M 1 12  EI a   6
2
u2 
2 6
 2    u 2  2 2  3.96 P  4.21Q
 La La  15
  4 
M 2 22'  EI b  2
 2   2 '   14.4 2  3.28 P  6.78Q
 Lb Lb 
MATRIX NOTATION

Matrix notation represents a compact notation for writing and solving sets of simultaneous
algebraic equations. It may therefore be inferred that the matrix formulation of the force and stiffness
methods is given in terms of simultaneous algebraic equations.

Consider forces X1,Y1,Z1,X2,Y2,Z2,…Xn,Yn,Zn acting at points (nodes) 1,2,…,n on an elastic structure.

Furthermore, let u1,v1,w1,u2,v2,w2,…un,vn,wn be the corresponding set of nodal displacements.

The column matrix of forces is then written as

 X1 
Y 
 1
 Z1 
 
X 2 
 Y2 
 
X   X   2 
Z
 . 
 . 
 
 . 
X 
 n
 Yn 
Z 
 n

And the corresponding column of displacements appear as

 u1 
v 
 1
 w1 
 
u2 
 v2 
 
u  u   2 
w
 . 
 . 
 
 . 
u 
 n
 vn 
w 
 n
In addition to the column matrices of nodal forces and displacements is the matrix

of stiffness influence coefficients K which may be expressed as

 k11 k12 ... k1n 

k k 22 ... k 2 n 
K   K   21
 ... ... ... ... 
 
k n1 k n2 ... k nn 

Nodal forces X and nodal displacements are directly related through the stiffness

matrix K. The applicable equation is

X  Ku 
 stiffness equation

Let us assume an arbitrarily elastic structure supported against a rigid-body

motion, and subjected to loads F1,F2,…Fn acting at nodes 1,2,…n. The

corresponding displacements is represented by 1,2,…n.

Consider the displacement i at node i. It is affected by full set of forces. For

linearly elastic structure we can write the following equation expressing the

contribution of each applied force to i

 i  ci1 F1  ci 2 F2  ...  cin Fn

Where ci1 = deflection produced at node i due to a unit load at node 1

(F1=1)

ci1F1 = contribution to i if the load at node 1 is F1 rather than

unity

ci1, ci2,…, cin = deflection influence coefficients; specify precisely how

much each unit load at any node contributes to i

Suppose we let i apply to nodes 1,2,…,n, there will be n equations and is gathered

into a single matrix equation as follows

 1   c11 c12 ... c1n   F1 

  c c 22 ... c 2 n   F2 
 2    21  
 ...   ... ... ... ...   ... 
   
 n  c n1 cn 2 ... c nn   Fn 

The equation may be written in compact matrix form

   C F  or   CF

Where C = matrix deflection influence coefficients or overall structure flexibility

matrix

If C is known

1. Nodal displacements  due to any set of prescribed nodal forces F can be

calculated
 2. Matrix calculations can be used to determine the vibration characteristics (deformations and
stresses) of the structure to predict structural responses under dynamic conditions
3. Algebraic equations can be solved for forces in terms of deflection

F  C 1

where C-1 = inverse of the matrix of deflection influence coefficients

C-1 = K = stiffness matrix

The expanded form of F  C 1 is as follows

 F1   k11 k12 ... k1n   1 

 F  k k 22 ... k 2 n   2 
 2    21  
 ...   ... ... ... ...   ... 
   
 Fn  k n1 k n2 ... k nn   n 

Now assume the structure to be forced into a displaced configuration defined

by 1=1, 2,…,n = 0, then

F1  k11 F2  k 21 F3  k n1

DIRECT STIFFNESS METHOD

The simple elastic spring is used to illustrate the basic procedure

STIFFNESS MATRIX – The Elastic Spring

A typical elastic spring is shown

X1,u1 k X2,u2

eeeeeeeeeeeeeeeeeeee
1 2

Forces (X1,X2) exist at nodes 1 and 2, and displacements (u1,u2) are measured at these same points.
Spring constant k represents the constant slope of the linear force- deformation curve.

The stiffness matrix for the spring relates the column of nodal forces to the column of nodal
displacements. By analogy, we note the form of this equation must be as follows:

 X 1   k11 k12  u1 

   
 X 2  k 21 k 22  u 2 

To find the elements kij of the K matrix, consider the total displacement states which the member can
experience

CASE A: u1  u1 ; u 2  0

Since node 2 is now supported, X2 becomes a reaction and X1 is the applied load. By definition of the
spring constant k, we can write

X1
k X 1  ku1
u1

Imposing equilibrium on the forces and requiring the above force deflection relation

X 1   X 2  ku1
CASE B: u1  0; u 2  u 2

Since node 1 is now supported, X1 becomes a reaction and X2 is the applied load. By definition of the
spring constant k, we can write

X2
k X 2  ku 2
u2

Imposing equilibrium on the forces and requiring the above force deflection relation

X 2   X 1  ku 2

There are no further distinct displacement states which this system can experience

Collecting results given by Case A and Case B into a single matrix equation yields

X 1  ku1  ku 2
X 2  ku 2  ku1   ku1  ku 2

 X1   k  k  u1 
   
 X 2   k k  u 2 

k  k  1  1
K    k 
 k k   1 1 
STIFFNESS MATRIX – The Elastic Spring Assemblage

A typical elastic spring assemblage is shown

ka kb

eeeeeeeeee eeeeeeeeee
1 2 3

Springs a and b have individual stiffness constants ka and kb. They are connected at node 2. This time we
have three nodes and consequently a total of three nodal forces and three nodal displacements

CASE A: u1  u1 ; u 2  u 3  0

X1 is the applied force. Hence

X 1  k a u1

Equilibrium requires that

X 1   X 2  k a u1
X3  0

Note that since u2 = u3 = 0, no force can occur at node 3

CASE B: u 2  u 2 ; u1  u 3  0

X 2   k a  k b u 2
X 1  k a u 2
X 3  k b u 2

CASE C: u 3  u 3 ; u1  u 2  0
X 3  kb u3
X 2  k b u 3
X1  0

Superimpose the three independent displacement states, we obtain the general case

