8/10/2022

Department of
Finite Element Method Mechanical Engineering
(01ME0701)
Unit No : 01
Mathematical Modeling Unit title : Mathematical
Modeling
Code : 01ME0701

Dr. Ramesh Bhoraniya

Course overview
❑ Teaching Scheme:
Lectures : 4 hrs per week
Laboratory : 2 hrs per week
Basic mechanics
Credit : 5
❖ Solid mechanics
❑ Examination Scheme:
❖ Fluid mechanics
Mid-Sem Exam = 30
CSE = 20 ❖ Heat transfer
End-Sem Exam = 70
Practical Exam = 25+25=50
Total Marks = 150
❑ Pre-requisite
Matrix algebra; Basic Mechanics;

1
 8/10/2022

Objectives & Course outcomes

❖ Course objectives :
➢ Problem formulation and solution of …
o 1D and 2D structural problems
o 1D heat transfer problems
o 1D fluid flow problems
❖ Course outcomes

➢ Apply the knowledge of FEM for …

o 1D stress analysis
o Modal analysis
o Heat transfer analysis
o Flow analysis

Objectives & Course outcomes

➢ Formulate and solve problems of . . .
o Trusses
o Beams
o Frames
➢ Develop 2-D FE formulations involving …
o Triangular
o Quadrilateral
o Higher order elements.
➢ Develop algorithms and write FE code for solving simple design problems
➢ Understand the use of commercial packages for complex problems.

2
 8/10/2022

Program outcomes

❑ Engineering knowledge ❑ Environment and sustainability

❑ Problem analysis: ❑ Ethics
❑ Design/development of solutions ❑ Individual and team work
❑ Conduct investigations of complex problems ❑ Communication
❑ Modern tool usage ❑ Project management and finance
❑ The engineer and society ❑ Life-long learning

Course content & Books

❖ Mathematical modeling for structural problems

❖ Bars, Truss and Beams
❖ Two dimensional problems
❖ Plate and shell elements

❖ Reference books
➢ Finite element methods in Engineering, S. S. Rao, Elsevier – Good description

➢ Introduction to finite element engineering, T R Chandrupatala, PHI - Numericals

➢ A Text book of Finite Element Analysis, P. Sheshu, PHI – Simple Math-- structural

➢ A first course in the finite element method, Logan , Thomson Publication - Reference

3
 8/10/2022

Introduction to FEM

❖ Types of Solution
❖ Exact solution (Analytical solution)
❖ Approximate solution (Numerical solution

5
 1;
4
5
 1.2;
4
5
= 1.25
4
Area of circle

Numerical Methods

❖ Numerical Methods
❖ Finite difference method
❖ Finite element method
❖ Finite volume method

❖ Mesh less method

❖ Continuum : Infinite
Vs
❖ Finite Element

4
 8/10/2022

FEM Process

Refine Analysis
Physical Model
Improve
Mathematical
Model

Mathematical Model Refine Mesh

Assessment of
FEM Solution Accuracy

Interpret Results

Procedure in FEM

Structure related
problems
σ,ε,q

Step 1 Step 2 Step 3 Step 4 Step 5

Solve global matrix

• Generate Local
Select appropriate Stiffness matrix for unknowns
Discretize the
shape function 1. Minimum potential • Incorporate
domain
energy approach Boundary 1. Elimination
1. Linear Conditions and Approach
1. 1D elements 2. Galerkin approach
2. Quadratic Initial
2. 2D elements
3. Higher degree Conditions 2. Penalty
3. 3D elements • Generate Global
of polynomial Approach
Stiffness Matrix

5
 8/10/2022

Applications of FEM

Stress analysis of the connecting road

Applications of FEM

Static analysis of the bracket - displacement

6
 8/10/2022

FEM Approaches

❑ Direct Approach: Equilibrium Nature of Problems

❑ Variational Approach :
❖ To determine the extremum or stationary value of the functional w. r. t.
variables of the problem.
❖ In solid mechanics it forms energy of the system
❖ In other problem, it forms integral form of governing equation
❖ Mathematically , 𝜕𝜋
𝜋 = න𝐹(𝑥, 𝜙′, 𝜙′′)𝑑𝑥 ; =0
𝜕𝑥

