FINITE ELEMENT METHOD

Formulation of Finite Element Method.

The purpose of formulating the mathematical and numerical modelling in finite element
method is to analyse complex engineering problems which cannot be solved by conventional
methods. Hence before analysing any complex engineering problem one has to know prior to the
analysis the following
1. Type of the Analysis
2. Type of Material
3. Type of Loading / Force
4. Type of Boundary Condition
5. Type of Solutions

Type of the Analysis:

Whether the analysis has to be carried by
1. One Dimensional Analysis – Using One Dimensional Element (Line Element)
2. Two-Dimensional Analysis – Using Two-Dimensional Element (Triangular, Quadrilateral)
3. Three-Dimensional Analysis – Using Three-Dimensional Element (Tetrahedron,
Hexahedron for Solid Mechanics and Plates and Shell for Structural Mechanics)

Material:
In the analysis of continuum mechanics, one has to know the following
1. Type of Medium – Solid, Liquid, Gas/Air
2. Type of Composition – Homogeneous, Heterogeneous, Composite
3. Type of Variation - Linear or Nonlinear (Material / Geometric), Rate Independent
4. Type of Behavior – Flexible / Rigid, Conductive / Insulate, Ideal / In-compressible

Loading / Force:
In any analysis of a continuum loading and forces acting on the continuum is one the key
factor for the behaviour on the functional variables involved, Hence knowledge of the different
types of loading/forces acting has be dealt in an appropriate manner otherwise the entire effort on

1
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
the analysis will be waste. Thus loading/forces are to be treated in different ways during the
analysis which can be considered as one of following case
Static – Concentrated, Distributed, Body Force, Temperature Load, Conduction Convection,
Radiation, Pressure, Discharge. Rate Independent.
Time Dependent - Dynamics: Transient, Steady State, Random, Rate Independent
Impact Loads
Crash Loads

Boundary Condition:
All continuum mechanics problems will be subjected to one or other type of boundary
conditions which is very important in the analysis. Unless and until an appropriate boundary is
applied the solutions will not be accurate. Thus during solving the problem different types of
boundary conditions which are commonly encountered in finite element analysis are
Static – In Solid / Structural Mechanics: Displacement, Slope, Bending Moment, Shear Force,
Rate Independent etc.
Dynamics – Initial Condition and final conditions: Displacement, Velocity, Acceleration, Rate
Independent
Heat Transfer – Conductive, Convective, Heat Flux etc.
Fluid Dynamics – Stream Function, Velocity Potential, Pressure.

Solutions:
Different type of solutions which are commonly carried out are
Static - Linear, Non-Linear, Composite
Dynamics – Modal, Transient, Random or Spectral Density Analysis
Heat Transfer – Conductive, Convective, Radiation
Fluid Dynamics – In-compressible fluid, Compressible Fluid, Open Flow, Closed Flow.

2
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD

General Description of the Finite Element Method

Any analysis in finite element method involves basically six steps which are

1. Discretization of the given Continuum.

2. Approximation of the Basic Unknown.
3. Element Characteristic Equation.
4. Assembly of Characteristic Equation for overall Continuum.
5. Solution of the basic unknown.
6. Computation of the other unknown.

Step -1: - Discretization Process

The Continuum domain is divided into finite number of small domains called elements
connected by nodes taking into account of any symmetry, different material, loading, boundary
condition and the type of analysis.
The collection of elements is called the finite element mesh. If all the elements are of the
same shape and size then the mesh is said to be uniform Mesh or else it is said to be non-uniform
mesh.
In static analysis the continuum can be treated under different type of symmetry taking into
consideration the geometry, material, loading and boundary conditions.