CASE D: u1  u1 ; u 2  u 2 ; u 3  u 3

X 1  k a u1  k a u 2
X 2   k a u1  (k a  k b )u 2  k b u 3
X 3  k b u 2  k b u 3

Writing these equations in matrix form

 X1   ka  ka 0  u1 
    
 X 2    k a k a  kb  k b  u 2 
X   0  kb k b  u 3 
 3 

Elements in K must conform to the ordering of elements in the force and displacement column matrices.
Hence, if K is to be written separately, the order in which elements appear in u or X must be clear

u1 u2 u3
 ka  ka 0 
K   k a k a  kb  k b 
 0  kb k b 

Assembling K by superposition

Superimpose the stiffness matrices of the individual structural members

u1 u2 u2 u3
k  ka  k  kb 
Ka   a Kb   b
 ka ka   k b k b 

The above matrices are not immediately subject to superposition since the columns in the two matrices
are not compatible. To correct this situation we expand each member stiffness matrix to the order of
the total structure stiffness matrix.
 u1 u2 u3 u1 u2 u3
 ka  ka 0 0 0 0 
K a   k a ka 0 
K b  0 k b  k b 
 0 0 0 0  k b k b 

K  Ka  Kb

u1 u2 u3
 ka  ka 0 
K   k a k a  kb  k b 
 0  kb k b 

Properties of the Stiffness Matrix

1. Sum of the elements in any column is zero.

2. K matrix is symmetric.
3. All terms along the main diagonal of K are positive.

General Treatment of Solution Procedure

No matter what the actual structure may be, it can be idealized into an assemblage of members
connected at node points. Assuming the stiffness matrix for each component member to be known, the
total structure stiffness matrix can be formed by superposition.

The total structure stiffness equation can be represented by

X  Ku

Where X = total set of nodal forces; includes bending and twisting moments, as well as

x,y,z components of rectilinear force

X = correspond to u and represent applied loads

X = correspond to u and represent unknown reactions

u = total set of nodal displacements

u = unknown nodal displacements

 u = prescribed displacements (due to boundary conditions)

 X    K   K   u 
    
       
 X  K  K   u  
    

By matrix multiplication

X   K  u  K  u 
X   K  u  K  u 

Unknown nodal displacements are obtained by inverting K

1
u  K  ( X   K  u  )

If u is a null matrix (as when it represents boundary conditions all of which are prescribed to be zero)

u   0
1
u  K  X 

Unknown reactions are obtained as

1 1
X   K  K  X   ( K  K  K   K  )u 

If u is a null matrix

u   0
1
X   K  K  X 

STRESS MATRIX
The stress matrix is defined as that matrix which directly yields internal forces ( in case of one
dimensional member), or, stresses(when more complex structural members are being considered), in
terms of the nodal displacements.

When all nodal displacements have been found, the deformation of each member of the assemblage is
known. These deformations can now be thought of as being caused by a fictitious set of equivalent
nodal forces, Xe. The member stiffness equation may be written as

X e  Ku

Where K = member stiffness matrix

u = actual nodal displacements for the member

Since the right-hand side of the equation is known, the equivalent nodal forces likewise become known.

For the simple case of the spring, equilibrium considerations immediately give the internal force as
being precisely equal to the equivalent nodal force.

 X i e   k  k   u i 
 e    
 X j   k k  u j 

From which

ui 
X i  k  k     X j
e e

u
 j

The internal force Sij in the spring is then

ui 
S ij  X i   k k    S  u ( )
e

u j 
APPLICATION OF THE DIRECT STIFFNESS METHOD FOR TRUSS ANALYSIS

Replace the spring by a truss member extending between nodes 1 and 2. Assume that the member has a
constant cross sectional area A and a modulus of elasticity E. If node 2 is fixed and force X1 is applied,
the force-deformation relation may be written as

X 1L AE
u1  X1  u1
AE L

AE
for the truss member is equivalent to k for the spring
L

The stiffness equation for the truss member is

 X 1  AE  1  1 u1 
   
X 2  L  1 1  u 2 

The above equation is referred to local coordinates( x, y ), that is with the axis lying along the member.
Another coordinate system called reference coordinates( x, y ) is oriented with respect to the overall
structure.

For displacements, forces and stiffness matrices written with respect to local coordinates, designate by
u , X , K , respectively. When written with respect to reference axes the bar is omitted, that is u , X , K

Y2
Y2

X2
 y 2
X2

L,AE

x
X1
1

Y1

At node 2: At node 1:

Local Coordinates Reference Coordinates

Force X2 X 2 ,Y2

Note: 1. Y 2 does not exist since the truss member cannot sustain a normal component of

Force

2. Two components of force appear for the member in terms of local coordinates,

a total of four components arise when reference coordinates are used

Given: Stiffness Equation referred to local coordinates

 X 1  AE  1  1  u1 
   
X 2  L  1 1  u 2 

Expand to order which is consistent with the reference coordinates

X1 1 0 1 0  u 1 
    
 Y 1  AE  0 0 0 0  v 1 
   
X 2  L  1 0 1 0 u 2 
Y 2   
  0 0 0 0 v 2 

The compact form is given by

X  Ku

Transformation of the Stiffness Matrix

From

X 2  X 2 cos   Y2 sin  X 1  X 1 cos   Y1 sin 

Y 2   X 2 sin   Y2 cos  Y 1   X 1 sin   Y1 cos 

If   cos    sin 

The above set of four equations may be written as

X1    0 0  X 1 
  
 Y 1     0 0   Y1 
   
X 2   0 0    X 2 
 Y 2   0 0 

   Y2 
 

X  TX X  T 1 X  TT X
u  Tu
from
X  Ku
TX  KTu
X  T 1 KTu  T T KTu  Ku

We therefore see that K  T T KT

   0 0  1 0 1 0    0 0
  0 0  AE  0
 0 0 0    0 0 
K
0 0     L  1 0 1 0  0 0  
    
0 0    0 0 0 0  0 0  

  0  0    0 0
 0   0 
AE   0   0
K
L   0  0  0 0  
  
  0  0  0 0  

u1 v1 u2 v2
 2   2   
 
AE   2    2
K
L   2   2  
 
   2   2 

This stiffness matrix for the truss member referred to x,y coordinates.

