Mathematical Modeling

IS6301

Acknowledgement: These lecture notes were prepared with reference to the lecture notes of
Prof. J. M. R. S. Appuhamy
Course Overview OUTLINE

Introduction to Mathematical Modeling Ordinary Differential Equations (ODEs)

01 04
What, why, and how of modeling Dynamic systems in time

Linear Programming Partial Differential Equations (PDEs)

02 05
Optimization with constraints Spatial and temporal phenomena

Probabilistic Models Qualitative Theory of Dynamical Systems

03 06
Uncertainty and randomness Equilibria, stability, bifurcations
Introduction to Mathematical
Modeling
From physical reality to mathematical equations
Lecture Roadmap OVERVIEW

This lecture is structured in five parts. Each builds on the previous.

What is a Mathematical Model?

A
Definition, purpose, and the three types of model (empirical, mechanistic, hybrid).

The 7-Step Modelling Process

B
A structured workflow from identifying the problem to presenting results.

From Words to Equations

C
How verbal physical statements translate directly into mathematical notation.

Engineering Model Types with Examples

D
Algebraic, differential, statistical, and optimisation models — worked examples.

Model Validity & Effectiveness

E
Checking dimensions, limits, agreement with data, and the simplicity trade-off.
P A R T A

What is a Mathematical Model?

Definition · Purpose · Classification

Defining Mathematical Modelling PART A

Mathematical modelling is the practice of translating physical (engineering) situations into precise mathematical forms
that can be analysed, solved, and interpreted.

A model can take several forms:

Empirical Formulae Algebraic Equations

Ready-made formulae from handbooks and standards. Equations and formulae from textbooks encoding known physical laws.
Example: Manning's equation for open-channel flow: Q = Example: Ohm's Law V = IR, or beam bending σ = My/I
(1/n)·A·R^(2/3)·S^(1/2) Valid within the assumptions of the underlying theory.
Derived by fitting equations to large experimental datasets.

Differential & Integral Equations Numerical / Computational Models

Equations relating rates of change — used when a quantity evolves over When closed-form solutions are not possible, discretise and solve
time or space. numerically.
Example: Newton's cooling dT/dt = −k(T − T∞) Example: Finite Element Method (FEM) for stress analysis, Finite
Require initial or boundary conditions to be fully determined. Difference for heat conduction.
Powerful but require careful validation.
Examples of Real Problems PART A

Problem 1 Why Mathematics is the key

A road bridge must carry vehicles of varying The beam obeys Newton's laws and material
weights at speed. Engineers need to know the constitutive relations exactly. These translate
maximum deflection at mid-span and the into a 2nd-order ODE (the Euler-Bernoulli
maximum bending stress in the beam before beam equation) whose solution gives
the bridge is built — to ensure it will not fail or deflection at every point along the span —
cause discomfort to drivers. something no physical intuition alone can
provide.

Mathematical Model Without Maths

𝑑2 𝑦
Build it, test it, rebuild if it fails —
𝐸𝐼. 4 = 𝑤(𝑥) catastrophically expensive and
𝑑𝑥
dangerous.
Examples of Real Problems PART A

Problem 2 Why Mathematics is the key

A new infectious disease enters a city of Individual human behaviour is unpredictable,

500,000 people. Health authorities need to but population-level rates obey consistent
know: how fast will it spread? How many proportionality laws. The mathematics model
people will be simultaneously ill (peak hospital captures these as a system of ODEs — turning
demand)? Will it die out on its own, or will an impossibly complex social problem into a
vaccination be needed — and how many tractable mathematical one whose solution
doses?. gives actionable numbers.

Mathematical Model Without Maths

𝑑𝑠 𝑆. 𝐼 react as it happens — always behind the

= −𝛽 curve, resources misallocated.
𝑑𝑡 𝑁
𝑑𝐼 𝑆. 𝐼
= −𝛽 − 𝑌. 𝐼
𝑑𝑇 𝑁
Why Do We Build Mathematical Models? PART A

ADDED VALUE — WHAT MATHEMATICS GIVES US THAT NOTHING ELSE CAN

Predict behaviour Design & optimise Understand mechanisms

Calculate how a bridge will Find the dimensions or

Reveal which physical processes
deflect under load before it is parameters that give the best
dominate in a given situation.
built. performance.