Symmetry conditions:
In case of static analysis symmetry reduces the model size and hence solution time depending upon
1) Planar Symmetry 2) Axisymmetric
3) Cyclic Symmetry 4) Repetitive Symmetry

To understand the step by step procedure of finite element method Consider an example
shown in figure -1
P
8m

Q1 Q2 Q3
1 2 3
Figure -1
3
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
Given:
Area A = 200 mm2
E = 2.00E+05 N/mm2
P= 4 kN
Analytical Solution:
Deformation Ux=L = PL/AE = [(4000*8000)/(200*2x105)] = 0.80 mm
Strain x = P/AE = [4000/(200*2x105)] = 1x10-4
Stress x = Ex or P/A = (1x10-4)*(2x105) or 4000/200 = 20 N/mm2
Reaction R = -P = -4 kN
The simple bar is divided into two element which has one degree of freedom at each node i.e. the
displacement along the x-direction identified or denoted as Qi as shown thus totally there will be
three degrees of freedom which is represented in the matrix form as {Q}T = {Q1 Q2 Q3}
For the example problem since the bar is discretise into two one dimensional bar element with two
nodes having one degrees of freedom at each node. Hence totally the bar will have three degrees
of freedom. The length of each element is 4m or 4000 mm.

Step-2: - Selection of Basic Unknown Approximation Function

The variation of the basic unknown variable (Displacement / Temperature / Velocity Potential etc.)
is selected by an approximate function which depends upon the type of analysis and the type of
element to be adopted for the analysis such as
For Simplex / Basic Element – Linear polynomial or trigonometric Functions and for Higher Order
Elements – Quadratic or Higher order function.
Example as
() = 0 + 1x - Simplex / Basic Elements
() = 0 + 1x + 2x2 + 3x3 + ……. - Higher Order Elements
{} = [P]{}
Where [P] is known as the poly nominal matrix and each of i for i varying from 0 to n the number
of constant are known as coordinate constant which are determined by considering at  = i the
unknown variable  = i then we get n equation which are in terms of i when written in the
matrix form will be of the form {i} = [C]{}.

4
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
Considering the unknown nodal variable {i can be represented in a general form by the matrix
{q}. Then the above equation can be written as {q} = [C]{}
Then we have {} = [C]-1{q} by substituting these in to the unknown variable function
Thus {} = [P]{} = [P] [C]-1{q} = [N]{q}
Hence {} = [N]{q}
Where [N] is known as the shape function matrix and {q} is the unknown nodal variable vector.
For the example considered:
Consider an isolated element as shown in figure -2

Isolated Element
q1 q2
1 2 
= − =+
x = 0 x = x1 x = x2 x
Figure-2

The element geometry is identified by the nodal coordinates x1 and x2 thus the length of the element
will be le = (x2-x1). Each node has one degree of freedom qi thus totally the element will have two
degrees of freedom. Hence the size of the element stiffness will be 2x2 related to nodal
displacement vector {q}T = {q1 q2}.
Consider this isolated element as a parent element shown in figure -2 in the natural  coordinate
system which varies from -1 to +1. Then the transformation from global x coordinates to natural 
coordinate system and vice versa can be achieved from the relation  = [2(x – x1)/(x2 – x1)] - 1
since (x2-x1) = le is the length of the element. Then we have
d = (2/le)dx or dx = (le/2)d or dx/d = le/2 or d/dx = 2/le
The variation of the any field variable () within the parent element can be approximated using a
linear polynomial in terms of the natural  coordinate system as () = 0 + 1 …….. (1)
Which can be written in the matrix form as {} = [1 ] 0
1
which is represented by matrix as {} = [P]{} ………(2)
Where 0 and 1 are the constants to be determined as follow.
At node 1  = -1  = 1 = 0 - 1 ……. (3)
At node 2  = +1  = 2 = 0 + 1 ……. (4)

5
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
Solving for 0 and 1 we get 0 = (1 + 2)/2 and 1 = (2 - 1)/2
substituting these into equation (1) we get
() = 0+1 = [(2 + 1)/2] + [(2 - 1)/2]
= [(1 – )/2] 1 + [(1 + )/2] 2 = N1 1 + N2 2
Thus () = N1 1 + N2 2 In matrix form {} = [N]{i} ....... (5)
Equation (3) and (4) can also be represented in the matrix form as
I = 1 -1 0
i 1 1 1 i.e. {i} = [C]{} or {} = [C]-1{i} ...... (6)