Stress Matrix

Stress matrix is to yield the internal forces, or stresses, in a member once its nodal

displacements have been found

X e  Ku

 X1   2   2    u1 
Y    
 1  AE   2     2   v1 
   
X 2  L   2   2   u 2 
 Y2   
   2   2   v 2 

Equivalent forces at 2 are then given by

X2 
e AE
L

 2 u1  v1  2 u 2  v 2 
AE
L
 
(u 2  u1 )2  (v 2  v1 ) 

Y2 
e AE
L

 u1   2 v1  u 2   2 v 2 
AE
L
 
(u 2  u1 )  (v 2  v1 )  2 
S1 2  X 2 cos   Y2 sin   X 2   Y2 
e e e e

S1 2 
AE
L

(u 2  u1 )2  (v 2  v1 )   
AE
L
 
(u 2  u1 )  (v 2  v1 )  2 

S1 2 
AE
L
 
(u 2  u1 ) 2  (v 2  v1 )2   (u 2  u1 ) 2  (v 2  v1 )  2

S1 2 
AE
L

(2   2 )  (u 2  u1 )  (v 2  v1 )  
S1 2 
AE
(u 2  u1 )  (v2  v1 )  
L

This equation may be written in the following matrix form

u  u1 
S1 2 
AE
   2 
L v 2  v1 

For the arbitrary member, nodes 1, 2 can be replaced by i,j. Equation then

becomes

u j  ui 
S ij  
AE
ij   ij  
L v j  vi 

Analysis of space truss

 y,v

2
L,AE

1
x,u

z,w

The truss member in three-dimensional space is shown. It will be assumed to have

a length L and a uniform axial stiffness factor AE. As for the two dimensional

case, let x,y,z represent reference axes and x, y, z represent local axes.

The local and reference axis systems are related through angles defined as

follows:

 x x  angle between x and x

 x y  angle between x and y

……………………………..

 z y  angle between z and y

Nodal forces referred to the two axes systems then satisfy the equations

X  X cos  x x  Y cos  y x  Z cos z x

Y  X cos  x y  Y cos  y y  Z cos z y

Z  X cos  x z  Y cos  y z  Z cos z z

It is now convenient to adopt the notation

 x  cos  x x  y  cos  x y  z  cos  x z

 x  cos  y x  y  cos  y y  z  cos  y z

 x  cos  z x  y  cos  z y  z  cos  z z

With this notations we can write

 X 1   x x x 0 0 0  X 1 
    y y 0 0 0   Y1 
Y1   y  
 Z 1    z z  z 0 0 0   Z 1 
   
X 2   0 0 0 x x  x  X 2 
Y 2   0 0 0 y y  y   Y2 
    
 Z 2   0 0 0 z z  z   Z 2 

This equation may be written in compact form as

X  TX

 
Where T 
  

 x x  x 
 
And A is given by    y y  y 
  z  z 
 z

Transform K into K

u1 v1 w1 u 2 v 2 w2
1 0 0 1 0 0
0 0 0 0 0 0

AE  0 0 0 0 0 0
K  
L  1 0 0 1 0 0
0 0 0 0 0 0
 
 0 0 0 0 0 0
 Using T as given above and applying transformation, the stiffness matrix for the truss member
referred to reference axes is

u1 v1 w1 u2 v2 w2
  2
SYM 
 
  2

AE    2 
K  
L   2  u   2 
    2    2 
 
     2    2 

Example: 2D Truss
 Y Y

1 3 5 7

5 ft

10ft
4 6

2 8

10ft 10ft 10ft

For all members : E = 29x103 ksi and A = 10.0 in2

Y Value : 5 kips

Member θ◦      AE/L

1-3 0 1.000 0.000 1.000 0.000 0.000 29000.00

1-4 333.44 0.894 -0.447 0.800 0.200 -0.400 25938.39

2-4 26.56 0.894 0.447 0.800 0.200 0.400 25938.39

2-3 45 0.707 0. 707 0.500 0.500 0.500 20506.10

3-4 270 0.000 -1.000 0.000 1.000 0.000 58000.00

3-6 333.44 0.894 -0.447 0.800 0.200 -0.400 25938.39

3-5 0 1.000 0.000 1.000 0.000 0.000 29000.00

4-6 0 1.000 0.000 1.000 0.000 0.000 29000.00

4-5 26.56 0.894 0.447 0.800 0.200 0.400 25938.39

5-7 0 1.000 0.000 1.000 0.000 0.000 29000.00

5-8 315 0.707 -0.707 0.500 0.500 -0.500 58000.00

 5-6 270 0.000 -1.000 0.000 1.000 0.000 25938.39

6-7 26.56 0.894 0.447 0.800 0.200 0.400 25938.39

6-8 333.44 0.894 -0.447 0.800 0.200 -0.400 25938.39

Stiffness Matrix for Each Member

u1 v1 u2 v2
 2   2   
 
AE   2    2
K
L   2   2  
 
   2   2 

u1 v1 u3 v3
1 0  1 0  29000 0  29000 0 u1
 0 0 0  0 0 v1
 AE   0 0 0
K1 3   
 L 1 3  1 0 1 0  29000 0 29000 0 u3
   
0 0 0 1  0 0 0 0 v3

u1 v1 u4 v4
 0.80  0.40  0.80 0.40   20750.71  10375.36  20750.71 10375.36  u1
 0.40  0.20   10375.36  5187.68  v1
 AE   0.40 0.20 5187.68 10375.36
K 1 4   
 L 1 4  0.80 0.40 0.80  0.40  20750.71 10375.36 20750.71  10375.36 u 4
   
 0.40  0.20  0.40 0.20   10375.36  5187.68  10375.36 5187.68  v 4

u2 v2 u4 v4
 0.80 0.40  0.80  0.40  20750.71 10375.36  20750.71  10375.36 u 2
 0.40 0.20  0.40  0.20  10375.36  10375.36  5187.68  v 2
 AE   5187.68
K 24   
 L  2  4  0.80  0.40 0.80 0.40   20750.71  10375.36 20750.71 10375.36  u 4
   
 0.40  0.20 0.40 0.20    10375.36  5187.68 10375.36 5187.68  v 4

u2 v2 u3 v3
 0.50 0.50  0.50  0.50  10253.05 10253.05  10253.05  10253.05 u2
 0.50  
0.50  0.50  0.50  10253.05 10253.05  10253.05  10253.05 v2
 AE  
K 23   
 L  2  3  0.50  0.50 0.50 0.50   10253.05  10253.05 10253.05 10253.05  u3
   
 0.50  0.50 0.50 0.50   10253.05  10253.05 10253.05 10253.05  v3
 u3 v3 u4 v4
0 0 0 0  0 0 0 0  u3
  