Safety & risk analysis Reduce cost & time Scale and extrapolate

Replace expensive physical tests

Predict failure modes, safety Extend small-scale lab results to
with virtual (computer)
factors, and extreme scenarios. full-scale engineering systems.
experiments.
Three Fundamental Types of Mathematical Model PART A

Models are classified by how they are derived — from data, from physics, or a combination of both.

Empirical (Data-Driven) Model Mechanistic (Physics-Based) Model Hybrid / Grey-Box Model

How derived: How derived: How derived:

Fitted to measured data. No physical law Derived from first principles — conservation Physical structure with data-fitted parameters,
assumed. laws, constitutive relations, field equations. or physics as a constraint on learning.

Advantages: Advantages: Advantages:

• Quick to develop
• Extrapolates beyond training data • Best of both worlds
• Works when physics is unknown
• Reveals cause-and-effect • Physical constraints prevent overfitting
• Handles complex nonlinearity via regression /
• Parameters have physical meaning • Works with limited data
ML

Limitations: Limitations: Limitations:

• Only valid within data range • Requires deep physical understanding • More complex to set up
• No physical insight • Simplifying assumptions may break down • Requires partial physical knowledge
• Cannot extrapolate reliably • Can be analytically intractable • Validation is more involved

Examples: Examples: Examples:

Traffic flow model fitted to sensor data. Navier-Stokes equations for fluid flow. Physics-Informed Neural Networks (PINNs).
Soil bearing capacity from site tests. Finite-element structural analysis. Semi-empirical turbulence models.
Neural-network fault detection in motors. Kirchhoff's laws in circuit design. Digital twins in manufacturing.
P A R T B

The Process of Mathematical

Modelling
The Mathematical Modelling Cycle PART A

Modelling is an iterative cycle, not a one-way process. We continually refine the model as we learn more.

Real-World Mathematical Analysis &

Problem Formulation Solution

Validation Interpretation
against Data & Prediction

Mathematical
Analysis & Solution Interpretation
Real-World Problem Formulation
Validation
Which equations encode
What quantity do I want to those laws? What does x(t) = … mean Do predictions match
predict? Can I solve analytically? physically? measured data?
What assumptions simplify
What are the relevant Do I need a numerical Does the solution Are units, signs, and
the problem?
variables? method? behaviour match intuition? magnitudes correct?
What are the initial /
What physical laws apply? What are the key What are the limiting Is the model the simplest
boundary conditions?
mathematical techniques? cases? that fits?
Overview: The 7-Step Modelling Process PART B

Regardless of the engineering domain, a rigorous modelling study always follows these seven steps.

Identify the real problem

1
Understand what is actually being asked; identify problem variables.

Construct relationships between variables

2
Use physical laws, conservation principles, or observations.

Create a model describing the problem

3
Choose the simplest equation or system that captures the physics.

Obtain the mathematical solution

4
Apply analytical or numerical techniques; write code if needed.

Interpret the mathematical solution

5
Translate the answer back into physical meaning and context.

Test validity & effectiveness

6
Compare with data; check signs, units, magnitude, and trends.

Present results / write a report

7
Communicate findings clearly, with appropriate level of detail.
P A R T B

Worked Examples
Worked Example 1

Problem: A radioactive substance initially contains 800 grams. The mass decays at a rate
proportional to the amount present. After 5 years, 200 grams remain. Find the amount remaining
after 8 years.
STEP 1 — Identify the Problem and Its Variables

y(t) Dependent variable t Independent variable

Time elapsed (years) since the start of observation.
Number of substance at time 𝑡
The substance decays as t increases.

k Growth rate constant y(0) = 800 Initial condition

Proportionality constant (year⁻¹). 800 substance present at t = 0.
Unknown — must be found from the given data point. This determines the unique solution.

y(5) = 200 Data condition What do we need to find?