Substituting equation (6) into equation (2) we get

{} = [P]{} = [P][C]-1{i}= [N]{i}
Thus {} = [N]{i} which is same as equation (5)
In which [N] = [N1 N2 ] = [(1 – )/2 (1 + )/2]
Where N1 = (1 – )/2 and N2 = (1 + )/2 are known as the shape functions which can be used to
represent any field variable within the element such as geometrical coordinate, cross sectional area,
displacement etc. which are assumed to vary linearly.
Thus, the variation of the displacement within the parent element can be expressed in terms of
nodal variables as
{U} = N1u1 + N2u2 = N1q1 + N2q2 = [N]{q}
Where N1 = (1 – )/2 and
N2 = (1 + )/2 are known as the shape functions

Step-3:- Derivation of The Elemental Characteristic Equation

The Equation to describe the behaviour of the Parent / Master Element are derived using any
one of the following basic principles.
Variational Formulation Methods
Weighted Residual Methods
Rayleigh – Ritz’s Method

6
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
Example in Solid / Structural Mechanics
In the second step we have derived the relationship between the unknown variable { to be
determined in terms of shape function [N] and unknown nodal variables {q} as
{} = [N]{q}
Thus in solid/structural mechanics using displacement approach if displacement is treated as the
basic unknown variable the { can be replaced by [U]
Thus, we have the unknown Displacement {U} = [N]{q}
It is observed that the unknown variable anywhere in the element domain is always written in
terms of the shape function [N] and nodal unknown variable {q}.
Since strains are the first derivatives of the displacement then strain in matrix form can be written
as { }= [L]{U} in which [L] is known as the differential operator matrix.
But the unknown displacement is written as
{U} = [N]{q}.
Substituting this in the strain matrix we have
{ } = [L]{U} = [L][N]{q} = [B]{q}

Thus { } = [B]{q} where [B] = [L][N]

Considering the analysis is for linear, homogeneous and isotropic material. From Hook’s law stress
is directly proportional to strain and the constant of proportionality in known as the modulus of
elasticity. Also form the basics of mechanics of materials for a linear, homogeneous and Isotropic
material subjected to three dimensional state of stress according to Hook’s law the state of stress
at any point is related to the state of strain with only two elastic constant (Young’s Modulus ‘E’
and poison’s ration  which is written in the matrix form as

Stress { } = [D]{} = [D][B]{q} where [D] is known as the elasticity matrix.

Displacements are caused by the application of loads. Also, these loads will do some work on the
continuum while undergoing the displacement. This work done on the continuum is stored as strain
energy in the continuum. Thus, graphically the area under the stress - strain curve is the strain
energy stored in the continuum.

7
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
T
Strain Energy Ue = 1/2 ∫v{} {}dv substituting for ‘’ and ‘’ we have
T T T
Ue = 1/2 ∫v{q} [B] [D] [B]{q} dv

Since {q} is the unknown variable (Displacement) to be determined hence can be taken out of the
integral and for a linear, homogeneous and isotropic material the elasticity matric is symmetric
T
hence [D = [D] Thus we have
T
Ue = 1/2{q} ( ∫v[B]T[D]T[B] dv) {q}
T
Ue = 1/2{q} [Ke]{q}

Where [Ke] = ∫v ([B]T[D]T[B])dv is known as Element Stiffness Matrix.

The elemental stiffness matrix [Ke] will be a square matrix and the size of the matrix depends
on the number of degrees of freedom (n) in the element. Thus the size of the elemental stiffness
matrix will be (n x n)
Since different types of forces/loads are acting on the continuum then the work done by these
loads are given by
T T n T n T
Work Done We = ∫v{U} {b}dv + ∫s{U} {t}ds + ∑i=1 {U}i {Pc}I + ∑i=1 d{U}i {M}i
T T T T n T T n T T
We= ∫v{q} [N] {b}dv + ∫s{q} [N] {t}ds + ∑i=1 {q} [N]i {Pc}I + ∑i=1 {q} d[N]i {M}i

