From Structural Analysis to Finite Element Method

Dhiman Basu
IIT Gandhinagar
-------------------------------------------------------------------------------------------------------------------Acknowledgement
Following text books were consulted while preparing this lecture notes:
Zienkiewicz, O.C. and Taylor, R.L. (2000). The Finite Element Method, Vol. 1: The Basis, Fifth
edition, Butterworth-Heinemann.
Yang, T.Y. (1986). Finite Element Structural Analysis, Prentice-Hall Inc.
Jain, A.K. (2009). Advanced Structural Analysis, Nem Chand & Bros.
--------------------------------------------------------------------------------------------------------------------1.0 Introduction
Analysis of a civil engineering structure, for example, a rigid jointed frame is often performed using centerline element model, wherein cross-sectional properties are lumped onto the center-line of the element.
Analysis procedure is usually displacement based, and direct stiffness method is generally adopted. Note
that center-line model is restricted to a very simple configuration and cannot be applied in most cases
including varying cross-sections, with opening/discontinuity and with two or three dimensional effects. In
such cases, finite element method (FEM) is widely used to capture the reasonable behavior. If FEM is
applied to those simple cases wherein center-line model provides reasonable solution, results will be
identical. Therefore, some similarity exists in principle between the center-line model (or conventional
structural analysis) and FEM. Objective of this lecture is to explore that similarity and illustrate the
transition path from conventional structural analysis to FEM.
In what follows next, first equilibrium of a beam element is considered and effect of orthogonal
transformation is illustrated. Second, conventional structural analysis using direct stiffness method is
discussed followed by a numerical illustration. Third, conventional structural analysis is revisited so as to
describe the passage to FEM. Fourth, concept of FEM is briefly discussed to address the similitude with
conventional structural analysis followed by a numerical example. Finally, a problem statement is made
that describe the generality of FEM approach.
2.0 Element Stiffness Matrix
Consider a two-noded beam element. Neglecting the axial deformation for now, joint displacements (in
generalized sense, including rotations as well) and joint forces (including moments also in a generalized
sense) are shown in their positive sense in Figure 1.
Using the principle of elementary structural mechanics, equilibrium of the beam element can be expressed
as

qe

Y1
M1
Y2
M2

K e ae
12
L2
6
EI L
L 12
L2
6
L

6
L
4
6
L
2

12
L2
6
L
12
L2
6
L

6
L
2

v1

(1)

6 v2
L 2
4

Axial deformation, if considered (Figure 2), under the assumption of small deformation remains uncoupled
with respect to others and the equilibrium of the element takes the form as follows:
qe

K e ae
EA
L

X1
Y1
M1
X2
Y2
M2

0
0
EA
L
0
0

12 EI
L3
6 EI
L2

6 EI
L2
4 EI
L

12 EI
L3
6 EI
L2

6 EI
L2
2 EI
L

EA
L
0
0
EA
L
0
0

12 EI
L3
6 EI
L2

6 EI
L2
2 EI
L

u2

u1
v1

(2)

v2
6 EI
2
L2
4 EI
L

12 EI
L3
6 EI
L2

Matrix K e is known as the stiffness matrix. Any element, say ij th , of the stiffness matrix indicates the
force developed along the direction of ith degree of freedom due to a unit displacement along the direction
of j th degree of freedom while all others degrees of freedom are held restrained.

Figure 1: Straight beam element without axial degrees of freedom

Figure 2: Straight beam element with axial degrees of freedom

3.0 Orthogonal Transformation
In previous section, degrees of freedom are chosen along and normal to the axis of the element. However,
constituting elements in a structure may have their axes oriented along any arbitrary direction (Figure 3).
In such cases, two different coordinate systems are usually considered namely, local and global. In local
coordinate system, one of the axes is chosen along the axis of the element and hence, every element has its
own local system. On the other hand, global coordinate system is unique to the overall structure and analysis
of deformation is carried out in this coordinate system. Note that, local coordinate system of any element
is related to the global coordinate system of the structure through a rotation only. Structural analysis often
requires transformation of vector from one coordinate system to another.

Figure 3: Frame with arbitrary orientation of element axis

Consider a local coordinate system which is rotated through an angle with respect to the global coordinate
system as shown in Figure 4. The coordinate of the point P is x, y and x ' , y ' in local and global systems,
respectively. By simple geometry, it can be shown that
x

cos

Since

sin

sin

0 x'

cos

0 y'

0
1

'

(3)

'

, this is also noted as orthogonal transformation.

