Module 3: Element Properties

Lecture 6: Isoparametric Formulation

40

3.6.1 Necessity of Isoparametric Formulation

The two or three dimensional elements discussed till now are of regular geometry (e.g. triangular
and rectangular element) having straight edge. Hence, for the analysis of any irregular geometry, it is
difficult to use such elements directly. For example, the continuum having curve boundary as shown
in the Fig. 3.6.1(a) has been discretized into a mesh of finite elements in three ways as shown.

(a) The Continuum to be discritized (b) Discritization using Triangular Elements (c)
Discritization using rectangular elements (d) Discritization using a combination of
rectangular and quadrilateral elements
 41

Fig 3.6.1 Discretization of a continuum using different elements

Figure 3.6.1(b) presents a possible mesh using triangular elements. Though, triangular elements can
suitable approximate the circular boundary of the continuum, but the elements close to the center
becomes slender and hence affect the accuracy of finite element solutions. One possible solution to
the problem is to reduce the height of each row of elements as we approach to the center. But,
unnecessary refining of the continuum generates relatively large number of elements and thus
increases computation time. Alternatively, when meshing is done using rectangular elements as
shown in Fig 3.6.1(c), the area of continuum excluded from the finite element model is
significantlyadequate to provide incorrect results. In order to improve the accuracy of the result one
can generate mesh using very small elements. But, this will significantly increase the computation
time. Another possible way is to use a combination of both rectangular and triangular elements as
discussed in section 3.2. But such types of combination may not provide the best solution in terms of
accuracy, since different order polynomials are used to represent the field variables for different
types of elements. Also the triangular elements may be slender and thus can affect the accuracy. In
Fig.3.6.1(d), the same continuum is discritized with rectangular elements near center and with four-
node quadrilateral elements near boundary. This four-node quadrilateral element can be derived
from rectangular elements using the concept of mapping. Using the concept of mapping regular
triangular, rectangular or solid elements in natural coordinate system (known as parent element) can
be transformed into global Cartesian coordinate system having arbitrary shapes (with curved edge or
surfaces). Fig.3.6.2 shows the parent elements in natural coordinate system and the mapped elements
in global Cartesian system.
 42

 
 43

(a) Natural Coordinate System (b) Global Coordinate System

Fig. 3.6.2 Mapping of isoparametric elements in global coordinate system

3.6.2 Coordinate Transformation

The geometry of an element may be expressed in terms of the interpolation functions as follows.
n
x  N1 x1  N 2 x2  ...  N n xn   N i xi
i 1
n
(3.6.1)
y  N1 y1  N 2 y2  ...  N n yn   N i yi
i 1
n
z  N1 z1  N 2 z2  ...  N n zn   N i zi
i 1

Where,
n=No.of Nodes
Ni =Interpolation Functions
x i ,yi ,zi =Coordinates of Nodal Points of the Element
One can also express the field variable variation in the element as
 44

n
  , ,     N i  , ,   i (3.6.2)
i 1

As the same shape functions are used for both the field variableand description of element geometry,
the method is known as isoparametricmapping. The element defined by such a method is known as
an isoparametric element. This method can be used to transform the natural coordinates of a point to
the Cartesian coordinate system and vice versa.

Example 3.6.1
Determine the Cartesian coordinate of the point P (ξ= 0.8, η= 0.9) as shown in Fig. 3.6.3.

Fig. 3.6.3 Transformation of Coordinates

Solution:
As described above, the relation between two coordinate systems can be represented through their
interpolation functions. Therefore, the valuesof the interpolation function at point P will be
(1   )(1  n) (1  0.8)(1  0.9)
N1    0.005
4 4
(1   )(1  n) (1  0.8)(1  0.9)
N2    0.045
4 4
(1   )(1  n) (1  0.8)(1  0.9)
N3    0.855
4 4
(1   )(1  n) (1  0.8)(1  0.9)
N4    0.095
4 4

Thus the coordinate of point P in Cartesian coordinate system can be calculated as

 45

4
x   N i xi  0.005  1  0.045  3  0.855  3.5  0.095 1.5  3.275
i 1
4
y   N i yi  0.005  1  0.045 1.5  0.855  4.0  0.095  2.5  3.73
i 1

Thus the coordinate of point P (ξ= 0.8, η= 0.9) in Cartesian coordinate system will be 3.275, 3.73.

Solid isoparametric elements can easily be formulated by the extension of the procedure followed for
2-D elements. Regardless of the number of nodes or possible curvature of edges, the solid element is
just like a plane element which is mapped into the space of natural co-ordinates, i.e,
  1,  1,   1 .

3.6.3 Concept of Jacobian Matrix

A variety of derivatives of the interpolationfunctions with respect to the global coordinates are
necessary to formulate the element stiffness matrices. As the both element geometry andvariation of
the shape functions are represented in terms of the naturalcoordinates of the parent element,some
additional mathematical obstacle arises. For example, in case of evaluation of the strain vector, the
operator matrix is with respect to x and y, but the interpolation function is with x and h . Therefore,
the operator matrix is to be transformed for taking derivative with x and h . The relationship between
two coordinate systems may be computed by using the chain rule of partial differentiation as
¶ ¶ ¶x ¶ ¶y ¶ ¶ ¶x ¶ ¶y
= + and = + (3.6.3)
¶ x ¶x ¶x ¶y ¶ x ¶h ¶ x ¶ h ¶y ¶h
The above equations can be expressed in matrix form as well.
ì ¶ üï é ¶x ¶y ù ìï ¶ üï ìï ¶ üï
ïï
ïï ¶x ïïï êê ¶x ú ïï ïï ïï ïï
ï ¶x ú ï ¶ ï ï ¶x ïï
í ï = ê úï í ï  = [ J ]ï
x
í  (3.6.4)
ï
ï ¶ ïï ê ¶x ¶y ú ïï ¶ï ï ï
ï ¶ ïï
ï
ï ï ê úï ï ï ï
ï ¶h ïï ëê ¶h
î
ï
¶h ûú î ï
ï ¶y ï ïî ¶y ïï
ï

é ¶x ¶y ù
ê ú n
ê ¶x
The matrix [J] is denoted as Jacobian matrix which is: ê
¶x ú
ú .As we know, x   N i xi
ê ¶x ¶y ú i 1
ê ú
êë ¶h ¶h úû
n
  N i xi
x n
N i
where, nis the number of nodes in an element. Hence, J11   i 1
 xi
  i 1 
Similarly one can calculate the other terms J12, J21and J22 of the Jacobian matrix. Hence,
 46

é n ¶N i n ù
¶N i
êå x å ¶x yi ú
ê i=1 ¶x i ú
[ J ] = êê n i =1
ú (3.6.5)
ê å ¶N i xi
n
¶N i úú
ê i=1 ¶h å
i =1 ¶h
yi ú
ë û

From eq. (3.6.4), one can write

ì ¶ ïü
ï ìï ¶ üï
ï ï ïï ïï
ïï ¶x ïï ï ¶x ï
ï
í ï = [ J ] ïí ï
- 1
(3.6.6)
ï
ï ¶ï ï ïï ¶ ïï
ï
ïï ï
ï ïï ïï
î ¶y ï îï ¶h ï

éJ* J12* ùú
Considering ê 11 are the elements of inverted [J] matrix, we may arise into the following
êJ* * ú
J 22
ë 21 û
relations.
¶ ¶ ¶
= J11* ⋅ + J12* ⋅
¶x ¶x ¶h
(3.6.7)
¶ * ¶ * ¶
= J 21 ⋅ + J 22 ⋅
¶y ¶x ¶h

Similarly, for three dimensional case, the following relation exists between the derivative operators
in the global and the natural coordinate system.

    x y z     
       
  x   x 
      
    x y z     
      J   (3.6.8)
         y   y 
    x y z     
       
         z   z 

Where,
 47

 x y z 
    

 x y z 
J    (3.6.9)
    
 x y z 
    

[J] is known as the Jacobian Matrix for three dimensional case. Putting eq. (3.6.1) in eq. (3.6.9) and
after simplifying one can get
 N i N i N i 
  xi 
yi
 
zi

n
 N N i N i 
 J     i xi (3.6.10)
 
yi zi
i 1
  
 N i N i N i 
  xi 
yi
 
zi


From eq. (3.6.8), one can find the following expression.

   
 x    
   
 1   
   J    (3.6.11)
 y    
   
   
 z    

 J11* J12* J13* 

 * * 
Considering  J    J 21
1 *
J 22 J 23  we can arrived at the following relations.
 J 31
* *
J 32 * 
J 33 

¶ ¶ ¶ ¶
= J11* ⋅ + J12* ⋅ + J13* ⋅
¶x ¶x ¶h ¶z
¶ * ¶ * ¶ * ¶
= J 21 ⋅ + J 22 ⋅ + J 23 ⋅ (3.6.12)
¶y ¶x ¶h ¶z
¶ * ¶ * ¶ * ¶
= J 31 ⋅ + J 32 ⋅ + J 33 ⋅
¶z ¶x ¶h ¶z
 Module 3: Element Properties
Lecture 7: Stiffness Matrix of Isoparametric Elements

48

3.7.1 Evaluation of Stiffness Matrix of 2-D Isoparametric Elements

For two dimensional plane stress/strain formulation, the strain vector can be represented as
ì ü ì ü
ïïï ¶u ïïï ï ï
ï J11* ⋅
¶u
+ J12* ⋅
¶u ï
ï
ï
ï ¶x ï ïï ï ¶x ¶h ï
ïìï ex ïüï ï ï ï ï ï
ï
ïï ïï ï ¶v ï ïï ï ï ¶ v ¶ v ïï
{e} = í e y  = í =í
*
J 21 ⋅ + J 22 ⋅*
 (3.7.1)
ïï ïï ïï ¶y ïï ïï ¶x ¶h ï
ï
ïîïg xy ïï ïï ï ï ïï ï
ï
ïï ¶ v ¶ u ïï ï ¶ v ¶ v ¶ u ¶ u ï
+ ï ï * *
J11 ⋅ + J12 ⋅ *
+ J 21 ⋅ + J 22 ⋅ ï
*
ï
ï ¶x ¶y ï
îï
ï
ï ï
ïî ¶x ¶h ¶x ¶h ï ï
ï


The above expression can be rewritten in matrix form

ìï ¶u ü
ïï ï ï
ïï ¶x ï ï
ïï ï ï
é J11* ¶u ï
ê J12* 0 0 ùú ïïï ï ï
ï
* ú ï ¶h ï
{e} = êê 0 ïí ï (3.7.2)
*
0 J 21 J 22 ú
êJ * ï ¶v ïï
ê 21
*
J 22 J *
11 J12* úú ï ï ï
ë û ï ¶x ïï
ï
ïï ï
ïï ¶v ï ï
ïï ï ï
îï ¶h ï
ï
n
For an n node element the displacementu can be represented as, u = å N i ui and similarly for v&w.
i =1

Thus,

ìï ¶u üï é ¶N1 ¶N n ù
ïï ï ï êê ¶x  0  0 ú
ïïï ¶x ïï ¶x ú ïìïu ü ï
ê ú ï 1 ïï
ïï ¶u ï ïï ê ¶N1 ¶N n ú  ïï
ï
ïï ïï ê  0  0 ú ïï ï
ï ¶ h ï ê ¶h ¶h úï ïun ïï
ï
íï ï = êê úí ï (3.7.3)
ï ¶v ï ¶N1 ¶N n úú ïï v1 ï
ïïï ïïï êê 0  0  ï ï ï
¶x ¶x ¶x úú ïï  ï
ïïï ïïï ê ï ï ï
ïïï ¶v ïïï ê 0
ê ¶N1 ï vn ï
¶N n úú îïï ï
ï

 0 
ïîï ¶h ïï êë ¶h ¶h úû

As a result, eq.(3.7.2) can be written using eq. (3.7.3) which will be as follows.
 49

é ¶N1 ¶N n ù ìïu1 üï
ê  0  0 ú ïï ïï
ê ¶x ¶x ú ïï: ïï
ê ú ïï ïï
é J11
* * ù ê ¶N ¶N n ú ï: ï (3.7.4)
ê J12 0 0 úê 1  0  0 ú ïï ïï
ê úïu ï
{e} = êê 0 0 J*21 J*22 úú ê ¶h ¶h
ú ïí n ï
ê J* * úê ¶N1 ¶N n ú ïïv1 ïï
ê 21 J*22 *
J11 J12 ú êê 0  0  úï ï
ë û ¶x ¶x ú ïï: ïï
ê úï ï
ê ¶N1 ¶N n ú ïï: ïï
ê 0  0  úï ï
ëê ¶h ¶h úû ïïïv n ïïï
î 
Or,
{e} = [ B]{d} (3.7.5)

Where {d} is the nodal displacement vector and [B] is known as strain displacement relationship
matrix and can be obtained as
é ¶N1 ¶N n ù
ê  0 0 ú 
ê ¶x ¶x ú
ê ú
é J11
* * ù ê ¶N ¶N n ú (3.7.6)
J12 0 0 úê 1  0  0 ú
ê
* ê ¶h ¶h ú
[ B] = êê 0 0 J 21 J 22 úú êê
*
ú
ê J* J * J * J* ú ê 0 ¶N1 ¶N n ú
ê 21 22 11 12 ú ê  0  ú
ë û ¶x ¶x ú
ê ú
ê ¶N1 ¶N n ú
ê 0  0  ú
êë ¶h ¶h úû
It is necessary to transform integrals from Cartesian tothe natural coordinates as well for calculation
of the elemental stiffness matrix in isoparametric formulation. The differential area relationship can
be established from advanced calculus and the elemental area in Cartesian coordinate can be
represented in terms of area in natural coordinates as:
dA = dx dy = J dx dh (3.7.7)

Here J is the determinant of the Jacobian matrix. The stiffness matrix for a two dimensional
element may be expressed as
T
[ k ] = òòò [ B] [ D ][ B]dW = t òò [ B] [ D ][ B]dxdy
T (3.7.8)
W A

Here, [B] is the strain-displacement relationship matrix and t is the thickness of the element. The
above expression in Cartesian coordinate system can be changed to the natural coordinate system as
follows to obtain the elemental stiffness matrix
 50

+1 +1
T
[k ] = t ò ò [B] [D ][B] J dxdh (3.7.9)
-1 -1

Though the isoparametric formulation is mathematically straightforward, the algebraic difficulty is

significant.

Example 3.7.1:
Calculate the Jacobian matrix and the strain displacement matrix for four nodetwo dimensional
quadrilateral elements corresponding to the gauss point (0.57735, 0.57735) as shown in Fig.3.6.4.

Fig.3.7.1 Two dimensional quadrilateral element

Solution:
The Jacobian matrix for a four node element is given by,
é n ¶N i n ù
¶N i
êå x å ¶x yi ú
ê i=1 ¶x i ú
[ J ] = êê n i =1
ú
ê å ¶N i xi
n
¶N i úú
ê i=1 ¶h å
i =1 ¶h
yi ú
ë û

For the four node element one can find the following relations.
(1 - x)(1 - h) ¶ N1 1 - h ¶ N1 1- x
N1 = , =- , =-
4 ¶x 4 ¶h 4
(1 + x)(1- h) ¶N 2 1 - h ¶N 2 1+ x
N2 = , = , =-
4 ¶x 4 ¶h 4
(1 + x)(1 + h) ¶N 3 1 + h ¶N 3 1 + x
N3 = , = , =
4 ¶x 4 ¶h 4
 51

(1- x)(1 + h) ¶N 4 1 + h ¶N 4 1 - x
N4 = , =- , =
4 ¶x 4 ¶h 4
Now, for a four node quadrilateral element, the Jacobian matrix will become
é ¶N ¶N 2 y1 ù
¶N 3 ¶N 4 ù é x1
ê 1 ú úê
ê ¶x ¶x y2 ú¶x ¶x ú ê x 2
[J] = ê ú úê
ê ¶N ¶N 2 y3 ú
¶N 3
ú ¶N 4 ú êê x 3
ê 1 ú
êë ¶h ¶h y 4 úû
¶h ¶h úû êë x 4
é 1- h 1- h 1 + h 1 + h ù éê x1 y1 ù
ú
ê- - úê
ê 4 4 4 4 ú êx2 y2 ú
=ê ú ú
ê 1- x 1 + x 1 + x 1 - x ú êê x 3 y3 ú
ú
ê- - ú
ë 4 4 4 4 û êë x 4 y 4 úû

Putting the values of ξ & η as 0.57735 and 0.57735 respectively, one will obtain the following.
¶N1 ¶N1
= -0.10566 = -0.10566
¶x ¶h
¶N 2 ¶N 2
= 0.10566 = -0.39434
¶x ¶h
¶N3 ¶N3
= 0.39434 = 0.39434
¶x ¶h
¶N 4 ¶N 4
= -0.39434 = 0.10566
¶x ¶h
4
¶N i
Hence, J11 = å xi = -0.10566´1 + 0.10566´ 3 + 0.39434´3.5 - 0.39434´1.5 = 1.0
i =1 ¶x
Similarly,J12 =0.64632, J21 =0.25462 and J22 =1.14962.
Hence,
é1.00000 0.64632ù
J=ê ú
êë 0.25462 1.14962 úû
Thus, the inverse of the Jacobian matrix will become:
é * J12* úù é 1.1671 -0.6561ù
é J * ù = ê J11 =ê ú
ëê ûú ê J * * ú
J 22 êë-0.2585 1.0152 úû
ë 21 û
Hence strain displacement matrix is given by,
 52

é ¶N1 ¶N n ù
ê  0  0 ú
ê ¶x ¶x ú
ê ú
é J11
* * ù ê ¶N ¶N n ú
J12 0 0 úê 1  0  0 ú
ê ê ú
[ B] = êê 0 0 J*21 J*22 úú ê ¶h ¶h
ú
ê J* * úê ¶N1 ¶N n ú
ê 21 J*22 *
J11 J12 ú êê 0  0  ú
ë û ¶x ¶x ú
ê ú
ê ¶N1 ¶N n ú
ê 0  0  ú
êë ¶h ¶h úû
é 1.1671 -0.6561 0 0 ùú
ê
= êê 0 0 -0.2585 1.0152 úú ´
ê-0.2585 1.0152 1.1671 -0.6561úú
ëê û
é-0.10566 0.10566 0.39434 -0.39434 0 0 0 0 ù
ê ú
ê ú
ê-0.10566 -0.39434 0.39434 0.10566 0 0 0 0 ú
ê 0 0 0 0 -0.10566 0.10566 0.39434 -0.39434úú
ê
ê ú
ë 0 0 0 0 -0.10566 -0.39434 0.39434 0.10566 û

é-0.0540 0.3820 0.2015 -0.5294 0 0 0 0 ù

ê ú
=ê 0 0 0 0 -0.0800 -0.4276 0.2984 0.2092 ú
ê ú
ê-0.0800 -0.4276 0.2984 0.2092 -0.0540 0.3820 0.2015 -0.5294ú
ë û

3.7.2 Evaluation of Stiffness Matrix of 3-D Isoparametric Elements

Stiffness matrix of 3-D solid isoparametric elements can easily be formulated by the extension of the
procedure followed for plane elements. For example, the eight node solid element is analogous to the
four node plane element. The strain vector for solid element can be written in the following form.
 53

 u 
 x 
 
 u 
 y 
 
 u 
 z 
  x  1 0 0 0 0 0 0 0 0   v 
   0  
 y  0 0 0 1 0 0 0 0   x 
  z  0 0 0 0 0 0 0 0 1   v 
   
 xy  0 1 0 1 0 0 0 0 0   y 
 yz  0 0 0 0 0 1 0 1 0   v 
    
 zx  0 0 1 0 0 0 1 0 0   z 
 w 
 x 
 
 w 
 y 
 
 w  (3.7.10)
 z 

The above equation can be expressed as

 u 
  
 
 u 
  
 u   
 x   u 
    
 v   
 y   J11* J12* J13* 0 0 0 0 0 0   v 
   
 w   0 0 0 *
J 21 *
J 22 *
J 23 0 0 0    
 
 z   0 * * * 
 v 
    u v    *
0 0 0 0 0 J 31 J 32 J 33
 
    J 21
*
J 22 *
J 23 J11* J12* J13* 0 0 0    
 y x   0 0 0 *
J 31 *
J 32 *
J 33 *
J 21 *
J 22 * 
J 23 v 
   *  
 v  w   J 31 J13    
* *
J 32 J 33 0 0 0 J11* J12* *

 z y   w 
   
 x  w    
 z x   w 
 
  
 v  (3.7.11)
 
  
 54

8
For an 8 node brick element u can be represented as, u = å N i ui and similarly for v&w.
i =1

u N
8
u N u N 8 8
  i ui ,   i ui &   i ui
 i 1   i 1   i 1 
v 8
N v 8
N v 8
N
  i vi ,   i vi &   i vi (3.7.12)
 i 1   i 1   i 1 

w 8 N i w 8 N i w 8
N
 wi ,  wi &   i wi
 i 1   i 1   i 1 

Hence eq. (3.7.11) can be rewritten as

 N i 
  0 0 
 
 N i 
J *
J *
J *
0 0 0 0 0 0   0 
0 

11 12 13
* * *   
0 0 0 J 21 J 22 J 23 0 0 0   N i 
8 
0 0  ui 
0 0 0 0 0 0 J* *
J 32 * 
J 33     (3.7.13)
    *    vi 
31

 J 21
* *
J11* J12* J13* 0 0 0  i 1  N i N i 
0   wi 
J 22 J 23
0 * * * * * *    
0 0 J 31 J 32 J 33 J 21 J 22 J 23  
 * 
 J 31
*
J 32 *
J 33 0 0 0 J11* J12* J13*   N i N i 
 0   
 
 N i N i 
  0
  
Thu, the strain-displacement relationship matrix [B] for 8 node brick element is
 55

 N i 
  0 0 
 
 N i 
J *
J *
J *
0 0 0 0 0 0   0 
0 

11 12 13
* * *   
0 0 0 J21 J22 J23 0 0 0   N i 
8 
0 0
0 0 0 0 0 0 *
J 31 *
J 32 * 
J 33  
 B   *    (3.7.14)
 J 21
*
J 22 *
J 23 J11* J12* J13* 0 0 0  i 1  N i N i 
  0 
0 0 0 *
J 31 *
J 32 *
J 33 *
J 21 *
J 22 * 
J 23 
 *   
 J 31
*
J 32 *
J 33 0 0 0 J11* J12* J13*   N i N i 
 0   
 
 N i N i 
  0
  

The stiffness matrix may be found by using the following expression in natural coordinate system.
+1 +1 + 1
T T
[ k ] = òòò [ B] [ D ][ B]dW = òòò [ B] [ D ][ B]dxdydz = ò (3.7.15)
ò ò [B] [D][B]dxdhdz J
T

W V -1 -1 -1
 Module 3: Element Properties
Lecture 8: Numerical Integration: One Dimensional

56

The integrations, we generally encounter in finite element methods, are quite complicated and it is
not possible to find a closed form solutions to those problems. Exact and explicit evaluation of the
integral associated to the element matrices and the loading vector is not always possible because of
the algebraic complexity of the coefficient of the different equation (i.e., the stiffness influence
coefficients, elasticity matrix, loading functions etc.). In the finite element analysis, we face the
problem of evaluating the following types of integrations in one, two and three dimensional cases
respectively. These are necessary to compute element stiffness and element load vector.

ò f (x) dx; ò f (x, h) dxdh; ò f (x, h, z) dxdhdz; (3.8.1)

Approximate solutions to such problems are possible using certain numerical techniques. Several
numerical techniques are available, in mathematics for solving definite integration problems,
including, mid-point rule, trapezoidal-rule, Simpson’s 1/3rd rule, Simpson’s 3/8th rule and Gauss
Quadrature formula. Among these, Gauss Quadrature technique is most useful one for solving
problems in finite element method and therefore will be discussed in details here.

3.8.1 Gauss Quadrature for One-Dimensional Integrals

The concept of Gauss Quadrature is first illustrated in one dimension in the context of an integral in
+1 x2
the form of I = ò f (x) dx from ò f (x)dx . To transform from an arbitrary interval of x1≤ x ≤ x2
-1 x1

to an interval of -1 ≤ ξ ≤ 1, we need to change the integration function from f(x) to ϕ(ξ) accordingly.
Thus, for a linear variation in one dimension, one can write the following relations.
1-x 1+ x
x= x1 + x 2 = N1x1 + N 2 x 2
2 2
1- (-1) 1 -1
so for x = -1, x = x1 + x 2 = x1
2 2
x = +1, x = x 2
x2 +1
\I=ò f (x)dx = ò f ( x ) dx
x1 -1

Numerical integration based on Gauss Quadrature assumes that the function ϕ(ξ) will be evaluated
over an interval -1 ≤ ξ ≤ 1. Considering an one-dimensional integral, Gauss Quadrature represents
the integral ϕ(ξ) in the form of
+1 n
I=ò f (x ) dx » å w i f (x i ) » w 1f (x1 ) + w 2f (x 2 ) + .. + w h f (x h ) (3.8.2)
-1
i=1
 57

Where, the ξ1, ξ2, ξ3, ..., ξn represents n numbers of points known as Gauss Points and the
corresponding coefficients w1, w2, w3, …, wn are known as weights. The location and weight
coefficients of Gauss points are calculated by Legendre polynomials. Hence this method is also
sometimes referred as Gauss-Legendre Quadrature method. The summation of these values at n
sampling points gives the exact solution of a polynomial integrand of an order up to 2n-1. For
example, considering sampling at two Gauss points we can get exact solution for a polynomial of an
order (2×2-1) or 3. The use of more number of Gauss points has no effect on accuracy of results but
takes more computation time.

3.8.2One- Point Formula

Considering n = 1, eq.(3.8.2) can be written as
1

ò-1
f(x)dx » w 1f(x1 ) (3.8.3)

Since there are two parameters w1 and x1 , we need a first order polynomial for ϕ(ξ) to evaluate the
eq.(3.8.3) exactly. For example, considering, f (x) = a 0 + a1x ,
1
Error = ò (a 0 + a1x)dx - w1f (x1 ) = 0
-1

 2a 0 - w1 (a 0 + a1x1 ) = 0

 a 0 (2 - w1 ) - w1a1x1 = 0 (3.8.4)

Thus, the error will be zero if w1 = 2 and x1 = 0 . Putting these in eq.(3.8.3), for any general ϕ, we
have
1
I = ò f ( x ) d x = 2f ( 0 ) (3.8.5)
-1

This is exactly similar to the well known midpoint rule.