0  1 0 58000 0  58000 v3
 AE  0 1
K 3 4   
 L  3  4 0 0 0 0  0 0 0 0  u4
   
0  1 0 1  0  58000 0 58000  v4

u3 v3 u6 v6
 0.80  0.40  0.80 0.40   20750.71  10375.36  20750.71 10375.36  u3
 0.40  0.20   10375.36  5187.68  v3
 AE   0.40 0.20 5187.68 10375.36
K 3 6   
 L 3 6  0.80 0.40 0.80  0.40  20750.71 10375.36 20750.71  10375.36 u6
   
 0.40  0.20  0.40 0.20   10375.36  5187.68  10375.36 5187.68  v6

u3 v3 u5 v5
 1 0 1 0  29000 0  29000 0 u3
 0  0 0 v3
 AE   0 0 0 0 0
K 3 5   
 L 3  5   1 0 1 0  29000 0 29000 0 u5
   
 0 1 0 1  0 0 0 0 v5

u4 v4 u6 v6
 1 0 1 0  29000 0  29000 0 u4
 0  0 0 v4
 AE   0 0 0 0 0
K 46   
 L  4  6  1 0 1 0  29000 0 29000 0 u6
   
 0 1 0 1  0 0 0 0 v6

u4 v4 u5 v5
 0.80 0.40  0.80  0.40  20750.71 10375.36  20750.71  10375.36 u4
 0.40 0.20  0.40  0.20  10375.36  10375.36  5187.68  v4
 AE   5187.68
K 45   
 L  4  5  0.80  0.40 0.80 0.40   20750.71  10375.36 20750.71 10375.36  u5
   
 0.40  0.20 0.40 0.20    10375.36  5187.68 10375.36 5187.68  v5

u5 v5 u7 v7
 1 0 1 0  29000 0  29000 0 u5
 0  0 0 v5
 AE   0 0 0 0 0
K57   
 L 5  7   1 0 1 0  29000 0 29000 0 u7
   
 0 1 0 1  0 0 0 0 v7

u5 v5 u8 v8
 0.50  0.50  0.50 0.50   10253.05  10253.05  10253.05 10253.05  u5
 0.50  0.50  10253.05 10253.05 10253.05  10253.05 v5
 AE   0.50 0.50
K 5 8   
 L 5 8  0.50 0.50 0.50  0.50  10253.05 10253.05 10253.05  10253.05 u8
   
 0 . 50  0 . 50  0 . 50 0 . 50   10253 . 05  10253 . 05  10253.05 10253.05  v8
 u5 v5 u6 v6
0 0 0 0  0 0 0 0  u5
  
0  1 0 58000 0  58000 v5
 AE  0 1
K56   
 L  5  6 0 0 0 0  0 0 0 0  u6
   
0  1 0 1  0  58000 0 58000  v6

u6 v6 u7 v7
 0.80 0.40  0.80  0.40  20750.71 10375.36  20750.71  10375.36 u6
 0.40 0.20  0.40  0.20  10375.36  10375.36  5187.68  v6
 AE   5187.68
K67   
 L 6  7  0.80  0.40 0.80 0.40   20750.71  10375.36 20750.71 10375.36  u7
   
 0.40  0.20 0.40 0.20    10375.36  5187.68 10375.36 5187.68  v7

u6 v6 u8 v8
 0.80  0.40  0.80 0.40   20750.71  10375.36  20750.71 10375.36  u6
 0.40  0.20   10375.36  5187.68  v6
 AE   0.40 0.20 5187.68 10375.36
K 6 8   
 L 6 8  0.80 0.40 0.80  0.40  20750.71 10375.36 20750.71  10375.36 u8
   
 0.40  0.20  0.40 0.20   10375.36  5187.68  10375.36 5187.68  v8

Boundary Conditions will trim down the 16x16 matrix to 8x8

At nodes 1,2,7 and 8, u1,v1,u2,v2,u7,v7,u8,v8 = 0

At nodes 4 and 6 due to symmetry, v4= v6

At nodes 3 and 5 due to symmetry, v3= v5

Solving for displacement, U from X=KU

Where X is the applied force

K is the stiffness matrix

U is the displacement

U = K-1X

The applied forces, X are as follows

X3 = 0 Y3 = -5
X4 = 0 Y4 = 0

Y5 = 0 Y5 = -5

X6 = 0 Y6 = 0
THE UNIFORM, RECTILINEAR BEAM

I. BEAM ELEMENT

A. Beam Element Stiffness Matrix (K)

M2,θ2
y
x

M1,θ1
1 2
L

Y1,v1 Y2,v2

Derive the stiffness matrix by considering separate displacement states for which known results are
available.

CASE A: v2 = θ2 = 0

Under the imposed boundary conditions, the beam becomes a cantilever supported at node 2.
Beam theory then gives displacements v1 and θ1 in terms of applied loads Y1 and M1

Y1 L3 M 1 L2
v1  
3EI 2 EI

Y1 L2 M 1 L
1  
2 EI EI

Solving these equations for forces in terms of displacements and expressing the results in matrix
form provides

 12 6
 L3 
 Y1  L2  v1   K v1 
   EI  6 4  1  11  
M 1   2  1 
 L L 

The reaction for this state can be found by applying equilibrium to the force system. Doing this
yields
 F y  Y1  Y2  0
M 2  Y1 L  M 1  M 2  0

From which

 Y2    1 0   Y1  Y  v 
     A 1   AK 11  1 
M 2   L  1 M 1  M 1  1 
 12 6
  1 0   L3  2  v 
 EI  L  1

 L  1  6 4  1 

 L2 L 
 12 6 
  L3 L2  v1 
 EI  6 2   
   1
 L2 L 
v 
 K 21  1 
1 

CASE B: v1 = θ1 = 0

The beam element now becomes a cantilever fixed at node 1. Beam theory then gives
displacements v2 and θ2 in terms of applied loads Y2 and M2

Y2 L3 M 2 L2
v2  
3EI 2 EI
 Y2 L2 M 2 L
2   
2 EI EI

Solving these equations for forces in terms of displacements and expressing the results in matrix
form provides

 12 6
 Y2   L3 L2  v 2   K v 2 
   EI  6 4   2  22  
M 2   2   2 
L L

The reaction for this state can be found by applying equilibrium to the force system. Doing this
yields