T

Figure 4: Orthogonal transformation

4.0 Element Equilibrium in Global Coordinate System
Consider an element with axis oriented at an angle with respect to the global coordinate system along
the counter clockwise direction as shown in Figure 5; over bar denotes the local coordinate system.
Displacements and forces acting at the joints in both coordinate systems are related as follows:
cos

sin

cos

X2

cos

v2

Y2

sin

M2

u1

X1

v1

Y1

sin

M1

u2

0 u1
0 v1

X1
M1

a eL

T e a eG

sin

0 1
0 u2

X2

q eL

T e q eG

cos

0 v2

Y2

1 2

M2

Y1

(4)

Employing Eq (4) into Eq (2), equilibrium relation in the global coordinate system can be derived as
follows:

q eL

K eL a eL

q eG

K eG a eG and K eG

T e q eG

K eL T e a eG
Te

q eG

Te

K eL T e a eG
(5)

K eL T e

Note that the size of the matrices remains same in this transformation.

5.0 Direct Stiffness Matrix Method of Analysis

Step-1: Element Equilibrium in Local Coordinate
For any typical element (say ith), develop the equilibrium equation in local coordinate:

q eL

i
6 1

K eL

Consider q eL

6 6

i
6 1

a eL

(6)

6 1

as the negative of the fixed end forces due to span loading.

Figure 5: Local to global coordinate transformation

Step-2: Element Equilibrium in Global Coordinate
Transform the equilibrium equation from local to global coordinate.
q eG
K

K eG

6 1

eG i

6 6

6 6

ei T
6 6

a eG
K

i
6 1

eL i
6 6

T ei

6 6

, q eG

i
6 1

T ei

T
6 6

q eL

i
6 1

, a eG

i
6 1

T ei

T
6 6

a eL

(7)

6 1

Step-3: Element Equilibrium in Expanded Global Coordinate

Note that the degrees of freedom considered in Eq (7) are oriented along the global coordinate system and
hence, a connectivity matrix must be formed that relates each of these to the global degrees of freedom.
With the help of connectivity matrix, Eq (7) can be expanded to the global size of the problem, which is 3N
and N is the number of nodes including that are restraint.
Exp

q eG

i
3N 1

Exp

K eG

i
3N 3N

Exp

a eG

(8)

3N 1

Step-4: Assemble Element Equilibrium in Expanded Global Coordinate

Assemble Eq (8) for elements. Let the number of elements be M. Then
5

M
Exp

q eG

Exp

3N 1

i 1

K eG

i 1

Exp

3N 3N

a eG

(9)

3N 1

After assembling Eq (9) and taking into account the externally applied joint forces

q*

3N 1

in global

coordinate, it may be written as

qG

q*

3N 1

Here, K G

KG

3N 1

3N 3N

qG

3N 3N

3N 1

aG

aG

(10)

3N 1

3N 1

represent the stiffness matrix, force vector associated with span

loading and deformation vector, respectively, in global coordinate but without accounting for the effect of
restraint.
Step-5: Effect of Restraints
Identify the restraint degrees of freedom and remove the associated rows and columns from the stiffness
matrix. Also remove the associated elements from the force and deformation vectors. Let the problem be
reduced to a size S and Eq (10) is more formally written as
q

S 1

S S

(11)

S 1

Step-6: Solution for Displacement

Eq (11) can be solved for the displacement vector using any standard procedure.
a

S 1

S S

(12)

S 1

Next, inserting zeros into the computed displacement vector for the restraint degrees of freedom, the
displacement vector a G

3N 1

is formed.

Step-7: Solution for Element Response

Using the same connectivity matrix as used in Step-3, nodal displacement vector of any element in global
coordinate, i.e., a eG

a eL

i
6 1

T ei

6 6

is extracted. Associated vector in local coordinate can be calculated as

6 1

a eG

(13)

6 1

Member end force for the element is then calculated as

F eL

i
6 1

K eL

i
6 6

a eL

i
6 1

q eL

(14)

6 1

Note that member end force vector includes the effect of span loading through the fixed end forces. In
order to calculate the member forces at any point within the span, superimpose the span loading on the
end forces calculated through Eq (14).
6

Step-8: Calculation of Reaction Forces

Instead of removing the rows and column associated with the restraint degrees of freedom, entire degrees
of freedom and equilibrium equations in global coordinate may be rearranged as
qr