200 grams after 5 years. The value of y(8) — the amount of substance at t = 8 years.
Used to calculate k (the growth rate). Two sub-problems: (i) find k, then (ii) evaluate y(8).
 Verbal law: "The growth rate of the colony is proportional to the number currently present.“
𝑑𝑦
∝𝑦
𝑑𝑡

𝑑𝑦 After 5 years remaining is 200g After 8 years amount is

= 𝑘𝑦
𝑑𝑡
i.e when 𝑡 = 5, 𝑦 = 200 𝑦 = 800𝑒 −0.277×8
ln 𝑦 = 𝑘𝑡 + ln 𝐴
𝑦 = 𝐴𝑒 𝑘𝑡 So that
200 = 800𝑒 5𝑘 = 87.23
𝑦(0) = 800
⇒ 𝐴 = 800 1
5𝑘 = ln
4 How long does it take for
So that 𝑦 = 800𝑒 𝑘𝑡
𝑘 = −0.277 the substance to reduce
to half of its initial
𝑦 = 800𝑒 −0.277𝑡 amount ?
Worked Example 2

Problem: A bacterial culture starts with 500 bacteria and grows at a rate proportional to its size.
After 3 hours there are 8 000 bacteria. Find the number of bacteria after 4 hours.

STEP 1 — Identify the Problem and Its Variables

y(t) Dependent variable t Independent variable

Number of bacteria in the culture at time t. Time elapsed (hours) since the start of observation.
This is the unknown function we want to find. The culture grows as t increases.

k Growth rate constant y(0) = 500 Initial condition

Proportionality constant (hr⁻¹). 500 bacteria present at t = 0.

Unknown — must be found from the given data point. This determines the unique solution.

y(3) = 8000 Data condition What do we need to find?

8 000 bacteria after 3 hours. The value of y(4) — the number of bacteria at t = 4 hours.
Used to calculate k (the growth rate). Two sub-problems: (i) find k, then (ii) evaluate y(4).
 𝑑𝑦 After 3 hours bacteria is 8000 After 4 hours bacteria is
∝𝑦
𝑑𝑡
i.e when 𝑡 = 3, 𝑦 = 8000 𝑦 = 500𝑒 0.9242×4
𝑑𝑦
= 𝑘𝑦
𝑑𝑡 So that
ln 𝑦 = 𝑘𝑡 + ln 𝐴 8000 = 500𝑒 3𝑘 = 20,159.03

𝑦 = 𝐴𝑒 𝑘𝑡 80
3𝑘 = ln
𝑦(0) = 500 5
𝐾 = 0.9242
⇒ 𝐴 = 500
So that 𝑦 = 500𝑒 𝑘𝑡 𝑦 = 500𝑒 0.9242𝑡
Worked Example 3

Problem: A cup of coffee is poured from a pot whose contents are at 950 𝐶 into a non-
insulate cup in a room at 200 𝐶. After a minute the coffee has cooled to 900 𝐶 . How
much time is required before the coffee reaches a drinkable temperature of 650 𝐶 .

𝑇 0 = 95
𝑇 1 = 90
𝑇(𝑡) Dependent variable
𝑇 𝑅 = 20
What do we need to find?
t Independent variable
𝑡 where 𝑇 𝑡 = 65
 Newton’s law of cooling
“The rate of change of temperature is proportional to the deference between the
temperature of the object and that of the surrounding environment’

𝑑𝑇(𝑡) After 1 minute Temperature When temperature is

∝ (𝑇 𝑡 − 𝑇(𝑅))
𝑑𝑡 is 900 𝐶 650 𝐶
𝑑𝑇
𝑑𝑡
= 𝑘(𝑇 − 𝑇𝑅 ) i.e when 𝑡 = 1, 𝑇 = 90 65 = 75𝑒 −0.06899𝑡 + 20
ln(𝑇 − 𝑇𝑅 ) = 𝑘𝑡 + ln 𝐴
So that 45
𝑇 − 𝑇𝑅 = 𝐴𝑒 𝑘𝑡 90 = 75𝑒 𝑘 + 20 −0.06899𝑡 = ln
75
𝑇 = 𝐴𝑒 𝑘𝑡 + 𝑇𝑅 70
𝑘 = ln
75
Applying the initial
= 7.4 (min)
conditions
𝑘 = −0.06899
𝑇 0 = 95, 𝑇𝑅 = 20
𝑻 = 𝟕𝟓𝒆𝒌𝒕 + 𝟐𝟎 𝑦 = 75𝑒 −0.06899𝑡 + 20
Excercises

Problem 01: A steel rod is heated to 300 °C in a furnace, then removed and placed in a
workshop at 25 °C. After 5 minutes it cools to 200 °C. How long does it take to reach a
safe handling temperature of 50 °C?.