3.8.3 Two-Point Formula

If we consider n = 2, then the eq.(3.8.2) can be written as
1

ò f (x) dx » w 1f (x1 ) + w 2f (x 2 ) (3.8.6)

-1

This means we have four parameters to evaluate. Hence we need a 3rd order polynomial for ϕ(ξ) to
exactly evaluate eq.(3.8.6).
Considering, f(x) = a 0 + a1x + a 2x2 + a 3x3
é 1 ù
Error = ê ò (a 0 + a1x + a 2x2 + a 3x3 ) dxú - éë w1f (x1 ) + w 2f(x2 )ùû
ëê -1 ûú
 58

2
 2a 0 + a 2 - w1 (a 0 + a1x1 + a 2 x12 + a 3x13 ) - w 2 (a 0 + a1x 2 + a 2 x 2 2 + a 3x 23 ) = 0
3
æ2 ö
 (2 - w 1 - w 2 ) a 0 - ( w 1x1 + w 2 x 2 ) a1 + çç - w 1x12 - w 2 x 2 2 ÷÷÷ a 2 - ( w 1x13 + w 2 x 2 3 ) a 3 = 0
çè 3 ø

Fig 3.8.1 One-point Gauss Quadrature

Requiring zero error yields

w1 + w 2 = 2
w1x1 + w 2 x 2 = 0
2 (3.8.7)
w1x12 + w 2x 22 =
3
3 3
w1x1 + w 2 x 2 = 0
These nonlinear equations have the unique solution as
w1 = w 2 = 1 x1 = -x2 = -1 3 = -0.5773502691 (3.8.8)
From this solution, we can conclude that n-point Gaussian Quadrature will provide an exact solution
if ϕ(ξ) is a polynomial of order (2n-1) or less. Table 3.8.1 gives the values of w1 and x1 for Gauss
Quadrature formulas of orders n = 1 through n = 6. From the table it can be observed that the gauss
 59

points are symmetrically placed with respect to origin and those symmetrical points have the same
weights. For accuracy in the calculation maximum number digits for gauss point and gauss weights
should be taken. The Location and weights given in the Table 3.8.1 must be used when the limits of
integration ranges from -1 to 1. Integration limits other than [-1, 1], should be appropriately changed
to [-1, 1] before applying these values.

Table 3.8.1 Gauss points and corresponding weights

Number of Gauss points, n Gauss Point Location, xi Weight, wi
1 0.0 2.0
2 ±0.5773502692 (= 1 3) 1.0

3 0.0 0.8888888889 (=8/9)

±0.7745966692 (=  0.6 ) 0.5555555556 (=5/9)
4 ±0.3399810436 0.6521451549
±0.861363116 0.3478548451
5 0.0 0.5688888889
±0.5384693101 0.4786286705
±0.9061798459 0.2369268851
6 ±0.2386191861 0.4679139346
±0.6612093865 0.3607615730
±0.9324695142 0.1713244924

Example 1:
1æ 2x ö÷
Evaluate I = ò ççe x - 2 ÷dx using one, two and three point gauss Quadrature.
ç
0 è x - 2 ÷ø

Solution:
Before applying the Gauss Quadrature formula, the existing limits of integration should be changed
from [0, 1] to [-1, +1].Assuming, x = a + bx , the upper and lower limit can be changed. i.e., at x =
0,ξ = -1 and at x = 1, ξ = +1. Thus, putting these conditions and solving for a & b, we get a = -1 and
b = 2. The relation between two coordinate systems will become x = 2x -1 and dx = 2dx .
Therefore the initial equation can be written as
 60

æ æ x +1ö÷ ö÷
çç æ ö 2 çç ÷
1 ç ççç
x+1 ÷
çç è 2 ÷ø÷ çè 2 ÷÷ø ÷÷÷
I = ò çe - ÷÷dx
-1 ç ÷
2
çç æ x + 1ö
÷ - 2 ÷÷
çç ÷ ÷
çè çè 2 ÷ø ø÷

1 1 æçç 2 4 (x +1) ö÷÷

(x+1)

2 ò-1çè
Or, I = ç e - ÷dx
(x +1) - 8 ÷÷ø
2

Using one point gauss Quadrature:

w1 = 2, x1 = 0 and
I » 2f (0)
æ1 æ 4 öö
Or I » 2 çç ççe0.5 + ÷÷÷÷÷÷ = 2.22015
ç
èç 2 è 7 øø
Using two point gauss Quadrature:
w1 = w 2 = 1
x1 = -0.5773502692
x 2 = 0.5773502692
Putting these values and calculating, I = 2.39831

Using three point gauss Quadrature:

w1  0.555555556
1  0.774596669
w2  0.888888889
2  0.000000000
w3  0.555555556
3  0.774596669
and I = 2.41024
This may be compared with the exact solution as Iexact = 2.41193
 Module 3: Element Properties
Lecture 9: Numerical Integration: Two and Three Dimensional
61

Numerical integrations using Gauss Quadrature method can be extended to two and three
dimensional cases in a similar fashion. Such integrations are necessary to perform for the analysis of
plane stress/strain problem, plate and shell structures and for the three dimensional stress analysis.

3.9.1 Gauss Quadrature for Two-Dimensional Integrals

For two dimensional integration problems the above mentioned method can be extended by first
evaluating the inner integral, keeping η constant, and then evaluating the outer integral. Thus,

1 1 1 é n ù n é n ù
I=ò
-1 ò-1
f ( x, h ) dx dh » ê å
ò-1 êë i=1 w i f ( x i , h )ú
úû
dh » å w j
ê å
êë i=1
w i f ( x i , h j )ú
úû
i=1

Or,
n n
I » åå w i w jf (xi , h j ) (3.9.1)
i=1 j=1

In a matrix form we can rewrite the above expression as

 1 ,1   1 , 2   1 , n    w1 
 
   2 ,1    2 , 2    2 , n    w2 
I   w1  wn  (3.9.2)
  
w2
 
  
   n ,1     n ,  2     n ,  
n   wn 

Example 1:
y=d = 4 x = b=3
2 2
Evaluate the integral: I = ò ò (1- x ) (2 - y) dxdy
y=c=-4 x =a = 2

Solution:
Before applying the Gauss Quadrature formula, the above integral should be converted in terms of
x and h and the existing limits of y should be changed from [-4,4] to [-1, 1] and that of x is from
[2,3] to [-1,1].
 62

(b - a ) (b + a ) (x + 5) dx
x= x+ = ; dx =
2 2 2 2
( d - c ) ( d + c )
y= h+ = 4h; dy = 4dh
2 2
h=+1 x=+1 2 h=+1 x=+1
æ 3 + x ö÷
I = 2 ò ò çç
2
çè 2 ÷ø ÷ ( 2 - 4h ) dxdh = ò ò f (x, h)dxdh
h=-1 x=-1 h=-1 x=-1
2
æ 3 + x ö÷
f (x, h) = 2 çç
2 2 2
where
çè 2 ø÷ ÷ ( 2 - 4h ) = 2 (3 + x ) (1 - 2h )

Fig. 3.9.1 Gauss points for two-dimensional integral

 63

1 1 1 1
x1 = - ; h1 = - ; x2 = ; h2 =
3 3 3 3
2 2
æ 1 ö æ 2 ö
f (x1 , h1 ) = 2 çç3 - ÷÷ çç1 + ÷÷ = 54.49857
çè ÷
3ø èç 3 ÷ø
2
æ ö
çç 3 + 1 ÷÷ 2

f (x 2 , h1 ) = çç
ç 3 ÷÷÷ æ ö
çç2 + 4 ÷÷ = 118.83018
çç 2 ÷÷÷ èç 3ø÷
çç ÷÷
è ø
2
æ ö
çç 3 + 1 ÷÷ 2
ç 3 ÷÷÷ æ 4 ö
f (x 2 , h 2 ) = çç
çç 2 ÷÷÷ ççç2 - ÷÷÷ = 0.61254
è 3ø
çç ÷÷
è ø
2
æ ö
çç 3 - 1 ÷÷ 2
çç ÷
3 ÷÷ æ 4 ö
f (x1 , h 2 ) = ç
çç 2 ÷÷÷ ççç2 - ÷÷÷ = 0.28093
è 3ø
çç ÷÷
è ø
é f (x1 , h1 ) f (x1 , h 2 )ù ïì w 1 ïü
I = {w 1 w 2 } êê ú ïí ï
f ( x , h ) f ( x , h )ú ïw ï
ë 2 1 2 2 ûîï 2 ï

é 54.49857 0.28093ù ìï 1ïü

= {1 1} ê ú íï ï
êë118.83018 0.61254úû ï
ïî1ï
ï

= 174.22222 agrees with the exact value 174.22222

3.9.2 Gauss Quadraturefor Three-Dimensional Integrals

In a similar way one can extend the gauss Quadrature for three dimensional problems also and the
integral can be expressed by.
1 1 1 n n n
I=ò ò ò f (x, h, z) dxdhdz » ååå w i w j w k f (xi , h j , z k ) (3.9.3)
-1 -1 -1
i=1 j=1 k =1

The above equation will produce exact value for a polynomial integrand if the sampling points are
selected as described earlier sections.
 64

3.9.3 Numerical Integration of Element Stiffness Matrix

As discussed earlier notes, the element stiffness matrix for three dimensional analyses in natural
coordinate system can be written as
+1 +1 + 1
T T
[ k ] = òòò [ B] [ D ][ B]dW = òòò [ B] [ D ][ B]dxdydz = ò (3.9.4)
ò ò [B] [D][B]dxdhdz J
T

W V -1 -1 -1

Here, [B] and [D] are the strain displacement relationship matrix and constitutive matrix respectively
and integration is performed over the domain. As the element stiffness matrix will be calculated in
natural coordinate system, the strain displacement matrix [B] and Jacobian matrix [J] are functions
of x, h and z . In case of two dimensional isoparametric element, the stiffness matrix will be
simplified to
+1 +1

[k ] = t ò (3.9.5)
ò [B] [D][B]dxdh J
T

-1 -1

This is actually an 8×8 matrix containing the integrals of each element. We do not need to integrate
elements below the main diagonal of the stiffness matrix as it is symmetric. Considering,
f (x, h) = t[B]T [D][B] J , the element stiffness matrix will become after numerical integration as
n n
[ k ] = åå w i w jf(xi , h j ) (3.9.6)
i=1 j=1

Using a 2×2 rule, we get

[ k ] = w12f (x1 , h1 ) + w1w 2f (x1 , h2 ) + w 2 w1f (x 2 , h1 ) + w 22f (h2 , h2 ) (3.9.7)

Where w1 = w2 = 1.0, x1 = h1 =-0.57735....,and x2 = h2 = +0.57735.... Here, wn is the weight

factor at integration point n. A suitable computer program can be written to calculate the element
stiffness matrix through the numerical integration. The process of obtaining stiffness matrix using
Gauss Quadrature integration will be demonstrated through a numerical example in module 5.

3.10.4 Gauss Quadrature for Triangular Elements

The procedure described for the rectangular element will not be applicable directly. The Gauss
Quadrature is extended to include triangular elements in terms of triangular area coordinates.
n
I = òò f (L1 , L 2 , L3 ) dA »å w if (Li1 , Li2 , Li3 ) (3.9.8)
A i=1

Where, L terms are the triangular area coordinates and the wi terms are the weights associated with
those coordinates. The locations of integration points are shown in Fig. 3.9.2.
 65

Fig. 3.9.2 Gauss points for triangles

The sampling points and their associated weights are described below:
For sampling point =1 (Linear triangle)
1
w1 = 1 L11 = L12 = L13 = (3.9.9)
3
For sampling points =3 (Quadratic triangle)
1 1
w1 = L11 = L12 = , L13 = 0
3 2
1 1
w2 = L21 = 0, L22 =L23 = (3.9.10)
3 2
1 1 1
w3 = L31 = , L32 = 0, L33 =
3 2 2
For sampling point = 7 (Cubic triangle)
27 1
w1 = L11 = L12 = L13 =
60 3
8 1
w2 = L21 = L22 = , L23 = 0
60 2
8 1
w3 = L31 = 0, L32 = L33 =
60 2
8 1
w4 = L41 = L43 = , L42 = 0 (3.9.11)
60 2
3
w5 = L51 = 1, L52 = L53 = 0
60
3
w6 = L61 = L63 = 0, L62 = 1
60
3
w7 = L71 = L72 = 0, L73 = 1
60
 66

3.10.5 Gauss Quadrature for Tetrahedron

The Gauss Quadrature for triangles can be effectively extended to include tetrahedron elements in
terms of tetrahedron volume coordinates.
n
I = òò f (L1 , L 2 , L3 , L 4 ) dA »å w if (Li1 , Li2 , Li3 , Li4 ) (3.9.12)
A i=1

Where, L terms are the volume coordinates and the wi terms are the weights associated with those
coordinates. The locations of Gauss points are shown in Fig. 3.9.3.

Fig. 3.9.3Gauss points for tetrahedrons

The sampling points and their associated weights are described below:
For sampling point =1 (Linear tetrahedron)
1
w1 = 1 L11 = L12 = L13 = L14 = (3.9.13)
4
For sampling points = 4 (Quadratic tetrahedron)
1
w1 = L11 = 0.5854102, L12 = L13 = L14 = 0.1381966
4
1
w2 = L22 = 0.5854102, L21 = L23 = L24 = 0.1381966
4
(3.9.14)
1
w3 = L33 = 0.5854102, L31 = L32 = L34 = 0.1381966
4
1
w4 = L44 = 0.5854102, L41 = L42 = L43 = 0.1381966
4
For sampling points = 5 (Cubic tetrahedron)
 67

4 1
w1 = - L11 = L12 = L13 = L14 =
5 4
9 1 1
w2 = L21 = , L22 = L23 = L24 =
20 3 6
9 1 1
w3 = L32 = , L31 = L33 = L34 = (3.9.15)
20 3 6
9 1 1
w4 = L43 = , L41 = L42 = L44 =
20 3 6
9 1 1
w5 = L54 = , L51 = L52 = L53 =
20 3 6
 68

Worked out Examples

Example 3.1 Calculation of displacement using area coordinates
The coordinates of a three node triangular element is given below. Calculate the displacement at
point P if the displacements of nodes 1, 2 and 3 are 11 mm, 14mm and 17mm respectively using the
concepts of area coordinates.

1 1 2 3
A= 1 1 5 4 = [(30-12) - (12-9) + (8-15)] = = 4
1 1 3 6

1 1 3 4
1 = 1 5 4 = [(30-12) - (18-12) + (12-20)] = = 2
1 1 3 6

Fig.Ex.3.1 Nodal coordinates of a triangular element

1 1 3 4
= 1 = 1 3 6 = [(9-12) - (9-8) + (18-12)] = = 1
1 1 2 3
 69

1 1 3 4
= 1 = 1 2 3 = [(8-15) - (12-20) + (9-8)] = = 1
1 1 5 4

= = = 0.5

= = = 0.25

= = = 0.25

u =   + +
= 0.5 x 11 + 0.25 x 14 + 0.25 x 17 = 13.25 mm
 70

Example 3.2 Derivation of shape function of four node triangular element

Derive the shape function of a four node triangular element.

Fig.Ex.3.2Degrading for four node element

The procedure for four node triangular element is the same as five node triangular element to derive
its interpolation functions. Here, node 5 and 6 are omitted and therefore displacements in these
nodes can be expressed in terms of the displacements at their corner nodes. Hence,
u 2 + u3 u1 + u 3
u '5 = and u '6 =
2 2 (3.11.1)
Substituting the values of u’5 and u’6 in eq.(3.3.8), the following relations can be obtained.
(u 2 + u 3 ) ( u 3 + u1 )
u = N1u1 + N 2 u 2 + N3u 3 + N 4 u 4 + N5 + N6
2 2
(3.11.2)
æ N ö æ N ö æ N + N 6 ÷ö
= çç N1 + 6 ÷÷÷ u1 + çç N 2 + 5 ÷÷÷ u 2 + çç N3 + 5 ÷÷ u 3 + N 4 u 4
èç 2 ø èç 2 ø çè 2 ø
Now, the displacement at any point inside the four node element can be expressed by its nodal
displacement with help of shape function.
u = N1¢u1 + N¢2 u 2 + N3¢ u 3 + N¢4 u 4 (3.11.3)
Comparing eq. (3.11.2) and eq. (3.11.3), one can find the following relations.
N6 4L L
N1¢ = N1 + = L1 (2L1 -1) + 3 1 = L1 (1- 2L 2 )
2 2
N 4L L
N ¢2 = N 2 + 5 = L 2 (2L 2 -1) + 2 3 = L 2 (1- 2L1 )
2 2 (3.11.4)
N + N6 4L L + 4L3L1
N 3¢ = N 3 + 5 = L3 (2L3 -1) + 2 3 = L3
2 2
N ¢4 = N 4 = 4L1L 2
Thus, the shape functions for the four node triangular element are
 71

N1¢ = L1 (1- 2L 2 )
N ¢2 = L 2 (1- 2L1 )
(3.11.5)
N 3¢ = L3
N ¢4 = 4L1L 2

Example 3.3 Numerical integration for two dimensional problems

3

ò (x + 11x - 32) dx using one, two and three point gauss Quadrature.
2
Evaluate the integral: I =
-2

Also, find the exact solution for comparison of accuracy.

Solution:
The existing limits of integration should be changed from [-2, +3] to [-1, +1]. Assuming, x = a + bx ,
the upper and lower limit can be changed. i.e., at x1 = -2, x1 =-1 and at x2 = 3, x2 = +1. Thus,
putting these limits and solving for a&b, we get a = -0.2 and b = 0.4. The relation between two
coordinate systems will become:
5x + 1
x = -0.2 + 0.4x or x = and dx = 2.5dx
2

Thus, the initial equation can be written as

3 éæ 5x + 1ö2+1 ù
I = ò ( x + 11x - 32) dx = 2.5ò êêçç
2 ÷÷ + 11æçç 5x + 1ö÷÷ - 32úd x
çè 2 ø÷ èç 2 ø÷ ú
-2 -1 êë úû

(i) Exact Solution:

3

I = ò ( x 2 + 11x - 32) dx
-2
3
é x 3 11x 2 ù
=ê + - 32 xú
ê3 2 ú
ë û -2
é 99 ù é 8 ù
= ê9 + - 96ú - ê- + 22 + 64ú
êë 2 úû êë 3 úû
= -37.5 - 83.33333 = -120.83333
Thus, Iexact = -120.83333
 72

(ii) One Point Formula:

+1

I = ò f(x) d x = w1f(x1 )
-1

For one point formula in Gauss Quadrature integration, w1 = 2, x1 = 0 . Thus,

éæ 5´ 0 + 1ö2 æ 5´ 0 + 1ö÷ ù
ê
I1 = 2 ´ 2.5 êçç ÷ ç
+ 11ç - 32úú
çè 2 ÷÷ø ç 2 ÷÷ø
è
êë úû
é 1 11 ù
= 5 ê + - 32ú = -131.25
êë 4 2 úû
Thus, % of error = (120.83333-131.25)×100/120.83333 = 8.62%

(iii) Two Point Formula:

Here, for two point formula in Gauss Quadrature integration,
1
w1 = w 2 = 1.0 and x1 = -x 2 = - . Thus,
3
I 2 = w1f(x1 ) + w2f(x 2 )
éæ -5 ö
2
æ -5 ö ù éæ 5 ö
2
æ 5 ö÷ ù
êç + 1÷
÷÷ ç + 1÷
÷÷ ú ê ç + 1÷
÷÷ ç + 1 ÷÷ ú
êçç  3 çç  3
÷÷ + 11ç ú ê
÷÷ - 32ú + 1.0 ´ 2.5´ êç
çç  3 çç  3
÷÷ + 11ç ÷÷ - 32úú
1.0 ´ 2.5´ êçç ÷ ç ÷ ç ÷ ç
êçç 2 ÷÷ çç 2 ÷÷ ú êçç 2 ÷÷ çç 2 ÷÷÷ ú
êç ÷ ç ÷ ú ê ç ÷ ç ÷ ú
êëè ø è ø úû êëè ø è ø úû

= (0.88996 -10.37713 - 32)´ 2.5 + (3.77671 + 21.3771- 32)´ 2.5

= -48.3333´ 2.5
= -120.83325
Thus, % of error = (120.83333-120.83325)×100/120.83333 = 6.62×10-05

(iv) Three Point Formula:

Here, for three point formula in Gauss Quadrature integration,
w1 = 0.8889, x1 = 0.0
w2 = 0.5556, x 2 = +0.7746
w3 = 0.5556, x3 = -0.7746
Thus,
I 3 = w1f(x1 ) + w2f(x2 ) + w3f(x3 )
 73

éæ 5´ 0 + 1ö2 5´ 0 + 1 ù
ê
I 3 = 0.8889 ´ 2.5´ êç ç ÷ + 11´ - 32úú
çè 2 ÷ø÷ 2
ëê ûú
éæ 5´ 0.7746 + 1ö2 5´ 0.7746 + 1 ù
+ 0.5556 ´ 2.5´ êç ê ç ÷
÷÷ + 11´ - 32úú
çè 2 ø 2
êë úû
éæ -5´ 0.7746 + 1ö2 -5´ 0.7746 + 1 ù
+ 0.5556 ´ 2.5´ êç ê ç ÷
÷÷ + 11´ - 32úú
çè 2 ø 2
êë úû
I 3 = 0.8889 ´ 2.5´[0.25 + 5.5 - 32]
+ 0.5556 ´ 2.5´[5.9365 + 26.8015 - 32]
+ 0.5556 ´ 2.5´[ 2.0635 -15.8015 - 32]
= 2.5´(-23.3336 + 0.4100 - 25.4120)
= -2.5´ 48.3356 = -120.839
Thus, % of error = (120.83333-120.839)×100/120.83333 = 4.69×10-03. However, difference of
results will approach to zero, if few more digits after decimal points are taken in calculation.

Example 3.4 Numerical integration for three dimensional problems

1 1 1 2 2 2
Evaluate the integral: I = ò
-1 ò ò
-1 -1
(1- 2x) (1- h) (3z - 2) dxdhdz

Solution:
Using two point gauss Quadrature formula for the evaluation of three dimensional integration, we
have the following sampling points and weights.
w1 = w 2 = 1
x1 = -0.5773502692
x 2 = 0.5773502692
h1 = -0.5773502692
h 2 = 0.5773502692
z1 = -0.5773502692
z 2 = 0.5773502692
2 2 2
Putting the above values, in f (x, h, z) = (1- 2x) (1-h) (3z - 2) one can find the following values
in 8 (i.e., 2 × 2 × 2) sampling points.
 74

f (x1 , h1 , z1 ) = 160.8886
f (x1 , h1 , z 2 ) = 0.8293
f (x1 , h 2 , z1 ) = 11.5513
f (x1 , h 2 , z 2 ) = 0.0595
f (x 2 , h1 , z1 ) = 0.8293
f (x 2 , h1 , z 2 ) = 0.0043
f (x 2 , h2 , z1 ) = 0.0595
f (x 2 , h2 , z 2 ) = 0.0003
2 2 2
Now, I = ååå w i w j w k f (xi , h j , z k )
i=1 j=1 k =1

Thus, I = w1w1w1f (x1 , h1 , z1 ) + w1w1w 2f (x1 , h1 , z 2 ) +  + w 2 w 2 w 2f (x 2 , h 2 , z 2 ) = 174.222,where

asIexact = 174.222.
 Module 4: Analysis of Frame Structures
Lecture 2: Analysis of Truss 7

4.2.1 Element Stiffness of a 3 Node Truss Member

Fig. 4.2.1 3-node truss member

Here, the displacement function using Pascal’s triangle can be expressed as:
 0 
 
u  x    0  1 x   2 x  1 x x  1 
2 2
(4.2.1)
 
 2

Applying boundary conditions:

At x= 0, u(0)= u 1 , x=L/2, u(L/2) = u and at x=L, u(L) = u
2 3

And solving for  0 ,  1 and  2

2u1  4u2  2u3
 0  u1 ,   3u1  4u2  u3 and  2 
1
L L2
Therefore,
 3x 2 x2   4x 4x2   x 2 x2 
u  x   1   2 
 1 
u  2  2   2
u   u 3   N u (4.2.2)
 L L   L L   L L 

Here, N is the shape function of the element and is expressed as:

 3 x 2 x 2   4 x 4 x 2   x 2x2  
   1   2    2 
N    2   (4.2.3)
 L L   L L   L L 

Now, the element stiffness matrix can be written as

 k     B   D  B  d 
T
(4.2.4)


d N   3 4x 4 8x 1 4x 
Where,  B      2   
dx  L L L L2 L L2 

So, the stiffness matrix will be:

 k     B   D  B  d   0  B  E  B  Adx
T L T


 8

3 4

4 8 3 4 4 8 1 4

1 4

16 24 40 32 16 16
9 12 3
40 32 64 64 24 32
12 16 4
16 16 24 32 8 16
3 4 1

(4.2.5)

After integrating the above equation, the stiffness matrix of the 3-node truss member will
become:
7 8 1
8 16 8 (4.2.6)
1 8 7

4.2.2 Worked Out Example

Analyze the truss shown below by finite element method. Assume the cross sectional area of
the inclined member as 1.5 times the area (A) of the horizontal and vertical members. Assume
modulus of elasticity is constant for all the members and is E.

Fig. 4.2.2 Plane truss

 9

Solution

The analysis of truss starts with the numbering of members and joints as shown below:

Fig. 4.2.3 Numbering of members and nodes

The member information for the truss is shown in Table 4.2.1. The member and node
numbers, modulus of elasticity, cross sectional areas are the necessary input data. From the
coordinate of the nodes of the respective members, the length of each member is computed.
Here, the angle  has been calculated considering anticlockwise direction. The signs of the
direction cosines depend on the choice of numbering the nodal connectivity.

Table 4.2.1 Member Information for Truss

Member Starting Ending Value Area Modulus of

No. Node Node of  Elasticity
1 1 2 90 A E
2 2 3 315 1.5A E
3 3 1 180 A E

Now, let assume the coordinate of node 1 as (0, 0). The coordinate and restraint joint
information are given in Table 4.2.2. The integer 1 in the restraint list indicates the restraint
exists and 0 indicates the restraint at that particular direction does not exist. Thus, in node no.
2, the integer 0 in x and y indicates that the joint is free in x and y directions.