F y  Y1  Y2  0
M 1  Y2 L  M 1  M 2  0

From which

 Y1   1 0   Y2   Y2  v 2 
      B    BK 22  
M 1   L  1 M 2  M 2   2 
 12 6 
 1 0   L3 L2  v 2 

 L  1  6
 EI  4   2 

 L2 L 
 12 6
 L3  L2  v 2 
 EI  6 2   
  2
 L2 L 
v 
 K 12  2 
 2 

Collecting results of the two possible distinct displacement states which can be imposed on the
beam element, a single matrix equation gives

 Y1   v1 
M  K 12  1 
 1   K 11
   
 Y2   K 21 K 22  v 2 
M 2   2 
 v1 1 v2 2
 12 6 12 6
 L3   
L2 L3 L2 
 6 4 6 2 
 2  Stiffness matrix for the
K  EI  L L L2 L  basic beam element
  12 6 12 6 
referred to local
 L3 L2 L3 L2 
 6 2 6 4  coordinates
 2 
 L L L2 L 

Alternative form

 Y1 
 M1   12  6  12  6  v1 
 EI  6  
 4 6 2  1 L 
 YL   3   12 12 6   v 2 
 2  L 
6
 
M 2    6 2 6 4   2 L 
 L 

If x, y are reference axes and x, y local axes, transform K . But prior to transformation, the
order of K must be increased
 u1 v1 1 u2 v2 2
0 0 0 0 0 0 
 12 6 12 6
0 L3  2 0  3  2
L L L
 6 4 6 2 
0  2 0 
K  EI 0 L L L2 L 
 0 0 0 0 0 
0  12 6
0
12 6 
 L3 L2 L3 L2 
 4 
0  62 2
0
6
 L L L2 L 

Transformation matrix, T

   0 0 0 0
   0 0 0 0

 0 0 1 0 0 0
T  
 0 0 0   0
 0 0 0   0
 
 0 0 0 0 0 1

Stiffness Matrix for the Basic Beam Element Referred to Reference Coordinates

K  T ' KT

u1 v1 1 u2 v2 2
 12 2 
 L3  SYM 
 12 12 2 
 3   
 L L3 
 6  6 4  Stiffness matrix for the
 2  2 
L basic beam element
K  EI  L L

12 2 12 6 12 2 referred to reference
 3    2  
 L L3 L L3  coordinates
 12  12
 3 2
6

12
 3 
12 2
 
 L3 L L2 L L3 
 6 6 2 6 6 4
 2  2  2  
 L L L L L2 L
B. Beam Element Stress Matrix (S)
The significant internal forces for the beam element are taken to be the shear and the bending
moment.

Consider beam element ij as one component of a gross beam structure. From the stiffness
equation of a beam element,

 12 6 12 6 
Y1  EI  3 v1  2 1  2 v 2  2  2 
L L L L 

 6 4 6 2 
M 1  EI   2 v1  1  2 v 2   2 
 L L L L 

At node i the equivalent nodal forces Yie and Mie may be written as

 12 6 12 6 
Yi  Yi  EI  3 vi  2  i  2 v j  2  j 
e

L L L L 

 6 4 6 2 
M i  M i  EI   2 vi   i  2 v j   j 
e

 L L L L 

Using these equivalent nodal forces, the shear V and bending moment M in the element can be
written as
 V  Yi
e

M  Yi x  M i
e e

In matrix form we write,

v 
 12 6 12  i 
6
V   3
     i 
   EI 12 L 6 L2 L3 L2
M  6 4 12 6 6 2  v 
 3 x 2  x  x 2  2 x   j 
L L L2
L L3
L L L   j 
 

 12 6 12  6
     Stress matrix for the
S  EI  L3 L2 L3 L2
12 6 6 4 12 6 6 2 basic beam element
 3 x 2  x  x 2  2 x  referred to reference
L L L2
L L3
L L L
coordinates

II. BEAM ASSEMBLAGE

1 2 3

La Lb
 v1 1 v2 2 v3 3
 12 
 L3 SYM 
 a 
 6 4 
 L 2 La 
 12a
6 12 
K a  EI a   3 
 La La
2
La
3

 6 2 6 4 
 2 2

 La La La La 
 0 0 0 0 0 
 
 0 0 0 0 0 0

v1 1 v2 2 v3 3
0 SYM 
0 0 
 
0 0
12 
 Lb
3 
 6 4 
K b  EI b 0 0  2 
 Lb Lb 
 12 6 12 
0 0  3 2 3

 Lb Lb Lb 
 6 2 6 4
0 0  2 2 
 Lb Lb Lb Lb 

By superimposing Ka and Kb we obtain the gross structure stiffness matrix

Example No. 1

Compute the displacements, reactions and internal forces for the beam shown in the figure.
Assume EI constant
 P P

1 2 3 4

a c a

Stiffness Matrix of each member

v1 1 v2 2
 12 6 12 6 v
 a3   
a2 
1
a2 a3
 6 4 6 2 
 2 
K 1 2  EI  a a a2 a  1
  12 6 12 6  v2
 a3 a2 a3 a2 
 6 2 6 4 
 2 
 a a a2 a  2

v2 2 v3 3
 12 6 12 6 v
 c3   
c2 
2
c2 c3
 6 4 6 2 
 2 
K 2 3  EI  c c c2 c  2
 12 6 12 6  v3
 c 3 c2 c3 c2 
 6 2 6 4 
 2 
 c c c2 c  3
 v3 3 v4 4
 12 6 12 6 v
 a3   
a2 
3
a2 a3
 6 4 6 2 
 2 
K 3 4  EI  a a a2 a 3
 12 6 12 6  v4
 a 3 a2 a3 a2 
 6 2 6 4 
 2 
 a a a2 a  4

Applying boundary conditions v1 = v4 = 0 , θ1 = θ4 = 0

And symmetry conditions v2 = v3 , θ2 = - θ3

v2 2 v3 3
 12 12   6 6 12 6 
 a 3  c 3   a 2  c 2    2  v2
   
3
c c 

 6 6   4 4  6 2 
 
 a 2 c 2   a c  c2 c  2
K  EI 
12 6  12 12   6 6  v
  3  3  3   2  2  3
 c c2 a c  c a 
 6 2  6 6  4 4 
  2  2  2     3
 c c c a  c a 

v2 2 v3 3
12c  12a
3 3
6c  6 a
2 2
12 6 
    v2
 a 3c 3 a 2c 2 c3 c2 
 6c  6 a 4c  4 a
2 2
6 2 
K  EI  a c
2 2
ac c2 c  2
2 
12 6 12a  12c 3
3
6 a  6c v 3
2
  3 
 c c2 a 3c 3 a 2c 2 
 6 2 6 a 2  6c 2 4 a  4c 
 2 
 c c a 2c 2 ac  3