K rr

K rs ar

qs

K sr

K ss

(15)

as

Here the subscript r and s denote the restraint and unrestraint degrees of freedom. Note, K ss in Eq (15) is
same as K
qr

S S

is Eq (11). Reactions can now be calculated as

(16)

K rs as

where as is computed in Eq (12). If displacements along the restraint degrees of freedom are specified nonzero,
as

K ss

qr

K rr ar

qs

K sr ar

(17)

K rs as

5.1 Brief Numerical Example for Direct Stiffness Method

Plane Frame with Inclined LegsProblem Description

Figure 6: Numerical example

Geometry and Material Properties

Transformation Matrices

Consider the frame as shown in Figure 6, EA=8000 kN/m2 and EI= 20000 kNm2.

Degrees of Freedom

Equivalent Joint Load (in Negative Sense)

Element Stiffness Matrix in Local Coordinate

Element-1

Element-2

Element-3

Element Stiffness in Global Coordinate

i K eg iT 66 i K eg iT 66
66
66
T

Element-1

10

Element-2

Element-3

Expansion and Assembly of Global Stiffness Matrix

11

Transformation of Force Vector from Local to Global Coordinate

Assembling of Force Vector in Global Coordinate

12

Solution for Unrestrained Dofs

Deformation and Reaction force at Restraint Dofs

Member end Forces in Local Coordinate

13

6.0 Revisiting Stiffness Matrix for Structural Analysis and FEM

In structural analysis, as defined earlier, one column of a stiffness matrix is directly computed by applying
unit displacement along the direction of a particular degree of freedom while restraining the others and then
calculating the forces developed along the direction of all degrees of freedom. In this section, we will revisit
the same concept from a different perspective.
6.1 Shape Functions
Considering only flexural deformation, equilibrium equation of a beam element of constant flexural rigidity
in the unloaded region is given by
4

v
x4

(18)

The solution of Eq (18) may be expressed through a cubic polynomial as

v x

1 2 x 3 x2

4 x3

(19)

where i s are arbitrary constants and can be evaluated through the boundary conditions. These boundary
conditions are:
v
x
v
v2 and
x

v1 and

1 at x

(20)

2 at x

Utilizing Eq (20) into Eq (19) and arranging the resulting equations in matrix form, it may be shown that
v1

1
v2

1
0 1 0
0 2
2
1 L L
L3 3
0 1 2 L 3L2 4

1 0

(21)

Inverting Eq (21),

1
2
3
4

1
L3

L3

L3
2

3L

2L

v1

1
2

3L

v2

H a

(22)

Substituting i s from Eq (22) into Eq (19) and thereafter rearranging the resulting equation, it may be
shown that

14

v x

1 f2 x

v 1 f1 x
x
L

f1 x

1 3

f2 x

x
x1 2
L

f3 x

f4 x

x
L

2
x
L

x
L

x
L
x
L
x
L

v 2 f3 x

2 f4 x

(23)

The functions fi s are essentially the shape functions and derived from the equilibrium of the unloaded
beam element of constant flexural rigidity. Variation of these shape functions over the length is shown in
Figure 7.

Figure 7: Shape functions for a beam with constant flexural rigidity

15

Now consider, for example, v 1 1.0,

0, v

0,

0, which corresponds to v x

f1 x . Recall

that, this is also the case of generating the first column of stiffness matrix. Therefore, f1 x represents the
displacement profile when unit displacement is applied along the first degree from freedom while others
are held restraint. Similar interpretation holds for other shape functions also. Hence, displacement at any
point over the span can be readily calculated if the nodal displacements and all the associated shape
functions are known.
6.2 Castiglianos Theorem
Consider an elastic system subjected to a set of conservative forces Pi , i

1, n. Let

denotes

displacement along the direction of Pi and at its point of application. Let us consider two different cases.
In case-1, only ith force Pi is applied and

along the direction of force as described in

denotes the resulting displacement at all the n points but

. In case-2, all the forces except ith are applied and

denotes the resulting displacement at all the n points but along the direction of force as described in

i
i

Now, if case-1 loading is applied first followed by case-2, the work done on the body or strain energy is
given by
U1

1
Pi
2

Pi i

1
Pj
1 2

(24)

j
j i

If an infinitesimal virtual deformation profile ai (compatible with the restraint) is applied in between case1 and case-2, then the strain energy will be given by
U2

1
Pi
2

Pi ai

Pi i

1
Pj
1 2

(25)

j
j i

Therefore, change in strain energy is

Pi ai

Pi

Considering Lt ai

Pi

U
ai

(26)

0,

U
ai

(27)

Eq (27) is known as the Castiglianos first theorem.