𝑇 𝑡 = 275𝑒 𝑘𝑡 + 25

𝑘 = 0.09039

𝑡 = 26.5
Dependent variable

Independent variable
Modelling Components
Important aspects to the modelling process

The process of encapsulation into a mathematical model

i A real-world problem is translated into mathematical form.
• Identify the key variables and parameters involved
• Make reasonable assumptions to simplify the situation.
• Express relationships using equations, functions, or inequalities.

The qualitative descriptions of behavior

ii Understand what the model predicts about the system’s behaviour.
• Study trends such as growth/decay, stability, or oscillations
• Understand how variables influence each other.

Verification
iii Validate the model against reality and improve it if necessary.
• Compare model predictions with real-world data.
• Ensure the model produces reasonable and consistent results

E
General types of models

Deterministic
i
Models that do not involve random variables .
Ex. Models in calculus, differential equations and linear algebra all introduce these types of models

Stochastic
ii Models involve random variables
Ex. Models in probability, statistics and stochastic processes will introduce these models

iii

E
Construct Relationships Between Variables PART B · STEP 2

Source the governing physical relationship — it might come from:

Conservation Laws Empirical Physical Laws

Conservation of energy, mass, or momentum — Well-established relationships from experiment that
always valid regardless of material. have been formalised.
Example: Energy in = Energy out + Energy stored → Newton's Law of Cooling: 'The rate of heat loss is
basis of heat equation. proportional to the temperature excess over
surroundings.'
Constitutive Relations Geometric / Kinematic Constraints
Material-specific relationships between stress-strain, Constraints from the physical setup: compatibility,
flow-pressure, etc. equilibrium of forces, connectivity.
Example: Fourier's Law of conduction: q = −k∇T, Example: In a beam, curvature κ = M / (EI)
ideal law of gas PV=nRT
Dimensional Analysis

Checking the mathematical model is physically consistent

What is Dimensional Analysis?

Dimensional analysis is the process of checking that every term in a mathematical model has the same physical
dimensions.

Every physical quantity Both sides of an equation Arguments of functions

1 2 3
has a dimension must match must be dimensionless
Dimension -Nature of the physical If LHS has dimension [M L T⁻²], You can only write e^(x), ln(x), sin(x)
quantity. then RHS must also have dimension [M if x has no physical dimension.
Unit — standard measurement to L T⁻²]. Example: e^(kt) requires [k][t] = T⁻¹ · T
express the dimension . A mismatch is a definite error — no =1✓
Examples for diamentions: exceptions.
Mass [M], Length [L], Time [T],
Temperature [Θ], Current [I].

Units (kg, m, s …) are choices;

dimensions are fundamental.
 The 7 SI Base Dimensions and Key Derived Quantities

The 7 base dimensions — everything is built from these

M L T Θ I N J
Amount of Luminous
Mass Length Time Temperature Electric current
substance intensity
kilogram kg metre m second s kelvin K ampere A mole mol candela cd

Key derived quantities used in engineering models

Quantity Symbol Dimensions SI unit Quantity Symbol Dimensions SI unit

Velocity v L T⁻¹ m/s Force F M L T⁻² N

Acceleration a L T⁻² m/s² Pressure/Stress σ M L⁻¹ T⁻² Pa

Energy E M L² T⁻² J Power P M L² T⁻³ W

Density 𝜌 (Rho) M L⁻³ kg/m³ Damping coeff. c M T⁻¹ kg/s

Voltage V M L² T⁻³ I⁻¹ V Resistance R M L² T⁻³ I⁻² Ω

Bending moment M M L² T⁻² N·m 2nd moment area I L⁴ m⁴

The Four Rules of Dimensional Analysis

Equality

1 Both sides of any equation must share the same

dimension.
F = ma → [F] = [m][a] = M · LT⁻² = MLT⁻² ✓

Addition & Subtraction

2 Only quantities of the same dimension may be added or

subtracted.
V = IR + V₀ → all three terms must have dimension M L² T⁻³ I⁻¹ (volts)