Table 4.2.2 Nodal Information for Plane Truss

Node No. Coordinates Restraint List

x y x y
1 0 0 1 1
2 0 L 0 0
3 L 0 1 1
 10

The stiffness matrices of each individual member can be found out from the stiffness matrix
equation as shown below.

 cos 2  cos  sin   cos 2   cos  sin  

 
 cos  sin  sin 2   cos  sin   sin 2  
k   AE  
L   cos 2   cos  sin  cos 2  cos  sin  
 
 cos  sin   sin 2  cos  sin  sin 2  


Thus the local stiffness matrices of each member are calculated based on their individual
member properties and orientations and written below.
1 2 3 4 3 4 5 6
0 0 0 0 1 1 1 1 1 3
   
   
0 1 0  1 1 1 1  1 4
AE 
2 3 AE 
k 1    k 2   
L 0 0 0 0 4 2 L  1 1 1  1 5
3
   
   
0 1 0 1  4  1 1 1 1  6
and

5 6 1 2
1 0 1 0 5
 
 
0 0 0  1 6
AE 
k 3   
L  1 0 1 0 1
 
 
 0 0 0 0  2
 11

Global stiffness matrix can be formed by assembling the local stiffness matrices into globally.
Thus the global stiffness matrix are calculated from the above relations and obtained as
follows:
1 2 3 4 5 6
1 0 0 0 1 0  1
0 1 0 1 0 0 
 3 3 3 3 2
0 0   
 4 2 4 2 4 2 4 2  3
AE  3 
K    0  1 
3 3 3
1  
L  4 2 4 2 4 2 4 2 4
 1 0  3 3 3
1 
3 
 4 2 4 2 4 2 4 2 5
 3 3 3 3 
0 0   
 4 2 4 2 4 2 4 2  6

 Fx1   0 
F   
 y1   0 
 F   2 P 
The equivalent load vector for the given truss can be written as:  F    x 2    
 Fy 2   P 
 Fx 3   0 
   
 Fy 3   0 

Let us assume that u and v are the horizontal and vertical displacements respectively at joints.
Thus the displacement vector will be expressed as follows:

u1   0 
v   0 
 1  
u  u 
d    2    2 
v 2   v 2 
u 3   0 
   
 v3   0 

Therefore, the relationship between the force and the displacement will be:
 12

1 0 0 0 1 0
0 1 0 1 0 0
 
 Fx1   3 3 3 3  0
F   0 0     
 y1   4 2 4 2 4 2 4 2  0
 2 P  AE  3 3 3 3  u2 
   0 1  1   
P L  4 2 4 2 4 2 4 2   v2 
 Fx 3   3 3 3 3  0
   1 0  1    
 Fy 3   4 2 4 2 4 2 4 2 0
 3 3 3 3 
0 0  
 4 2 4 2 4 2 4 2 

From the above relation, the unknown displacements u2 and v2 can be found out through
computer programming. However, as numbers of unknown displacements in this case are
only two, the solution can be obtained by manual calculations. The above equation may be
rearranged with respect to unknown and known displacements in the following form:

 F   k k   d 
   
 F   k k  d  

Thus the developed matrices for the truss problem can be rearranged as:

2P  3 3 3  3
4 2 0 0
 4 2 4 2 4 2  u2
P   3 3 3 3 
1 0 1 v2
4 2 4 2 4 2 4 2 
 
Fx1 AE   0
 1
L 
0 0 1 0 0 
Fy1  0 1 0 1 0 0  0
 3 3 3 3 
 1 0 1  
Fx3 4 2 4 2 4 2 4 2 0
 3 3 3 3 
4 2 0 0 
4 2 
Fy3 0
 4 2 4 2
.
The above relation may be condensed into following

 3 3 
 2 P  AE  4 2 4 2  u 
    2
 P  L  3 1
3   v2 
4 2 4 2 

 13

The unknown displacements can be derived from the relationships expressed in the above
equation.
1
 3 3   3 3 

u2  AE 4 2 4 2  2 P  4 2 L 1  4 2 4 2  2 P 
        
 
v L  3 3   P  3 AE  3 3  P 
2
4 2 1
 4 2   4 2
 4 2 

Thus the unknown displacement at node 2 of the truss structure will become:

 8 2
u2  PL 3  
   3 
 v2  AE  3 
 

Support Reactions:

The support reactions {Ps} can be determined from the following relation:
Ps    Pcs    K   d 

Where, {Pcs} correspond to equivalent loadings at supports. Thus, the support reaction of the
present truss structure will be:

 0 0 
 0  0 
0  1 
0    8 2    3P 
AE 
Ps        3 3  PL 3   
 AE  3   2 P 
0  L  4 2 4 2
 3   2 P 
0   3 3   
 
4 2 4 2

Member End Actions:

Now, the member end actions can be obtained from the corresponding member stiffness and
the nodal displacements. The member end forces are derived as shown below.

Member –1

 0 
 Fmx1  0 0 0 0   0 
F     0   
 my1  AE 0 1 0 1   PL 3P 
       
 Fmx 2  L 0 0 0 0  3  8 2  AE  0 
 Fmy 2    3 
0 1 0 1   3P 
 3 
 14

Member – 2
 8 2
 Fmx 2   1 1 1 1  3    2P 
F   1 1 1 1  3   
 my 2  3 AE     PL  2 P 
    3    
 Fmx 3  4 2 L  1 1 1 1  0  AE  2 P 
 Fmy 3   
 1 1 1 1   0   2 P 
 

Member –3
 Fmx 3  1 0  1 0 0 0
F  
 my 3  AE  0
  
0 0 0 0 PL 0
     
 Fmx1  L  1 0 1 0 0 AE 0
 Fmy1   
0 0 0 0 0 0

Thus the member forces in all members of the truss will be:
 3P   3P 
   
Fm     2 P    2 P    2 2 P 
2 2

   0 
 0   

The reaction forces at the supports of the truss structure will be:
 0 
 3P 
FR    
2 P 
 2 P 

Thus the member force diagram will be as shown in Fig. 4.2.4.

Fig. 4.2.4 Member Force Diagram

 Module 4: Analysis of Frame Structures
Lecture 3: Stiffness of Beam Members 15

4.3.1 Introduction
A beam is a structural member which is capable of withstanding load primarily by resisting
bending. The primary tool for analysis of beam is the Euler–Bernoulli beam equation. Other
methods for determining the deflection of beams include "slope deflection method" and
"method of virtual work". For calculation of internal forces of beam include "moment
distribution method", force or flexibility method and stiffness method. However, all these
methods have limitations if either of geometry, loading, material properties or boundary
conditions becomes arbitrary in nature. Finite element techniques can well handle such cases
and relieve the analyzer of making simplifications to arrive approximate solutions.

4.3.2 Derivation of Shape Function

The degrees of freedom at each node for a beam member will be (i) vertical deflection and
dv
(ii) rotation. For a beam member, the slope of the elastic curve θ is given by:   , where
dx
the variable v is the displacement function of the beam. As the beam has two degrees of
freedom at each node, the variation of v will be cubic and can be expressed using Pascal’s
triangle as:
 0 
 
v x   0  1 x   2 x 2   3 x 3  1 x x 2 x 3  1  (4.3.1)
 2 
 3 
and
 0 
 

dv
  

 0 1 2 x 3x 2  1  (4.3.2)
dx  2 
 3 

Fig. 4.3.1 Beam element

 16

Now, applying boundary conditions, the following expressions from the above relations can
be obtained:

At x=0:
 0   0 
   
 
V1  1 0 0 0   1  ; 1   0 1 0 0  1  ;
 2   2 
 3   3 

At x=L:
 0   0 
   
 
V2  1 L L2 L3   1  ;  2  0 1 2 L 3L2   1 
 2   2 
 3   3 

Thus combining the above expressions one can write:

V1  1 0 00   0 
  0
 1  1 0 0  1 
      A  (4.3.3)
V2  1 L L2 L3   2 
 2  0 
1 2 L 3L2   3 

So,
1
 0  1 0 0  V1   1
0 0 0 0 
V1 
  0  0   
 1  1 0  2   0
0 1 0
    3 2 3 1  1  (4.3.4)
L L2 L3  V2   L2    
 2  1 L L2 L  V2 
 
 3  0 1 2 L 3L2  2   2 1 2 1   
 L3  3  2
L2 L L2 

Therefore,
 1 0 0 0
 0 V1  V1 
 3 1 0 0     
1 1 

v x   1 x x2 
x 3  2 
2 3
     N1 N2 N3
 
N 4  1  (4.3.5)
 l l l2 l  V2  V2 
 2 1 2 1     2 
 l 3  3  2
l2 l l 2 
Where,
3 2 2 3 2 2 x3 3x 2 2 x 3 x2 x3
N1  1  x  x ; N 2  x  x  ; N 3   and N 4   
L2 L3 L L2 L2 L3 L L2
(4.3.6)
 17

N is called shape function which interpolates the beam displacement in terms of its nodal
displacements.

4.3.3 Derivation of Element Stiffness Matrix

Now, the strain displacement relationship matrix [B] can be expressed from the following
expressions with the help of eq. (4.3.1):
 0 
 
d 2v  1 
 0 6 x    B    B A d 
1
 0 2 (4.3.7)
dx 2  2 
 
 4 

1 0 0 
0 V1 
0 1 0 0   
 
Where,  B    0 0 2 6 x  ;  A   ; d    2 
1 L L 2
L 
3
V2 
 2  2 
0 1 2 L 3L 

From the moment curvature relationship, we can write:

d 2v
 EI B  A d 
1
M  EI  EI 2
(4.3.8)
dx
Strain energy,

 d  A  B  B A d dx

L L
1 T
0 2   M dx = 2
EI 1 T 1
U  (4.3.9)
T T

Thus,
U
  B B A d dx
L
F    EI  A 1
T T 1
(4.3.10)
d  0

So, the stiffness matrix will be:

k   EI  A 1 T B T B A 1 dx  EI A 1 T  B T B dxA1

L L
(4.3.11)
0 0

0 0 0 0 0  0 0 0 0 
L   
L
0 
 
L
0 0 0 0 0 0 0 0 
Now, B T B dx    0 0 2 6 x dx    dx   
0 0
2 0
0 0 4 12 x  0 0 4L 6 L2 
6 x     
0 0 12 x 36 x 2 
0 0 6 L2 12L3 
(4.3.12)
 18

So,
0 0 0 0 
 
0 0 0 0  1
 
k   EI A1 T   A
0 0 4L 6 L2 
 
0 0 6 L2 12 L3 

 3 2 
1 0  0  1 0 0 0 
L2 L3  0 0 0
 2 1   
0 1   0 0 0 0   0 1 0 0 
k   EI  L L2    3 2 3 1
  
0 0
3 2
 3  0 0 4L 6 L2   L2 
L L2 L
 L2 L   2
  1 2 1 
1 1  0 0 6 L2 12 L3   3  3
0 0    L L2 L L2 
 L L2 

 12 6 12 6 
 1 0 0 0   L3 L2 L3 L2 
0 0 0 6  0   
 1 0 0   6 4 6 2 
0 0 2 0   3  2
2 3 1  L2 L 
 EI       EI 
L L
0 0 0 6  L2 L L2 L   123 6 12 6 
    2  2
0 0 2 6l   2 1 2 1   L L L3 L 
 3  6
 L3 L2 L 2 
L  2 6 4 
 2  2 
 L L L L 

Thus, the element stiffness of a beam member is:

 12 6 L 12 6 L 
 6 L 4 L2 6 L 2 L2 
EI
 k   3  12 6 L 12 6L  (4.3.13)
L
 2 
 6 L 2 L 6 L 4 L 
2

4.3.4 Generalized Stiffness Matrix of a Beam Member

Consider a beam member making an angle ‘’ with X axis as shown in Fig 4.3.2 below. By
resolving the forces along local X and Y direction, the following relations are obtained.
 19

Fx1  Fx1 cos   Fy1 sin 

Fx 2  Fx 2 cos   Fy 2 sin 
Fy1   Fx1 sin   Fy1 cos 
(4.3.14)
Fy 2   Fx 2 sin   Fy 2 cos 
M1  M1
M2  M2

Where, Fx1 and Fx 2 are the axial forces along the member axis X . Similarly, Fy1 and Fy 2 are

the forces perpendicular to the member axis X . M1 and M 2 are the moment about its axis at
node 1 and 2 respectively.

Fig. 4.3.2 Inclined beam member

The relationship expressed in eq. (4.3.14) can be rewritten in matrix form as follows:

 Fx1   cos  sin 0 0 0 0   Fx1 

  
 Fy1    sin cos  0 0 0 0   Fy1 
 M 1   0 0 1 0 0 0   M 1  (4.3.15)
   
 Fx 2   0 0 0 cos  sin 0   Fx 2 
 Fy 2   0 0 0  sin cos  0   Fy 2 
    
 M 2   0 0 0 0 0 1   M 2 

Now, the above equation can be expressed in short as:

F   T F  (4.3.16)
 20

Similarly, the displacement vector in local coordinate system ( , ) may be transformed to

global ( , ) coordinate system by the following relation.

d   T d  (4.3.17)

The force-displacement relation in local coordinate system may be expressed as:

0 0 0 0 0 0 
 6 EI   u 
 Fx1   0 12 EI 6 EI
0
12 EI
 3 1
   L3 L2 L L2   
 Fy1   6 EI 4 EI 6 EI 2 EI   1 
v
  0 0  2   
 M1   L2 L L L  1 (4.3.18)
   
 Fx2   0 0 0 0 0 0  u2 
   12 EI 6 EI 12 EI

6 EI  
 Fy2   0  3  2 0  2   v2 
M   L L L3 L  
 2  
6 EI 2 EI 6 EI 4 EI   2 
0 0  2 
 L2 L L L 

The matrices in the above equation are written with respect to the member axis. Now, the eq.
(4.3.18) can be rewritten as follows with the use of eqs. (4.3.16) and (4.3.17).

T F   k T d  (4.3.19)

Or,
F   T   k  T d 
1
(4.3.20)

Here, the transformation matrix [T] is orthogonal. Thus, from the above relationship, the
generalized stiffness matrix can be expressed as:

k   T T k T  (4.3.21)

Considering   cos  and   sin  the above expression can be written as follows:
 21

0 0 0 0 0 0 
 12 6 12 6 
  0  0 0  3   0
L2  
0 0 0 0 0 0
   L3 L2 L
  0 0 0 0  6 4 6 2    0 0 0 0 
 0 0  2 
0 0 1 0 0 
0  L2 L  0 0 1 0 0 0
 k   EI  
L L
 
0 0 0   0  0 0 0 0 0 0  0 0 0   0
0  
0 0   0  12 6 12 6  0 0 0   0
  0  3  2 0  2  
0 0 0 0 0 1  L L L3 L  0 0 0 0 0 1
 6 2 6 4 
0 0  2 
 L2 L L L 
(4.3.22)

Thus, the generalized stiffness matrix of a beam member is derived as:

 12  2 12  6 12  2 12  6 
 L3    
 L3 L2 L3 L3 L2 
 12  12 2 6 12  12 2 6 
  L3 L3 L2 L3
 3
L L2 
 
  6 6 4 6 6
 2
2 
 2
L2 L2 L 
 k   EI  L 2 L L
 (4.3.23)
  12  12  6 12  2 12 
 3
6 
 L3 L3 L2 L3 L L2 
 
 12  12 2
 3
6
 2
12 
 3
12 2 6
 2
 L3 L L L L3 L 
 6 6 2 6 6 4 
   2 
 L2 L2 L L2 L L 
 Module 4: Analysis of Frame Structures 22
Lecture 4: Analysis of Continuous Beam

4.4.1 Equivalent Loading on Beam Member

In finite element analysis, the external loads are necessary to be acting at the joints, which
does not happen always; as some forces may act on the member. The forces acting on the
member should be replaced by equivalent forces acting at the joints. These joint forces
obtained from the forces on the members are called equivalent joint loads. These joint loads
are combined with the actual joint loads to provide the combined joint loads, which are then
utilized in the analysis.

4.4.1.1 Varying Load

Let a beam is loaded with a linearly varying load as shown in the figure below. The
equivalent forces at nodes can be expressed using finite element technique. If w(x) is the
function of load, then the nodal load can be expressed as follows.
Q     N  w x dx
T

(4.4.1)
The loading function for the present case can be written as:
w 2  w1
w x   w1  x
(4.4.2)
L

Fig. 4.4.1 Varying load on beam

From eqs. (4.4.1) and (4.4.2), the equvalent nodal load will become
 L  2 x 3 3x 2   
  3  2  1 w x dx   7 w1  3w2  L 

 F1   0L 3 
L L   20 20  
    x 2x 2     w1 w2  2 
 M 1     L2  L  x  w x dx    20  30  L 
Q      0L 

 
 (4.4.3)
 2

 F2     2 x  3x  w x dx   1  2  L 
3
 3 w 7 w
   0  L3 L2    20 20  
M 2   L 3   w w  2 
x x 2

  2   wx dx    30  20  L 
1 2

 0  L L     
 23

Now, if w1=w2=w, then the equivalent nodal force will be:

 wL 
 2 
 
 wL 
2

 
Q   12  (4.4.4)
 wL 
 2 
 2
 wL 
 12 

4.4.1.2 Concentrated Load

Consider a force F is applied at a point is regarded as a limiting case of intense pressure over
infinitesimal length, so that p(x)dx approaches F. Therefore,
Q     N  x dx N 
T T
p  *
F
(4.4.5)

Fig. 4.4.2 Concentrated load on beam

Here, [N*] is obtained by evaluating [N] at point where the concentrated load F
is applied. Thus,

 2 x3 3x 2   2a 3 3a 2 
 3   1  3  2  1
 L3 L2   L3 L 
 x 2x 2
  a 2a 2 
 L2  L  x   L2  L  a 
 N *
T
 3 2 
at dista nce a   3 2  (4.4.6)
  2 x  3x    2a  3a 
 L3 L2   L3 L2 
 3 2   3 2 
 x x   a a 
 L2 L   L2 L 
 24

ïìïæç 2a 3 3a 2 ö ïü
ïç - 2 +1÷÷÷ Fïï
ïïçè L3
L ø÷ ïï
ïï ïï
ìï F1 üï ïï æç a 3 2a 2 ö÷ ïï
ïïï ïïï ïïï çç L2 - L + a ÷÷÷ F ïïï
Therefore, {Q} = ïí M1 ï = ïí è ø ï

ïï F2 ïï ïï æ 2a 3 3a 2 ö ïï (4.4.7)
ç ÷
ïïM ïï ïïï çèç- L3 + L2 ÷ø÷÷ F ïïï
îï 2 ï ï ïï
ïï ïï
ïï æç a 3 a 2 ö÷ ïï
ïï çç 2 - ÷
÷ F
ïî è L L ÷ø ïï


Now, if load F is acting at mid-span (i.e., a=L/2), then equivalent nodal load will be

 F 
 2 
 
 FL 
 
Q   8  (4.4.8)
 F 
 2 
 FL 
 
 8 

With the above approach, the equivalent nodal load can be found for various loading function
acting on beam members.

4.4.2 Worked Out Example

Analyze the beam shown below by the finite element method. Assume the moment of inertia
of member 2 as twice that of member 1. Find the bending moment and reactions at supports
of the beam assuming the length of span, L as 4 m, concentrated load (P) as 15 kN and udl, w
as 4 kN/m.

Fig. 4.4.3 Example of a continuous beam

 25

Solution
Step 1: Numbering of Nodes and Members

The analysis of beam starts with the numbering of members and joints as shown below:

Fig. 4.4.4 Numbering of nodes and members

The member AB and BC are designated as (1) and (2). The points A,B,C are designated
by nodes 1, 2 and 4. The member information for beam is shown in tabulated form as shown
in Table 4.4.1. The coordinate of node 1 is assumed as (0, 0). The coordinate and restraint
joint information are shown in Table 4.4.2. The integer 1 in the restraint list indicates the
restraint exists and 0 indicates the restraint at that particular direction does not exist. Thus, in
node no. 2, the integer 0 in rotation indicates that the joint is free rotation.

Table 4.4.1Member Information for Beam

Member Starting node Ending node Rigidity modulus

number

1 1 2 EI
2 2 3 2EI

Table 4.4.2 Nodal Information for Beam

Node No. Coordinates Restraint List

x y Vertical Rotation
1 0 0 1 1
2 L 0 1 0
3 2L 0 1 0
 26

Step 2: Formation of member stiffness matrix:

The local stiffness matrices of each member are given below based on their individual
member properties and orientations. Thus the local stiffness matrix of member (1) is:

1 2 3 4

 12 EI 6 EI 12 EI 6 EI 
 L3  1
L2 L3 L2 
 6 EI 4 EI 6 EI 2 EI 
  2  2
k 1   12
L2 L L L 
 EI 6 EI
 2
12 EI
 2 
6 EI 3
 L3 L L3 L 
 6 EI 2 EI 6 EI 4 EI 
  2  4
 L2 L L L 

Similarly, the local stiffness matrix of member (2) is:

3 4 5 6
 24 EI 12 EI 24 EI 12 EI 
 L3  3
L2 L3 L2 
 12 EI 8 EI 12 EI 4 EI 
  2  4
k 2  L
2
L L L 
 24 EI 12 EI
 2
24 EI
 2 
12 EI 5
 L3 L L3 L 
 12 EI 4 EI 12 EI 8 EI 
  2  6
 L2 L L L 

Step 3: Formation of global stiffness matrix:

The global stiffness matrix is obtained by assembling the local stiffness matrix of members
(1) and (2) as follows:

1 2 3 4 5 6

 12 EI 6 EI 12 EI 6 EI 
 L3  0 0  1
L2 L3 L2
 6 EI 4 EI 6 EI 2 EI 
  2 0 0  2
 L 
2
L L L
 12 EI 6 EI
 2
36 EI 6 EI

24 EI 12 EI 
3
 L2 
K 
3
 L L L L2 L3
6 EI 2 EI 6 EI 12 EI 12 EI 4 EI 
 2
 2  4
 L L L2 L L L 
 0 0
24 EI
 3
12 EI
 2
24 EI
 2 
12 EI
5
 L L L3 L 
 12 EI 4 EI 12 EI 8 EI 
 0 0  2  6
 L2 L L L 
 27

Step 4: Boundary condition:

The boundary conditions according to the support of the beam can be expressed in terms of
the displacement vector. The displacement vector will be as follows

 0
 0
 
 0
d    
 2 
 0
 
 3 

Step 5: Load vector:

The concentrated load on member (1) and the distributed load on member (2) are replaced by
equivalent joint load. The equivalent joint load vector can be written as

Fig. 4.4.5 Equivalent Load

 P 
  
2
 PL 
  
 8 
  P wL  
   2  2  
F    PL wL2 
  
 8 12 
 wL 
  
 2
2

 wL 
 12 

Step 6 : Determination of unknown displacements:.

The unknown displacement can be obtained from the relationship as given below:
 28

 F    K  d 
d    K   F 
1

1  P 
 12 EI 6 EI 12 EI 6 EI    
 L3  0 0  2

L2 L3 L2
  PL 
 0   6 EI 4 EI 6 EI 2 EI   
 2 0 0   8 
 0   L2 L L L 
   12 EI 12 EI    P wL  
    2  
6 EI 36 EI 6 EI 24 EI
  2 
 0   L3 L L L2 L3 L2     2
    6 EI 4 EI   PL wL2 
 2  
2 EI 6 EI 12 EI 12 EI
 2    
0  L  8 12 
2
L L2 L L L 
   24 EI 12 EI 24 EI 12 EI  
 3   0 0  3  2  2   
wL

L L L3 L 
 12 EI 4 EI 12 EI 8 EI   2 
 2
2
 0 0   wL 
 L2 L L L   12 

The above relation may be condensed into following

1
12 EI 4 EI   PL wL2   PL wL2 
   
 2  L
 L   8 12  L  2 1  8 12 

         
3   4 EI 8 EI   20 EI  1 3   wL
2 2
 wL 
 L L   12   12 

 PL wL 2 
 2  L  4  
4
    

 3 20 EI  PL wL 2


 8 3 

PL 2 wL 3
2  
80 EI 80 EI
PL wL3
3   
160 EI 60 EI

Step 7: Determination of member end actions:

The member end actions can be obtained from the corresponding member stiffness and the
nodal displacements. The member end actions for each member are derived as shown below.

Member-(1)
 29

 12 EI 6 EI   3P 3wL 
 40  40 
6 EI 12 EI
 L3 
L2 L3 L2   0   2 
 F1   6 EI 2 EI  
M  
4 EI 6 EI
 2  0   PL  wL 
 1 L  L2 L    40 40 
   3P 3wL 
 L L
   0
 F2  20 EI  3  2   PL wL2  
12 EI 6 EI 12 EI 6 EI 
 2 
M 2   L L L3 L     40 40 
 6 EI 2 EI 6 EI 4 EI   4 4   PL wL2 
  2   20  20 
 L2 L L L 

Member-(2)
 24 12 24 12   wL 3P 
 
 L2 L L2 L   0   20 40 
 F2  
 12   PL wL  3PL wL2 
2
12
M    4     
4    40
8
 2  EI  L L    4 30 
   
 F3  L  2
24 12 24 12 

0
  wL 3P 
   
 L   PL  wL   20 40 
2
 M 3  L L L2
 
 12   8 3   wL2
8  
12
 4  
 L L   12 

Actual member end actions:

Member (1)
 3P 3wL   P 
 40  40    23P 3wL 
  
 F1    
2
 40 40 
PL wL2
PL  6 PL wL2 
      
 M 1   40 40   8    
 
   3P 3wL   P    40 40 

 F2        17 P  3wL 
M   40 40   2   40 40 
 2   PL wL2   PL 
 3 PL wL2 
 20  20   8   40  20 
 
Member (2)
 wL 3P   wL 
 20  40   2  11wL 3P 
 F2        
 
2
 
2 20 40
  3 PL wL wL  3PL wL2 

 M 2   40 30   12    
       40 20 
 F3    wL  3P   wL   9wL 3P 
   20 40   2    
M 3   2   2  20 40 
 wL   wL   0 
 12   12 
 23P 3wL 
 40  40 
 A 
R 
The support reactions at the supports A, B and C are  FR    RB    25wL  P 
 R   40 2 
 C   9wL 3P 
  
 20 40 
 30

Putting the numerical values of L, P and w (P=15, L=4, w=4) the member actions and support
reactions will be as follows:

Member end actions:

 F2  9.925  F1  7.425
       
M 2   7.7   M 1   7.4 
   ,   
 F3  6.075  F2  7.575
       
 M 3   0  M 2    7.7 

Support reactions:
 R A  7.425
FR    RB    17.5 
 R  6.075
 C  
 Module 4: Analysis of Frame Structures 31
Lecture 5: Plane Frame Analysis

4.5.1 Introduction
The plane frame is a combination of plane truss and two dimensional beam. All the members
lie in the same plane and are interconnected by rigid joints in case of plane frame. The
internal stress resultants at a cross-section of a plane frame member consist of axial force,
bending moment and shear force.

4.5.2 Member Stiffness Matrix

In case of plane frame, the degrees of freedom at each node will be (i) axial deformation, (ii)
vertical deformation and (iii) rotation. Thus the frame members have three degrees of
freedom at each node as shown in Fig. 4.5.1 below.

Fig. 4.5.1 Plane frame element

Therefore, the stiffness matrix of the frame in its local coordinate system will be the
combination of 2-d truss and 2-d beam matrices:
u1 v1 θ1 u2 v2 θ2

0 0 0 0
0 0
0 0
= (4.5.1)
0 0 0 0
0 0
0 0

4.5.3 Generalized Stiffness Matrix

In plane frame the members are oriented in different directions and hence it is necessary to
transform stiffness matrix of individual members from local to global co-ordinate system
 32

before formulating the global stiffness matrix by assembly. The generalized stiffness matrix
of a frame member can be obtained by transferring the matrix of local coordinate system into
its global coordinate system. The transformation matrix can be expressed as:

cos sin 0 0 0 0
sin cos 0 0 0 0
0 0 1 0 0 0
[T]= (4.5.2)
0 0 0 cos sin 0
0 0 0 sin cos 0
0 0 0 0 0 1

Now, the generalized stiffness matrix of the member can be obtained from the relation of
. Thus considering cos and sin the stiffness matrix in global
coordinate system can be written as follows:
AE AE
0 0 0 0
L L
12EI 6EI 12EI 6EI
0 0
λ μ 0 0 0 0 L L L L
μ λ 0 0 0 0 6EI 4EI 6EI 2EI
0 0
0 0 1 0 0 0 L L L L
K EI 
0 0 0 λ μ 0 AE AE
0 0 0 0
0 0 0 μ λ 0 L L
0 0 0 0 0 1 12EI 6EI 12EI 6EI
0 0
L L L L
6EI 2EI 6EI 4EI
0 0
L L L L
λ μ 0 0 0 0
μ λ 0 0 0 0
0 0 1 0 0 0

0 0 0 λ μ 0
0 0 0 μ λ 0
0 0 0 0 0 1

EA 12EI EA 12EI 6EI EA 12EI EA 12EI 6EI

λ μ λμ λμ μ λ μ λμ λμ μ
L L L L L L L L L L
EA 12EI EA 12EI 6EI EA 12EI EA 12EI 6EI
λμ λμ μ λ λ λμ λμ μ λ λ
L L L L L L L L L L
6EI 6EI 4EI 6EI 6EI 2EI
μ λ μ λ
L L L L L L
EA 12EI EA 12EI 6EI EA 12EI EA 12EI 6EI
λ μ λμ λμ μ λ μ λμ λμ μ
L L L L L L L L L L
EA 12EI EA 12EI 6EI EA 12EI EA 12EI 6EI
λμ λμ μ λ λ λμ λμ μ λ λ
L L L L L L L L L L
6EI 6EI 2EI 6EI 6EI 4EI
μ λ μ λ
L L L L L L
 33

(4.5.3)

4.5.4 Worked Out Example

Analyse the plane frame shown below. Assume the modulus of elasticity of the horizontal
member is 1.5 times that of the vertical member and length of the vertical member is 1.5
times that of horizontal member. Find the bending moment and reactions at support
assuming the length, cross section area and modulus of elasticity of vertical member as
3.0 m, 0.4 x 0.4 m2 and 2 x 1011 N/mm2, respectively.