6.3 Application of Castiglianos Theorem
Assuming only flexural deformation, strain energy of the beam element shown in Figure 1 can be expressed
as
16

EI
2

v
dx
x2

(28)

Noting
v '' x

v 1 f1'' x

1 f 2'' x

2 f 4'' x

v 2 f 3'' x

(29)

and employing Castiglianos theorem we may write, for example,

L

U
v1

Y1

EI
0

v
x2

v
dx
x2

v1

1 f 2'' x

v 1 f1'' x

EI

1 EI

f1'' x f1'' x dx
0

2 f 4'' x f1'' x dx

v 1 EI

v 2 f 3'' x

f1'' x f 2'' x dx
0

2 EI

f1'' x f 3'' x dx

v 2 EI
0

f1'' x f 4'' x dx (30)

0

K11v 1 K12 1 K13v 2 K14 2

Similarly, other three nodal force can also be derived leading to the ij th element of stiffness matrix as
L

Kij

fi '' x f j'' x dx

EI

(31)

Using the shape functions developed in Eq (23) into Eq (31) and carrying out the necessary integration over
the length, one may derive the stiffness matrix which is same as that shown in Eq (1). For example,
L

K11

EI
0

x
1 3
L

x
2
L

3 ''

x
1 3
L

x
2
L

3 ''

dx

EI
0

6
L2

12 x
dx
L3

12 EI
L3

(32)

6.4 Application of Rayleigh Ritz Method

Expression of strain energy involves derivative/integration of displacement profile and hence that of shape
functions, which becomes complicated for more complex finite elements. To overcome such problem,
Rayleigh Ritz method is illustrated below with respect to the same beam element. Assuming the same cubic
polynomial displacement profile as considered before and substituting its second derivative into the
expression of strain energy, we may write
v x
U

1 2 x 3 x2
EI
2

4 x3

v '' x

2 3

6 4 x

2 3

6 4 x dx
2

EI 2 32 L

6 3 4 L2

6 42 L3

Eq (33) is quadratic of the coefficient i s and can be expressed as

17

(33)

1
2
0 0
0
0
2
0 0 4 EIL 6 EIL 3
0 0 6 EIL2 12 EIL3 4

0 0
U

1
1 2 3 4
2

(34)

Here, elements of k are obtained through

2

kij

(35)

i j

Substituting from Eq (22) into Eq (34), strain energy can be expressed as

1
a
2

k H

(36)

Another way to formulate the strain energy is to use the nodal force and displacement vectors as follows:
Y1
U

1
v 1 1 v2 2
2

M1

1
a
2

Y2

K a

(37)

M2

Here K is the element stiffness matrix, which can be expressed by equating the strain energy from Eq
(36) and Eq (37) as
K

(38)

k H

Substituting H from Eq (22) and k from Eq (34) into Eq (38), element stiffness matrix can be obtained
as
L3

3L

2 L2

3L

1
L3

EI
L

12
L2
6
L
12
L2
6
L

L
6
L
4
6
L
2

2
12
L2
6
L
12
L2
6
L

L2

4 EIL

6 EIL

6 EIL2
3

12 EIL

1
L3

L3

L3

3L

2 L2

3L
2

L2
L

(39)

6
L
2
6
L
4

18

The stiffness matrix is same as that obtained using elementary structural mechanics or by using
Castiglianos theorem.
While deriving the stiffness matrix using either Castiglianos theorem or Rayleigh Ritz method, axial
deformation is not considered. Under the assumption of small deformation problem, axial degrees of
freedom are uncoupled with flexure and shear, and can be included in the stiffness matrix with an additional
pair of shape functions.

7.0 Finite Element Method: A Preliminary Revisit

7.1 Displacement Functions
Consider for simplicity a plane stress (or two dimensional in a loose sense) problem as shown in Figure 8.
The structure is discretized into several area elements and a typical element is denoted as e. Nodal points
of the element are indicated by i, j, m etc. A typical node has two displacements, one along each orthogonal
direction, and for the ith node, it is denoted as ai

u xi

u yi

. Here over bar denotes a vector quantity to

make distinction from a scalar. Similarly, the displacement at any point within the element may be denoted
as u

u xi x, y

u yi x, y

Displacement u may be approximated as

e

ai

e
k

Nk a

Ni

Nj

.... a j

Na e

(40)

.
.

