Mathematical

Modeling
Introduction
One of the most important things engineers and
scientist do is to model physical phenomena.
Virtually every phenomenon in nature, whether
civil, aerospace, biological, chemical, geological,
electrical, or mechanical can be described, with the
aid of laws of physics or other fields in terms of
algebraic, differential, and/or integral equations
relating various quantities of interest.
Introduction – Continued…
Analytical descriptions of physical phenomena and
processes are called mathematical models
Mathematical models of a process are developed
using:
◼ Assumptions concerning how the process works
◼ Appropriate axioms or laws governing the process
They are often characterized by very complex
differential and/or integral equations posed on
geometrically complicated domain
Introduction – Continued…
The processes to be studied, until the advent of
electronic computation, were drastically simplified
so that the governing equations can be solved
analytically
Over the last four decades, however, computers
have made it possible, with the help of suitable
mathematical models and numerical methods, to
solve many practical engineering applications
Introduction – Continued…
The use of numerical methods and a computer to
evaluate the mathematical model of a process and
estimate its characteristics is called numerical
simulation.
There now exist a new growing body of
knowledge connected with the development of
mathematical models and use of numerical
simulations of physical systems, and it is known as
computational mechanics.
Introduction – Continued…
Any numerical simulation, such as the one by the
Finite Element Method (FEM), is not an end in itself
but rather an aid to analysis, design, and
manufacturing.

Why any engineer or a scientist should study a

numerical method, especially the Finite Element
Method?
Introduction – Continued…
1. Most practical problems involve complicated
domains that forbid the development of
analytical solutions → approximate solutions.
2. A numerical method can be used to investigate
the effects of various parameters ↔ multitude
physical experiments.
Introduction – Continued…
3. Because of the power of numerical methods and
computers, it is possible to include all relevant
features in a mathematical model of a physical
process without worrying about its solution by
exact means.
Introduction – Continued…
4. The correct use of a computer program requires
a deep understanding of the problem to be
analyzed and it may be found that it is difficult
to interpret or explain computer-generated
results or even to develop proper input data.
5. The Finite Element Method and its
generalizations are the most powerful computer
oriented methods ever devised to analyze
practical engineering problems.
Introduction – Mathematical Models
A mathematical model can be broadly defined as a
set of equations that expresses the essential features
of a physical system in terms of variables that
describe the system.
The mathematical models of physical phenomena
are often based on fundamental scientific laws of
physics such as the principle of conservation of
mass, conservation of linear momentum, and
conservation of energy.
Introduction – Mathematical Models
Example 1: A Dynamic Problem
A simple pendulum consist of a bob of mass
m (kg) attached to one end of a rod of length l
(m) and the other end is pivoted to a fixed
point O.
In order to derive the governing equation of the problem,
we must make certain assumptions concerning the system
consistent with the goal of the analysis.
Introduction – Mathematical Models
Example 1: A Dynamic Problem
Introduction – Mathematical Models
for linear motion of the pendulum, we assume
1. The bob as well as the rod are rigid.

2. The rod is massless compared to the bob.

3. No friction at the pivot point O.

4. The resistance of the surrounding medium

is negligible.
Introduction – Mathematical Models
Using the principle of conservation of linear
momentum ( Newton’s second law)
The vector sum of externally applied forces
on a system is equal to the time rate of change
of the linear momentum (mass times velocity)
of the system
Introduction – Mathematical Models
 d  
F = ( mv ) = ma
dt
dvx
Fx = m
dt

d 2 d 2 g
− mg sin  = ml 2 or 2
+ sin  = 0
dt dt l
Introduction – Mathematical Models
The linear version of the equation takes the form

d 2 g
2
+ =0 (4)
dt l
The following boundary conditions are required
for the solution of equations (3) and (4)

dθ
(0) = 0 , = v0 (5)
dt
Introduction – Mathematical Models
The problem described by Eqs. (4) and (5) is
called an initial-value problem and has the following
analytical solution
(t ) = A sin t + B cos t (6)

with
g v0
= , A= , B = 0 (7)
l 
Introduction – Mathematical Models
The solution to the linear problem is

v0
(t ) = sin t + 0 cos t (8)

For zero initial velocity and nonzero initial position

(t ) = 0 cos t (9)

which represent a simple harmonic motion
Introduction – Mathematical Models
If we were to solve the nonlinear equation (3)
subject to the conditions (5), we may consider using
a numerical method because it is not possible to
solve it exactly for large values of .
Introduction – Mathematical Models
Example 2: A Solid Mechanics Problem
This example introduces the mathematical
formulation of the axial deformation of a bar
of variable cross section
The term bar is used in solid and structural
mechanics for a structural element that carries
only axial loads
Introduction – Mathematical Models
Example 3: A Solid Mechanics Problem
Introduction – Mathematical Models
We assume:
1. One dimensional case