Fig. 4.5.2 Plane frame

Solution

Step 1: Numbering of Nodes and Members

The numbering of members and joints of the plane frame are as shown below:

Fig. 4.5.3 Numbering of Nodes and Members

The members AB and BC are designated as (1) and (2). The points A, B and C are designated
by nodes 1, 2 and 3. The member information for the frame is shown in tabulated form as
shown in Table 1(a). The coordinate of node 1 is assumed as (0,0). The coordinate and
restraint joint information are shown in Table 1(b). The integer 1 in the restraint list indicates
 34

the restraint exists and 0 indicates the restraint at that particular direction does not exist.
Thus, in node no. 2, the integer 0 all the restraint type indicates that the joint is in free all the
three directions.

Table 4.5.1 Member Information for Beam

Member number Starting node Ending node Rigidity modulus
1 1 2 EI
2 2 3 1.5EI

Table 4.5.2 Nodal Information for Beam

Node no. Coordinates Restraint list
X Y Axial Vertical Rotation
1 0 0 1 1 1
2 0 1.5L 0 0 0
3 L 1.5L 1 1 1

Step 2: Formation of member stiffness matrix:

The individual member stiffness matrices can be found out directly from eqn. shown above.
Thus the stiffness matrices of each member in global coordinate system are given below
based on their individual member properties and orientations. Thus the stiffness matrix of
member (1) is:

1 2 3 4 5 6

0 0 1
. . . .

0 0 0 0 2
. .

0 0 3
. . . .
=
0 0 4
. . . .

0 0 0 0 5
. .

0 0 6
. . . .

Similarly, the stiffness matrix of member (2) is :

4 5 6 7 8 9
 35

. .
0 0 0 0 4
. . . .
0 0 5
. . . .
0 0 6
. .
0 0 0 0 7
. . . . 8
0 0
. . . . 9
0 0

Step 3 : Formulation of global stiffness matrix:

The global stiffness matrix is obtained by assembling by assembling the local stiffness matrix
of member (1) and (2) as follows:
1 2 3 4 5 6 7 8 9

0 0 0 0 0 1
0 0 0 0 0 0 0 2

0 0 0 0 0 3

0
.
0
.
0 0 4

[K] = 0 0 0 0 5

0 0 6

0 0 0
.
O O
.
0 0 7

0 0 0 0 0 8

0 0 0 0 0 9
Step 4: Boundary conditions:
The boundary conditions according to the support of the frame can be expressed in terms of
the displacement vector. The displacement vector will be as follows:
0
0
0

0
0
0
Here, , and indicate the displacement in X-direction, displacement in Y-direction
and rotation at point B.

Step 5: Load vector:

 36

The distributed load on member (2) can be replaced by its equivalent joint load as shown in
the figure below.

Fig. 4.5.4 Equivalent Joint Loads

Thus, the equivalent joint load vector can be written as

0
0
0
0

12
0

12
Step 6: Determination of unknown displacements:
The unknown displacements can be obtained from the relationship of {F} = [K]{d} or
{d} =[k]-1 {F}. Now eliminating the rows and columns in the stiffness matrix and force
matrix, corresponding to zero elements in displacement matrix, the reduced matrix will be as
follows.
32 1.5 8
0 0
9 3
2 18 9
0 2
3
8 9 8 6
12
3 3
Thus, the unknown displacements will be:
1 0.04327
1.7127
10
5.4978
Step 7: Determination of member end actions:
The member end actions can be obtained from the corresponding member stiffness and the
nodal displacements. The member end actions for each member are derived as shown below.
 37

Member – (1)
In case of member (1), the member forces will be:
56.17 0 126.4 56.17 0 126.4
0 7110 0 0 7110 0
126.4 0 379.2 126.4 0 189.6
10
56.17 0 379.2 56.17 0 126.4
0 7110 0 0 7110 0
126.4 0 189.6 126.4 0 379.2
0
0
0
4.327 10
1.7127 10
5.4978 10

0.0697
1.2177
0.10479
=
0.06925
1.21661
0.20793

It is to be noted that {Fm} are the end actions due to joint loads. Hence it must be added to the
corresponding end actions in the restrained structure in order to obtain the end actions due to
the loads. Therefore, {Fm}actual are the true member end actions due to actual loading system
can be expressed as
= {Fm} + {Ffm}

Where, {Ffm} are the end actions in the restrained structure. Since there is no load acting on
member (1), the actual end actions will be:

0.0697 0 0.0697
1.2177 0 1.2177
0.10479 0 0.10479
{Fm}actual =
0.06925 0 0.06925
1.21661 0 1.21661
0.20793 0 0.20793

Member (2)
In similar way, the member forces in member (2) will be {Fm}(2) = [K](2){d}(2)
 38

16 0 0 16 0 0 4.327 10
0 0.284 0.426 0 0.284 0.426 1.7127 10
0 0.426 0.853 0 0.426 0.426 5.4978 10
10
16 0 0 16 0 0 0
0 0.284 0.426 0 0.284 0.426 0
0 0.426 0.426 0 0.426 0.853 0
0.069232
0.28325
0.54215
=
0.06923
0.283245
0.3076

The actual member forces in the member (2) will be:

0.069232 0 0.0692
0.28325 1.5 1.2167
0.54215 0.75 0.2078
{Fm}actual =
0.06923 0 0.0692
0.283245 1.5 1.7832
0.3076 0.75 1.0576
 Module 4: Analysis of Frame Structures
Lecture 6 Analysis of Grid and Space Frame 39

4.6.1 Introduction
The property of a grid member is basically a combination of 2-d beam with torsional effect.
The plane frame is assumed to be loaded in its own plane where as loading in the grid is
normal to its plane. As a result torsional effects are included in the grid analysis. Thus the
grid member can withstand bending moment, shear force as well as torsional moment.

4.6.2 Element Stiffness Matrix for Grid Members

The degrees of freedom at each node of the grid member will be (i) vertical deformation and
(ii) rotation in two different directions.

Fig. 4.6.1 Degrees of freedom of grid element

Therefore, the stiffness matrix of the grid in its local coordinate system will be:

θx1 θz1 v1 θx2 θz2 v2

 GI x GI x 
 L 0 0  0 0  Mx1
L
 4 EI y 6 EI y 2 EI y 6 EI y 
 0  0  Mz1
 L L 2
L L2 
 6 EI y 12 EI y 6 EI y 12 EI y 
 0  0    Fy1
k    GI L2 L3
GI x
L2 L3  (4.6.1)
 x 0 0 0 0  Mx2
 L L 
 2 EI y 6 EI y 4 EI y 6 EI y 
 0  0  Mz2
2
 L L L L2 
 0 6 EI y 12 EI y 6 EI y 12 EI y 
 0 Fy2
 L2 L3 L2 L3 
 40

Here, the G is the modulus of torsional rigidity.

4.6.3 Generalized Stiffness Matrix
The generalized stiffness matrix of a grid member can be obtained by transferring the matrix
of local coordinate system into its global coordinate system. The transformation matrix can
be expressed as:

cos sin 0 0 0 0
sin cos 0 0 0 0
0 0 1 0 0 0
[T]=
0 0 0 cos sin 0
0 0 0 sin cos 0
0 0 0 0 0 1

Now, the generalized stiffness matrix of the member can be obtained from the relation
of . Thus considering cos and sin the stiffness matrix in
global coordinate system can be written as follows:

0 0 0 0
0 0 0 0 0 0
0 0 0 0
0 0 1 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0
0 0 0 0 0 1 0 0
0 0
0 0 0 0
0 0 0 0
0 0 1 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0 0 1
 41

 GI x 2 4 EI y 2  GI x 4 EI y  6 EI y GI x 2 4 EI y 2  GI 2 EI y  6 EI y 
    
 L  L 
        x    
 L L   L 2
L L  L L  L 2

  GI x 4 EI y  GI x 2 4 EI y 2 6 EI y  GI 2 EI y  GI 2 EI y 2 6 EI y 
           x    x 2    
  L L  L L L2  L L  L L L2

 6 EI y 6 EI y 12 EI y 6 EI y 6 EI y 12 EI y 
   2    2   
 L2 L L3 L2 L L3 
 GI x 2 4 EI y 2  GI x 2 EI y  6 EI y GI x 2 4 EI y 2  GI x 4 EI y  6 EI y 
 L   L          
 L  L 
  2 
 L L  L 2
L L   L
 
  GI x 2 EI y  GI 2 EI y 2 6 EI y  GI x 4 EI y  GI x 2 4 EI y 2 6 EI y 
   
  x 2     
 L  L 
     
  L L  L L L2   L L L2 
 6 EI y 6 EI y 12 EI y 6 EI y 6 EI y 12 EI y 
  2     2   
 L L2 L3 L L2 L3 

(4.6.2)

4.6.4 Worked Out Example

Analyze the grid shown below and draw the shear force and bending moment diagram
assuming the cross sectional area and modulus of elasticity of each member as 0.30.3 m2
and 21011 N/m2 respectively. Assume EI = 3GJ. The length of member AB and BC is 4 m
and 5 m respectively.

Fig. 4.6.2 Grid structure

Solution

Step 1: Numbering of Nodes and Members

The numbering of members and joints of the plane frame are as shown in the figure below:
 42

Fig. 4.6.3 Numbering of nodes and members

The member AB and BC are designated as (1) and (2). The points A, B and C are designated
by nodes 1, 2 and 3. The member information for the grid is shown in tabulated form as
shown in Table 4.6.1. The coordinate of node 1 is assumed as (0, 0). The coordinate and
restraint joint information are shown in Table 4.6.2. The integer 1 in the restraint list indicates
the restraint exists and 0 indicates the restraint at that particular direction does not exist.
Thus, in node no. 2, the integer 0 all the restraint type indicates that the joint is free in all the
three directions.

Table 4.6.1 Member Information

Member number Starting node Ending node
1 1 2
2 2 3

Table 4.6.2 Member Coordinates

ode coordinates Restraint list
No x z Vertical Rotation Rotation
1 0 0 1 1 1
2 4 0 0 0 0
3 4 5 1 1 1

Step 2: Formation of member stiffness matrix:

The individual member stiffness matrices can be found out directly. Thus the stiffness
matrices of each member in global coordinate system are given below based on their
individual member properties and orientations. As the member AB is horizontal, i.e.,  = 0,
the values of Cos  = 1 and Sin  = 0. Thus the stiffness matrix of member (1) is:
 43

1 2 3 4 5 6
 GJ GJ  1
 L 0 0  0 0 
L
 12 EI 6 EI 12 EI 6 EI 
 0 0   2
 L3 L2 L3 L2 
 0 6 EI 4 EI 6 EI 2 EI 
0  2 3
 L 
k AB 
GJ
L2 L
GJ
L

 0 0 0 0  4
 L L 
 0 12 EI 6 EI 12 EI 6 EI
  0  2  5
 L3 L2 L3 L 
 6 EI 2 EI 6 EI 4 EI 
 0 0  2  6
 L2 L L L 

Assuming EI=3GJ=3K, the above equation can be written as

1 2 3 4 5 6
 K K  1
 L 0 0  0 0 
L
 36 K 18 K 36 K 18K 
 0 0   2
 L3 L2 L3 L2 
 0 18K 12 K 18 K 6K 
 0  2 3
k AB  L2 L L L 
K K 
 0 0 0 0  4
 L L 
 0 36 K 18K 36 K 18K 
 3  2 0  2
 L L L3 L  5
 18K 6K 18 K 12 K 
 0 0  2 
 L2 L L L  6

As the member BC member is also horizontal, the value of Cos  = 1 and Sin  = 0 and thus,
the stiffness matrix will be:
6 5 4 7 8 9
 K K  6
 L 0 0  0 0 
L
 36 K 18K 36 K 18K 
 0 0   5
 L3 L2 L3 L2 
 0 18K 12 K 18K 6K 
0  2 4
 L 
k BC   K
2
L L L
K 
 0 0 0 0  7
 L L 
 0 36 K 18K 36 K 18K
 3  2 0  2 
 L L L3 L  8
 18K 6K 18K 12 K 
 0 0  2 
Step 3: Formation
 ofLglobal
2 stiffness
L matrix: L L  9
 44

The global stiffness matrix can be obtained by assembling the local stiffness matrix of
members (AB) and (BC). Now looking at the grid structure, the displacements at the fixed
supports, are known and all are equal to zero. Only the displacement at co-ordinates 4, 5, 6
are unknown. So the global system stiffness matrix, corresponding to the displacement at co-
ordinate 4, 5, 6 will be:

2.65 0.72 0
= K 0.72 0.8505 1.125
0 1.125 3.2

Step 4: Boundary condition:

The boundary conditions according to the support of the grid structure can be expressed in
terms of the displacement vector. The displacement vector will be as follows

0
0
 
0
 
d 4 
d    d5 
d 
 6
0
 
0
 0 
Here, d4, d5, and d6 indicate the displacement vectors at point B.

Step 5: Load vector:

The distributed load on member (1) can be replaced by its equivalent joint load as shown in
the figure below.
 45

Fig. 4.6.4 Equivalent load

Thus the equivalent load vector will be:

 0 
 wL 
 2 
 wL2 
 
 12 
 0 
P    wL 
 22 
 wL 
 12 
 0 
 
 0 
 0 
 

Step 6: Determination of unknown displacements:.

The unknown displacements can be obtained from the relationship of  F    K d  or

d    K   F  .
1
Now, eliminating the rows in the force matrix, corresponding to zero
element in displacement matrix, the reduced matrix will be as follows.
0
2.65 0.72 0 4
k 0.72 0.8505 1.125 2
0 1.125 3.2 16
12
0
1 0.662 1.047 0.368
2
1.047 3.856 1.3556 4
0.368 1.355 0.789
3

Thus, the unknown displacements will be:

 46

1 1.603
5.905
— 1.658

Step 7: Determination of member end actions:

The member end actions can be obtained from the corresponding member stiffness and the
nodal displacements. The member end actions for each member are derived as shown below.

Member - AB
In case of member (AB), the member forces will be: {Fm}(AB) = [K](AB) {d}(AB)
1 1
0 0 0 0
4 4
36 18 36 18
0 0
4 4 4 4 0
18 12 18 6 0
0 0 1
4 4 4 4 0
1 1 1.603
0 0 0 0 5.905
4 4
36 18 36 18 1.658
0 0
4 4 4 4
18 6 18 12
0 0
4 4 4 4

Thus,
0.4
1.456
4.156
0.4
1.456
1.70

It is to be noted that {Fm} are the end actions due to joint loads. Hence it must be added to the
corresponding end actions in the restrained structure in order to obtain the end actions due to
the loads. Therefore, {Fm} Actual are the true member end actions due to actual loading system
and can be expressed as
{Fm} Actual = {Fm} + {Ffm}

Where, {Ffm}are the end actions in the restrained structure. Since there is no load acting on
member (1), the actual end action will be:
 47

0
4
0.4 2 0.4
1.456 4 3.46
4.156 12 5.49
0.4 0 0.40
1.456 4 0.54
1.70 2 0.34
4
12

Member - BC

In similar way, the member forces in member (BC) will be: {Fm}(BC) = [K](BC) {d}(BC)
1 1
0 0 0 0
5 4
36 18 36 18
0 0
5 5 5 5 1.658
18 12 18 6 5.905
0 0 1
5 5 5 5 1.603
1 1 0
0 0 0 0 0
5 5
36 18 36 18 0
0 0
5 5 5 5
18 6 18 12
0 0
5 5 5 5

Thus,

0.33
0.55
0.40
0.33
0.55
2.33

Since there is no load acting on member (BC), the actual end action will be:
0.33 0 0.33
0.55 0 0.55
0.40 0 0.40
0.33 0 0.33
0.55 0 0.55
2.33 0 2.33
 48

Thus, the reaction forces at the support and load at the joints will be:

0.4
3.46
5.49
0
0
0
0.33
0.55
2.33

4.6.5 Analysis of Space Frame

Space frames are an increasingly common architectural technique especially for large roof
spans in commercial and industrial buildings. The rigid jointed frames such as building
frames are usually three dimensional space structures. Thus in case of certain structures like
multi-storeyed buildings, it is necessary consider 3-dimensional effects for analysis. The
space frame constitutes the final step of increasing complexity. It consists of plane frame and
grid actions. The displacement and rotation vector associated with each joint have three
components in case of space frame structures. There are six equilibrium equations associated
with each joint. The degrees of freedom at each node of the space frame member will be (i)
displacement in three perpendicular directions and (ii) rotations in three different directions.
Therefore, the degrees of freedom in each node of the member will be six as shown in the
figure below. The stiffness matrix in local coordinate system considering all possible degrees
of freedom will be as given in Table 4.6.1.

Fig. 4.6.5 Degrees of freedom for space frame member

Table 4.6.1 Stiffness matrix of space frame member

 49

1 2 3 4 5 6 7 8 9 10 11 12

 EAX EAX 
1  L 0 0 0 0 0  0 0 0 0 0 
L
 
 0 12 EI Z 6 EI Z 12 EI Z 6 EI Z 
0 0 0 0  0 0 0
2  L3 L2 L3 L 
2

 12 EIY 6 EIY 12 EIY 6 EIY 

 0 0 0  0 0 0  0  0 
3  L3 L2 L3 L2 
 GI X GI X 
 0 0 0 0 0 0 0 0  0 0 
4  L L 
 6 EIY 4 EIY 6 EIY 2 EIY 
5  0 0  0 0 0 0 0 0 
 L2 L L2 L 
 6 EI Z 4 EI Z 6 EI Z 2 EI Z 
6  0 L2
0 0 0
L
0 
L2
0 0 0
L 
 
  EAX 0 0 0 0 0
EAX
0 0 0 0 0 
7  L L 
 12 EI Z 6 EI Z 12 EI Z 6 EI Z 
8  0  0 0 0  0 0 0 0  2
 L3 L2 L3 L 
 12 EIY 6 EIY 12 EIY 6 EIY 
9  0 0  0 0 0 0 0 0 
 L3 L2 L3 L2 
 GI X GI X 
10  0 0 0 
L
0 0 0 0 0
L
0 0 
 
 6 EIY 2 EIY 6 EIY 4 EIY 
11  0 0  0 0 0 0 0 0 
L2 L L2 L
 
 0 6 EI Z 2 EI Z 6 EI Z 4 EI Z 
12 0 0 0 0  0 0 0
 L2 L L L 

The generalized stiffness matrix of a rigid jointed space frame member can be obtained by
transferring the matrix of local coordinate system into its global coordinate system. The
transformation matrix will become a square matrix of size 12×12 in this case as the degrees
of freedom for each node/joint is six.
 Module 5: FEM for Two and Three Dimensional Solids
Lecture 1: Constant Strain Triangle 1

The triangular elements with different numbers of nodes are used for solving two dimensional solid
members. The linear triangular element was the first type of element developed for the finite element
analysis of 2D solids. However, it is observed that the linear triangular element is less accurate
compared to linear quadrilateral elements. But the triangular element is still a very useful element for
its adaptivity to complex geometry. These are used if the geometry of the 2D model is complex in
nature. Constant strain triangle (CST) is the simplest element to develop mathematically. In CST,
strain inside the element has no variation (Ref. module 3, lecture 2) and hence element size should
be small enough to obtain accurate results. As indicated earlier, the displacement is expressed in two
orthogonal directions in case of 2D solid elements. Thus the displacement field can be written as
ïìuïü
{d } = ïí ï (5.1.1)
ïîïvïï
Here, u and v are the displacements parallel to x and y directions respectively.

5.1.1 Element Stiffness Matrix for CST

A typical triangular element assumed to represent a subdomain of a plane body under plane
stress/strain condition is represented in Fig. 5.1.1. The displacement (u, v) of any point P is
represented in terms of nodal displacements
u  N1u1  N2u2  N3u3
(5.1.2)
v  N1v1  N2v2  N3v3
Where, N1, N2, N3 are the shape functions as described in module 3, lecture 2.

Fig. 5.1.1 Linear triangular element for plane stress/strain

 2

The strain-displacement relationship for two dimensional plane stress/strain problem can be
simplified in the following form from three dimensional cases (eq.1.3.9 to1.3.14).

¶u 1 êéæç ¶u ö÷ æç ¶v ö÷ úù
2 2

ex = + ç ÷ +ç ÷
¶x 2 êêëçè ¶x ÷ø çè ¶x ÷ø úúû

¶v 1 éêæç ¶u ö÷ æç ¶v ö÷ ùú
2 2

ey = + ç ÷ +ç ÷ (5.1.3)
¶y 2 êêçè ¶y ÷÷ø çè ¶y ÷÷ø úú
ë û
¶v ¶u ¶u ¶u ¶v ¶v ù
é
g xy = + +ê + ú
¶x ¶y êë ¶x ¶y ¶x ¶y úû

In case of small amplitude of displacement, one can ignore the nonlinear term of the above equation
and will reach the following expression.
u
x 
x
v
y  (5.1.4)
y
v u
 xy  
x y
Hence the element strain components can be represented as,
 u N1 N N
 x   u1  2 u2  3 u3
 x x x x
 v N N N
   y   1 v1  2 v2  3 v3
 y y y y
 u v N1 N N N N N
 xy    u1  2 u2  3 u3  1 v1  2 v2  3 v3
 y x y y y x x x
Or,

 N1 N 2 N 3   u1 
 0 0 0  u 
  x   x x x  2
   N1 N 2 N 3  u3 
   y    0 0 0   (5.1.5)
   y y y   v1 
 xy   N N 2 N 3 N1 N 2 N 3   v2 
 
1

 y y y x x x   v3 

Or,    B  d  (5.1.6)
 3

In the above equation [B] is called as strain displacement relationship matrix. The shape functions
for the 3 node triangular element in Cartesian coordinate is represented as,
ì
ï ü
ï
1 é
( x y - x y ) + ( y - y ) x + ( x - x ) y ùïï
ï
ï 2A ë 2 3 3 2 2 3 3 2 ûïï
ìï N1 üï ï ï
ï ï ï ï
ïí N ï = ïïí 1 é( x y - x y ) + ( y - y ) x + ( x - x ) yù ïï
ïï 2 ïï ïï 2A ë 3 1 1 3 3 1 1 3 ûï
ïï
ïîï N3 ïï ïï ï
ï 1 é( x1y 2 - x 2 y1 ) + ( y1 - y 2 ) x + ( x 2 - x1 ) yù ïï
ï
ï ë ûï
îï 2A ï
Or,
ìï 1 ü
ïï [a1 + b1x + g1y ] ïïï
ìï N1 üï ïï 2A ïï
ïï ïï ïï 1 ïï
í N  = í [a + b2 x + g 2 y ]ï
ï (5.1.7)
ïï 2 ïï ïï 2A 2 ïï
ïîï N 3 ïï ïï ï
ïï 1 [a + b x + g y ]ïïï
ïïî 2A 3 3 3
ïï
Where,
a1 = ( x 2 y 3 - x 3 y 2 ) , a 2 = ( x 3 y1 - x1y3 ), a 3 = ( x1y 2 - x 2 y1 ),
b1 = ( y 2 - y3 ), b2 = ( y3 - y1 ), b3 = ( y1 - y 2 ),
(5.1.8)
g1 = ( x 3 - x 2 ), g 2 = ( x 2 - x1 ), g 3 = ( x 2 - x 1 ),
Hence the required partial derivatives of shape functions are,
N1 1 N 2  2 N 3 3
 ,  ,  ,
x 2 A x 2A x 2 A
N1  1 N 2  2 N 3  3
 ,  ,  ,
y 2 A x 2A x 2 A
Hence the value of [B] becomes:
 N1 N 2 N 3 
 0 0 0 
 x x x 
 N1 N 2 N 3 
 B   0 0 0 
 y y y 
 N1 N 2 N 3 N1 N 2 N 3 
 
 y y y x x x 
 1 2 3 0 0 0
1 
Or,  B   0 0 0 1  2  3  (5.1.9)
2A
  1 2 3 1  2 3 
 4

According to Variational principle described in module 2, lecture 1, the stiffness matrix is

represented as,
 k     B   D  B  d 
T
(5.1.10)


Since, [B] and [D] are constant matrices; the above expression can be expressed as

 k    B   D  B   dV   B   D  B V
T T
(5.1.11)
V

For a constant thickness (t), the volume of the element will become A.t . Hence the above equation
becomes,
 k    B   D  B  At
T
(5.1.12)
For plane stress condition, [D] matrix will become:
 
1  0 
E  
 D  2 
 1 0  (5.1.13)
1  
1  
0 0 
 2 
Therefore, for a plane stress problem, the element stiffness matrix becomes,
 1 0 1 
 0  2   
 2 1  0   1  2 3 0 0 0
 3 3   
0   0 0  3 
Et 0
k   2    1 0 1  2 (5.1.14)
4 A 1     0 1 1  
1      1  2 3 1  2  3 
0 2 2   0 0 
  2 
 0 3 3 
Or,
 2 1      
 1  C 1 1 2  C 1 2 13  C 1 3 1 2  C  2 1 1 3  C 3 1 
2
1 1
 2 

  C
2 2
 2 3  C 2 3  2 1  C 1 2
1      
 2 3  C 3 2  (5.1.15)
 2 2
2
2 2

Et  
k    1      
4 A 1    
2
32  C 32 3 1  C 1 3 3 2  C  2 3 3 3 
 2 
  12  C 12  1 2  C 1 2  1 3  C 13 
 Sym.  22  C  22  2 3  C  2 3 
 
  32  C 32 

Where, C 
1   
2
Similarly for plane strain condition, [D] matrix is equal to,
 5

 
1     0 
E  
 D    1    0  (5.1.16)
1   1  2   1  2 
 0 0 
 2 
Hence the element stiffness matrix will become:
 M 12   12 M 1 2   1 2 M 13   1 3 (   1) 1 1 1 2   2 1 1 3  3 1 
 
 M  2
2
2
2 M  2  3   2 3  2 1  1 2    1  2 2  2 3  3 2 
 M 32   32 3 1  1 3 3 2  C  2 3    1 3 3 
(5.1.17)
Et
k    
2 A 1     M  12  12 M  1 2  1 2 M  1 3  13 
 Sym. M  22   22 M  2 3   2 3 
 
 M  32  32 

Where M  1   

5.1.2 Nodal Load Vector for CST

From the principle of virtual work,

ò d {e} {s}dW = ò d {u} {FG }dG + ò d {u} {FW }dW

T T T
(5.1.18)
W G W

Where, F, and F are the surface and body forces respectively. Using the relationship between
stress-stain and strain displacement, one can derive the following expressions:
    D Bd ,      B d and  u   N   d (5.1.19)
Hence eq. (5.1.18) can be rewritten as,
é Sù
T
(5.1.20)
ò d {d} [B] [D][B]{d}dW = ò d {d} ëê N ûú {FG }dG + ò d {d} [ N ] {FW }dW
T T T T T

