1
MATHEMATICAL
MODELLING
Dr. E.O. Diemuodeke
Department of Mechanical Engineering
University of Port Harcourt
Consult:
Oko, C.O.C Mathematical Modelling and Operations
Research
2
Introduction
Model is an approximate representation of a given part of
the material world.
Modelling is the process of model building and analysis
that usual culminates in a valid decision making in respect
of the modelled entity.
Physics of the material world and mathematical analysis
are involved in modelling.
3
Three Main Types of Engineering Model
• Mathematical model: Is a description of a physical system
using mathematical concepts and language. The process
of developing a mathematical model is termed
mathematical modeling.
• Physical model: Is a smaller or larger physical copy of the
physical system. The object being modelled may be small
(for example, an atom) or large (for example, the Solar
System).
• Computer model: Is a computer program that is designed
to simulate a physical system.
4
Physical System
Physical system is the part of the material world with
physical entity isolated for study; major attribute is cause-
and-effect relationship.
Cause-and-Effect Relationship
E S R
Problem Classification
Given: E and S to determine R => analysis problem
Given: E and R to determine S => synthesis problem
Given: S and R to determine E => control problem
5
Mathematical Modelling Sequence
• Identification of problem
• Understanding the physics of the problem
• Translating the physics into mathematical form
• Approximation of the mathematics
• Mathematical analysis
• Solution generation
• Verification
• Validation
• Simulation
• Interpretation
• Monitoring and maintenance
6
Modelling Methodology
Separability: It involves isolation of the physical
system from the surrounding with boundary.
Selectivity: This involves the selection of
significant parameters that describe the physical
system.
Causality: It involves the identification of the
cause-and-effect relationship between input and
output. Input and output are fundamentally related.
7
Information for Building MM
• Physical laws governing the physical system
(deduction)
• Structure of the physical system (boundary
conditions imposed on the system)
• Parameters influencing the physical system —
Experimental data from constructed physical
observation of the physical system (induction)
8
Three Classes of MM
1) Distributed Parameter (or micro) models:
Relevant independent variables are maintained
in continuous form (i.e. PDE).
2) Lump Parameter (or macro) models: All space
variables are discretised, whereas time
variable appears continuous (ODE).
3) Discretise Time (or meta) models: All space
and time variables are discretised; they usually
yield algebraic equations (normally
encountered in operations research).
9
Building Mathematical Model
𝑣
Conservation of mass
y
𝑢
𝑤
dy
In: 𝜌𝑢 𝑑𝑦𝑑𝑧 𝜕 𝜌𝑢
Out: 𝜌𝑢 + 𝑑𝑥 𝑑𝑦𝑑𝑧
𝜕𝑥
x
dz
dx
z
10
Same for y and z directions; hence, net mass flow out of CV must
equal time rate of decrease of mass in the CV.
𝜕 𝜌𝑢 𝜕 𝜌𝑣 𝜕 𝜌𝑤 𝜕𝜌
+ + 𝑑𝑥𝑑𝑦𝑑𝑧 = − 𝑑𝑥𝑑𝑦𝑑𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑡
Therefore, the conservation of mass can be expressed
mathematically
𝜕𝜌 𝜕 𝜌𝑢 𝜕 𝜌𝑣 𝜕 𝜌𝑤
+ + + =0
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
This is also known as the CONTINUITY equation
11
Momentum Equation
Newton 2nd law of motion: if the momentum of a
fluid changes it must be a result of a net force
acting on the fluid.