Since {q} is the unknown variable (Displacement) to be determined hence can be taken out of the
integral which is common in the work done expression thus we have
T T T n T n T
We= ∫v{q} [[N] {b}dv + ∫s[N] {t}ds + ∑i=1 [N]i {Pc}I + ∑i=1 d[N]i {M}i]
T T
We = {q} [ {fb} + {ft} + {fc} + {fm}] Thus We = {q} {fe}
Where {fe} is known as the elemental force/load vector in which
T
{fb} = ∫v[N] {b}dv is the elemental force vector due to body force
T
{ft} = ∫s[N] {t}ds is the elemental force vector due to traction force
n T
{fc} = ∑i=1 [N]i {Pc}i is the elemental force vector due to concentrated force
n T
{fm} = ∑i=1 d[N]i {M}i is the elemental force vector due to Moments

Apart from these physical forces/loads if there is a change in temperature due to which the load
vector is {fint} = ∫v [B]T[D]{int}dv is elemental force vector as a result of temperature variation.

8
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
The total potential in the element is given by  = Ue ± We.
Since the work is done on the system then the work done will be negative thus
 = 1/2{q}T[Ke]{q} - {q}T{fe}
By the Principal of Minimum Potential Energy for the system to be equilibrium the total potential
should be at its minimum. Thus ∂p/∂{q} = 0
∂/∂{q} = [Ke]{q} – {fe} = 0
Thus, we have [Ke]{q} = [fe] is the characteristic equation for the element
Which is very well-known equation relating the stiffness and displacement to force.
In which [Ke] is element stiffness matrix whose size will be equal to (n x n) where n is the number
of degrees of freedom in the element.
{q} is the unknown nodal displacement vector whose size will be (n x 1)
{fe} is the elemental force/load vector whose size will be (n x 1)

For the example considered

Displacement is given by {U} = [N]{q} = [N1 N2 ] {qi}
U() = [(1 – )/2]q1 + [(1 + )/2]q2
The strain-displacement relation is given by x = u/ x.
Strain-Displacement relation can also be written in the matrix form as
{}= [/x]{U}= [L]{U} = [L][N]{q} = [B]{q}
where [B] = [L][N] = [/x][N1 N2] = [N1x N2/x]
Since Ni’s are the Shape functions which are functions of the natural coordinates x whereas the
strains are the cartesian derivatives hence by chain rule of differentiation we have Ni/x =
[Ni/x]*[x/x].
Since dx/dx = 2/le Thus we have Ni/x = 2/le [Ni/x]
Thus [B]= [N1/x N2/x] = 2/le[{(1 – )/2}/ {(1 + )/2}/]
[B] = 2/le[(-1/2) (1/2)] = 1/le[-1 1]
Hence [B] = 1/le [-1 1]
Thus {} = [B]{q} = 1/le [-1 1] {q1 q2}T
The Stress - Strain relation is given by x= E x
Stress - Strain relation can also be written in the matrix form as
{}= [D] {} since {} = [B]{q} Then we have {} = [D][B]{q}
9
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD

Strain Energy Ue = ½ ∫v{}T{}dv = ½∫v [{q}T[B]T[D]T[B]{q}]dv

Since {q} is the unknown nodal displacement vector to be determined hence can be taken out of
the integral. For a linear, homogenous, and isotropic material the elasticity matrix is symmetric
hence [D]T = [D]
Thus Ue = ½ {q}T[ ∫v [B]T[D][B] dv]{q} = ½ {q}T[ke]{q}
In which [ke] = ∫v [B]T[D][B]dv is known as the elemental stiffness matrix.
Then [ke] = ∫v [B]T[D][B]dv is derived by substituting the terms for [B] and [D] we get

E (A1 + A2) 1 -1
ke =
2le -1 1
is the elemental stiffness matrix for tapered 1-D bar element.
If the bar is of uniform cross-section then we have A1 = A2 = A. Then the elemental stiffness
matrix for a uniform cross section bar will be

EA 1 -1
ke = -1 1
le

For the example problem

Work done is given by we = {q}T {fe}
In which {fe} = {fc} is the elemental force vector where
n
{fc} = ∑i=1 [N]iT{Pc}i is elemental force vector due to concentrated force