2. Static loading

3. Presence of body forces

Using Newton’s second law and a uni-axial

stress-stain (constitutive) relation
Introduction – Mathematical Models
Newton’s second law F = 0 in the x direction
− [ A x ]x + [ A x ]x + x + gAx = 0 (23)
where g(x) is the weight per unit volume.
Dividing by x and taking the limit x→0
[ A x ]x [ A x ]x +x gAx
lim − + + =0
x →0 x x x
d
( A x ) + Ag ( x) = 0 (24)
dx
Introduction – Mathematical Models
Eq. (24) represent the equilibrium in the x-
direction, using Hook’s law
du
 x = E x , x = (25)
dx
And substituting (25) in (24), we obtain
d  du 
 EA  + gA = 0, 0 xL (26)
dx  dx 
Introduction – Mathematical Models
Eq. (26) is a second-order differential
equation that can be solved using known
boundary conditions at x = 0 and x = L.
The boundary conditions of a bar involve
specifying either displacement u or the force
Ax at a boundary point.
The known boundary conditions for the given
problem are
Introduction – Mathematical Models
 du 
u ( L) = 0,  EA dx  = − p (27)
  x =0
Equations (26) and (27) may not admit an
analytical solution when a = a(x), requiring us
to seek approximate solutions using numerical
methods.
For the special case in which EA  a and f  gA
Introduction – Mathematical Models
The general solution of (26) in this case is
1 f 2 
u ( x) =  − x + C1 x + C2  (28)
a 2 
Using the prescribed boundary conditions, the
solution becomes
1 g 2 P 
u ( x) =  ( L − x ) + ( L − x) 
2
(29)
E 2 A 
Introduction – Numerical simulations
Numerical simulations
While the derivation of the governing
equations for most problems is not extremely
difficult, their solution by exact methods of
analysis is often difficult due to geometric and
material complexities
In such cases, numerical methods of
analysis provide alternative means of finding
solutions
Introduction – Numerical simulations
By numerical simulation of a process, we mean
the solution of the governing equations (or
mathematical model) of the process using a
numerical method and a computer
Numerical methods typically transform
differential equations governing a continuum to a set
of algebraic equations of a discrete model of the
continuum that are to be solved using computers
Introduction – Numerical simulations
Problem Classification, Modeling, & Discretization
1. Classification
◼ What are the more important physical phenomena involved?
◼ Is the problem time-independent or time-dependent (is the
problem static or dynamic)?
◼ Is nonlinearity involved? What results are sought from
analysis?
◼ Etc., ……..
Introduction – Numerical simulations
Problem Classification, Modeling, & Discretization
2. Modeling
◼ A model is a simplified and idealized representation of the
actual physical problem.
◼ The geometric model of the physical problem becomes a
mathematical model when its behavior is described, or
approximated, in mathematical terms.
◼ The mathematical model may incorporate restrictions such as
homogeneity, isotropy, constancy of material properties, and
smallness of strains and rotations.
Introduction – Numerical simulations
Problem Classification, Modeling, & Discretization
3. Discretization
◼ In FEM, the mathematical model is discretized by dividing
it into a mesh of finite elements.
◼ The continuous field of the model is represented by
piecewise continuous fields over each element.
◼ Each piecewise continuous field defined by a finite number
of nodal quantities and simple interpolation within each
element.
Introduction – Numerical simulations
Modeling and FE analysis of a tapered support post
Introduction – Numerical simulations
Example: Elements, Interpolation, Nodes, and DOF
1. Elements, Nodes, and DOF
 Introduction – Numerical simulations
Elements, Interpolation, Nodes, and DOF
2. Interpolation
u = f1u2 + f 2u3
 x x
u =  1 −  u2 + u3
 L L
u =  1u2 +  2u3

Linear Shape Functions

Introduction – Numerical simulations
Elements, Interpolation, Nodes, and DOF
3. Displacements
u1 = 0,
PL
u2 = u1 + ,
6 EA
PL
u3 = u2 + ,
4 EA
PL
u4 = u3 +
2 EA
Introduction – Numerical simulations
Elements, Interpolation, Nodes, and DOF
4. Stresses