Where,
x = independent variable (direction)
 = displacement
 = Potential energy

August 10, 2022

FEM Approaches
❑ Weighted Residual Method: The approximate solution in which the
residual is forced to zero in some average sense over a domain.
❖ Mathematically, 𝐼 = න𝑊 𝑥 𝑅 𝑥 𝑑𝑥 = 0
W(x) – Weight function
R(x) - Residual of solution
❑ Types of Weighted residual method
❖ Collocation method
❖ Sub domain method
❖ Least square method
❖ Galerkin method

7
 8/10/2022

Direct Approach

Force = Stiffness  Net displacement

F = Force or load
K = Stiffness of spring
u = displacement

Variational Approach
(Minimum Potential Energy Approach)

For an elastic member under action of external load,

Potential Energy () = Strain Energy (U) + Work Potential (WP)

  EA
1 1
The SE of an elastic material is U =  T  dV = 2
dx
2 V 2
x
Where,
WP = −   fdV −  Tds −  i Pi
V S i U=Total Potential Energy

WP = −   fAdx −  Tdx −  i Pi = Elemental stress

x S i = Elemental strain
= Nodal displacement
f = Body force per unit volume
T= Traction force per unit surface area

8
 8/10/2022

Minimum Potential Energy Approach)

   fAdx −  Tdx −  
1
 = AE 2 dx − i pi
2 e
e e e

 d 
2


1
U =
e
2
AE   dx
 dx    fAdx = 
e
i fAl

2
 1 − 2 
 2 AE  x  Tdx =  Tl
1
=  dx i
e 1 − x2  e

 ( −  ) 2    P = ( P +  2 P2 )

1 i i 1 1
= AE   dx
1 2
2  l 2 
e  
1 AE
1 AE  = (12 − 12 + 2 2 ) − i fAl − iTl − (1 P1 + 2 P2 )
= (12 − 212 − 2 2 ) 2 l
2 l

Minimum Potential Energy Approach)

  AE  1 −1 1   P1 + F1 + T1 
= 0; =0    = 
1 2 l  −1 1  2   P2 + F2 + T2 

𝜕𝜋 1 𝐴𝐸
= (2𝜙1 − 2𝜙2 ) − 𝑓𝐴𝑙 − 𝑇𝑙 − 𝑃1
𝜕𝜙1 2 𝑙
Element stiffness Nodal Load Vector{f}
𝜕𝜋 1 𝐴𝐸 matrix [k] displacement{}
= (2𝜙2 − 2𝜙1 ) − 𝑓𝐴𝑙 − 𝑇𝑙 − 𝑃2
𝜕𝜙2 2 𝑙
[K]  =  P
For minimum potential energy  = 0
1
f Al 1 Tl 1 B = ( −1 1)
F= e ee   T = e   l
2 1 2 1  
 =  1
2 
1  
 = B = ( −1 1)  1 
l 2 
 = E = EB

9
 8/10/2022

Weighted Residual approach

(Galerkin Method)

❑ Axial loaded bar

Essential BC : Explicitly imposed on the solution

Natural BC : Automatically satisfied by the solution

Weighted Residual approach

(Galerkin Method)

❑ The WR formulation is given as

𝐼 = න𝑊 𝑥 𝑅 𝑥 𝑑𝑥 = 0

❑ The governing equation for a 1-D elastic member under load is

d 
2
EA 2
=0
dx

❑ Let assume (x) is the solution of the above equation, then

d 2 Here,
Rd = EA
dx 2
W = weight function
❑ The weighted Residual statement is
E = Modulus of elasticity
L L
d 2
I =  W ( x) R
0
d ( x ) dx =  W (x)EA dx
0
2  = displacement
A = cross-section area
R = Residual

10
 8/10/2022

Weighted Residual approach

(Galerkin Method)

 uvdx = u  vdx −  dx (  vdx )dx

❑ Now Integration by part du
L
d 2
I =  W (x)EA dx 2
dx
0 u differentiable
d 2 dW  d 2 
= W ( x ) EA dx 2
dx −  
dx   EA
dx 2
dx  dx

v Integration
 
d 
L 2
d  d
L
 u = W ( x ); v =

dW
= W ( x ) EA − EA dx
dx 
2
 0 0
dx dx dx

❑ Weight functions are taken as shape functions (N1, N2)

x x
W1 = N1 = 1 − ; W2 = N 2 =
L L
dW1 1 dW2 1
=− ; =
dx L dx L

Weighted Residual approach

(Galerkin Method)