Thus, for the element 1 as there is no force is acting, we have

{f1} = Pci N1 = 0
N2  =  1

and for element 2 force of Pc is acting at node 2 where  = +1. Then we ha

{f2} = Pci N1 = Pci (1 – )/2 = P (1 – 1)/2 = P 0
N2  =  (1 + )/2  =  (1 + 1)/2  = + 1

Thus, the elemental characteristic equation for the element 1 is

EA 1 -1 q1
q2 = 0
le1 -1 1 0
10
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD

The elemental characteristic equation for the element 2 is

EA 1 -1 q2
q3 = 0
le1 -1 1 P

Substituting the values from the given data we have the elemental stiffness matrix
1 2
5
EA 1 -1 2x10 *200 1 -1 4 1 -1 1
k1 = =
le1 -1 1 4x10
3 -1 1 = 10 1 1 2

2 3
5 1 -1
k2 = EA 1 -1 2x10 *200 1 -1 4 2
= = 10 -1 1
le1 -1 1 -1 1
3 3
4x10

Elemental force vector {f1} = 0 1 {f2} = 4x103 0 2

0 2 1 3

Step-4: - Assembly of Equations for The Overall Continuum

Since the elemental characteristic equation is derived for an arbitrarily parent/master element,
it holds for any element from the finite element mesh. If the continuum domain is divided into ‘n’
number of elements with N degrees of freedom which are connected through nodes, then to express
the correspondence mathematically we label the nodal variables as of ‘qi’ at the global nodes with
‘Qi’ for i = 1, 2, 3, …..N. Then we have the following correspondence between the local (element)
nodal variables and the global nodal variables. These relations are called as inter-element
continuity conditions. The Element Equations are then assembled / combined to give the overall
system equations as

n n
∑e=1 [ke]{q} = ∑e=1 [fe] which can be represented as

[K] {Q} = {F} is the characteristic equation for the overall continuum
(NxN) (Nx1) (Nx1)

11
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
For the example considered:
The overall stiffness matrix for the problem [K] = [k1]+[k2]
1 2
4 1 -1
5
k1 = EA 1 -1 2x10 *200 1 -1 1
= = 10 -1 1
le1 -1 1 -1 1
3 2
4x10

2 3
5
4 1 -1
k2 = EA 1 -1 2x10 *200 1 -1 2
= = 10 -1 1
le1 -1 1
3 -1 1 3
4x10

Elemental force vector {f1} = 0 1 {f2} = 4x103 0 2

0 2 1 3

Thus, the overall stiffness matrix will be

1 2 3 1 2 3
1 -1 0 1 1 -1 0 1
K 4
-1 1+1 -1 -1 2 -1
= 10 2
= 10
4 2
0 -1 1 3 0 -1 1 3

0 1 0 1
3 3
Overall force vector will be {F} = 4x10 0+0 2 = 4x10 0 2
0 3 1 3

The characteristic equation for the overall continuum is given by

[K]{Q} = {F}
1 2 3
1 -1 0 1 Q1 0 1
3
-1 2 -1 2 Q2 = 4x10 0 2
0 -1 1 3 Q3 1 3

Step - 5: - Determination / Solution for The Basic Unknown

The overall system equations are solved by the incorporating the specified boundary conditions.
Then the systems of equations are modified to take into account of the constraints and the system
equations are solved to arrive at the unknown nodal variables (Displacement / Temperature /
Velocity Potential etc.). The boundary conditions can be treated by the any of the three methods
1) Elimination method 2) Penalty method 3) Multipoint Constraint Method
Elimination approach is simple and direct method for manual calculation but difficult to write
a general purpose algorithm and to implement in the software.
12
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
Penalty approach can be easily implemented in the software since the method is direct and a
general-purpose algorithm can be written in a simplified manner.
Multipoint constraint approach is the difficult method among the three but practically this
method is important in solving many industry related problems.
Among the above three method penalty approach is most commonly and widely used method