u2 − u1
 1−2 = E ,
L
u3 − u2
 2−3 = E ,
L
u4 − u3
 3−4 = E
L
Introduction – Numerical simulations
There exist a number of numerical methods, many of
which are developed to solve differential equations
1. Finite difference method
In the finite difference approximation of a
differential equation, the derivatives in the latter are
replaced by difference quotients that involve the
values of the solution at discrete mesh points of the
domain. The resulting algebraic equations are solved
for the values of the solution at the mesh points after
imposing the boundary conditions
Introduction – Numerical simulations
2. Classical variational method
In the solution of a differential equation by a
classical variational method, the equation is put into
an equivalent weighted-integral form and then the
approximate solution over the domain is assumed to
be a linear combination (j cj j) of appropriately
chosen approximation functions j and undetermined
coefficients cj.
The coefficients cj are determined such that the
integral statement equivalent to the original
differential equation is satisfied
Introduction – Numerical simulations
Various variational methods , e. g., the Ritz, Galerkin,
collocation, and least-squares methods, differ from each other
in the choice of the integral form, weight functions, and/or
approximation functions
The classical variational methods, which are meshless
methods, are powerful methods that provide globally
continuous solutions but suffer from the disadvantage that the
approximation functions for problems with arbitrary domains
are difficult to construct (the modern meshless methods seem
to provide a way to construct approximation functions for
arbitrary domains)
Introduction – Finite Element Method
3. The Finite Element Method
1. The Basic Idea
The finite element method is a numerical
method that is general and powerful in its
application to real-world problems that
involve complicated physics, geometry,
and/or boundary conditions
Introduction – Finite Element Method
In finite element method, a given domain is
viewed as a collection of subdomains, and over each
subdomain the governing equation is approximated
by any of the traditional variational methods
The main reason behind seeking approximate
solution on a collection of subdomains is the fact that
it is easier to represent a complicated function as a
collection of simple polynomials
Introduction – Finite Element Method

Piecewise approximation of a function

Introduction – Finite Element Method
2. The Basic Features
The method is endowed with three distinct
features that account for its superiority over other
competing methods.
A. Geometrically complex domains are represented as a
collection of geometrically simple subdomains, called
finite elements. Each finite element e is viewed as
independent domain by it self. The “domain” refers to
the geometric region over which the equations are
solved.
Introduction – Finite Element Method

Representation of a 2-D domain by a collection of

triangles and quadrilaterals
Introduction – Finite Element Method
B. Over each element, algebraic equations among
the quantities of interest are developed using the
governing equations of the problem
C. The relations from all elements are assembled (
elements are put back into their original position
of the whole domain) using certain interelement
relationship.
Introduction – Finite Element Method
In finite element, approximations enter engineering
analyses at several stages:
1. The division of the whole domain into finite
elements may not be exact (i.e., the assemblage
of elements, h, does not match the original
domain ), introducing error in the domain
being modeled
Introduction – Finite Element Method
2. When element equations are derived. Typically,
the dependent unknowns (u) of the problem are
approximated using the basic idea that any
continuous function can be represented by a
linear combination of known functions j and
undetermined coefficients cj (u ≈ uh = j cj j).
Algebraic equations among the undetermined
coefficients cj are obtained by satisfying the
governing equations, in a weighted-integral
sense, over each element.
Introduction – Finite Element Method
2. (continued….)

Linear interpolation functions and linear FE approximation

Introduction – Finite Element Method
2. (continued…) The approximation functions j are often
taken to be polynomials, and they are derived using
concepts from interpolation theory. Therefore, they are
called interpolation functions (shape functions). Thus ,
approximation errors in the second stage are introduced
both in representing the solution (u) as well as in
evaluating the integrals.
3. Finally, errors are introduced in solving the assembled
system of equations. Obviously, some of the mentioned
errors can be zero. When all the errors are zero, we
obtain the exact solution of the problem (which is not the
case in most problems)
Introduction – Finite Element Method

Approximation of a curved surface by a planar surface

Introduction – Summary
Scientists and engineers develop conceptual and mathematical
models of phenomena and systems that they wish to
understand. The understanding may be used to develop
and improve systems that contribute to the good being of
human kind.
Mathematical models are develop using axioms or laws of
nature that govern the phenomena. Mathematical models
consist of algebraic, differential, and/or integral
equations, which are often difficult to solve for desired
variables of the system for variety of input
Introduction – Summary
(called data), necessitating the use of numerical
simulations of the phenomena
In a numerical simulation of physical processes, we employ a
numerical method and computer to evaluate the
mathematical model of the process. The finite element
method is a powerful numerical method for solving
algebraic, differential, and integral equations, and it is
devised to study complex physical processes. The method
is characterized by three features:
Introduction – Summary
1. The domain of the problem is represented by a
collection of simple subdomains, called finite
elements. The collection of the finite elements is
called the finite element mesh
2. Over each finite element, the physical process is
approximated by functions of the desired type
(polynomials or otherwise), and algebraic
equations relating physical quantities at selective
points, called nodes, of the element are
developed
Introduction – Summary
3. The element equations are assembled using continuity
and/or “balance” of physical quantities
In the finite element method, we seek an approximation uh of
u in the form
u ≈ uh = j=1,..,n uj j + j=1,…,m cj j