❑ The final statement is

For weight function W1 For weight function W2
l l
 d  dW1 d l
d 
l
d
 0 0 
W1 ( x) EA dx  − dx EA
dx
dx = 0  dW2

W2 ( x) EA dx  − dx EA dx dx = 0
 0 0
d 
l
 d

x 1
d 
l
 (1 − l ) EA dx  − − l EA dx dx = 0  x d

1
 0  ( l ) EA dx  − l EA dx dx = 0
 0
❑ Apply Boundary conditions and evaluate at x=0 & x=L

 d  1  − 1  d  1  − 1
 EA dx 
 
− − EA 2
 x =0 e l l
dx = 0  EA dx 
 
− EA 2
 x =l e l l
dx = 0

 d  1  − 2 d  − + 2
 EA dx 
 
− EA 1
 x =0 e l l
dx = 0 
 EA dx 
  x =l e 
1
− EA 1
l l
dx = 0

11
 8/10/2022

Weighted Residual approach

(Galerkin Method)

For weight function W1 For weight function W2

EA
P0 −
EA
(1 − 2 ) = 0 Pl − ( −1 + 2 ) = 0
l l
EA   EA  
1 −1  1  = P0 1 −1  1  = Pl
l 2  l 2 

EA  1−1
[k ] =  1 
❑ Above equations together in matrix
form can be written as l  −1
   P0   P1 
EA  1

−1 1   P0 
 =   = 1   P =  Pl  =  P2 
l  −1 1 2   Pl   2
Where,
[K]  =  P K = Element stiffness matrix
 = Nodal displacement vector
P = Load vector

Strong and weak form of equations

❑ Strong form : (Difficult to solve)

❖ It is a conventional differential form of equation
❖ It imposes continuity and differentiability conditions on the solution
❖ All solution of strong form are solution of weak form
❖ Mathematically : Differential form of equations

❑ Weak form : (Comparatively easy to solve)

❖ It is an alternate representation of differential equation
❖ It relaxes continuity and differentiability certain extent on the solution
❖ All solution of weak from are not the solution of strong form
❖ Mathematically : Integral form of equations

12
 8/10/2022

Road map of FEM

Strong form Weak form

equations equations

Discrete form of
equations

Approximation
function

Rayleigh Ritz method

❑ Variational Approach

𝜕𝜋 𝜕𝜋
𝜋 = න𝐹(𝑥, 𝜙′, 𝜙′′)𝑑𝑥 ; =0 OR 𝜋 = න𝐹(𝑥, y′, y′′)𝑑𝑥 ; =0
𝜕 𝜕𝑦

2
1  dy 
F (x, y, y , y ) = −  + 1000 x 2 y
2  dx 

Functional form Differential Equ. form

1
 1  dy  2 
 = 
0
−  
 2  dx 

+ 1000 x 2 y dx

 Key point : Order of derivative is
less in weak form of equations

13
 8/10/2022

Rayleigh Ritz method

❑ There are three basic steps in Rayleigh Ritz method

❑ Assume a displacement field.
 ( x) +  c N , i = 1, 2,3,....where Ni is a shapefunction
i i

& ci are to be determined

❑ It should satisfy essential boundary conditions and compatibilities.
❑ Evaluate the total potential  .
❑ Setup and solve the system of equations. The total potential will be zero
with small variation in displacement.
 p
= 0, where i = 1, 2, 3
ci
❑ The constant ci can be evaluated from the above.

FEM formulation problems

Example 01 : Setup the potential energy functional for a truss element subjected to
a uniform traction as shown below.
q
1 2

 2 dV
1
As per potential energy approach,  = U + WP U =

 2 AE
1
=E and dV = A dx Thus, U = 2
dx
2
du  du 
  
1
But, = U = AE   dx WP = − (qdx) u = − (qu) dx
dx 2  dx 
1  du 
2 
 =   AE 
 2

 dx 
− qu dx


14
 8/10/2022

FEM formulation problems

Example 02 : The functional for a physical phenomenon is given below

 1
1  dy  
2
 =

0 

1000 x 2 y −
2  dx
 dx   
Assume the following approximation function for y with one Ritz parameter.
y = a  x − x5 
 

Where, a is Ritz parameter. Show that the given y satisfy the boundary condition.
Find the value of the Ritz parameter by Rayleigh-Ritz method.