From X   K U 
 v2 2 v3 3
12c  12a
3 3
6c  6 a
2 2
12 6 
   
 a 3c 3 a 2c 2 c3 c2 
Y2   P   v2 
 6c  6 a 4c  4 a
2 2
6 2 
M  0 
 2   a 2c 2 ac c2 c   2 
   EI   
Y3   P  12 6 12a  12c 3
3
6 a 2  6c 2   v 3  v 2 
  3 
 M 3  0   c c2 a 3c 3 a 2 c 2   3   2 
 6 2 6 a 2  6c 2 4 a  4c 
  2 
c c a 2c 2 ac

12c 3  12a 3 6c 2  6 a 2   12 6
Y2   P  
a 3c 3 2 2
a c
 v 2   c 3 
c 2   v2 
   EI      EI 
2    2 
M2  0  6c  6 a 4c  4a   2 
2 2
6
 2 
 a 2 c 2 ac   c c 

12c 3  12a 3 12 6c 2  6 a 2 6 
 Y2   3 3
 3  2  v 
  2 a c c a 2c 2 c  2
  EI  
M 2   6 c  6 a 2
6 4c  4a 2   2 
 2 
 a 2 c 2 c ac c 

 12 6 
 Y2   a3 a 2  v 2 
   EI  6 4c  2a   2 
M 2   2 
a ac 

Solve for displacement and rotation

1
 12 6 
v 2   a3 a 2   Y2 
   EI  6 4c  2a  M 2 
 2   2 
a ac 

Solve for inverse

  4c  2 a 6
 ac  2
a
1  12   a 3  2c  a  a 2c 
 12 6    2
6
   
 
 a3 a2   a a3  6 2a  c 22a  c  
6    2 
4c  2 a   12  4c  2a   6  6   a c ac 
 2   3     2  2   
a ac   a  ac   a  a   2 2 a  c  2 a  c 


 a 3  2c  a  a 2c 
  
v 2  1  6  2a  c  
22a  c   Y2   P 
     
 2  EI   a c  M 2  0 
2
ac
 22a  c  2a  c 


 a 3  2c  a  
 P   v3
1
v2   
EI  6  2a  c  

 
 22a  c   P    3
1 a 2c
2 
EI  

To solve for reactions and moments, consider the total stiffness matrix

12 EI   P a 3  2c  a  6 EI  P a 2c 
Y1         P
a 3  EI 6  2a  c  a 2  EI 22a  c  

12 EI   P a 3  2c  a  6 EI  P a 2c 
Y4   3      P
a  EI 6  2a  c  a 2  EI 22a  c  

6 EI   P a 3  2c  a  2 EI  P a 2 c   aa  c P
M1        
a 2  EI 6  2a  c  a  EI 22a  c   2a  c

6 EI   P a 3  2c  a  2 EI  P a 2 c  aa  c P
M4   2      
a  EI 6  2a  c  a  EI 22a  c   2a  c
III. TAPERED AND CURVED BEAMS

A. Tapered beam-theoretical solution

P A
w

2
1
x h(x)

A
A-A
l

As shown, the beam has a uniform width w but is tapered in depth as indicated by h(x). Assume
a square section at the free end (x = 0). The taper is taken into account by prescribing the
moment of inertia I in bending to be defined by

 x
I ( x )  I 0 1  n 
 l

Where n = arbitrary numerical factor

I0 = moment of inertia of section at x = 0

The basic differential equation for this problem may be written as

d 2v
EI ( x)  Px
dx 2

Where v(x) is the displacement in the y direction and I(x) is as previously given

dv
To integrate, put   , so that
dx

d P x

dx EI 0 x
1 n
l
Integrating yields

Pl 2  n   n 
 1  l x   log1  l x   A
n 2 EI 0    

Where A is a constant of integration

dv
Replacing θ by and integrating leads to
dx

Pl 2  n 2 l  l  n   n 
v  2l x  2 x    1  x  log1  x   Ax  B
n 2 EI 0  n n l   l 

Where B is the second constant of integration

Imposing the boundary conditions v   0 at xl

Pl 2
A log1  n   1  n 
n 2 EI 0

Pl 3 1 1
 n log1  n   2  1  n 
n
B
n 2 EI 0

Using the above forms for A and B and selecting n = 7, the v and θ equation at the free end of
the beam are
 Pl 2  n 2 l  l  n   n 
v  2l x  2 x  n   n 1  l x  log1  l x   Ax  B
n 2 EI 0      

Pl 3  1  1 Pl 3
v1  log 8  2 . 5 
49 EI 0  7  17.55 EI
0

Pl 2  n   n 
 1  l x   log1  l x   A
n 2 EI 0    

Pl 2 2
1  log 8  7   1 Pl
49 EI 0 9.95 EI 0

B. Tapered Beam –stiffness solution

A tapered beam is characterized by a variable cross section along its length. This in turn is
reflected in a non uniform bending moment of inertia; that is

I = I(x) for the tapered beam.

The simplest idealization of the tapered beam is represented by the single

Element having some average moment of inertia I. A better approximation can be achieved if
two elements are used, each element will have its own value for I.