W G W

Or, [ B ] [ D ][ B ]{d }d W = é N S ù {F G }d G +
T
[ N ] {FW }d W (5.1.21)
ò ò ò
T T
ëê úû
W G W
s
Here, [N ] is the shape function along the boundary where forces are prescribed. Eq.(5.1.21) is
equivalent to [ k ]{d} = {F} , and thus, the nodal load vector becomes

{F} = ò [ N ] {F }dG + ò [ N ] {F }dW

T T
S
G W

G W (5.1.22)
For a constant thickness of the triangular element eq.(5.1.22) can be rewritten as

{F} = t ò [ N ] {F }ds + t ò [ N ] {F }dA

T T
S
G W (5.1.23)
S A

For the a three node triangular two dimensional element, one can represent FW and FG as,
 6

ïìFWx üïï ïìFGx ï ü

{FW } = ïí  and {FG } = ïí ï 
ïîïFWy ïï ïîïFGy ï
ï

ìFWx ï
ï ü ì ï 0 ü ï
For example, in case of gravity load on CST element, {FW } = ï
í ï
 = íï ï
ï
ïFWy ï
î ï î
 ïï-rgïï
For this case, the shape functions in terms of area coordinates are:
é L1 L 2 L3 0 0 0ù
[ N ] = êê ú (5.1.24)
ë0 0 0 L1 L 2 L3 úû
As a result, the force vector on the element considering only gravity load, will become,
éL 0ù ïïì 0 ïïü ïïì 0 ïïü
ê 1
ú ïï 0 ïï ïï 0 ïï
êL 0ú
ï ï ïï ïï
ê 2
ú ïï ïï
êL 0 úì 0 ü ï0ï
ú ïí ïdA = t ò ïí
0 ï
{F} = t ò ê dA = -rgt ò ï í ïdA
3
(5.1.25)
ê0 L úï ï ïï-L rg ïï ïï L ïï
A
ê 1
ú îï-rgï ï
A
ï
1
ïï ïï
A 1

ê0 L ú ïï-L rgïï ïïL ïï

ê 2
ú ïï ïï 2 2

ê0 L úû ïïî-L rgïï ïïL ïï

ë 3 3 ïî ï 3

The integration in terms of area coordinate is given by,

p! q! r !
A L1 L2 L3dA  ( p  q  r  2 )! 2 A
p q r
(5.1.26)

Thus, the nodal load vector will finally become

ïìï 0 ïüï ì ï 0 üïï
ïï ïï ï ï ï
ïï 0 ïï ï ï 0 ïï ìï0üï
ïï 0 ïï ï ï ïï ïï ïï
ïï ïï ïï
0 ï
ïï ïï0ïï
ïï ïï ï
ï ï ï ï (5.1.27)
rgAt ïïï0ïïï
1!0!0! r gAt
ï 2Aï ïï- ïï
{F} = -rgt í (1 + 0 + 0 + 2)!  = í 3 ï = - 3 íï1ï
ïï ï
ï ïï
ïï 0!1!0! ï
ï ïï rgAt ïïï ïï ïï
ïï1ïï
ïï 2A ï ïï - ï
ïï (0 + 1 + 0 + 2)! ï ï
ï ï
ï
3 ïï
ï
ïï ïï
ïîï1ïï
ïï 0!0!1! ï
ï ïï rgAt ïï
ï 2Aï ïï- ï
ïîïï (0 + 0 + 1 + 2)! ïïï ï î 3 ïï
 Module 5: FEM for Two and Three Dimensional Solids
Lecture 2: Linear Strain Triangle
7

5.2.1 Element Stiffness Matrix for LST

In case of CST, it is observed that the strain within the element remains constant. Though, these
elements are able to provide enough information about displacement pattern of the element, but it is
unable to provide adequate information about stress inside an element. This limitation will be
significant enough in regions of high strain gradients. The use of a higher order triangular element
called Linear Strain Triangle (LST) significantly improves the results at these areas as the strin
inside the element is varying. The LST element has six nodes (Fig. 5.2.1) and hence, twelve degrees
of freedom. Thus the displacement function can be chosen as follows.
u   0  1 x   2 y   3 x 2   4 xy   5 y 2
(5.2.1)
v  a6  a7 x  a8 y  a9 x 2  a10 xy  a11 y 2

Fig. 5.2.1 Linear strain triangle element

Therefore, the element strain matrix is obtained as

¶u
ex = = a1 + 2a 3 x + a 4 y
¶x
¶v (5.2.2)
ey = = a 8 + a10 x + 2a11y
¶y
¶v ¶u
g xy = + = (a 2 + a 7 ) + (a 4 + 2a 9 )x + (2a 5 + a10 )y
¶x ¶y
 8

In the area coordinate system as discussed in module 3, lecture 3 we can write the shape function for
the six node triangular element as
N1 = L1 (2L1 - 1) N 2 = L 2 (2L 2 - 1) N 3 = L3 (2L3 -1)
(5.2.3)
N 4 = 4L1L 2 N 5 = 4L 2 L3 N 6 = 4L3L1
The displacement (u,v) of any point within the element can be represented in terms of their nodal
displacements with the use of interpolation function.
6
u   Ni ui
i 1
6
(5.2.4)
v   Ni vi
i 1

Using eq.(5.2.4) we can rewrite eq.(5.2.2) as,

 u1 
u 
 2
u3 
 
 N1 N 2 N 3 N 4 N 5 N 6  u4 
 0 0 0 0 0 0  u 
 x x x x x x  5
 N1 N 2 N 3 N 4 N 5 N 6  u6 
  0 0 0 0 0 0  
 y y y y y y   v1 
 N1 N 2 N 3 N 4 N 5 N 6 N1 N 2 N 3 N 4 N 5 N 6   v2 
  
 y y y y y y x x x x x x   v3 
v 
 4
 v5 
v 
 6
Or,
   B  d  (5.2.5)
Where,
 N1 N 2 N 3 N 4 N 5 N 6 
 0 0 0 0 0 0 
 x x x x x x 
 N1 N 2 N 3 N 4 N 5 N 6  (5.2.6)
 B   0 0 0 0 0 0 
 y y y y y y 
 N1 N 2 N 3 N 4 N 5 N 6 N1 N 2 N 3 N 4 N 5 N 6 
 
 y y y y y y x x x x x x 

Using Chain rule,

N1 N1 L1 N1 L2 N1 L3
 .  .  .
x L1 x L2 x L3 x
As discussed in module 3, lecture 1, we can write the above expression as,
 9

N1 b1 N1 b2 N1 b3 N1

 .  .  .
x 2 A L1 2 A L2 2 A L3
N1 b1
 .  4 L1  1
x 2 A
Similarly we can evaluate expressions for other terms and can be written as,
N1 b1 N 2 b2 N 3 b3
 .  4 L1  1  .  4 L2  1  .  4 L3  1
x 2 A x 2 A x 2 A
N 4 N 5 N 6
 4  L2b1  L1b2   4  L3b2  L2b3   4  L1b3  L3b1 
x x x
And,
N1 a1 N 2 a2 N3 a3
 .  4 L1  1  .  4 L2  1  .  4 L3  1
y 2 A y 2 A y 2 A
N 4 N5 N 6
 4  L2 a1  L1a2   4  L3a2  L2 a3   4  L1a3  L3a1 
y y y
Where,
a1  x2  x3 a2  x3  x1 a3  x1  x2
b1  y2  y3 b2  y3  y1 b3  y1  y2

The stiffness matrix of the element is represented by,

 k     B   D  B  d 
T
(5.2.7)


The, [D] matrix is the constitutive matrix which will be taken according to plane stress or plane
strain condition. The nodal strain and stress vectors are given by,
 n    x1  x 2  x 3  y1  y 2  y 3  xy1  xy 2  xy 3 
T
(5.2.8)

 n    x1  x 2  x 3  y1  y 2  y 3  xy1  xy 2  xy 3 
T
(5.2.9)

  Bn1   0 
 n     0  Bn 2  d  (5.2.10)
 Bn 2   Bn1  
Referring to section 3.3.1, using proper values of area coordinates in [B] matrix, one can find
 3b1 b2 b3 4b2 0 4b3 
1 
 Bn1    b1 3b2 b3 4b1 4b3 0  (5.2.11a)
2A
 b1 b 3b3 0 4b2 4b1 
And,
 10

 3a1  a2  a3 4a2 0 4a3 

 Bn 2    a1 3a2 a3 4a1 4a3 0 
1 
(5.2.11b)
2A
 a1 a 3a3 0 4a2 4a1 
Thus, the element stiffness can be evaluated by putting the values from eq. (5.2.11) in eq. (5.2.7).

5.2.2 Nodal Load Vector for LST

Similar to 3-node triangular element, the load will be lumped at each node which can be computed
using the earlier expression,

{F} = ò [ N ] {F }dG + ò [ N ] {F }dW

T T
S
G W (5.2.12)
G W

And for element with constant thickness,

{F} = t ò [ N ] {F }ds + t ò [ N ] {F }dA
T T
S
G W
(5.2.13)
S A

5.2.3 Numerical Example using CST

Determine the displacements at the nodes for the following 2D solid continuum considering a
constant thickness of 25 mm, Poisson’s ratio,  as 0.25 and modulus of elasticity E as 2 x 105
N/mm2. The continuum is discritized with two CST plane stress elements.

Fig. 5.2.2 Geometry and discretization of the continuum

The element 1 is connected with node 1, 3 and 4 and let assume its Cartesian coordinates are (x1, y1),
(x3, y3) and (x4, y4) respectively. If we consider nodes 1, 3 and 4 are similar to node 1, 2 and 3 in
eq.(5.1.9) then the [B] can be written as
 11

 1  2 3 0 0 0 
 B    0 0 0  1  2  3 
1 
2A
  1  2  3 1  2  3 
By introducing values of β & γ discussed in previous lecture note, we can get value of [B] as
 0 1 1 0 0 0 
1 
 B  0 0 0 3 0 3 
1500 
 3 0 3 0 1 1
For plain stress problem, putting the values of E and µ one can find the following values.
 
1  0  16 4 0 
E   4  104  
 D   1 0 
1   2  3  4 16 0 
1    0 0 6 
0 0 
 2 
Therefore the stiffness matrix for the element 1 will be
 k 1  tA  B   D  B 
T

Putting values of t, A, [B] & [D]we will get,

 750 0 -750 0 -250 250 

0 222.2222 -222.2222 -166.6667 0 166.6667 

 -750 -222.2222 972.2222 166.6667 250 -416.6667 
 k 1  4 103  0 -166.6667 166.6667 2000 0 -2000

 
 -250 0 250 0 83.3333 -83.3333 
 
 250 166.6667 -416.6667 -2000 -83.3333 2083.3333
Similarly element 2 is connected with nodes 1, 2 and 3 and global coordinates of these nodes are (x1,
y1), (x2, y2) and (x3, y3) respectively. For this element, by proceeding in a similar manner to element
1 we can calculate [B] matrix as,
 1 1 0 0 0 0 
1 
 B  0 0 0 0 3 3
1500 
 0 3 3 1 1 0 
Hence, the elemental stiffness matrix becomes,
 12

 222.2222 -222.2222 0 0 166.6667 -166.6667 

-222.2222 972.2222 -750 250 -416.6667 166.6667 

0 -750 750 -250 250 0 
 k 2  4 10  
3

0 250 -250 83.3333 -83.3333 0 
166.6667 -416.6667 250 -83.3333 2083.3333 -2000 
 
-166.6667 166.6667 0 0 -2000 2000 

By assembling the stiffness matrices into global stiffness matrix [K],

972.2222 -222.2222 0 -750 0 166.6667 -416.6667 250 
-222.2222 972.2222 -750 0 250 -416.6667 166.6667 0 
 
0 -750 972.2222 -222.2222 -416.6667 250 0 166.6667 
 
3 -750 0 -222.2222 972.2222 166.6667 0 250 -416.6667 
 K  410  0 250 -416.6667 166.6667 2083.3333 -83.3333 0 -2000 
 
166.6667 -416.6667 250 0 -83.3333 2083.3333 -2000 0 
-416.6667 166.6667 0 250 0 -2000 2083.3333 -83.3333 
 
250 0 166.6667 -416.6667 -2000 0 -83.3333 2083.3333

Now, applying equation  F    K d  , the following expression can be written.

 Fu1  972.2222 -222.2222 0 -750 0 166.6667 -416.6667 250  u1 
F  -222.2222 972.2222 -750 0 250 -416.6667 166.6667 0  u 
 u2    2
 Fu 3  0 -750 972.2222 -222.2222 -416.6667 250 0 166.6667  u3 
    
Fu 4  3 -750 0 -222.2222 972.2222 166.6667 0 250 -416.6667  u4 
    
  v1 
4 10
0
 Fv1  
250 -416.6667 166.6667 2083.3333 -83.3333 0 -2000
 
 Fv 2  166.6667 -416.6667 250 0 -83.3333 2083.3333 -2000 0  v2 
  -416.6667  
F
 v3  166.6667 0 250 0 -2000 2083.3333 -83.3333  v3 
 
F  2083.3333 v4 
 v4  250 0 166.6667 -416.6667 -2000 0 -83.3333
Putting boundary conditions u1  v1  u2  u4  v4  0 and adopting elimination technique for
applying boundary condition we get expression,
 0  972.2222 250 0  u3 
    
 0   4  10   250 2083.3333 2000   v2 
3

  2083.3333  v3 
 25000   0 -2000
Solving the above expression, the unknown nodal displacements may be obtained as follows.
v2  0.0606 mm, u3  0.0156 mm and v3  0.0612 mm.
 Module 5: FEM for Two and Three Dimensional Solids
Lecture 3: Rectangular Elements 13

The rectangular elements are widely used for solving two dimensional continuums. The main
advantage of this type of element is the easy formulation and easy development of computer code.
The element stiffness of such elements is derived here using the concept of isoparametric
formulation.

5.3.1 Computation of Element Stiffness

In case of a four node rectangular element, the geometry and displacement filed can be expressed in
terms of their nodal values with the help of interpolation function. As the formulation will be
isoparametric, the interpolation function will become same for expressing both the variables. Thus,
coordinates and displacements at any point inside the element (Fig. 5.3.1) can be expressed as
x = N1 x1 + N2 x2 + N3 x3 + N4 x4
(5.3.1)
y = N1 y1 + N2 y2 + N3 y3 + N4 y4
And
u = N1u1 + N2u2 + N3u3 + N4u4
(5.3.2)
v = N1v1 + N2v2 + N3v3 + N4v4

The above equations can be written in matrix form as

ì
ï x1 ü
ï
ï
ï ï
ï
ï
ï y ï
1ï
ï
ï ï
ï
ï
ï x2ï
ï ï ï
ìïï xüïï é N1 0 N2 0 N3 0 N4 0 ù ïï y2 ïï
í  = êê úí  (5.3.3)
ïîï xïï ë 0 N1 0 N2 0 N3 0 N 4 úû ïï x3 ïï
ï
ï ï
ï y3 ï
ï
ï
ï ï
ï
ï
ï x4ï
ï
ï
ïy ï ï
ï 4 ï
î
And
ìu1 ï
ï ü
ï
ï ï
ï
ï
ï v1 ïï
ï
ï ï
ï
ï
ï u 2 ï
ï
ï ï
ìïïuüïï é N1 0 N2 0 N3 0 N4 0 ù ïïv2 ïï
í = ê úí  (5.3.4)
ïîïv ïï êë 0 N1 0 N2 0 N3 0 N 4 úû ïïu3 ïï
ï
ï ï
ï v3 ïï
ï
ï ï
ï
ï
ï u 4ï
ï
ï
ï ï
ï
ï
ï ï
î v4 ï
 14

The shape function four node rectangular element is derived and shown in module 3, lecture 4.
However the shape functions are reproduced here for easy reference for the derivation of the
stiffness matrix.
ïìï (1- x )(1- h ) ïüï
ïï ïï
ïï 4 ïï
ïìï N1 ïüï ïï (1 + x )(1- h )ïï
ïï ïï ïï ïï
ï N2 ï ï 4 ï
Ni = í  = í  (5.3.5)
ïï N 3 ïï ïï(1 + x )(1 + h )ïï
ïï ïï ïï ïï
ïî N 4 ï ïï 4 ï
ïï 1- x 1 + h ïïï
ïï ( )( )ï
ïï
ïîï 4 ï

Fig. 5.3.1 Four node rectangular element

The strain-displacement relationship for two dimensional plane stress/strain problem can be
simplified in the following form from three dimensional cases (eq.1.3.9 to1.3.14).
 15

¶u 1 éêæç ¶u ö÷ æç ¶v ö÷ ùú
2 2

ex = + êç ÷÷ + ç ÷÷ ú
¶x 2 ëêçè ¶x ø èç ¶x ø ûú

¶v 1 éêæç ¶u ö÷ æç ¶v ö÷ ùú
2 2

ey = + ç ÷ +ç ÷ (5.3.6)
¶y 2 êêçè ¶y ø÷÷ èç ¶y ø÷÷ úú
ë û
¶v ¶u é ¶u ¶u ¶v ¶v ù
g xy = + +ê + ú
¶x ¶y êë ¶x ¶y ¶x ¶y úû
In case of small amplitude of displacement, one can ignore the nonlinear term of the above equation
and will reach the following expression.
u
x 
x
v
y  (5.3.7)
y
v u
 xy  
x y
Using the shape function the above expression can be written as
 u1 
 
 N1 N 2 N 3 N 4   v1 
 0 0 0 0  u 
  x   x x x x  2
   N1 N 2 N 3 N 4   v2 
y    0 0 0 0      B  d  (5.3.8)
   y y y y  u3 
 xy   N N1 N 2 N 2 N 3 N 3 N 4 N 4   v3 
1
  
 y x y x y x y x  u4 
v 
 4
Here, [B] is known as strain displacement relationship matrix. The derivatives of the shape functions
are calculated using the chain rule.
Ni Ni x Ni y
 .  .
 x  y 
(5.3.9)
Ni Ni x Ni y
 .  .
 x  y 
Here, i is referred to number of nodes in an element and will be 4 in this case. Converting above
expression in matrix form
ìï ¶Ni ü ï é ¶x ¶y ù ïì ¶Ni ïü ïì ¶Ni ïü
ïï ï
ï ê úï ï ï ï
ïï ¶x ïï ê ¶x ¶x ú ïï ¶ ïï ï
ï ¶ ï
ï
ú ïí ï = [ J ]ïí ï
x x
í = ê (5.3.10)
ïï ¶Ni ï
ï ê ¶x ¶y ú ï
ï ¶N iïï ï
ï ¶N iï
ï
ïï ï ê úï ï ï ï
ï ï
ï ¶y ï ï
ï ¶y ï
îï ¶h ï ëê ¶h ¶h ûú î ï
 î ï

 16

The matrix [J] is referred to Jacobian matrix which is discussed in Lecture 7, module 3. Using eq.
(5.3.1) one can write
¶x ¶N1 ¶N ¶N ¶N
= x1 + 2 x2 + 3 x3 + 4 x4 (5.3.11)
¶x ¶x ¶x ¶x ¶x
Putting the values of the nodal coordinates and shape functions of the four node element in the above
equation the following relations will be obtained.
¶x 1´(1- h ) 1´(1- h ) 1´(1 + h ) 1´(1 + h ) a
=- ´0 + ´a + ´a - ´0 = (5.3.12a)
¶x 4 4 4 4 2
Similarly,
¶y 1´(1- h ) 1´(1- h ) 1´(1 + h ) 1´(1 + h )
=- ´0 + ´0 + ´b - ´b = 0 (5.3.12b)
¶x 4 4 4 4
¶x 1´(1- x ) 1´(1 + x ) 1´(1 + x ) 1´(1- x )
=- ´0 - ´a + ´a + ´0 = 0 (5.3.12c)
¶h 4 4 4 4
¶y 1´(1- x ) 1´(1 + x ) 1´(1 + x ) 1´(1- x ) b
=- ´0 - ´0 + ´b + ´b = (5.3.12d)
¶h 4 4 4 4 2
Substituting above values in Jacobian matrix the following relations will be obtained.
éa ù é2 ù
ê 0ú ê 0ú
ê ú ê ú
[ J ] = ê 2 ú and [ J ] = ê a ú
-1
(5.3.13)
ê bú ê 2ú
ê0 ú ê0 ú
ëê 2 ûú ëê b ûú
Thus, eq.(5.3.10) can be written as
ïïì ¶Ni ïïü ì ¶N ü ì ¶N ü ì 2 ¶Ni üï
ïïï i ïïï éê 2 0 ùú ïïï i ïïï ï ï
ï . ïï
ï ¶x ïï
ï ï ¶ x ï ê ú ï ¶ x ï ï a ¶ x ï
íï ï = [ J ] íï ï = ê a ú ïí ï = íï ï
-1
(5.3.14)
ï
ïï ¶N ï
ïï ï
ïï ¶N ï
ïï ê 0ê 2 ú ï ¶N ï ï 2 ¶N ï
úï ï ï . ï
i
ï ï
ïï ï ï
i i i

îïï ¶y ïï ïîï ¶h ïï ëê b ûú ïï

î ¶ h  î
ï
ï b ¶ h ï
ï
After derivation of the shape functions expressed in eq.(5.3.5), the following values will be obtained.
¶N1 2 ¶N1 (1- h ) ¶N 2 (1- h ) ¶N 3 (1 + h ) ¶N 4 (1 + h )
= . =- ; = ; = ; =-
¶x a ¶x 2a ¶x 2a ¶x 2a ¶x 2a (5.3.15)
¶N1 2 ¶N1 (1- x ) ¶N 2 (1 + x ) ¶N 3 (1 + x ) ¶N 4 (1- x )
= . =- ; =- ; = ; =
¶y b ¶h 2b ¶y 2b ¶y 2b ¶y 2b
So, the strain displacement relationship matrix, [B] will become as follows.
 17

 1    1    1    1    
 0 0 0  0 
 2a 2a 2a 2a 
 1    1    1    1     (5.3.16)
 B   0  0  0 0 
 2b 2b 2b 2b 
 1    1    1    1    1    1    1    1    
    
 2b 2a 2b 2a 2b 2a 2b 2a 

The element stiffness matrix will become

 k   t   B   D  B  dx dy  t   B   D  B  J d  d
T T
(5.3.17)
It is seen that the above is expressed in terms of ξ and η and hence can be numerically integrated by
the Gauss Quadrature rule. The stiffness matrix for each element can be found which needs to be
globally assembled for getting the global stiffness matrix to obtain the solution. The stiffness matrix
of higher order rectangular element can be derived in a similar fashion. For example, in case of eight
node rectangle element, the size of [B] matrix will become 16 × 3 which was 8 × 3 for four node
element. Thus the size of element stiffness for eight node element will become 16 × 16.

5.3.2 Computation of Nodal Loads

If a distributed load acts on a side of a four node rectangular element, the nodal load vector can be
calculated the similar procedure as discussed in case of triangular element. If an element as shown
below is subjected to a linearly varying intensities of load at its one side, then the magnitude of this
at any point on the side can be expressed by its interpolation function as follows.
1   1    qx 2 
qx     (5.3.18)
 2 2   q x 3 
Here, qx2 and qx3 are the force intensities per unit length at nodes 2 and 3 respectively. The load at
nodes can be calculated from the following expression.
Fx     N 2S  q x d 1 (5.3.19)
As ξ=1 along the side 2-3, the interpolation function will become
 1   1    
   0 
 4   
 1   1      1    
  
2 
N S
 4
  (5.3.20)
 1   1      1    
2

 4   2 
   
 1   1      0 
 4 
 18

If the element thickness is t, then dΓ1 =t.dl. Thus the eq.(5.3.19) can be replaced as
 0 

 1    
1
  1  1    qx 2 
 x    2  
F  t  dl (5.3.21)
1  1     
2 2   q x 3 
 2 
 
 0 

Fig. 5.3.2 Varying load on a four node element

After integrating the above expression, the nodal load vector along x direction will become as
follows.
 0 
 2q  q 
Fx    x 2 x 3 
t
(5.3.22)
3  q x 2  2q x 3 
 0 
 Module 6: FEM for Plates and Shells 11 
Lecture 2: Finite Element Analysis of Thin Plate

6.2.1 Triangular Plate Bending Element

A simplest possible triangular bending element has three corner nodes and three degrees of freedom per
nodes , , as shown in Fig. 6.2.1.

Fig. 6.2.1 Triangular plate bending element

As nine displacement degrees of freedom present in the element, we need a polynomial with nine
independent terms for defining, w(x,y) . The displacement function is obtained from Pascal’s triangle by
choosing terms from lower order polynomials and gradually moving towards next higher order and so
on.
  1

x y

x2 xy y2

x3 x2y xy2 y3

x4 x3 y x2 y2 xy3 y4

Thus, considering Pascal triangle, and in order to maintain geometric isotropy, we may consider the
displacement model in terms of the complete cubic polynomial as,
w  x, y    0  1 x   2 y   3 x 2   4 xy   5 y 2   6 x 3   7  x 2 y  xy 2    8 y 3 (6.2.1)
 12 
 

Corresponding values for , are,

w
x    2   4 x  2 5 y   7  x 2  2 xy   3 8 y 2 (6.2.2)
y
w
y    1  2 3 x   4 y  3 6 x 2   7  2 xy  y 2  (6.2.3)
x
In the matrix form,
 0 
 
 1
 2 

 w  1 x y x
2
xy y 2 x3  x y  xy 
2 2
y 3   
 3
     (6.2.4)
 x   0 0 1 0 x 2y 0  x  2 xy 
2
3 y 2   4 
    
 y  0 1 0 2 x  y 0 3 x 2
   2 xy  y 
2
0   5 
 6 
 
 7
 8 
Putting the nodal displacements and rotations for the triangular plate element as shown in Fig. 6.2.1 in
the above equation, one can express following relations.
 w1  1 0 0 0 0 0 0 0 0 
 0 
  0 0 1  0 0 0 0 0 0   
 x1   1
 y1  0 1 0 0 0 0 0 0 0  
  1 0 y2  2   2 
0 0 y2 0 0 y23   
 w2   3
  0 0 1 0 0 2 y2 0 0 3 y22   
 x 2      4  (6.2.5)
  0 1 0 0  y2 0 0  y22 0  
5
 y 2  1 x y x32 x3 y3 y32 x33  x32 y3  x3 y32  y32   
 w3   3 3
6
  0 0 1 2
3 y3   7 
 x 3   0 x3 2 y3 0  2 x3 y3  x32 
 y 3    
  0 1 0 2 x3  y3 0 3x32   y32  2 x3 y3  0   8 
Or,
     di 
1
(6.2.6)
Further, the curvature of the plate element can be written as

= (6.2.7)

Again, from eq. (6.2.1) the following equations can be obtained.