For the sample problem considered:

The characteristic equation for the overall continuum is given by [K]{Q} = {F}

1 2 3
1 -1 0 1 Q1 0 1
3
-1 2 -1 2 Q2 = 4x10 0 2
0 -1 1 3 Q3 1 3

Apply the specified boundary condition i.e Q1 = 0 since it if fixed. Thus, by elimination approach
eliminating the first row and column from the overall characteristic equation we get
2 3
3
2 -1 2 Q2 = 4x10 0 2
-1 1 3 Q3 1 3

solving for Q2 and Q3

We get 2Q2 – Q3 = 0 then Q3 = 2Q2
-Q2 + Q3 = 0.4 or -Q2 + 2Q2 = 0.4 or Q2 = 0.4
Since Q3 = 2Q2 = 2x0.4 = 0.8
Solution for the nodal displacement are Q1 = 0, Q2 = 0.4 and Q3 = 0.8

Step - 6: - Computation of Other Unknowns

The other required unknowns (Gradients/derivatives of the unknown variable) such as Strains,
Stresses and Reactions in case of Solid / Structural Mechanics are computed. Heat Flow rate in
case of Heat Transfer. Velocity in case of Fluid Dynamics are computed.
Simplex Element – Evaluation at Nodes
Higher Order Elements – Evaluation at Integration Points and then extrapolate to Nodes.
The calculated Nodal results are averaged and tabulated or plotted as Graphs or as
Contour Plots.

13
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD
For the sample problem:
The solutions are Q1 = 0, Q2 = 0.4 and Q3 = 0.8
Other unknowns are computed element by element.
Thus, for the element -1 we have q1 = Q1 = 0 and q2 = Q2 = 0.4
We know {U} = [N]{q} = N1*q1 + N2*q2 = N1*0 + N2* 0.4 = 0.4(1 + )/2
Thus U() = 0.2(1 + ) so U(-1) = 0, u(1) = 0.4 mm and U(0) = 0.2 mm
Strain {} = [B]{q} = 1/le [ -1 1]{q1, q2}T = (1/4000)[-1 1]{ 0 0.4}T = 1x10-4
Stress {} = [D]{} = E = 2x105 *1x10-4 = 20 N/mm2
Also, stress {} = P/A = 4x103/2x102 = 20 N/mm2

For the element -2 we have q1 = Q2 = 0.4 and q2 = Q3 = 0.8

We know {U} = [N]{q} = N1*q1 + N2*q2 = N1*0.4 + N2*0.8 = 0.4(1 + )/2
Thus U(x) = 0.4[(1 - )/2] + 0.8[(1 + )/2] = 0.2(3 + )
so U(-1) = 0.4, u(1) = 0.8 mm and U(0) = 0.6 mm
Strain {} = [B]{q} = 1/le[ -1 1]{q1, q2}T = (1/4000)[-1 1]{ 0.4 0.8}T = 1x10-4
Stress {} = [D]{} = E = 2x105 *1x10-4 = 20 N/mm2
Also, stress {} = P/A = 4x103/2x102 = 20 N/mm2

Reaction Forces are determined from the eliminated row of the overall characteristic equation.

Q1
4
10 1 -1 0 Q2 = R1
Q3

0
Thus R = 104 1 -1 0 0.4
0.8

4
R = 10 [1*0 -1*0.4 + 0*0.8] = -4000 N or -4 kN

14
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru
 FINITE ELEMENT METHOD

Advantages of FEM?
1. Handles non-homogeneous and anisotropic complex materials
2. Handles complex and irregular geometries
3. Handles multiple loading conditions
4. Handles various kinds of boundary conditions
5. Handles structural interactions
6. Handles linear and non-linear problems
7. Handles coupled field problems.

Disadvantages of FEM?
Error Free Input Data
Interpretation of Output Results
Economically Costlier

15
Dr. Jagadish. T, Former Professor and Head, Department of Mechanical Engineering, BIT, Bengaluru
Presently Director, Research and Development, DHIO Research and Engineering Pvt. Ltd., Bengaluru