The functions Ni , N j should be chosen as to give appropriate displacements at the respective nodes. In
other words
Ni x j , y j

ij

1 i

0 i

(41)

Understanding that nodal displacement at a particular node is a vector quantity involving displacement
along any set orthogonal directions, their spatial derivative etc, we may now drop the over bar from the
representation of vector quantities and Eq (40) is restated as

Nae

(42)

Further, unless otherwise specifically stated, we do not restrict the discussion within the domain of plane
stress problem.

19

Figure 8: A plane stress problem

7.2 Strain-Displacement Relation
In most cases, using the principle of mechanics, it is possible to express the strain tensor at any point within
a finite element in terms of the spatial derivative of displacements at that point. In a plane stress problem,
for example, strain vector (constituted from the independent elements of the strain tensor) can be expressed
as

xx
yy
xy

ux
x
uy

x
0

y
ux
y

0
ux

(43)

y uy

uy
x

Here, u x and u y denote the displacement at the point along two orthogonal directions. In general, Eq (43)
may be written as

(44)

S u

and upon substituting Eq (42)

20

S N

S N ae

S u

B ae

(45)

7.3 Constitutive Relation

Assuming linear elastic behavior, general stress strain relation can be expressed as

Here,

(46)

is the strain vector (constituted from the independent elements of the stress tensor) and, 0

and 0 stand for the residual stress and initial strain, respectively. For a plane stress problem,

xx
yy and D
xy

E
1 2

(47)

1 2

0 0

where, E and are the Youngs Modulus and Poissons Ratio, respectively.
7.4 External Loading
Three different types of loadings are considered. First, distributed body forces of intensity b ; this force is
considered acting on per unit volume of the material. Second, distributed traction of intensity t ; this force
is considered acting on per unit surface area. Third, external concentrated forces acting on the nodal points.
7.5 Derivation of Element Equilibrium
The simplest procedure of establishing element equilibrium is to apply an admissible virtual displacement
and equate the external and internal work done. Let a e be the virtual displacement vector at the nodes.
At any point within the element, the associated displacement and strain vectors may be expressed as

N a e and

B ae

(48)

Work done per unit volume by the internal stress and body force may be expressed as

b . Further, work done by the surface loading per unit surface area is

t .

Noting that total work done over the entire element is zero, we write

Ve

b dVe

t dAe

Ae

Substituting Eq (48), Eq (49) can be simplified to

21

(49)

ae

ae

ae

b dVe

Ve

t dAe

Ae

(50)

e T

b dVe

Ve

t dAe

Ae

Therefore, element equilibrium can be expressed as

b dVe

Ve

t dAe

(51)

Ae

Further substituting the constitutive relation Eq (46) and subsequently Eq (45), equilibrium equation can be
rewritten as
0

D dVe

Ve

D 0 dVe

Ve

0 dVe

Ve

b dVe

Ve

t dAe

Ae

(52)
=

D B dVe

Ve

D 0 dVe

Ve

0 dVe

Ve

b dVe

Ve

t dAe

Ae

The element equilibrium equation can be expressed in more formal way as follows:
qe

K e ae

Ke

(53)

D B dVe

Ve

qe

D 0 dVe

Ve

Ve

0 dVe

Ve

b dVe

t dAe

Ae

7.6 Overall Analysis

Eq (53) is of the same form as discussed in conventional structural analysis (Eq (1) or Eq (2)) . Therefore,
all the steps described in direct stiffness method of conventional structural analysis including local to global
transformation, expansion in global coordinate, assembly (taking into account the externally applied
concentrated nodal force vector q* ), and applying physical restraints to the global degrees of freedom
will be followed after Eq (53) to get the complete solution for unrestraint nodal displacements. Once the
nodal displacement vector a e associated with an element is extracted, stress vector at any point within
the element may be calculated as

D B ae

D 0

(54)

Therefore, FEM follows essentially the same principle as that of the conventional structural analysis but
the procedure of formulating governing equilibrium equations, for example stiffness matrix etc., is different.
Use of approximate shape functions while formulating the equilibrium equation allows FEM to be