 13 
 

2w
 2 3  6 6 x  2 7 y
x 2
2w
 2 5  2 7 x  6 8 y
y 2 (6.2.8)
 w
2
  4  2 7  x  y 
x y

The above equation is expressed in matrix form as

 0 
 
 1
 2 
 
  x  0 0 0 2 0 0 6 x 2 y 0   3 
    
  y   0 0 0 0 0 2 0 2 x 6 y   4  (6.2.9)
   0 0 0 0 2 0 0 4 x  y 0   5 
 xy    
 6 
 
 7
 8 
Or,
    B  
Thus,
    B    d 
1

(6.2.10)
Further for isotropic material,
 
 Mx  1  0   x 
  Et 3   
My   2 
 1 0   y  (6.2.11)
 M  12 1     1      xy 
 xy  0 0 
 2 
Or
M    D     D  B    d 
1
(6.2.12)
Now the strain energy stored due to bending is
(6.2.13)
Hence the force vector is written as
(6.2.14)
Thus, [k] is the stiffness matrix of the plate element and is given by
 14 
 

(6.2.15)
For a triangular plate element with orientation as shown in Fig. 6.2.2, the stiffness matrix defined in
local coordinate system [k] can be transformed into global coordinate system.
(6.2.16)
Where, [K] is the elemental stiffness matrix in global coordinate system and [T] is the transformation
matrix given by
1 0 0 0 0 0 0 0 0 
0 l m 0 0 0 0 0 0 
 x x 
0 l y my 0 0 0 0 0 0 
 
0 0 0 1 0 0 0 0 0 
T   0 0 0 0 lx mx 0 0 0  (6.2.17)
0 0 0 0 l y my 0 0 0 
0 0 0 0 0 0 1 0 0 
 
 0 0 0 0 0 0 0 l x mx 
0 0 0 0 0 0 0 l m 
 y y

Here, (lx , mx) and (ly, my) are the direction cosines for the lines OX and OY respectively as shown in Fig.
6.2.2.

Fig. 6.2.2 Local and global coordinate system

 15 
 

6.2.2 Rectangular Plate Bending Element

A rectangular plate bending element is shown in Fig. 6.2.3. It has four corner nodes with three degrees
of freedom , , at each node. Hence, a polynomial with 12 independent terms for defining w(x,y)
is necessary.

Fig 6.2.3 Rectangular plate bending element

Considering Pascal triangle, and in order to maintain geometric isotropy the following displacement
function is chosen for finite element formulation.
w(x, y) = a 0 + a1x + a 2 y + a 3 x 2 + a 4 xy + a 5 y 2 + a 6 x 3 + a 7 x 2 y
(6.2.18)
2 3 3 3
+ a 8 xy + a 9 y + a10 x y + a11xy
Hence Corresponding values for , are,
¶w (6.2.19)
qx = = a 2 + a 4 x + 2a 5 y + a 7 x 2 + 2a 8 xy + 3a 9 y 2 + a10 x 3 + 3a11xy 2
¶y
¶w (6.2.20)
qy = - = -a1 - 2a 3 x - a 4 y - 3a 6 x 2 - 2a 7 xy - a 8 y 2 - 3a10 x 2 y - a11y3
¶x
The above can be expressed in matrix form
3 ùï
ì a 0 üï
ìï w üï é1 x y x 2 xy y 2
x 3
x 2
y xy 2
y 3
x 3
y xy ï ï
ïïï ïïï êê ú ïï a ïï
íq x  = ê0 0 1 0 x 2y 0 x2 2xy 3y 2 x3 3xy 2 úú ïí 1 ï
ïï ïï ê ï  ï
ïîïq y ïï ëê0 -1 0 -2x -y 0 -3x 2 -2xy -y 2 0 -3x 2 y -y3 úúûïïï ïïï
îïa11ï
(6.2.21)
 16 
 

In a similar procedure to three node plate bending element the values of {α} can be found from the
following relatios.
ì w1 ï
ï ü é1 0 0 0 0 0 0 0 0 0 0 0 ù ìï a 0 ïü
ï
ï ï
ï ê ú ïï ïï
ï
ï q ï
ï ê 0 0 1 0 0 0 0 0 0 0 0 0 ú ïï a1 ïï
ï x1
ï ê úï ï
ïï q ïï ê0 -1 0 0 0 0 0 0 0 0 0 0 ú ïï a 2 ïï
ï
ï
y1
ï
ï ê ú ïï ïï
ï w ï ê 1 0 b 0 0 b 2
0 0 0 b 3
0 0 úïa ï
ï
ï
2ï
ï ê ú ïï 3 ïï
ïq ï ê 2
0 úú ïï a 4 ïï
ï
ï x2ï
ï ê0 0 1 0 0 2b 0 0 0 3b 0
ï ïï
ï
ïq ï
ï ê0 -1 0 3 úï
ú ïï a 5 ïï
2
ï y2 ï = ê 0 - b 0 0 0 - b 0 0 - b
í  ê úí 
ï
ï w3ï ï ê1 a b a
2
ab b 2 a3 a 2 b ab 2 b3 a 3b ab3 ú ïï a 6 ïï
ï
ï ï ê úï ï
ï q x3 ïï ê 0 0 1 0 a 2b 0 a2 2ab 3b 2 a3 3ab3 ú ïï a 7 ïï
ï
ï ï
ï ê 3 úï
ï ïï
ï
ï q y3 ïï ê 0 -1 0 - 2a - b 0 - 3a 2
- 2ab - b 2
0 - 3a 2
b - b ú ïï a 8 ïï
ï
ï ï ê ú ï ï
ï w4ï ï ê1 a 0 a 2
0 0 a 3
0 0 0 0 0 ú ïï a 9 ïï
ï
ï ï
ï ê ú ïï ïï
ï
ïq ï
ï ê 0 0 1 0 a 0 0 a 2
0 0 a 3
0 ú ïïa10 ïï
ï
x 4
ï ê úï ï
ïq
ï y4  ï ê
ï ë 0 -1 0 - 2a 0 0 - 3a 2
0 0 0 0 0 úû ïîïa11 ïï
î
(6.2.22)
Thus,
      d 
1
(6.2.23)
Further,

= (6.2.24)

Where,
¶2w
= 2a 3 + 6a 6 x + 2a 7 y + 6a10 xy
¶x 2
¶2w (6.2.25)
= 2a 5 + 2a 8 x + 6a 9 y + 6a11xy
¶y 2
¶2w
= a 4 + 2a 7 x + 2a 8 y + 3a10 x 2 + 3a11y 2
¶x¶ y
Thus, putting values of eq. (6.2.25) in eq. (6.2.24), the following relation is obtained.
ìa ï
ï ü
ïïìc x ïïü é 0 0 0 -2 0 0 -6x -2y 0 0 -6xy 0 úù ïï 0 ïï
ê ï a1 ï
ïíïc ïï = ê 0 0 0 0 0 2 0 0 - 2x -6y 0 -6xyú ï í 
ï (6.2.26)
ïï ïï ê
y ê ú ï  ï
ïïc xy ïï ê 0 0 0 0 2 0 0 4x 4y 0 6x 2 6y 2 úú ïï ï ï
î  ë û ïïa11ïï
î 
 17 
 

Or,
{c} = [ B ] éëêF-1 ùûú {d } (6.2.27)
Further in a similar method to triangular plate bending element we can estimate the stiffness matrix for
rectangular plate bending element as,
 
 Mx  1  0   x 
  Et 3   
My   2 
 1 0   y  (6.2.28)
  12 1     1      xy 
 M xy  0 0 
 2 
Or
M    D     D  B    d 
1
(6.2.29)
The bending strain energy stored is
(6.2.30)
Hence, the force vector will become
(6.2.31)
Where, [k] is the stiffness matrix given by
(6.2.32)
The stiffness matrix can be evaluated from the above expression. However, the stiffness matrix also can
be formulated in terms of natural coordinate system using interpolation functions. In such case, the
numerical integration needs to be carried out using Gauss Quadrature rule. Thus, after finding nodal
displacement, the stresses will be obtained at the Gauss points which need to extrapolate to their
corresponding nodes of the elements. By the use of stress smoothening technique, the various nodal
stresses in the plate structure can be determined.
 Module 6: FEM for Plates and Shells 18 
Lecture 3: Finite Element Analysis of Thick Plate

6.3.1 Introduction
Finite element formulation of the thick plate will be similar to that of thin plate. The difference will be
the additional inclusion of energy due to shear deformation. Therefore, the moment curvature relation
derived in first lecture of this module for thick plate theory will be the basis of finite element
formulation. The relation is rewritten in the below for easy reference to follow the finite element
implementation.
1 0 0 0 χ
1 0 0 0 χ
0 0 0 0 χ (6.3.1)
0 0 0 0
0 0 0 0
Or,
M   D   0    
    (6.3.2)
 Q    0  Ds    
The above relation is comparable to stress-strain relations.
 P  C P  P (6.3.3)
Where,
χ
χ
χ (6.3.4)

Where [B] is the strain displacement matrix and {di} is the nodal displacement vector. Thus, combining
eqs. (6.3.3) and (6.3.4), the following expression is obtained.
 P  C P  Bdi  (6.3.5)

6.3.2 Strain Displacement Relation

Let consider a four node isoparametric element for the thick plate bending analysis purpose. The
variation of displacement w and rotations, θx and θy within the element are expressed in the form of
nodal values.
4
w   N i wi
i 1
4
 x   N i xi (6.3.6)
i 1
4
 y   N i yi
i 1
 19 
 

Where, the shape function for the four node element is expressed as,
1
N i  1   i 1  i  (6.3.7)
4
Here, i and i are the local coordinates  and  of the ith node. Using eq. (6.3.6), eq. (6.3.4) can be
rewritten as
4
N i
 x    yi
i 1 x
4
N
 y    xi i
i 1 y
4
N i 4 N i
 xy    yi    xi (6.3.8)
i 1 y i 1 x
4
N i 4
 x   wi    yi N i
i 1 x i 1
4
N 4
 y   wi i    xi N i
i 1 y i 1
The above can be expressed in matrix form as follows:

 N i 
 0 0
x 
 
N i
 x   0  0 
   y 
 y   N N i 

 P    xy    0
 i d  (6.3.9)
   x y  i

 x   N i
  y   0 Ni 
x 
 
 N i
 Ni 0 
 y 
Here, for a four node quadrilateral element, the nodal displacement vector {di} will become
, , , , , , , , , , , (6.3.10)
Thus, the strain-displacement relationship matrix will be
 20 
 

 N1 N 2 N 3 N 4 
 0 0 0 0 0 0 0 0
 x x x x 
 0 N1 N 2 N 3 N 4
  0 0  0 0  0 0  0 
y y y y
 
N N1 N N 2 N N 3 N N 4 
 B    0  1
x y
0  2
x y
0  3
x y
0  4
x y 
 
 N1 N 2 N 3 N 4 
 x 0 N1 0 N2 0 N3 0 N4 
x x x
 
 N1  N1 0
N 2
 N2 0
N 3
 N3 0
N 4
 N4 0 
 y y y y 
(6.3.11)
Or,
 B    B1 53  B2 53  B3 53  B4 53  (6.3.12)

Now, eq. (6.3.5) can be expressed using above relation as

4
 P  C P  B di   C P   B di  (6.3.13)
i 1

Or,
C P  B   CP B1 53 CP B2 53 CP B3 53 CP B4 53  (6.3.14)

Considering the ith sub-matrix of the above equation,

   t 2  N i  t 2  N i 
   
 0 1    y  1    x 
 t 2  N i  t  N i  
2
  
 0 1    x  1    y  
 
Et t 2  N i  t 2  N i  
CBi   
    (6.3.15)
12 1     0 2  x  2  y  
 
  N i  
6  x 
 6 N i 
 0 
  N i  
6   
  y  6 N i 0 

The bending and shear terms form above equation are separated and written as
 21 
 

é é -mt 2 æ ¶N ö t 2 æç ¶N i ÷öùú ù
êê çç i ÷÷ ú
ê ê 1 - m çè ¶y ÷ø 1 - m ççè ¶x ÷÷øú é 0 0 0 ùú ú
ê ê0 ú ê ú
êê ú ê 0 0 ú
0 ú úú
-t 2 æç ¶N i ö÷ mt æç ¶N i ö÷ú ê
2
êê
êê ççè ÷÷ çç ÷ú ê 0 0 0 úú úú
Et ê ê0 1 -m ¶x ø 1 - m è ¶y ÷øú ê
[CBi ] = êê ú + 6a ê ¶N úú
12(1 + m ) ê ê - t 2
æ ¶N ö t 2æ
¶ N ö ú ê i úú
êê çç i ÷÷ ç
çç
i ÷
÷÷ ú ê ¶x 0 Ni ú ú
ê ê0 ç
2 è ¶x ø ÷ 2 è ¶y ø ú ê úú
êê ú ê ¶N i úú
êê ú ê úú
ê ê0 0 0 ú êë ¶y -N i 0 úû ú
êê ú ú
ëê ë0 0 0 û úû
(6.3.16)
The above expression can be written in compact form as
[CBi ] = [CBi ]b + [CBi ]s (6.3.17)
Here, the contributions of bending and shear terms to stress displacement matrix is denoted as [CBi]b and
[CBi]s respectively. Generally, the contribution due to bending, [CBi]b in eq. (6.3.17) is evaluated
considering 2×2 Gauss points where as the shear contribution [CBi]s is evaluated considering 1×1 Gauss
point.

6.3.3 Element Stiffness Matrix

The expression for element stiffness matrix is
T
[ k ] = ò òA[ B] [C]p [ B]dx dy (6.3.18)

In natural coordinate system, the stiffness matrix is expressed as

T
[ k ] = ò òA[ B] [C]p [ B] J dx dh (6.3.19)
T
Using value of [C]P form eq. 6.3.1 and [Bi] from 6.3.11, the product of [ B] [C]p [ B] is evaluated as
 22 
 

é ¶N i ¶N i ù
ê 0 0 0 ú
ê ¶x ¶y ú
ê ú
ê ¶N i ¶N i ú
é k ù = é[ B] [C] [ B]ù = ê 0
T
- - 0 -N i ú
ë û êë p úû ê ¶y ¶x ú
ê ú
ê ¶N ¶N i ú
ê i 0 Ni 0 ú
ê ¶x ¶y ú
ë û i=1,2,3,4
é ¶N i ù
ê 0 0 ú
é ù êê ¶x ú
é ù ú
ê ê ú é 0 0ù ú ê ¶N ú
ê ê1 u 0 ú ê ú ú ê 0 - i 0 ú
ê Et 3
ê ú ê 0 0ú ú ê ¶y ú
ê 2 êu 1 0 ú ê ú ú ê ú
ê l (1 - u) ê ú ê 0 0ú ú ê ¶N ¶N i ú
´ê ê 1- u ú êë úû ú´ê 0 - i ú
ê ê0 0 ú ú ê ¶x ¶y ú
ê ë 2 û ú ê ú
ê ú ¶N
ê é0 0 0ùú Eta é1 0ù ú êê i 0 Ni ú
ú
ê ê ê ú ú ê ¶x ú
êë êë 0 0 0úû 2(1 + u) êë 0 1úû úû ê ú
ê ¶N i -N 0 úú
ê i
êë ¶y úû i=1,2,3,4
Or in short, (6.2.20)
é ék ù é k12 ù é k13 ù é k14 ù ù
ê ë 11 û ë û ë û ë ûú
ê ú
êék ù é k 22 ù é k 23 ù é k 24 ù ú
é k ù = [ B]T [C] [ B] = ê ë 21 û ë û ë û ë ûú (6.2.21)
ë û p êé ù é k 32 ù é k 33 ù é k 34 ù úú
ê ë k 33 û ë û ë û ë ûú
ê
êék ù é k 42 ù é k 43 ù é k 44 ù ú
êë ë 41 û ë û ë û ë û úû
Where,
é ù
ê ú
ê ú
ê ú
ê ú
ê é ¶N i ¶Ni ¶N i ¶N i ù ¶N i ¶N i ú
ê6a ê + ú -6a Nj 6a Nj ú
ê
ê ë ¶x ¶x ú
¶y ¶y û ¶y ¶y ú
ê ú
ê æ t 2 é ¶N ¶N j ù ö æ mt 2 æ ¶N ¶N j ö÷öú
ê ççç ê i ú ÷÷ çç- ç ÷÷ú
çç 1 - u èçç ¶y ¶x ø÷÷÷÷÷ú
i
ê 1 -m ëê ¶y ¶y ûú ÷
é k ij ù = Et ê ¶N i ççç ÷÷÷ çç ÷÷ú
´ - 6a N ÷
ë û 12(1 + m ) êê i
¶y ççç t 2 êé ¶N i ¶N j úù ÷÷÷ çç t 2 æ ¶N ¶N ö ÷÷÷úú
j÷
ê ç+ + 6aNi N j ÷÷ çç- çç i ÷ ÷÷ú
ê èç 2 ëê ¶x ¶x ûú ÷ø çè 2 èç ¶x ¶y ÷ø÷ ÷÷øú
ê ú
ê æ mt 2 é ¶N ¶N j ù ö÷ æ t 2 æ ¶N ¶N j öö÷ ú
ê ççç- ê i ú ÷÷ ççç çç i ÷÷÷ ú
ê ê ç ÷÷ø÷÷ ú
ê ¶N i çç 1 -m ë ¶x ¶y úû ÷÷÷ çç 1 - m è ¶x ¶ x ÷÷ ú
ê -6aN i çç 2 ÷ çç 2 ÷ú
ê ¶x çç t êé ¶N i ¶N j ùú ÷÷÷ çç t æç ¶N i ¶N j ö÷ ÷÷÷ ú
ê ç- ÷ çç+ çç ÷÷ ÷ ú
ê
ë èç 2 êë ¶y ¶x ûú ÷÷ø è 2 è ¶y ¶y ø÷ ÷÷ø úû
 23 
 

(6.3.22)
By separating the bending and shear terms from above equation,
é ù
ê ú
ê0 0 0 ú
ê ú
ê ú
æ ö æ ö
ê çç t êé ¶N i ¶N j úù + t êé ¶N i ¶N j úù ÷÷ çç- mt æçç ¶N i ¶N j ö÷÷ - t æçç ¶N i ¶N j ö÷÷÷÷úú
2 2 2 2
é k ij ù = Et
ë û 12(1 + m ) ´ êê 0 ççè1 - m ê ¶y ¶y ú 2 ê ¶x ¶x ú ÷÷ø ççè 1 - m çè ¶y ¶x ÷÷ø 2 çè ¶x ¶y ÷÷øø÷÷ú
ê ë û ë û ú
ê æ mt é ¶N ¶N j ù t é ¶N ¶N j ù ö÷
2 2 æ t 2 æ ¶N ¶N j ö t 2 æ ¶N ¶N j öö÷ úú
ê çç- ê i ú- ê i ú÷ çç çç i ÷÷ + çç i ÷÷÷
ê0 ççè 1 - m êë ¶x ¶y úû 2 êë ¶y ¶x úû ÷÷ø ççè1 - m çè ¶x ¶x ÷÷ø 2 çè ¶y ¶y ø÷÷÷÷ø úú
êë û
é é ¶N ¶N ¶N i ¶N i ù ¶N ¶N i ùú
êê i i
+ ú - i Nj Nj
ê ê ¶x ¶x ¶y ¶y úû ¶y ¶x ú
êë ú
ê ¶ N ú
+6a êê Ni i
Ni N j 0 úú
ê ¶y ú
ê ¶ N ú
ê i
N 0 N N ú
ê ¶x
j i j ú
êë úû
(6.3.23)
Thus, the matrix é k ù can now be written as the sum of bending and shear contributions
ë û
é kù = é kù + é kù (6.2.24)
ë û ë û b ë ûs
Or,
é ék ù é k12 ù é k13 ù é k14 ù ù é é k11 ù é k12 ù é k13 ù é k14 ù ù
ê ë 11 û b ë ûb ë ûb ë û b ú ê ë ûs ë ûs ë ûs ë ûs ú
ê ú ê ú
ê é k 21 ù é k 22 ù é k 23 ù é k 24 ù ú ê é k 21 ù é k 22 ù é k 23 ù é k 24 ù ú
ékù = êë û b ë ûb ë ûb ë ûb ú êë ûs
+
ë ûs ë ûs ë ûs ú (6.2.25)
ë û êék ù é k 32 ù é k 33 ù é k 34 ù úú êê é k 33 ù é k 32 ù é k 33 ù é k 34 ù úú
ê ë 33 û b ë ûb ë ûb ë û b ú êë ûs ë ûs ë ûs ë ûs ú
ê
êék ù é k 42 ù é k 43 ù é k 44 ù úú êê é k 41 ù é k 42 ù é k 43 ù é k 44 ù úú
êë ë 41 û b ë ûb ë ûb ë ûb û ëë ûs ë ûs ë ûs ë ûs û
The stiffness matrix [k ] can be evaluated from the following expression by substituting
é k ù for [ B]T [C ] [ B] in eq. (6.3.19) and is given as
ë û p

+1 +1
[ k ] = ò-1 é k ù J d x dh
ò -1 ë û (6.2.26)

Here, J is the determinate of the Jacobian matrix. The Gauss Quadrature integration rule is used to
compute the stiffness matrix [k].

6.3.4 Nodal Load Vector

Considering a uniformly distributed load q on the plate, the equivalent nodal load vector can be
calculated for finite element analysis from the flowing expression.
 24 
 

ì Fx ï
ï ü ìïqüï ìïqüï
ï
ï ï ï ïï ïï
1 1
ïï ïï
{Qi } = íM x  = ò Ni í0 dA = ò ò Ni í0 J dξ dη (6.3.27)
ï
ï ï
ï ï
ï ï
ï ï
ï ï
ï M ï
îï x ï
A
ï
î ïï
ï 0 -1 -1
ïîï0ï
ïï
Using Gauss Quadrature integration rule the above expression can be evaluated as,
ì
ï Ni üï
n n ï ï
{Q} = q åå wi w j J ïí 0 ï (6.3.28)
i=1 i=1
ï
ï ïï
ï
î ïïi=1,2,3,4
ï 0
The nodal load vectors from each element are assembled to find the global load vector at all the nodes.
 25 
Module 6: FEM for Plates and Shells
Lecture 4: Finite Element Analysis of Skew Plate

6.4.1 Introduction
Skew plates often find its application in civil, aerospace, naval, mechanical engineering structures.
Particularly in civil engineering fields they are mostly used in construction of bridges for dealing
complex alignment requirements. Analytical solutions are available for few simple problems. However,
several alternatives are also available for analyzing such complex problems by finite element methods.
Commonly used three discretization methods for skew plates are shown in Fig 6.4.1.

(b) Discretization using combination of

(a) Discretization using rectangular plate
rectangular and triangular plate
elements
elements

(c) Discretization using skew plate element

Fig 6.4.1 Discretization of a skew plate

 26 
 

If the skew plate is discretized using only rectangular plate elements, the area of continuum excluded
from the finite element model may be adequate to provide incorrect results. Another method is to use
combination of rectangular and triangular elements. However, such analysis will be complex and it may
not provide best solution in terms of accuracy as, different order of polynomials is used to represent the
field variables for different types of elements. Another alternative exist using skew element in place of
rectangular element.

6.4.2 Finite Element Analysis of Skew Plate

Let consider a skew plate of dimension “2a” and “2b” as shown in Fig. 6.4.2. Let the skew angle of the
element be “ϕ”. It is possible for the parallelogram shown in Fig. 6.4.3 to map the coordinate from
orthogonal global coordinate system to a skew local coordinate system. If the local coordinates are
represented in the form of ξ, η, then the relationship can is represented as,
x = x + h cos f, y = h sin f (6.4.1)
Hence,
h = ycosecf, x = x - ycot f (6.4.2)

Fig. 6.4.2 Skew plate in global coordinate system

 27 
 

Fig. 6.4.3 Point “P” in global and local coordinate system

It is important to note that the terms (ξ, η) in the above equations represent the absolute coordinate of the
point “P” in skew coordinate system, not the natural coordinates. Since the above element has four
corner nodes and each node have three degrees of freedom present, a polynomial with minimum 12
independent terms are necessary for defining the displacement function w(x,y). Considering Pascal
triangle, and in order to maintain geometric isotropy, the displacement function may be considered as
follows:
w = a 0 + a1x + a 2h + a 3x 2 + a 4xh + a 5h 2 + a 6x3 + a 7x 2h
(6.4.3)
+ a 8xh 2 +a 9h3 + a10x3h + a11xh3
Hence corresponding values for , are,
¶w
qx = = (a 2 + a 4x + 2a 5h + a 7x 2 + 2a 8xh + 3a 9h 2 + a10x3 + 3a11xh 2 ) (6.4.4)
¶h
¶w
qh =- = -(a1 + 2a 3x +a 4h + 3a 6x2 + 2a 7xh +a8h2 + 3a10x2h +a11h3 ) (6.4.5)
¶x
Or, in matrix form,
3 ùï
ì a 0 ïü
ïìï w ïüï é1 x h x 2 xh h 2
x 3
x 2
h xh 2
h 3
x 3
h xh ï ï
ïï ïï êê ú ïï a ïï
íq x  = ê 0 0 1 0 x 2h 0 x2 2xh 3h2 x3 3xh 2 úú ïí 1 ï
ïï ïï ê ï  ï
ïîïq h ïï êë 0 -1 0 -2x -h 0 -3x 2 -2xh -h 2 0 -3x 2h -h3 úúûïïï ïïï
ïîa11ï
(6.4.6)
Or,
 28 
 

d      (6.4.7)
The value of [α] can be determined using value of , , at four nodes as
ïìï w1 ï ü é1 0 0 0 0 0 0 0 0 0 0 0 ù ìï a 0 ïü
ïï ï ï
ï
ê
ê0
ú ïï ïï
ïï ï q 0 1 0 0 0 0 0 0 0 0 0 ú ïï a1 ïï
x1
ï ê úï ï
ïï q ïï ê0 -1 0 0 0 0 0 0 0 0 0 0 úú ïï a 2 ïï
ïï ï y1
ê ïï ïï
ïïw 2 ï ï ê1 0 b 0 0 b2 0 0 0 b3 0 0 úú ïï a 3 ïï
ïï ï ï ê
ê0 ï ï
ïï x 2 ï q ï
ê 0 1 0 0 2b 0 0 0 3b 2 0 0 úú ïï a 4 ïï
ïïq ï ï ï ï
ï y2 ï = ê0 ï ê -1 0 0 -b 0 0 0 -b 2 0 0 -b3 úú ïïï a 5 ïïï
í  ê úí 
ïï w 3 ï ê1 a b a2 ab b2 a3 a 2b ab 2 b3 a 3b ab3 ú ïï a 6 ïï
ïï ï ï ê úï ï
ïïq x3 ï ï
ï ê0 0 1 0 a 2b 0 a2 2ab 3b 2 a3 3ab3 ú ïï a 7 ïï
ïïq ï ê úï ï
ïï y3 ï ï ê0 -1 0 -2a -b 0 -3a 2 -2ab -b 2 0 -3a 2 b -b3 ú ïïï a 8 ïïï
ïïw 4 ï ï ê úï ï
ï ê1 a 0 a2 0 0 a3 0 0 0 0 0 ú ïï a 9 ïï
ïï ï ï ê úï ï
ïïq x 4 ï ê0 0 1 0 a 0 0 a2 0 0 a3 0 ú ïïïa10 ïïï
ïï ï ï
ï
ê úï ï
q ê
îï y4 ï ë0 -1 0 -2a 0 0 -3a 2 0 0 0 0 0 úû ïîïa11 ïï
(6.4.8)
Or,
     di 
1
(6.4.9)
Further, considering eq. (6.4.2)
¶h ¶h üï
= 0, = cos ecfïï
¶x ¶y ïï
 (6.4.10)
¶x ¶x ï
= 1, = - cot f ïï
¶x ¶y ïï
Again,