2 3
1

l=2L

 x
The Moment of Inertia is calculated from I ( x )  I 0 1  n 
 l

If n = 7

11 25
I12  I0 I 2 3  I0
4 4
 v1 1 L v 2 2L
 12  6  12  6
EI  6 4 6 2 
K 3 
L   12 6 12 6 
  6 2 6 4 

v1 1 L v2 2L
 33.0  16.5  33.0  16.5 v1
EI  16.5 11.0 16.5 5.5  1 L
K 1 2  30 
L   33.0 16.5 33.0 16.5  v 2
  16.5 5 .5 16.5 11.0   2 L
v2 2L v3 3L
 75.0  37.5  75.0  37.5 v 2
EI  37.5 25.0 37.5 12.5   2 L
K 2 3  30 
L   75.0 37.5 75.0 37.5  v3
  37.5 12.5 37.5 25.0   3 L

Forming the total stiffness equation

v1 1 L v2 2L v3 3L
 33.0  16.5  33.0  16.5 0 0  v1
  16.5 11.0 16.5 5 .5 0 0  1 L

EI a  33.0 16.5 108.0  21.0  75.0  37.5 v 2
K  3  
L   16.5 5 .5  21.0 36.0 37.5 12.5   2 L
 0 0  75.0 37.5 75.0 37.5  v3
 
 0 0  37.5 12.5 37.5 25.0   3 L

Boundary conditions are given by

v3   3  0

 Y1 
 M1   33.0  16.5  33.0  16.5   v1 
  EI  16.5 11.0 16 . 5 5 . 5   L 
 Y   3   33.0 16.5
L  1 
0

L  108  21 . 0   v2 
 2 
M
 2   16.5 5.5  21.0 36.0   L 
 L   2 

Y1 = -P

Y2 = 0

M1 = 0

M2 = 0

 v 2  0.446  0.206  v1 
   
 2 L  0.719  0.273 1 L 

And therefore

 P  EI 0   33.0  16.5  33.0  16.5 0.446  0.206  v1 

   3     
 0  L   16.5 11.0   16.5 5.5  0.719  0.273 1 L 

 v1  L3  6 .1 5 .2    P 
  5.2 6.5  0 
1 L  12.4 EI 0   

1 Pl 3 1 Pl 3
v1   VS v1 
16.3 EI 0 17.55 EI 0

1 Pl 2 1 Pl 2
1   VS 1  
9.55 EI 0 9.95 EI 0

If a single element is used to represent the beam with I equal to the value given by
 x
I ( x)  I 0  (1  n  at x =l/2, we obtain
 l

1 Pl 3
v1  
13.5 EI 0
 1 Pl 2
1  
9 EI 0

C. Curved-beam element – stiffness matrix

P2,u2

M2,θ2


2 Q2,v2
P1,u1
S
1
M1,θ1
Q1,v1
Nodes exist at the end points. Positive nodal forces and directions are taken as shown in the
figure. v is always radial (positive inward) whereas u is always tangential. Flexural stiffness is
represented by EI, calculated for a radial cross section. Coordinate S lies along the arc of the
circle. Angle  subtends the entire element, and  locates an arbitrarily point S on the elastic
curve.

Strain Energy takes the form

1
U
2 EI  M 2 dS

Where M  M 1  P1 R (1  cos  )  Q1 RSin

Introducing Rd= dS

R
2 EI 
U M 2 d

Applying Castigliano’s theorem gives the displacements at node 1 as

U R  M
u1  
P1 EI 0
M
P1
d

U R  M
v1  
Q1 EI  0
M
Q1
d

U R  M
1  
M 1 EI  0
M
M 1
d

With M  M 1  P1 R (1  cos  )  Q1 RSin and collecting terms into a single matrix equation
the gives

 3 sin 2  sin 2  
  2 sin   cos    1   sin  
u1   2 4 2   P1 R 
  R2  sin 2   sin 2   
v1   cos   1  cos   1  Q1 R 
  EI  2 2 4  
 1    sin  cos   1   M 1 
 
 

Let
a    sin 
sin 2 
b  cos   1
2
3 sin 2 
c  2 sin  
2 4
 sin 2 
d 
2 4
e  cos   1

u1  c b a   P1 R 
  R   
2

v1   b d e  Q1 R 
  EI
 1 a e    M 1 

 C 
 A B
  u 
R
 1
P  1
  EI  E  
 Q1   3  B D R v1 
M  R G    
 1 C E F 2  1 
 R R R 
    

Where

e2 ae
A d B b
 
a2
C  ad  be D c

E  ce  ab F  b 2  cd
 ae   e 2  a2d
G  b b  2   c  d  
     

 C 
 A B
 
R

EI  E 
K 11  3 B D R
R G  
C E F 2
 R R R 
    

At node 2, forces can now be found by imposing equilibrium conditions

P1  P2 cos   Q2 sin   0

Q1  P2 sin   Q2 cos   0

M 1  P1 R(1  cos  )  Q1 R sin   M 2  0

The first two equations may be solved for P2 and Q2

P2   P1 cos   Q1 sin 

Q2  P1 sin   Q1 cos 

The third equilibrium equation gives M2 directly in terms of forces at node 1

M 2   M 1  P1 R(1  cos  )  Q1 R sin 

Collecting these results provides

 P2    cos   sin  0   P1 
    
 Q2    sin   cos  0   Q1 
M   R (cos   1) R sin   1 M 1 
 2 

 P2  u1 
   
 Q2   AK 11 v1 
M   
 2  1
  
 
 ( A cos   B sin  )  ( B cos   D sin  )  (C cos   E sin  ) R 
   u 
 P2   R  
1
  EI
 Q2   3  A sin   B cos  B sin   D cos  (C sin   E cos  )  v1 
M  R G    
 2   A(cos   1)   B(cos   1)   1 
   C (cos   1)  R 2

 R  B sin   C  R 
 E sin   F   
 D sin   E   
       

 
 
 ( A cos   B sin  )  ( B cos   D sin  )  (C cos   E sin  ) R 
 
EI  R 
K 21  3  A sin   B cos  B sin   D cos  (C sin   E cos  ) 
R G  
  A(cos   1)   B (cos   1)  
     C (cos   1)  R 2 
 R  B sin   C  R
 D sin   E   E sin   F   
        


The total stiffness equation for the curved-beam element may be written as

 P1   u1 
Q  v 
 1  1
 M 1   K 11 K 12  1 
   
 P2   K 21 K 22  u 2 
 Q2  v 2 
   
M 2   2 

K11 and K22 have already been found. Also K21 = K’12

 C 
 A B
 
R

EI  E 
K 22  3 B D R
R G  
C E F 2
 R  R R 
    

For the special case  = 90◦

 u1 v1 1 u2 v2 2
 42.901 39.394 9.940 R  39.394 42.901  12.996 R 
 39.394 42.901 12.996 R  42.901 39.394  9.940 R 

EI  9.940 R 12.996 R 5.462 R 2  12.996 R 9.490 R  1.956 R 2 
K 3 
R   39.394  42.901  12.996 R 42.901  39.394 9.490 R 
 42.901 39.394 9.490 R  39.394 42.901  12.996 R 
 
 12.996 R  9.9490 R  1.956 R  12.996 R 5.462 R 2 
2
9.490 R

CURVED-BEAM

Problem 1. Clamped Quarter Circle

M
Y
X
1
R

X = P1, Y = Q1 and M = M1

u2 = v2 = 0

X   42.901 39.394 9.940 R  u1 

  EI   
 Y   3  39.394 42.901 12.996 R  v1 
M  R 9.940 R 12.996 R 5.462 R 2   
    1 

Matrix Inversion yields the following equations for the unknown displacements
u1   0.356  0.500 0.571   X 
  R   
3

v1     0.500 0.785  1.000 R   Y 

  EI  0.571  1.000 R 1.570 R   M 
 1   
R

Problem 2: Closed Ring

y
P/2
θ
P
A 1

u
v
B
x
2

P P/2

It is desired to find defections at points A and B, as well as the internal moments at these points for the
closed ring when loaded as shown.