22

applicable to all possible problems, whereas conventional structural analysis uses exact shape function and
its applicability is restricted to such simple cases for which exact shape functions exist.
7.7 Finite Element Method without Assembling Element Equilibrium
Unlike direct stiffness method, equilibrium of the entire structure without assembling the element
equilibrium equations is often considered in conventional method. Such is a case, usually applies to a
relatively simple configuration, wherein stiffness matrix of the entire structure based on unrestraint degrees
of freedom is derived in global coordinate. FEM can also be derived from that perspective without
assembling the element equilibrium equations.
In order to illustrate that, a set of admissible discretization as in the previous case is considered. All the
concentrated loads are assumed to be applied through the nodal points only. Let a denotes a vector listing
the displacement of all the nodal points and N

denotes the associated shape functions such that

displacement at any point is given by

u

(55)

N a

These shape functions are different than what were assumed previously. Consider, for example, j as a
nodal point common to a set of elements e . Shape function associated with the nodal displacement at j
is N j . If the point at which displacement to be approximated lies outside all the elements included in e ,
then N j

N ej where the point belongs to the element e and N ej is the shape function

0 . Otherwise, N j

of the j th node of element e , as defined in previous approach. As before,

S u

S N a

(56)

B a

We now drop the over bar with an understanding that the quantities like shape functions etc. are defined
over the whole region.
Applying any admissible virtual displacement a and equating the external and internal work done, we
may write

q*

b dV

t dA

dV

(57)

Now taking the variation of Eq (55) and Eq (56) for the virtual quantities, and substituting into Eq (57),
we write

q*

N
V

b dV

t dA

dV

Cancelling out the virtual deformation and, substituting the constitutive relation and Eq (56),

23

(58)

q*

b dV

or ,

t dA

A
T

D B dV

D B dV

q*

q*

D 0 dV

b dV

or , K a

a
T

0 dV

t dA

D 0 dV

0 dV

(59)
Clearly, Eq (59) represents the similar form that obtained after assembling Eq (53) for all the elements,
wherein i) q is same as the assembled equivalent nodal force vector, ii) assembly of integration over
elements is same as the integration over the whole region and iii) K is same as the assembled element
stiffness matrices.
7.8 Finite Element Formulation from the Minimization of Total Potential Energy
Virtual work principle is used in formulating the FEM in previous two cases. In this section, admissible
virtual displacement is considered as the variation of the real displacement. Therefore, a and hence,

u and are the variation of real quantities. Denoting W and U as the potential energy of the external
load and strain energy, respectively, Eq (57) can be rewritten as

q*

b dV

or , W

t dA

dV

(60)

Here is the total potential energy, which is stationary per Eq (60). Therefore, finite element formulation
can also be derived by setting

a1

a2

. .

(61)

This is the well known Rayleigh Ritz method and illustrated earlier in context with conventional structural
analysis of a beam element.
7.9 Example of FEM Formulation on a Beam Element
Assuming only flexural deformation governs, stress-strain relation can be considered in as generalized
d 2v
sense as moment-curvature relation. Hence, generalized strain is
and generalized stress is
dx 2
d 2v
M
EI 2 . Clearly, D with D EI . The beam is discretized and consider an element e
dx
with nodes i and j . Displacement at any point within this element can be approximated as u

24

N ae

dv
be
dx
continuous between the element, which can be enforced by incorporating them into the nodal displacement
. Since the strain involves second derivative of displacement, it is necessary that both v and

vector. Therefore, aie

dv
dx

vi i

Since, the element has two nodes each with two variables, assuming a cubic polynomial displacement
profile, as shown in Eq (19), the shape functions can be derived as shown in Eq (23). Therefore,

Ni

f1 x , f 2 x , N j

f3 x , f 4 x

Consequently,
Bi

f1'' x , f 2'' x , B j

Noting that B

Bi

f 3'' x , f 4'' x
f1'' x

Bj

f 2'' x

f 3'' x

f 4'' x , the element stiffness matrix is given

by
Ke

D B dVe

Ve

EI B dx

Any element of this is given by

Ke

fi '' x

EI B dx

EI

f j'' x dx

f i '' x f j'' x dx

EI
L

This is same as what has been derived with conventional structural analysis.
8.0 General Problem Statement
Consider a domain enclosed by a boundary . A set of coupled differential equations of the form
Au

A1 u , A2 u ,...

within the domain defines the problem statement together with

certain boundary conditions of the form B u

B1 u , B2 u ,...

defined on the boundary .

The objective is to find an unknown solution u that satisfies the differential equations and the boundary
conditions. FEM attempts to find the solution of this problem using a variety of techniques.

25