= (6.4.11)

The values in right hand side of the eq. (6.4.11) can be calculated by using chain rule as,
¶w ¶w ¶x ¶w ¶h ¶w
= . + ⋅ = (6.4.12)
¶x ¶x ¶x ¶h ¶x ¶x
Therefore,
¶2w ¶2w
= 2 (6.4.13)
¶x 2 ¶x
 29 
 

Similarly,
¶w ¶w ¶x ¶w ¶h ¶w ¶w
= ⋅ + ⋅ = - cot f + cosecf
¶y ¶x ¶y ¶h ¶y ¶x ¶h
Thus, further derivation provides
¶2w 2
2 ¶ w 2 ¶2w ¶2w
= cosec f 2 + cot f 2 - 2cot f cosecf (6.4.14)
¶y 2 ¶h ¶x ¶x¶h
And,
¶2w ¶ æç ¶w ö÷ ¶2w 2 ¶2w
= ç ÷ = - cot f 2 + cosec f (6.4.15)
¶x¶y ¶x çè ¶y ÷ø ¶x ¶x¶h
Hence eq. (6.4.11) is converted to
ïìï ¶ 2 w ïüï ïìï ¶ 2 w ïüï
ïï- 2 ïï -
ï ¶x ï é ïïï ¶x 2 ïïï
ùï
ïï 2 ïï
ïï ¶ w ïï êê 2
1 0 0
ú ïï ¶ 2 w ïïï
í- 2  = êcot f cosec f -2cotfcosecfúú ïí- 2 ï
2
(6.4.16)
ïï ¶y ïï ï ¶h ï
ïï 2 ïï êë cotf 0 cosec2f úû ïïï 2 ïïï
ïï 2¶ w ïï ïï 2¶ w ïï
ïï ïï ïï ï
îï ¶x¶y ï îï ¶x¶h ïï
Or,
{c x ,y } = éë H (f)ùû {c x,h } (6.4.17)
Further, by partial differentiation of eq. (6.4.3),
¶2w ü
ï
- ï
= -2a 3 - 6a 6x - 2a 7 h - 6a10xh
ï
¶x 2 ï
ï
2
ï
ï
¶w ï
- 2 = -2a 5 - 2a 8x - 6a 9h - 6a11xh ï  (6.4.18)
¶h ï
ï
2
ï
ï
¶w 2ï
2 = 4a 7 x + 4a 8h + 6a10x + 6a11h ï
2
ï
¶x¶h ï
ï

Or, in matrix form,
 30 
 

ì
ï ¶2w ü ï
ï
ï - 2ï ï
ï
ïï ¶x ï ïï é0 0 0 -2 0 0 -6x -2h
ï ï ê 0 0 -6xh 0 ùìï a 0 üï
ï ¶ 2
w ï ú ïï ïï
ï
í - ï = ê 0 0 0 0 0 -2 0 0 -2 x -6 h 0 - 6xh úí  
2  ê úï ï
ï
ïï ¶h ï
ïï ê0 0 0 0 0 0
ï 2 ï ë 0 4x 4h 0 6x 2
6h ûîïïa11ïï
2 úï
ï
ï
ï 2¶ w ï
ï
ï
ï ï
ï ¶x¶h ï
î ï

(6.4.19)
Or,
{c x,h } = [B]{a } = [B] éêëF-1 ùúû {d i }
Or,
{c x,y } = éë H (f)ùû [B]éêëF-1 ùúû {d i } (6.4.20)
Again,
{M x,y } = [D ]{c x,y } (6.4.21)
Where [D] for plane stress condition is
é ù
ê ú
3 ê1 m 0 ú
Eh ê ú
[ D] = 2 ê
m 1 0 ú (6.4.22)
l2(1 -m ) ê ú
ê 1 -m ú
ê0 0 ú
ë 2 û
Using eq. (6.4.20) in eq. (6.4.21),
{M x,y } = [D ]éë H (f)ùû [B] éëê A-1 ùûú {d} (6.4.23)
The expression for bending strain energy stored,
a b
1
ò {c x,y } {M x,y }dxdy
T
U= ò (6.4.24)
2 0 0

Hence force vector,

a b
¶U T
é -1 ù T
é -1 ù
êëF ûú [ B] ëé H (f)ûù [ D ] ëé H (f)ûù [ B] ëêF ûú {d} dxdy
T

¶ {d} ò0 ò
=
0
(6.4.25)
a b
-1 T T
ò ëêéF ù é -1 ù
úû [ B] ëé H (f)ûù [ D ] ëé H (f)ûù [ B] êëF úû J dxdh {d}
T
=ò
0 0

Where,
 31 
 

é ¶x ¶y ù
ê ú
é ¶ ( x, y)ù ê ¶x ¶x ú é 1 0 ù
[ J ] = êê ú=ê
ú ê ¶x
ú=ê ú = sin f (6.4.26)
êë ¶ ( x, h ) úû ê ¶y ú êëcos f sin fúû
ú
ëê ¶h ¶h ûú

From expression in eq. (6.4.25)

¶U
{F} = = [ k ]{d} (6.4.27)
¶ {d}
Hence,
a b
T T T
[ k ] = sin W⋅ éêëF-1 ùúû ò ò éë B(x, h)ùû é H (f)ù [ D ] é H (f)ù é B(x, h)ù dxdh éêF-1 ùú
ë û ë ûë û ë û
0 0

(6.4.28)
Thus, the element stiffness matrix of a skew element for plate bending analysis can be evaluated from
the above expression using Gauss Quadrature numerical integration.
 32 

Module 6: FEM for Plates and Shells

Lecture 5: Introduction to Finite Strip Method

6.5.1 Introduction
The finite strip method (FSM) was first developed by Y. K. Cheung in 1968. This is an efficient tool for
analyzing structures with regular geometric platform and simple boundary conditions. If the structure is
regular, the whole structure can be idealized as an assembly of 2D strips or 3D prisms. Thus the
geometry of the structure needs to be constant along one or two coordinate directions so that the width
of the strips or the cross-section of the prisms does not change. Therefore, the finite strip method can
reduce three and two-dimensional problems to two and one-dimensional problems respectively. The
major advantages of this method are (i) reduction of computation time, (ii) small amount of input (iii)
easy to develop the computer code etc. However, this method will not be suitable for irregular geometry,
material properties and boundary conditions.

6.5.2 Finite Strip Method

To understand the finite strip method, let consider a rectangular plate with x and y axes in the plane of
the plate and axis z in the thickness direction as shown in Fig. 6.5.1. The corresponding displacement
components of the plate are denoted as u, v and w.

Fig. 6.5.1 Finite strip in a plate

The strips are assumed to be connected to each other along a discreet number of nodal lines that coincide
with the longitudinal boundaries of the strip. The general form of the displacement function in two
dimensions for a typical strip is given by
, ∑ (6.5.1)
Here, the functions ƒm(y) are polynomials and the functions Xm(x) are trigonometric terms that satisfy the
end conditions in the x direction. The functions Xm(x) can be taken as basic functions (mode shapes) of
the beam vibration equation.
 33 
 

0 (6.5.2)
Here L is the length of beam strip and µ is a parameter related to material, frequency and geometric
properties. The general solution of the above equation will become
sin cos sinh cosh (6.5.3)
Four conditions at the boundaries are necessary to determine the coefficients C1 to C4 in the above
expression.

6.5.2.1 Boundary conditions

According to different end conditions eq. (6.5.3) can be solved. Solution of the above equation is
evaluated for few boundary conditions in the below.
(a) Both end simply supported
For simply supported end, following conditions will arise:
(i) At one end (say at x =0) displacement and moment will be zero: x (0 ) = x" (0 ) = 0
(ii) At other end (at x =L) displacement and moment will be zero: x ( L) = x" ( L ) = 0
Thus, considering above boundary conditions, eq. (6.5.3) yields to the following mode shape function:
m
nπx æ μ πx ö
X m ( x ) = å sin
l
= sin çç m ÷÷÷
çè l ø (Where, m m = p, 2p.......upto n t h term) (6.5.4)
n=1

Since the functions Xm are mode shapes, they are orthogonal and therefore, they satisfy the following
relations:
0 (6.5.5)
And
" " 0 (6.5.6)
The orthogonal properties of Xm(x) result in structural matrices with narrow bandwidths and thus
minimizing computational time and storage. Using relation in eq. (6.5.4), eq. (6.5.1) can now be written
as
, ∑ . sin (6.5.7)
(b) Both end fixed supported
In case of fixed supported end at both the side, the following boundary conditions will be adopted:
(i) At one end (say at x =0) displacement and slope will be zero: x (0 ) = x' (0 ) = 0
(ii) At other end (at x =L) displacement and slope will be zero: x ( L) = x' ( L) = 0
For the above boundary conditions, eq. (6.5.3) yields to the following:
æ μ xö æ μ yö é æ μ yö æ μ y öù
X m ( x) = sinçç m ÷÷÷ - sin hçç m ÷÷÷ - αm êcos çç m ÷÷÷ - cos h çç m ÷÷÷ú (6.5.8)
èç L ø èç L ø êë çè L ø çè L øú
û
 34 
 

æ 2m +1 ÷ö sinμm - sinhμm
Where çç μm = 4.73,7.8532,10.996........ π ÷÷ and αm =
çè 2 ø cosμm - coshμm
(c) One end simply supported, other end fixed
(i) At simply supported end (say at x =0) displacement and moment will be zero: x (0 ) = x" (0 ) = 0
(ii) At fixed end (say at x =L) displacement and slope will be zero: x ( L) = x' ( L) = 0
Thus, the solution of eq. (6.5.3) will become
æ μ xö æ μ xö
X m ( x ) = sin çç m ÷÷÷ - αm sin h çç m ÷÷÷ (6.5.9)
çè l ø çè l ø
4m + 1 sinμm
where, μm = 3.9266,7.0685,10.2102....... p and αm =
4 sinhμm
(d) Both end free
If both the end of the strip element is free, the following boundary conditions will be assumed:
(i) At one end (say at x =0) moment and shear will become zero: x" (0 ) = x"' (0 ) = 0
(ii) At other end (at x =L) moment and shear will become zero: x" ( L) = x'" ( L) = 0
Thus, for the above end conditions, eq. (6.5.3) yields to the following:
2x
X 1 = 1, μ1 = 0 & X 2 = 1 - ,μ2 = 1
L
(6.5.10)
æ μ xö μ x æ μ x μ xö
X m ( x) = sin çç m ÷÷÷ + sin h m - αm ççcos m + cosh m ÷÷÷
çè l ø l èç l l ø
sinμm - sin hμm 2m - 3
Where, αm = and μm = 4.73,7.8532, 10.996....... π, for m = 3,4,- - - µ
cosμm - coshμm 2

6.5.3 Finite Element Formulation

In this section, finite element solution for a finite strip will be evaluated considering simply supported
conditions at both the end. As a result, the functions ƒm(Y) in eq. (6.5.7) can be expressed for the
bending problem as
∝ ∝ ∝ ∝ (6.5.11)
Applying boundary conditions of the strip plate of width b, the following relations will be obtained.
1 0 0 0
0 1 0 0
(6.5.12)
1
0 1 2 3
Thus, the nodal displacement can be written in short as
(6.5.13)
Thus, the unknown coefficient α are obtained from the following relations.
(6.5.14)
 35 
 

The formulation of the finite strip method is similar to that of the finite element method. For example,
for a strip subjected to bending, the moment curvature relation will become

0
0 (6.5.15)
0 0 2
2
Where Mx , My and Mxy are moments per unit length and [D] is the elasticity matrix. From eq. (6.5.7)
the following expressions are evaluated to incorporate in the above equation.
∑

∑ 2 6 (6.5.16)

2 ∑ 2 2 3
The above expression is written in matrix form in the below

∑
∑ 2 6 (6.5.17)
∑2 2 3
2
Here, is denoted as N. Rearranging the above expression, one can find the following.

sin 0 0
0 sin 0 2 6
0 0 cos 2 2 3
2

sin 0 0
= 0 sin 0 0 0 2 6 (6.5.18)
0 0 cos 0 2 4 6
Thus, in short, the curvature and moment equation will become
(6.5.19)
(6.5.20)
Now, the strain energy for the bending element can be written similar to plate bending formulation.
1 1
2 2
(6.5.21)
Thus the force vector can be derived as
 36 
 

(6.5.22)
Thus, the stiffness matrix of a strip element can be obtained from the following expression.

(6.5.23)
The stiffness matrix [k] can be simplified by integrating the term as follows.

sin 0 0 0 sin 0 0
0 sin 0 0 0 sin 0
0 0 cos 0 0 2 0 0 cos
sin sin 0 sin 0 0
sin sin 0 0 sin 0
0 0 2 cos 0 0 cos
sin sin 0
sin sin 0 (6.5.24)
0 0 2 cos
Here, the terms Dx, Dy and Dxy are constant and not varied with x or y. Following integrations are carried
out to simplify the above expression further.
Now

Putting, finally will become

Similarly;
Thus,
0
0 (6.5.25)
0 0 2
Using eq. (6.5.25), the expression for stiffness matrix [k] in eq. (6.5.23) is simplified as follows.

0 0
0
0 2
0 0 0 2 6
2 4
0 0 2 0 2 4 6
6 6
 37 
 

0
4
0 0 2 6
2 2 2 8
0 2 4 6
6 6 12

2 6
8 2 16 6 24
4 2 6

4 32 12 48
2 2 2 8
6 6
6 6 24
2 12 6 36
48 72
(6.5.26)
Thus, by putting the assumed shape function, the stiffness matrix of a strip element can be evaluated
numerically using Gaussian Quadrature or other numerical integration methods.
 Module 6: FEM for Plates and Shells 38 
Lecture 6: Finite Element Analysis of Shell

6.6.1 Introduction
A shell is a curved surface, which by virtue of their shape can withstand both membrane and bending
forces. A shell structure can take higher loads if, membrane stresses are predominant, which is primarily
caused due to in-plane forces (plane stress condition). However, localized bending stresses will appear
near load concentrations or geometric discontinuities. The shells are analogous to cable or arch structure
depending on whether the shell resists tensile or, compressive stresses respectively. Few advantages
using shell elements are given below.
1. Higher load carrying capacity
2. Lesser thickness and hence lesser dead load
3. Lesser support requirement
4. Larger useful space
5. Higher aesthetic value.

The example of shell structures includes large-span roof, cooling towers, piping system, pressure vessel,
aircraft fuselage, rockets, water tank, arch dams, and many more. Even in the field of biomechanics,
shell elements are used for analysis of skull, Crustaceans shape, red blood cells, etc.

6.6.2 Classification of Shells

Shell may be classified with several alternatives. Depending upon deflection in transverse direction due
to transverse shear force per unit length, the shell can be classified into structurally thin or thick shell.
Further, depending upon the thickness of the shell in comparison to the radii of curvature of the mid
surface, the shell is referred to as geometrically thin or thick shell. Typically, if thickness to radii of
curvature is less than 0.05, then the shell can be assumed as a thin shell. For most of the engineering
application the thickness of shell remains within 0.001 to 0.05 and treated as thin shell.

6.6.3 Assumptions for Thin Shell Theory

Thin shell theories are basically based on Love-Kirchoff assumptions as follows.
1. As the shell deforms, the normal to the un-deformed middle surface remain straight and normal
to the deformed middle surface undergo no extension. i.e., all strain components in the direction
of the normal to the middle surface is zero.
2. The transverse normal stress is neglected.
Thus, above assumptions reduce the three dimensional problems into two dimensional.

6.6.4 Overview of Shell Finite Elements

Many approaches exist for deriving shell finite elements, such as, flat shell element, curved shell
element, solid shell element and degenerated shell element. These are discussed briefly bellow.
 39 
 

(a) Flat shell element

The geometry of these types of elements is assumed as flat. The curved geometry of shell is obtained by
assembling number of flat elements. These elements are based on combination of membrane element
and bending element that enforced Kirchoff’s hypothesis. It is important to note that the coupling of
membrane and bending effects due to curvature of the shell is absent in the interior of the individual
elements.

(b) Curved shell element

Curved shell elements are symmetrical about an axis of rotation. As in case of axisymmetric plate
elements, membrane forces for these elements are represented with respected to meridian direction
as , , and in circumferential directions as , , . However, the difficulties associated with
these elements includes, difficulty in describing geometry and achieving inter-elemental compatibility.
Also, the satisfaction of rigid body modes of behaviour is acute in curved shell elements.

(c) Solid shell element

Though, use of 3D solid element is another option for analysis of shell structure, dealing with too many
degrees of freedom makes it uneconomic in terms of computation time. Further, due to small thickness
of shell element, the strain normal to the mid surface is associated with very large stiffness coefficients
and thus makes the equations ill conditioned.

(d) Degenerated shell elements

Here, elements are derived by degenerating a 3D solid element into a shell surface element, by deleting
the intermediate nodes in the thickness direction and then by projecting the nodes on each surface to the
mid surface as shown in Fig. 6.6.1.

(a) 3D solid element (b) Degenerated Shell element

6.6.1 Degeneration of 3D element

 40 
 

This approach has the advantage of being independent of any particular shell theory. This approach can
be used to formulate a general shell element for geometric and material nonlinear analysis. Such element
has been employed very successfully when used with 9 or, in particular, 16 nodes. However, the 16-
node element is quite expensive in computation. In a degenerated shell model, the numbers of unknowns
present are five per node (three mid-surface displacements and two director rotations). Moderately thick
shells can be analysed using such elements. However, selective and reduced integration techniques are
necessary to use due to shear locking effects in case of thin shells. The assumptions for degenerated
shell are similar to the Reissner-Mindlin assumptions.

6.6.5 Finite Element Formulation of a Degenerated Shell

Let consider a degenerated shell element, obtained by degenerating 3D solid element. The degenerated
shell element as shown in Fig 6.6.1(b) has eight nodes, for which the analysis is carried out. Let ,
are the natural coordinates in the mid surface. And ς is the natural coordinate along thickness direction.
The shape functions of a two dimensional eight node isoparametric element are:
(1- x )(1- h )(-x - h -1) (1 + x )(1- x )(1- h )
N1 = N5 =
4 2
(1 + x )(1- h )(x - h -1) (1 + x )(1 + h )(1- h )
N2 = N6 =
4 2 (6.6.1)
(1 + x )(1 + h )(x + h -1) (1 + x )(1- x )(1 + h )
N3 = N7 =
4 2
(1- x )(1 + h )(-x + h -1) (1- x )(1 + h )(1- h )
N4 = N8 =
4 2
The position of any point inside the shell element can be written in terms of nodal coordinates as
ì x ïü
ï ìï ïìï xi ïüï ì xi ï
ï ü üï
ï ï ïï ï ï ïï
ïí yï =
8
ï 1+ V ï ï 1- V ï ï ï
ïï ï å N i ( x , h )í
ï 2
í yi  +
ï ï 2
í yi 
ï ï ï
 (6.6.2)
ïî ï i = ïï ï ï ï ï ï
ï zï
1
ï îï ïïî zi ïïtop ï
ï zi ï
î ïbottom ï
 ï
Since, ς is assumed to be normal to the mid surface, the above expression can be rewritten in terms of a
vector connecting the upper and lower points of shell as
ìï x üï ì
ï ì
ïì xi üï ì xi üï ü
ï ì
ïì xi üï ì xi üï ü
ïü
ï
ïï ïï ï ïï
ï ï ï
ï ï ï ïï
ï ï ï
ï ï ïï
8 ï
ï ï
1 ïï ï ï ï ï V ïï ï ï ï ïï
ï
ï ï ï
í y = å Ni (x , h )í íí yi  + í yi   + íí yi  - í yi  
ïï ïï i=1 ï
ï 2 ïï
ïï ï ï ï ï 2 ïï ï ï ï ïï
ïîï z ïï ï
ïî ïïï
ï
î zi ïïïtop ïïîï zi ïïïbottom ïïï ïï
ïï
ïï
î zi ïïïtop ïïîï zi ïïïbottom ïï
ïï
ïï
î  î 
ï
Or,
 41 
 

ïìï xïüï ìïïì xi ïü ü

ïïï ï V ïïï
ïí yï =
8
ï
N i (x , h )ïïíí yi ï + V3i ï
ïï ïï å
(6.6.3)
i=1
ïï
ïï ïï 2 ïï
ïîï z ïï ïï
îïîï ïï
zi ïï
Where,
ìï xi üï ìïì x ü ìï xi üï üï ìï xi üï ìï xi üï
ïï ïï 1 ïïïïï i ïïï ïï ïï ïï ïï ïï ïï ïï
ï ï and, V = y
í yi  = íí yi  + í yi   í  - í y (6.6.4)
ïï ïï 2 ïï ïï ïï ïï ïï ïï 3i
ïï i ïï ïï i ïï
ïïî zi ïï ïï zï ïz ï
îïïî i ïtop îï i ïbottom ï
ï ïïî zi ïïtop îïï zi ïïbottom

Fig. 6.6.2 Local and global coordinates

For small thickness, the vector V3i can be represented as a unit vector tiv3i:
ìï xüï ìïìï xi üï üï
ïï ïï ïïï ï V ïï
8
ï ï
í y = å N i (x , h )ïï íí yi  + ti v3i ï (6.6.5)
ïï ïï i=1 ïï
ïï ïï 2 ïï
îïï z ïï îïïïîï zi ïï ïï
th
Where, ti is the thickness of shell at i node. In a similar way, the displacement at any point of the shell
element can be expressed in terms of three displacements and two rotation components about two
orthogonal directions normal to nodal load vector V3i as,
ïìï u ïüï ìïïì ui ïü üï
ïïï ï V t ì
ïa ü
ï ïï
ïí v ï =
8
ïï ï ï ï i ïï

ïï ïï å
N ( x , h ) v
íí i  + i
[ v -v ]
2 i í 
(6.6.6)
i=1
i
ïï
ïï ïï 2 1i ïîïbi ïï
ï
ï
ïîïwïï ïï
îïîïwi ïï ïï
Where, , are the rotations of two unit vectors v1i & v2i about two orthogonal directions normal to
nodal load vector V3i.The values of v1i and v2i can be calculated in following way:
The coordinate vector of the point to which a normal direction is to be constructed may be defined as
x = xiˆ + yjˆ + zkˆ (6.6.7)
 42 
 

In which, iˆ, ˆj , kˆ are three (orthogonal) base vectors. Then, V1i is the cross product of iˆ & V3i as shown
below.
V1i = iˆ´V3i & V2i = V3i ´V1i (6.6.8)
and,
V1i & v2i = V2i
v1i = (6.6.9)
V1i V2 i

6.6.5.1 Jacobian matrix

The Jacobian matrix for eight node shell element can be expressed as,
é 8 ¶N 8
* ¶N i
8
¶N ù
ê å ( xi + txi* ) i å ( y + tyi ) å ( zi + tzi* ) i ú
ê i=1 ¶x
i
¶x ¶x ú
ê i =1 i=1
ú
ê 8 ¶N 8
¶N 8
¶N ú
[ J ] = êê å ( xi + txi* ) i å ( yi + tyi* ) i å ( zi + tzi* ) i úú (6.6.10)
ê i=1 ¶h i =1 ¶h i=1 ¶h ú
ê 8 8 8 ú
ê ú
ê å Nx *
i å Ny *
i å Nz *
i ú
ëê i =1 i =1 i =1 ûú

6.6.5.2 Strain displacement matrix

The relationship between strain and displacement is described by
{e} = [ B ]{d } (6.6.11)
Where, the displacement vector will become:
T
{d } = {u1 v1 w1 v11v21  u8 v8 w8 v18v28 } (6.6.12)
And the strain components will be
ì
ï ¶u ü ï
ï
ï ï
ï
ï
ï ¶ x ï
ï
ï
ï ¶ v ï
ï
ï
ï ï
ï
ï
ï ¶y ï
ï
ï
ï ï
ï
ï
ï ¶ u ¶ v ï
ï
[ ] í
e = +  (6.6.13)
ï
ï ¶y ¶x ï ï
ï
ï ï
ï
ï ¶ v ¶ w ï
ï
ï + ï
ï
ï
ï ¶ z ¶y ï
ï
ï
ï ï
ï
ï ¶w ¶ u ï
ï
ï + ï
ï ¶x ¶z ï
î ï
Using eq. (6.6.6) in eq. (6.6.13) and then differentiating w.r.t. , , the strain displacement matrix
will be obtained as
 43 
 

é ¶u ¶v ¶w ù ìï ¶N i üï ì
ï ¶N i ü
ï ì
ï ¶N i üï
ê ú ïï ï ïV ï ïV ï
ê ¶x ¶x ¶x ú ïï ¶x ïïï ï
ï
ï ¶x ï
ï
ï T
ï
ï
ï ¶ x ï
ï
ï é a ùT
ê ú é b ù
8 ï ïï ¶N ïïï ï
ï ï
ï ê ú ï ï
ê ¶u ¶w ú ti v1i ï ¶N i ï êê úú
ï ï
1 1
¶v 8
ti v2i ï ¶N i ï ê ú 8
ê ú = å í i [ui vi wi ]- å íV ´ ê b2 ú + å íV ´ a
ê ¶h ¶h ¶h ú i=1 ïï ¶h ïï 2 ï ï ¶h ï ï ï
i =1 2 ï ¶h ï ê
ï ê 2ú
ê ú i =1 ê ú ï ë a3 úû i
ê ¶w úú
ï
ïïï 0 ïïï
ï ï
ï N ïï ï ë b3 û i ï
ï Ni ï
ê ¶u ¶v
ïï ïï
ï
ï
ï
i ï
ï
ï
ï
ï
ï
ï
ï
êë ¶V ¶V ¶V úû î  ï
î ï
 ï
î ï
(6.6.14)

6.6.5.3 Stress strain relation

The stress strain relationship is given by
{s} = [ D]{e} (6.6.15)
Using eq. (6.6.11) in eq. (6.6.15) one can find the following relation.
{s } = [ D ][ B ]{d } (6.6.16)
Where, the stress strain relationship matrix is represented by
é1 m 0 0 0 ù
ê ú
êm 1 0 0 0 ú
ê ú
ê 1- m ú
ê0 0 0 0 ú
E ê 2 ú (6.6.17)
[ D] = ê ú
1- m 2 ê a (1- m) ú
ê0 0 0 0 ú
ê 2 ú
ê ú
ê a (1- m)ú
ê0 0 0 0 ú
êë 2 úû
The value of shear correction factor a is considered generally as 5/6. The above constitutive matrix can
be split into two parts ([Db] and [Ds] )for adoption of different numerical integration schemes for
bending and shear contributions to the stiffness matrix.
é[ Db ]  [0] ùú
ê
[ D ] = êê    úú (6.6.18)
ê [ 0]  [ Ds ]úúû
êë
Thus,
é ù
ê ú
ê1 m 0 ú
E ê ú
[ Db ] = 2 ê
m 1 0 ú (6.6.19)
1- m ê ú
ê 1- m ú
ê0 0 ú
êë 2 úû
and
E a é 1 0ù
[ Ds ] = ê ú (6.6.20)
2 (1 + m) êë 0 1úû
 44 
 

It may be important to note that the constitutive relation expressed in eq. (6.6.19) is same as for the case
of plane stress formulation. Also, eq. (6.6.20) with a multiplication of thickness h is similar to the terms
corresponds to shear force in case of plate bending problem.