where uj are the values of uh at the element nodes, j are the

interpolating functions, cj are coefficients that are not
associated with nodes, and j are associated approximation
functions. Direct substitution of such approximations onto the
governing differential equations does not always result, for an
arbitrary choice of the data of the problem, in a necessary and
Introduction – Summary
sufficient number of equations for undetermined coefficients
uj and cj. Therefore, a procedure here is required to obtain the
necessary and sufficient number of equations is needed. One
such procedure is provided by a weighted-integral form of the
governing differential equation.

Although the Ritz-Galerkin method and polynomial

approximations are used frequently to generate the finite
element equations, any appropriate method and
approximations can be used, in principle, to generate the
algebraic equations.
Introduction – Summary
In this spirit, the collocation method, subdomain
method, boundary integral methods, etc., can be used
to generate the algebraic equations among discrete
values of the primary and secondary variables.
Introduction – Mathematical Models
Example 3: A Heat Transfer Problem
We will try to derive the governing
equations of steady-state heat transfer through
a cylindrical bar of non-uniform cross section.

The bar is subject to a known temperature T0 (oC)

at the left end and exposed, both on the surface and
at the right end , to a medium at temperature T∞
Introduction – Mathematical Models
Example 2: A Heat Transfer Problem
Introduction – Mathematical Models
We assume:
1. Temperature is uniform at any section of
the bar T = T(x)
2. There is convective heat transfer across
the surface of the body and at the right end
(due to the difference between the
temperature of the bar and the surrounding
medium).
Introduction – Mathematical Models
Using the principle of conservation of energy
(or the second law of thermodynamics)
The principle requires that the rate of
change (increase) of internal energy is equal
to the sum of heat gained by conduction,
convection, and internal heat generation
(radiation not included).
For a steady process, the time rate of internal energy is zero.
Introduction – Mathematical Models
The energy balance of a volume element of
length x and a cross section A(x) (m2) normal
to the axis gives

[ Aq ]x − [ Aq ]x +x + Px(T − T ) + gAx = 0 (10)

q = heat flux ( heat flow per unit area W/ m2)

Introduction – Mathematical Models
 = film conductance (formed between the body and
the medium around [W/(m2. oC)]

P = perimeter (m)

Dividing throughout by x,

[ Aq ] x + x − [ Aq ] x
− + P(T − T ) + gA = 0
x
Introduction – Mathematical Models
And taking the limit x → 0, we obtain
d
− ( Aq ) + P(T − T ) + Ag = 0 (11)
dx
Relating the heat flux q (W/ m2 ) to the temperature
gradient (Fourier Law)
dT
q( x ) = − k (12)
dx
Introduction – Mathematical Models
Substitute (12) in (11), we arrive at the heat
conduction equation
d dT
( kA ) + P(T − T ) + gA = 0 (13)
dx dx
or
d dT
− ( kA ) + P(T − T ) = gA (14)
dx dx
which is a linear, second-order differential equation
Introduction – Mathematical Models
The boundary conditions at the ends of the bar are
 dT 
T (0) = T0 , kA dx ) + P(T − T ) = 0
(15)
x=L
The second condition represents the balance of the
heat due to conduction [kA(dT/dx)] and convection
[A(T- T∞)]
Introduction – Mathematical Models
The problem described in (14) and (15) is called a
boundary-value problem
If we assign:
  T- T∞ , a  kA [W.m/ oC],
c  bP [W/(m. oC)], f  Ag (W/m) (16)
Equations (14) and (15) can be written as
d  d 
− a  + c = f (17)
dx  dx 
Introduction – Mathematical Models
 d 
 (0) = T0 − T , a dx + A  = 0
(18)
x=L

If a and c are constant (homogeneous bar of constant

cross section) and g = 0 (no internal heat generation),
Eqs. (17) and (18) become
d 2 c
− 2  = 0, where 0  x  L (19)
dx a
 d  
 (0) = T0 − T  0 ,  dx + k   = 0 (20)
x=L
 Introduction – Mathematical Models
The general solution of (19)
c P
 ( x ) = C1 cosh mx + C2 sinh mx , m  = (21)
−x
a − x kA
e −ex x
e +e
where sinh x = and cosh x =
2 2
 cosh m( L − x ) + (  / mk ) sinh m( L − x ) 
 ( x ) =  (0)   (22)
 cosh mL + (  / mk ) sinh mL 