Solution : The trial solution is given as , y = a  x − x 5 

 
When, x = 0 y=0

When, x = 1 y=0

FEM formulation problems

y = a  x − x5  dy
= a 1 − 5 x 4 
   
dx
Substituting it in the governing equation
1

 1000 x
1 2
( ) 
2
 = 2
a(x − x 5 ) − a 1 − 5x4 dx
0
2 

1000a 80a 2
 = − Thus, the approximate solution will be
8 90
 1000 160a y = 70.3  x − x 5 
90000  
= − =0a= = 70.3
a 8 90 1280
Home work : If trial function is following, then find Ritz parameter

y = a  x − x4 
 

15
 8/10/2022

FEM formulation problems

FEM formulation problems

Example 04: A uniform column is fixed at the bottom and is free at the top. It carries
a compressive load at the free end. Investigate the critical load of the column by
assuming a second degree polynomial as
y = a0 + a1 x + a2 x 2
And total potential is
1 2
 d2y 
1 2
P  dy 
 
EI
 =  2 
dx −
2  dx 
dx
2 
0  dx  0

Solution :
Let consider origin as a fixed end. Substitute these two condition in above polynomial

At x = 0, y = 0 a0 = 0 a1 = 0 y = a2x2

At x = 0 dy q
=0 y= x2
dx When x = l, y = q a2 = q/l2 2
l

16
 8/10/2022

FEM formulation problems

  4 EI 4P 
= 3 − q = 0, q may not zero.
1 2
 d2y 
1 2
3l 
P  dy 
 
EI
 =  2 
dx − q  l
2  dx 
dx
2 
0  dx  0

q dy 2qx d 2 y 2q
y= x2 = 2 , and = 2 4 EI
=
4P 3EI
P= 2
l2 dx l dx 2 l l 3
3l l
1 2 1 2
EI  2qx  P  2q 
 =  0
2  l 2 
dx − 
0
2  l 2 
dx

 2 EIq 2 2 Pq 2 
 = − 
 l3 3l 

Quiz- Unit-1

1. When weighted residual approach in known as Galerkin method

a. Weight function is linear
b. Shape function is considered as weight function
c. Both of above
d. Non of above
2. Minimum potential energy approach is a which type of approach ?

a. Direct approach
b. Weighted Residual approach
c. Both of above
d. Non of above
3. Residual is zero for which type of solution ?

a. Numerical solution
b. Exact solution
c. Both of above
d. Non of above

17
 8/10/2022

Quiz- Unit-1
4. It is a which type of FEM formulation ?
a. Variation approach
b. Weighted residual approach
c. Galerkin approach
d. All of above
5. FEM problem formulation results which type of equations ?

a. Logical equations
b. Linear Algebraic equations
c. Differential equations
d. Non of above

6. Which equations easy to solve numerically ?

a. Weak form
b. Strong form
c. Differential equations
d. Non of above

Least square method

(Weighted Residual Approach)

❑ It is a type of Weighted Residual method

❑ The integral of the square of the residual to be minimized.

I =  [ R(x, a
2
1, a 2 )] dx

I = f (a1, a 2 )

f f
= 0; =0
a1 a2

18
 8/10/2022

Example : Least square method

Problem 05: The following differential equation is available for a physical phenomenon.
d2y
+ 100 = 0 For 0  x  10
dx 2

The boundary conditions are y(0) = 0, and y(10) = 0. Using the following trial
function obtain the parameter a1 by Least square method and Galerkin method.
Ni = a1x (10 - x)

Solution : First Consider least square method

Verify boundary conditions; y(0), Ni=0 and y(10); Ni=0
10 10
I
I =  R 2 dx R = −2a1 + 100
a1
=  [−2a
0
1 + 100][ −2] dx = 0
0 10
I R I
R=
2
( a1 x(10 − x) ) + 100 a1
=  2 R da dx
a1
=  −2a1 + 100 x 0 = 0
10

x 2 0 1

a1= 50 y = 50 (10-x)

Example : Galerkin method

Solution : Consider now Galerkin method

10

 w( x) R(x)dx = 0
2a1= 100
I =
0 a1 = 50

w( x ) = a1 x (10 − x )
The trial function is y = 50x(10-x)

R ( x) = −2a1 + 100
10
I =  a [10 x − x ][ −2 a1 + 100]dx = 0
2
1
0

19