Consider half-circle represented by segments 1-2 and 2-3. Symmetry provides the following conditions
on the nodal displacement components

u1  u 3  0 1   3  0 v1  v3 u2 = 0 θ2 = 0

Q1  EI  42.901 39.394   v1 

  3   
Q2  R 2(39.394) 2(42.901) v 2 

Inverting gives

 v1  R 3  0.146  0.067  Q1   P 

    2
v 2  EI  0.134 0.073   Q2  0 

The displacements are therefore found to be

 PR 3 PR 3
v1  0.073 v 2  0.067
EI EI

The theoretically correct answers are

 1   PR  1 1  PR
3
PR 3 3
PR 3
v1      0.076 v2      0.067
  8  EI EI   4  EI EI

The bending moments at A and B can now be calculated

M1 
EI
12.996 R.v1  9.490 R.v2 )  0.317 PR 
R3

M2 
EI
 9.490 R.v1  12.996 R.v2 )  0.181PR 
R3

The theoretical solution gives numerical factors of 0.317 and 0.183 for M1 and M2

IV. GENERAL CASE OF THE ONE-DIMENSIONAL MEMBER

A. Stiffness matrix –torsion

 Mt1,1 1 2 Mt2,2

Case A: 2 = 0

The member is assumed to be supported at node 2. The displacement 1 at node 1 is known from
structural theory

Mt1 L GJ
1  M t1  1
GJ L

Where J = torsional stiffness factor for the element cross section

E
G = shearing modulus =
21  v 

Equilibrium conditions

M t 2   M t1

Case B: 1 = 0

The member is assumed to be supported at node 1. The displacement 2 at node 2 is known from
structural theory

Mt2L GJ
2  M t2  2
GJ L

Equilibrium conditions

M t1   M t 2

Collecting Case A and B into a single matrix equation

 M t 1  GJ  1  1 1 
    
M t 2  L  1 1   2 

The stiffness matrix for the one dimensional member in torsion is

GJ  1  1
K  1 1 
L  

B. Bending plus torsion

Mb1,θ1 Mb2,θ2
Mt1,1 1 2

Mt2,2

L
Y1,v1 Y2,v2

Stiffness Matrix for the member is obtained by simply superimposing the individual cases of bending
and torsion
 v1 1 1 v2 2 2
 12 EI 
 L3 SYM 
 6 EI 4 EI 
 2 
 L L 
 0 0
GJ 
 L 
K 
12 EI 6 EI 12 EI
 3 0 
 L L2 L3 
  6 EI 2 EI
0
6 EI 4 EI 
 L2 L L2 L 
 GJ GJ 
 0 0  0 0 
 L L 

C. Bending plus axial loading

Mb1,θ1 Mb2,θ2
X1,u1 1 2

X2,u2

L
Y1,v1 Y2,v2

Stiffness Matrix for the member is obtained by simply superimposing the individual cases of bending
and axial loading
 u1 v1 1 u2 v2 2
 AE 
 L SYM 
 12 EI 
 0 
 L3 
 0 6 EI 4 EI  Stiffness matrix for the
 2
 L L  basic beam element
K  
AE AE referred to local
 0 0 
 L L  coordinates
 0 
12 EI 6 EI
0
12 EI 
 L3 L2 L3 
 6 EI 2 EI 6 EI 4 EI 
 0  2 0 
 L L L2 L 

Stiffness matrix for the member referred to reference coordinates

u1 v1 1 u2 v2 2
 12 I 2 
 A  L2 
2
SYM 
 
  A  2 
12 I 12 I
A 2  2 2 
  L  L 
 6I 6I 
  
E 
4I
K  L L 
L   A2  12 I  2    A  12 I  6I
  A2 
12 I 2
 
  L 2
  L 
2
L L2 
  12 I   12 I  6 I  12 I  12 I 2 
   A  2    A 2  2 2    A  2  A 2   
  L   L  L  L  L2 
 6I 6I 6I 6I 
    2I    4I 
L L L L 

C. Bending about two axes

y
 1 2

Consider x z plane (due to M y1 )

v1  y1 v2  y2
 12 6 12 6
 L3   
L2 L3 L2 
 6 4 6 2 
 2 
K y  EI y  L L L2 L 
  12 6 12 6 
 L3 L2 L3 L2 
 6 2 6 4 
 2 
 L L L2 L 

where I y is the moment of inertia of cross section about the y principal axis

Consider x y plane (due to M z1 )

w1  z1 w2  z2
 12 6 12 6 
 L3 
L2 L3 L2 
 6 4 6 2 
 2  2 
K z  EI z  L L L L 
 12 6
 2
12
 2 
6
 L3 L L3 L
 6 2 6 4 
 2  2 
 L L L L 
 v1  y1 w1  z1 v 2  y2 w2  z2
 12 I y 
 
 L3 
 6I y 4I y 
  L2 L 
 12 I z 
 0 0 
 L3 
 6I z 4I z 
 0 0 
K  E L2 L 
 12 I y 6I y 12 I y 
 L3 L2
0 0
L3 
 6I 2I y 6I y 4I y 
 y 0 0 
 L2 L L2 L 
 12 I z 6I z 12 I z 
 0 0   0 0 
 
3 2
L L L3
 6I z 2I z 6I z 4I z 
 0 0 0 0 
L2 L L2 L 

To transform the stiffness matrix to arbitrary reference axes,

 
   x x vx 
     
T   y y vy 
  
   z v z 
  z
 

D. The General Case (Axial loading, bending in each of the two principal planes,
and torsion)
1 2