12
13
Doing same for y and z directions, considering the CONTINUITY equation,
thus,
𝜕(𝜌𝑢) 𝜕 𝜌𝑢𝑢 𝜕 𝜌𝑣𝑢 𝜕 𝜌𝑤𝑢 𝜕𝑝 𝜕𝜏𝑥𝑥 𝜕𝜏𝑦𝑥 𝜕𝜏𝑧𝑥
+ + + =− + + + + 𝜌𝑓𝑥
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕(𝜌𝑣) 𝜕 𝜌𝑢𝑣 𝜕 𝜌𝑣𝑣 𝜕 𝜌𝑤𝑣 𝜕𝑝 𝜕𝜏𝑥𝑦 𝜕𝜏𝑦𝑦 𝜕𝜏𝑧𝑦
+ + + =− + + + + 𝜌𝑓𝑦
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑦 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕(𝜌𝑤) 𝜕 𝜌𝑢𝑤 𝜕 𝜌𝑣𝑤 𝜕 𝜌𝑤𝑤 𝜕𝑝 𝜕𝜏𝑥𝑧 𝜕𝜏𝑦𝑧 𝜕𝜏𝑧𝑧
+ + + =− + + + + 𝜌𝑓𝑧
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧
convective change in
Viscous
momentum
net change body
local time
rate of pressure force
change of force
momentum These are the NAVIER-STOKES EQUATIONS
They are a VECTOR equation for the three components of
momentum
They are in CONSERVATION form
14
Classification of PDEs
General PDE
𝜕2𝑢 𝜕2𝑢 𝜕2𝑢 𝜕𝑢 𝜕𝑢
𝐴 2 + 2B +C 2 =𝐷 +𝐸 + 𝐹𝑢 + 𝐺
𝜕𝑥 𝜕𝑥𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑦
x and y may be both space coordinates, or space and time. The relative
magnitudes of A, B, and C determine the nature of the equation
Let the discriminant be defined as ∆≡ 𝐵2 − 𝐴𝐶
o If 𝐴𝐶 > 𝐵2 (∆< 0): elliptic PDE
o If 𝐴𝐶 + 𝐵2 (∆= 0): parabolic PDE
o If 𝐴𝐶 < 𝐵2 (∆> 0): hyperbolic PDE
15
Uniqueness (or Boundary) Conditions
First kind boundary condition (or the Dirichlet B.C): In this kind
of B.C the potential is specified for the boundary surface.
Namely 𝑢 = 𝜑𝑖 𝑥, 𝑦, 𝑧, 𝑡 ;on 𝑆𝑖 , 𝑖 = 1,2, … , 𝑛
Second kind boundary condition (or the Neuman B.C): The
normal derivative of the potential (flux) is prescribed at the
boundary.
𝜕𝑢
Namely, = 𝜑𝑖 𝑟, 𝑡 ; 𝑟 ⊂ 𝑆𝑖
𝜕𝑛
Third kind boundary condition (or Mixed B.C or Dirichlet-
Neuman B.C): Here the flux and potential are linked and
prescribed at the surface.
𝜕𝑢
Namely, 𝛼𝑖 + 𝛽𝑖 𝑢 = 𝜑𝑖 𝑟, 𝑡 ; 𝑟 ⊂ 𝑆𝑖
𝜕𝑛
16
Types of PDE
• Homogenous, e.g. 𝛻 2𝑇 = 𝛼1 𝜕𝑇
𝜕𝑡
• Non-homogeneous, e.g. 𝛻 2
𝑇+
𝑑
𝑘
𝑔(𝑟, 𝑡) =
1 𝜕𝑇
𝛼 𝜕𝑡
𝜕2 𝑇
• Linear, e.g. 𝜕𝑥 2
=
1 𝜕𝑇
𝛼 𝜕𝑡
• Non-linear, e.g. 𝜕
𝜕𝑥
𝑘(𝑇)
𝜕
𝜕𝑥
𝑇= 𝜌𝑐
𝜕𝑇
𝜕𝑡
17
Linear Operator
𝜕 𝜕2 𝜕2 𝜕2 𝜕𝑢 𝜕2 𝑢 𝜕2 𝑢 𝜕2 𝑢
∟≡ + + + ; ∟[𝑢] ≡ + + + 2
𝜕𝑡 𝜕𝑥 2 𝜕𝑥𝜕𝑦 𝜕𝑦 2 𝜕𝑡 𝜕𝑥 2 𝜕𝑥𝜕𝑦 𝜕𝑦
Linear operator satisfies the following relations:
∟ 𝑎𝑢 = 𝑎∟[𝑢]
∟ 𝑢+𝑣 =∟ 𝑢 +∟ 𝑣
∟ 𝑎𝑢 + 𝑏𝑣 = 𝑎∟ 𝑢 + 𝑏∟ 𝑣
Example
𝜕2 𝑇 1 𝜕𝑇
= , let 𝑇1 and 𝑇2 be solutions. What of 𝑇3 ?