6.6.5.4 Element stiffness matrix

Finally, the stiffness matrix for the shell element can be computed from the expression
T
[ k ] = òòò [ B ] [ D ][ B ] d W (6.6.21)
However, it is convenient to divide the elemental stiffness matrix into two parts: (i) bending and
membrane effect and (ii) transverse shear effects. This will facilitate the use of appropriate order of
numerical integration of each part. Thus,
[ k ] = [ k ]b +[ k ]s (6.6.22)
Where, contribution due to bending and membrane effects to stiffness is denoted as [k]b and transverse
shear contribution to stiffness is denoted as [k]s and expressed in the following form.
T T
[ k ]b = òòò [ B ]b [ D ]b [ B ]b d W and [ k ]s = òòò [ B ]s [ D ]s [ B ]s d W (6.6.23)
Numerical procedure will be used to evaluate the stiffness matrix. A 2 ×2 Gauss Quadrature can be used
to evaluate the integral of [k]b and one point Gauss Quadrature may be used to integrate [k]s to avoid
shear locking effect.
 Module 7: Applications of FEM
Lecture 1: Finite Elements for Elastic Stability

1 

7.1.1 Introduction
There are two types of structural failure associated: namely (i) material failure and (ii) geometry or
configuration failure. In case of material failure, the stresses exceed the permissible values which may
lead to formation of cracks In configuration failure, the structure is unable to maintain its designed
configuration under the external disturbance even though the stresses are in permissible range. The
stability loss due to tensile forces falls in broad category of material instability. The loss of stability
under compressive load is termed as structural or geometric instability which is commonly known as
buckling. Thus, buckling is considered by a sudden failure of a structural element subjected to large
compressive stress, where the actual compressive stress at the point of failure is less than the ultimate
allowable compressive stresses. Buckling is a wide term that describes a range of mechanical
behaviours. Generally, it refers to an incident where a structural element in compression deviates from a
behaviour of elastic shortening within the original geometry and undergoes large deformations involving
a change in member shape for a small increase in force. Various types of buckling may occur such as
flexural buckling, torsional buckling, torsional flexural buckling etc. Here, the use of finite element
technique for elastic stability analysis of various structural members will be briefly discussed.

7.1.2 Buckling of Truss Members

For a truss member, the axial strain can be expressed in terms of its displacements
¶u 1 éêæç ¶u ÷ö çæ ¶v ÷ö ùú
2 2

ex = + êç ÷÷ + ç ÷÷ ú (7.1.1)
¶x 2 êëèç ¶x ø èç ¶x ø úû
Here, u and v are the displacements in local XandY directions respectively. Here, the strain-displacement
relation is nonlinear. The total potential energy in the member with uniform cross sectional area and
subjected to external forces can be written as
L
EA
Π= ò εx 2 dx - ( Pv
1 1 + P2 v2 ) (7.1.2)
2 0
Where, P1 and P2 are the external forces in nodes 1 and 2, v1 and v2 are the vertical displacements in
nodes 1 and 2, E and A are the modulous of elasticity and cross sectional area respectively (Fig.
7.1.1).The length of the member is considered to be L.
 2 
 

Fig.7.1.1 Truss element in local coordinate system

The above expression can be rewritten in terms of displacement variables as

2
EA æç ¶u 1 éêçæ ¶u ÷ö çæ ¶v ÷ö ùú ÷ö÷
L 2 2

+ ç ÷ + ç ÷ ÷ dx - ( P1v1 + P2 v2 )
2 ò0 çèç ¶x 2 êëêçè ¶x ÷ø çè ¶x ø÷ úûú ø÷÷
Π= ç

L ìæ ü
EA ï ï
3ö 2
2 2 é 2 2ù
ïíççæç ¶u ÷÷ö + çæ ¶u ÷÷öæç ¶v ÷÷ö + çæ ¶u ÷÷ö ÷÷+ 1 êçæ ¶u ö÷÷ + æç ¶v ö÷÷ ú
ï ïï
 dx - ( Pv
1 1 + P2 v2 )
2 ò0 ïïçèççè ¶x ÷ø çè ¶x ÷øèç ¶x ÷ø çè ¶x ÷ø ÷ø÷ 4 ëêêçè ¶x ø÷ èç ¶x ø÷ úúû
= ç ç ç ç ÷ ç ç ïï
ï
î ï (7.1.3)
Neglecting higher order terms and considering EA ¶u = P the above expression can simplify to the
¶x
following form.
L 2 L 2
EA æç ¶u ö÷ P æ ¶v ö
Π= ò çèç ø÷÷ dx+ ò èççç ø÷÷÷ dx - ( P1v1 + P2 v2 ) (7.1.4)
2 0 ¶x 2 0 ¶x
Considering nodal displacements at nodes 1 and 2 as u1, v1 and u2, v2, the displacement at any point
inside the element can be represented in terms of its interpolation functions and nodal displacements.
The shape function for a two node truss element is

N   1  x x
(7.1.5)
 L L 
Equation expressed in eq. (7.1.5) is expressed in terms of the nodal displacement vectors {ui} and {vi}
with the help of interpolation function.
L L
1 T T P T T
Π = ò {ui } [ N '] EA[ N ']{ui } dx+ ò {vi } [ N '] [ N ']{vi } dx - ( Pv
1 1 + P2 v2 ) (7.1.6)
2 0 2 0
Where,
¶u ¶ ¶v ¶
= ([ N ]{ui }) = [ N' ]{ui } and = ([ N ]{vi }) = [ N' ]{vi } (7.1.7)
¶x ¶x ¶x ¶x
Making potential energy (eq. 7.1.6) stationary one can find the equilibrium equation as
 3 
 
L L
T T
{F } = ò [ N '] EA[ N '] dx {ui }+ P ò [ N '] [ N '] dx {vi } (7.1.8)
0 0

Or,
{F } = [ k A ]{ui } + P [ kG ]{vi } (7.1.9)
Where, [kA] is the axial stiffness of the member and [kG] is the geometric stiffness of the member in its
local coordinate system and can be derived as follows.
ì 1ï
ï ü
ï
ï- ï
ï ïï é 1 1 ù AE é 1 -1ù
L

[ k A ] = ò [ N '] EA[ N '] dx = AE ò0 í L  ê-

T L
ú dx = ê ú (7.1.10)
ï
ï 1 ïê
ï ë L L úû L êë-1 1 úû
0
ï
ï ï
ïLï
î ï


ì
ï 1üï
ï
ï- ï
ï L ïï é 1 1 é 1 -1ù
L
T L 1ù
[ kG ] = ò [ N '] [ N '] dx = ò0 í  ê- ú dx = ê ú (7.1.11)
ï
ï 1 ïê
ï ë L L úû L êë-1 1 úû
0
ï
ï ï
ï L ï
î ï
However, in a generalised form to accommodate both the direction, these stiffness matrices can be
written as
é 1 0 -1 0ù é0 0 0 0 ù
ê ú ê ú
AE êê 0 0 0 0úú 1 êê0 1 0 -1úú
[kA ]= and [ kG ] = (7.1.12)
L êê-1 0 1 0úú L êê0 0 0 0 úú
ê 0 0 0 0ú ê0 -1 0 1 ú
ë û ë û

The generalised stiffness matrix of a plane truss member in global coordinate system can be derived
using the transformation matrix as derived module 4, lecture 1. The transformation matrix can be
recalled and written here as follows.
 cos  sin 0 0 
  sin cos  0 0 
T   
 (7.1.13)
0 0 cos  sin 
 
 0 0  sin cos  
 
Thus, the stiffness matrices with respect to global coordinate system will become
T T
[ K A ] = [T ] [ k A ][T ] and [ KG ] = [T ] [ kG ][T ]   (7.1.14)
Here, [KA] and [KG] are the axial stiffness and geometric stiffness of the member in global coordinate
system. The force-displacement relationship in global coordinate system can be written from eq. (7.1.9)
as

{F } = [ K A ]{d } + P [ KG ]{d } = éë[ K A ] + P [ KG ]ùû {d } (7.1.15)

 4 
 

Where, {d} is the displacement vector in global coordinates. If the external force is absent in eq.
(7.1.15), the value of P will be considered to be undetermined as.
 K A d   P  K G d   0 (7.1.16)
The above equation can be solved as an eigenvalue problem to calculate buckling load P.

7.1.3 Buckling of Beam-Column Members

Let consider a pin ended column under the action of compressive force P. The elastic and geometric
stiffness matrices can be developed from 1st principles for a beam-column element which can be used in
the linear elastic stability analysis of frameworks.Considering small deflection approximation to the
curvature, the total potential energy is computed from the following.

    (7.1.17)

Here, w is the transverse displacement and I is the moment of inertia of the member.

Fig. 7.1.2 Beam-column member

Considering nodal displacements at nodes 1 and 2 as w1, θ1 and w2, θ2, the displacement at any point
inside the element can be represented in terms of its interpolation functions and nodal displacements.

(7.1.18)

Thus, the above energy equation can be rewritten as

1 Lé ù éê N " ùú EI éê N " ùú {d } + {d } P éê N ' ùú T éê N ' ùú {d }ù dx
T T
{ }
2 ò0 ë
Π= d û ë û ë û ë û ë û úû (7.1.19)

Where,
d 2w d 2 dw d
2
= 2 ([ N ]{d }) = [ N" ]{d } and = ([ N ]{d }) = [ N' ]{d } (7.1.20)
dx dx dx dx
 5 
 

Applying the variational principle one can express

F  

 d  0
T

   N "  EI  N "  P  N '  N ' dx d 
l T
 (7.1.21)

Thus, the stiffness matrix will be obtained as follows which have two terms.
 k   0
l
  N 
" T

EI  N " dx P 
0
l
 N '  N ' dx   k
T
F   P  kG  (7.1.22)

The first term resembles ordinary stiffness matrix for the bending of a beam. So this matrix is called
flexural stiffness matrix. The second matrix is known as geometric stiffness matrix as it only depends on
the geometrical parameters. Thus, the flexural stiffness matrix [kF] and geometric stiffness matrix [kG]
can be derived from the following expressions.
 k F   0
l
  N 
" T

EI  N " dx and  kG   0  N '  N ' dx
l T
(7.1.23)

The above matrices can be derived from the assumed interpolation function. For example, the
interpolation functions for a two node beam element are expressed by the following equation.
  1 3 2 , 2 , 3 2 , (7.1.24)
Now, the first and second order derivative of the above function will become
′
  6 6 , 1 4 3 , 6 6 , 2 3 (7.1.25)
"
  12 , 6 , 12 , 6 (7.1.26)
Using the above expressions in eq. (7.1.23), flexural and geometric stiffness matrices can be derived and
obtained as follows.
12 6 12 6
6 4 6 2 (7.1.27)
12 6 12 6
6 2 6 4

36 3 36 3
3 4 3 (7.1.28)
36 3 36 3
3 3 4

Due to external forces {F}, one can find the displacement vectors from the following equation.
 F    k   d    k F   d   P  k G  d  (7.1.29)
The above is the beam-column equation in finite element form. In the absence of transverse load, the
member will become column and P will be considered to be undetermined.
 k F  d   P  k G  d   0 (7.1.30)
Denoting P as -λ we can reach to the familiar eigenvalue problem as given below.
 k F d     kG d  (7.1.31)
 6 
 

Solving above equation, values of λ and associated nodal displacement vectors can be obtained.
Mathematically this can be expressed as
  k     k   d   0
F G (7.1.32)
Thus,
 k F     kG   0 (7.1.33)
For the matrix of size n, one can find n+1 degree polynomial in λ. The smallest root of the above
equation will become the first approximate buckling load. From this value λ, one can find a set of ratios
for the nodal displacement components. From this, the first buckling mode shape can be calculated.
Higher mode approximations can also be found in a similar process. The procedure to determine the
critical load by the above method is illustrated in the following example.

Example 7.1.1
Consider a column with one end clamped and other end free as shown in Fig. 7.1.2.

Fig. 7.1.3 Column with one end fixed and other end free

Now, the finite element equation of the column considering single element will be

12 6 12 6 36 3 36 3
6 4 6 2 1 3 4 3
λ 0
12 6 12 6 30 36 3 36 3
6 2 6 4 3 3 4

The boundary conditions for this member are given by

 7 
 

d1
dw
At x = 0, = 0 and q1 = =0
dx
Thus, according to the above boundary conditions, the first and second rows as well as columns of the
equation are deleted and rewritten in the following form.
12 6 36 3
0
6 4 30 3 4
Or
4 2 12 6 6
0
15 5 10
Solving the above expression, the critical value of λcr and thus the critical value of force Pcr will become
as

2.486
It is important to note that the exact value for such clamped-free column is
p2 EI p2 EI EI
Pcr = 2 = 2
= 2.467 2
Le (2L) L
The finite element result has slight deviation from the exact result. This difference can be minimized by
the increase of number of elements in the column as we know, more we subdivide the continuum, better
we can obtain the result close to the exact one.

7.1.4 Buckling of Plate Bending Elements

The elastic stability analysis of rectangular plates is discussed in this section. The total potential energy
for plate are expressed as

w 2 1

2 (7.1.34)
Here, Fx,Fy,and Fxy, are the in-plane edge load and compressive load is considered as positive.For, finite
element formulation the deflection in above expression needs to convert in terms of nodal displacements
in the element. Following the derivations in beam-column member (eqs. 7.1.18 to 7.1.22), the flexural
and geometric stiffness for the plate element can be derived. Thus, the above expression can be derived
to the following form using interpolation functions.
(7.1.35)
Here is the flexural stiffness matrix. The other stiffness matrices are analogous to the geometric
stiffness matrices of the plate and can be expressed as
∬ ′ ′
∬ ′ ′ (7.1.36)
 8 
 

∬ ′ ′
Where ′ and ′ indicate partial derivative of with respect to xand y respectively. Thus, the
equation of buckling becomes
0 (7.1.37)
If the in-plane loads have a constant ratio to each other at all time during their buildup, the above
equation can be expressed as follows
∗
(7.1.38)
The term ∗ is called the load factor, and , , and are constants relating the in-plane loads in the plate
member. Solving the above expression, the buckling mode shapes are possible to determine.
 Module 7: Applications of FEM
Lecture 3: Dynamic Analysis

16 

The finite element method is a powerful device for analyzing the dynamic response of structures as in
case of static analyses. The responses such as displacements, velocities, strains, stresses become time
dependent in dynamic analysis.The dynamic equation of motion of a structure can be derived from
Hamilton’s principle using Lagrange equation.

7.3.1 Hamilton’s Principle

Hamilton’s principle is a simple but powerful tool to derive discritized dynamic system equations. It can
be stated as “Of all the possibletime histories of displacement the most accurate solution makes the
Lagrangian functional a minimum provided the following conditions are satisfied.”
 the essential or the kinematic boundary conditions,
 the compatibility equations, and
 the conditions at initial (t1) and final time (t2).
First condition of the above ensures that the displacement constraints are fulfilled. Second condition
makes sure that the displacements are continuous in the problem domain and the third condition
necessitates the displacement history to satisfy the constraints at the initial and final times. Hamilton’s
principle allows one to assume any set of displacements, as long as it satisfies the above three
conditions. It is basically a variational formulation method. Just like in the other variational
formulations,a functional is used here. Mathematically, Hamilton’s principle states:
t2

  Ldt  0 (7.3.1)
t1

The Langrangian functional, L, is defined as follows using a set of permissible time histories of
displacement.
L = T-P (7.3.2)
Where T is the kinetic energy and P is the potential energy which is basically elastic strain energy. The
kinetic energy, T is defined in the integral form as
1
T = ò x T rx dW (7.3.3)
2 W
Here Ω represents the whole volume of the domain,ρ is the mass density of structure and x is the set of
timehistories of velocities.The strain energy in the entire domain of elastic structure can be expressed as
1 1
P = ò eT s dW = ò eT D e dW (7.3.4)
2 W 2 W
Here ε are the strains obtained from the set of time histories of displacements. Here, L can be expressed
in terms of the generalized variables x1 , x 2  x n and x 1 , x 2  x n . Here, xi is the displacement and
the velocity is expressed by
d
x i = ( x i ) (7.3.5)
dt
 17 
 

Considering xias generalised displacement, theLagrange’sequation of motion are expressed as

d æç ¶L ö÷ ¶L
ç ÷- =0 where i = 1 to n (7.3.6)
dt çè ¶x ÷÷ø ¶x
i i

7.3.2 Finite Element Form in MDOF System

Consider a two-degree of freedom system as shown in the figure below.

Fig. 7.3.1 Two Degree Freedom System

The kinetic and potential energy can be expressed by following form:

1 1
T = m1x 12 + m2 x 22
2 2 (7.3.7)
1 1 2
P = k1x12 + k 2 ( x 2 - x1 )
2 2
Now, substituting the values in Lagrangean L, we can obtain
1 1 1 1 1
L = T -P = m1x 12 + m2 x 22 - k1x12 - k 2 x 22 + k 2 x1x 2 - k 2 x12 (7.3.8)
2 2 2 2 2
Thus,
d æçç ¶L ö÷÷ ¶L
÷- = m1x1 + k1x1 - k 2 x 2 + k 2 x1
dt ççè ¶x 1 ÷÷ø ¶x1
= m1x1 + k1x1 - k 2 ( x 2 - x1 ) = 0
(7.3.9)
d æçç ¶L ö÷÷ ¶L
÷- = m 2 x 2 + k 2 x 2 - k 2 x1
dt ççè ¶x 2 ÷÷ø ¶x 2
= m 2 x 2 + k 2 ( x 2 - x1 ) = 0
Above equations can be written in the matrix form as shown below:
 18 
 

é  ïü é k + k 2 ) -k 2 úù ìïï x1 üïï ïïì0ïïü

ê m1 0 ú ïí 1 ï + êê( 1
ù ïì x
ê 0 m ú ïx  ú íï ï = í  (7.3.10)
êë 2 ûú ï
î  ë2
ï ê -k
ï 2 k ú ïx 2 ï
2 ûî  î ïï0ï ï

Thus, for a multi degree freedom system, the above expression can be written as
[ M ]{x} +[ K ]{x} = {0} (7.3.11)
For a steady state condition, one can consider x= a sin wt . Thus, x=
 -w2 a sin wt = -w2 x . As a result,
the above equation can be re-written as
[ K ]{x } -w 2 [ M ]{x } = {0}
é[ K ]- w 2 [ M ]ù {x } = {0}
êë úû
éw2 ù = [ M ]-1 [ K ] (7.3.12)
ëê ûú
Thus, for a single degree of freedom system, the fundamental frequency can be found as
K
w= (7.3.13)
M

7.3.3 Solid Body with Distributed Mass

The velocity vector x can be expressed in terms of the elemental nodal velocities with the help of
interpolation functions as x = Nx . Here, x is the nodal velocity, and N is the interpolation function. Thus
the expression of kinetic energy can be re-written as
1 1
T = ò x T rx dW = x T ò N T rN dW x (7.3.14)
2 W 2 W
In the above equations, themass matrix can be obtained as
M = ò N T rN dW (7.3.15)
W

Volume integralhas to be taken over the volume of the element to find the mass matrix of that element.

7.3.3.1 Mass matrix for truss element

In case of truss member, the degrees of freedom at each node is one because of its only axial
deformation. Thus for a two node truss element, the interpolation function will become
[ N ] = [1- x x ]
Here, x = x / L and thus, dx = Ldx . Hence, the mass matrix can be expressed as
 19 
 
 l
[ M ] = r ò N T NdW = rA ò0 N T Ndx = rAL ò0 N T Ndx
W

l é(1-x)ù lé x 1-x)ùú
ú é(1-x) xùdx = rAL ê (1-x) (
2

= rAL ò ê ò0 êx (1-x) x2 údx (7.3.16)

0 ê x úë û
ë û ëê ûú
1
é -(1-x)3 x 2 x3 ù é1 1ù
ê - úú ê ú
ê 3 2 3 ê3 6 ú rAL éê 2 1ùú
= rAL ê ú = rAL ê ú=
ê x 2 x3 x 3 ú ê1 1ú 6 êë 1 2úû
ê - ú ê ú
ê 2 3 3 úû 0 ëê 6 3 ûú
ë
Thus, the consistent mass matrix will be
rAL é 2 1ù
[M] = ê ú (7.3.17)
6 êë 1 2úû
Whereas the lumped mass matrix can be expressed as follows.
rAL é1 0ù
[M] = ê ú (7.3.18)
2 êë 0 1úû
Now, to obtain a generalised mass matrix of a two dimensional truss member, consider Fig. 7.3.2.

Fig. 7.3.2 Two Degree Freedom System

The displacement vector and the shape function for such case are expressed as

ì
ï u1 ü
ï
ï
ï ï
ï v1 ï
ï éN 0 N2 0 ùú
{d} = ïí ï and [ N ] = êê 1 (7.3.19)
ï
ï u2ïï ë0 N1 0 N 2 úû
ï
ï ï
ï
ïv 2 ï
ï
î ï


Thus, the mass matrix can be derived as given below.

 20 
 

é1-x 0 ù
ê ú
lê 0 1- xú é1- x 0 x 0ù
[ M ] = r ò N NdW = rAL ò0 êê
T úê
ú ê 0 1-x 0 xú
údx (7.3.20)
W ê x 0 úë û
ê 0 x ú
ë û
After performing integration, the mass matrix will be obtained as
é 2 0 1 0ù
ê ú
pAL êê 0 2 0 1úú
[M] = (7.3.21)
6 êê 1 0 2 0úú
ê 0 1 0 2ú
ë û

The lumped mass matrix of two dimensional truss member will become

é1 0 0 0ù
ê ú
pAL êê 0 1 0 0ú
ú
[M ]= (7.3.22)
2 êê 0 0 1 0úú
ê0 0 0 1úû
ë

7.3.3.2 Mass matrix for beam element

The beam element has two degrees of freedom at each node, one transverse displacement (v) and the
other one is an angle of rotation ( ). So, a beam with two nodes has a total of four degrees of freedom.
From earlier notes in module 4, lecture 3, the interpolation function of a beam is obtained as
3 2 2 3 2 2 x3
N1 = 1 - x + x , N 2 = x - x + 2,
L2 L3 L L (7.3.23)
2 3 2 3
3x 2x x x
N 3 = 2 - 3 and N4 = - + 2
L L L L
The consistent mass matrix can be following earlier process.
l T
[ M ] = r ò N T NdW = rA ò0 [ N ] [ N ]dx
W

Substitution for the interpolation functions and performing the required integrations, one can arrive the
following mass matrix for the beam element.
156 22 54 13
22 4 13 3
[M] = (7.3.24) 
54 13 156 22
13 3 22 4

Similarly, the lumped mass matrix of the beam element is

 21 
 

1 0 0 0
0 1 0 0
[M] = (7.3.25) 
0 0 1 0
0 0 0 1

7.3.3.3 Mass matrix for plane frame element

The plane frame element has three degrees of freedom at each node: (i) axial displacement (u), (ii)
transverse displacement (v) and (iii) angle of rotation ( ). So it has a total of six degrees of freedom for
a two node plane frame element and its shape functions are given by
x 3 2 2 3 2 x3
N1 = 1 - , N 2 = 1-
2
x + 3x , N3 = x - x 2 + 2 ,
L L L L L (7.3.26)
2 3 2 3
x 3x 2x x x
N4 = , N 5 = 2 - 3 and N6 = - + 2
L L L L L
Similar to earlier case, one can derive the consistent mass matrix of a two node plane frame member
which will become as given below.

1/3 0 0 1/6 0 0
13/35 11 /210 0 9/70 13 /420
/105 0 13 /420 /140
[M] = (7.3.27)
1/3 0 0
. 13/35 11 /210
/105

The lumped mass matrix of such two dimensional frame element will be
1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
[M] = (7.3.28)
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1

7.3.3.4Mass matrix for space frame element

In case of space frame, each node has six degrees of freedom: (i) three displacementsand (ii) three
rotations. Therefore, for a two node space frame element there will be twelve degrees of freedom and
the consistent mass matrix of the element in the local xyz system can be derived as
 22 
 

é1 ù
ê ú
ê3 ú
ê ú
ê 13 ú
ê0 ú
ê 35 ú
ê ú
ê0 13 ú
ê 0 ú
ê 35 ú
ê J ú
ê0 0 0 ú
ê 3A ú
ê ú
ê 11 L2 ú
ê0 0 - L 0 ú
ê 210 105 ú
ê ú
ê 11 L2 ú
ê ú
ê0 210
L 0 0 0
105 ú
ê
[ M ] = rAL ê ú
1 1 ú
ê 0 0 0 0 0 ú
ê6 3 ú
ê ú
ê 9 13 13 ú
ê0 0 0 0 L 0 ú
ê 70 420 35 ú
ê ú
ê 9 13 13 ú
ê0 0 0 - L 0 0 0 ú
ê 70 420 35 ú
ê J J ú
ê0 0 0 0 0 0 0 0 ú
ê 6A 3A ú
ê ú
ê 13 L2 11 L2 ú
ê - ú
ê0 0
420
L 0
140
0 0 0
210
L 0
105 ú
ê ú
ê L2 2 ú
ê 0 - 13 L 0 0 0 - 0 -
11
L 0 0 0
L ú
ê 420 140 210 105 úû
ë

(7.3.29)

In a similar process, one can derive the mass matrix of other types of structures such as two dimensional
plane stress/strain element, three dimensional solid element, plate bending element, shell element etc.

7.3.4 Time History Analysis

The dynamic equation of motion of a multi-degree freedom system can be written as
[ M ]{x} + [C]{x } + [ K ]{x} = {F( t )} (7.3.30)
Where, [C] is the damping matrix and {F(t)} is the time dependent force vector. The element matrices of
the above equation can be expressed as follows.
 23 
 
T
[ M ] = ò [ N ] r [ N ] dW
W
T
[ C ] = ò [ N ] m [ N ] dW (7.3.31)
W
T
[ K ] = ò [ B] [ D] BdW
W

In a linear dynamic system, these values remain constant throughout the time history analysis. Damping
is a mechanism which dissipates energy, causing the amplitude of free vibration to decay with time. The
various types of damping which influences structural dynamical behaviour are viscous damping,
hysteresis damping, radiation damping etc. Viscous damping exerts force proportional to velocity, as
exhibited by the term. A formulation for this kind of problem was developed by Rayleigh. In this case,
the energy dissipated is proportional to frequency and to the square of amplitude. Viscous damping is
provided by surrounding gas or liquid or by the viscous damper attached to the structure. Radiation
damping refers to energy dissipation to a practically unbounded medium, such as soil that supports
structure. Among the various types of physical damping, viscous damping is easy represent
computationally in dynamic equations. Fortunately, damping in structural problems is usually small
enough regardless of its actual source. Its effect on structural response is modelled well by regarding it
as viscous. The Rayleigh damping defines the global damping matrix C  as a linear combination of the
global mass and stiffness matrices.
C    M    K  (7.3.32)
Here,   and   are the stiffness and mass proportional damping constants respectively. The
relationship between the fraction of critical damping ratio,  ,  and  at frequency ωis given by
1 
      (7.3.33)
2 
Damping constants   and   are determined by choosing the fractions of critical damping ( 1 and  2 )
at two different frequencies (ω1 and ω2) and solving simultaneously as follows
2 w1w 2 (x¢2 w 2 -x1¢w1 )
a¢ =
(w22 -w12 )
(7.3.34)
2 (x1¢w 2 -x ¢2 w1 )
b¢ =
(w22 -w12 )
There are several numerical integration schemesare available to obtain the time history response of the
structural system. Amongst, the implicit type of methods such as the Newmark’s method isone of the
most popular one.