𝜕𝑥 2 𝛼 𝜕𝑡
18
Mathematical Analysis Schemes
Mathematical Analysis
Numerical Analytical
Examples Examples
Finite difference method Separation of variables
Finite element method Perturbation method
Finite volume method Superposition principle
Volume of fluid method Integral transform
Method of lines Change of variables
19
Method of Separation of Variable
Examples 1
𝜕2 𝑇 1 𝜕𝑇
Separate = (1)
𝜕𝑥 2 𝛼 𝜕𝑡
Let 𝑇(𝑥, 𝑡) = 𝜃(𝑥)𝜗(𝑡), therefore, Eq.(1) becomes
𝑑2 𝜃 𝜃(𝑥) 𝑑𝜗 1 𝑑2 𝜃 1 𝑑𝜗
𝜗(𝑡) 2 = ⟹ = (2)
𝑑𝑥 𝛼 𝑑𝑡 𝜃(𝑥) 𝑑𝑥 2 𝛼𝜗(𝑡) 𝑑𝑡
1 𝑑2 𝜃 1 𝑑𝜗
Let = = −𝛽2 (3)
𝜃(𝑥) 𝑑𝑥 2 𝛼𝜗(𝑡) 𝑑𝑡
𝑑𝜗 𝑑2 𝜃
Thus + 𝛼𝛽2 𝜗(𝑡) = 0 and + 𝛼𝛽2 𝜃 𝑥 = 0
𝑑𝑡 𝑑𝑥 2
20
Examples 2
𝜕2 𝑇 𝜕2 𝑇 1 𝜕𝑇
+ = (1)
𝜕𝑥 2 𝜕𝑦 2 𝛼 𝜕𝑡
Let 𝑇(𝑥, 𝑦, 𝑡) = 𝜃(𝑥)𝜑(𝑦)𝜗(𝑡), therefore, Eq.(1) becomes
𝑑2 𝜃 𝑑2 𝜑 𝜃(𝑥)𝜑(𝑦) 𝑑𝜗
𝜗 𝑡 𝜑 𝑦 +𝜗 𝑡 𝜃(𝑥) 2 = (2)
𝑑𝑥 2 𝑑𝑦 𝛼 𝑑𝑡
Divide through Eq. (2) by 𝜃(𝑥)𝜑(𝑦)𝜗(𝑡) to obtain
1 𝑑2 𝜃 1 𝑑2 𝜑 1 𝑑𝜗 2
+ = = −𝛽 (3)
𝜃(𝑥) 𝑑𝑥 2 𝜑(𝑦) 𝑑𝑦 2 𝛼𝜗(𝑡) 𝑑𝑡
𝑑𝜗 2 1 𝑑2 𝜃 2 1 𝑑2 𝜑
Thus + 𝛼𝛽 𝜗(𝑡) = 0; hence = −𝛾 and = −𝜇2
𝑑𝑡 𝜃 𝑑𝑥 2 𝜑 𝑑𝑦 2
2 2 2
where 𝛾 + 𝜇 = 𝛽
𝑑2 𝜃 2 𝑑2 𝜑
⇒ and + 𝛾 𝜃 𝑥 = 0 and + 𝜇2 𝜑 𝑦 = 0
𝑑𝑥 2 𝑑𝑦 2