Multiplication & Division

3 Dimensions combine algebraically — multiply and divide

like symbols.
[velocity] = [length] / [time] = L / T = LT⁻¹ → unit: m/s ✓

Transcendental Functions
e^(kt): [kt] = T⁻¹ · T = 1 (dimensionless) ✓
4 Arguments of e^(·), ln(·), sin(·), cos(·) must always be
dimensionless.
sin(ωt): [ωt] = T⁻¹ · T = 1 (dimensionless) ✓
Dimensional Check — Newton's Cooling & Exponential Models PART F

Apply dimensional checking to two ODEs already studied in this course:

Example 1 — Newton's cooling Example 2 — Exponential growth (bacteria)

dT/dt = −k (T − T∞) y(t) = y₀ · e^(kt)

LHS [dT/dt] = Θ · T⁻¹ (°C/s) Rule Exponent of e must be dimensionless

(T − T∞) temperature difference → [Θ] [k] T⁻¹ (growth rate, hr⁻¹)

k · (T − T∞) [k] · Θ must equal Θ T⁻¹ [t] T (time, hours)

∴ [k] = T⁻¹ (s⁻¹ or hr⁻¹) ✓ [kt] T⁻¹ · T = 1 (dimensionless) ✓

Check Θ T⁻¹ = Θ T⁻¹ ✓ Consistent [y₀ · e^(kt)] [y₀] × 1 = count ✓

✓ Both models pass dimensional analysis — every term is consistent and the exponent is dimensionless.
Dimensional Check — Mass-Spring-Damper & Catching an Error

Example 3 — All terms must have dimension of Force [M L T⁻²]

m·ẍ + c·ẋ + k·x = F(t)

m·ẍ c·ẋ k·x F(t)

M · LT⁻² = MLT⁻² [c]·LT⁻¹ → [c] = MT⁻¹ [k]·L → [k] = MT⁻² Applied force MLT⁻²

kg · m/s² = N kg/s · m/s = N N/m · m = N N (directly)

✓ All four terms share dimension [N = kg·m/s²] — consistent. Also confirms: c has units kg/s, k has units N/m.

Example 4 — Catch the error (deliberately wrong equation)

V = I · R + t ← is this valid?

I·R I × ML²T⁻³I⁻² = ML²T⁻³I⁻¹ t dimension T (time, seconds)

volts ✓ NOT volts ✗
Dimensional Analysis — Why It Matters & How to Apply It PART F

A practical 4-step procedure — apply to every model you write

1 Write the LHS dimension

Identify the dimension of the quantity on the left-hand side of the equation.

2 Expand each RHS term

Replace every symbol on the right-hand side with its fundamental dimensions.

3 Simplify and compare

Cancel dimensions algebraically. Check LHS = RHS. If not — there is an error.

4 Find unknown dimensions

If a parameter's units are unknown, dimensional matching reveals what they must be.

Why engineers perform dimensional analysis on every model

Catch errors instantly Reveal unknown units Scale & simplify models
Groups of dimensionless numbers
A wrong formula, unit conversion mistake, or If you derive k from an ODE, dimensional
(Buckingham π theorem) reduce complex
missing factor will fail dimensional check in analysis tells you immediately what units k
multi-variable problems to simpler universal
seconds — before any computation. must carry — you don't have to guess.
laws.
Definition

An equation is said to be dimensionally homogeneous if it does not depend on

fundamental unit. That is the units of LHS and RHS are same

4
Worked Example 4

Use the concept of dimensional analysis to show that

there are no small animals in the polar regions
The key physical constraint comes from how heat loss and heat generation scale with size
Let 𝐴 – Heat loss, 𝑉- Volume 𝐿- Length 𝑀-Mass

𝐴 ∝ 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑎𝑟𝑒𝑎
Then

𝐴 𝐿2
∝ 3
𝑇(𝑡) 𝑉 𝐿
Dependent variable
𝐴 1
∝
𝑉 𝐿
When 𝐿 is small the heat loss for a unit volume becomes high
Worked Example 5

Problem: Show that the volume 𝑉 of the crater (a cavity made by the impact of an
explosion) produced by a surface explosion of energy E in soil of density 𝜌 under gravity
𝑔 is defined by

3
𝐸 4
𝑉=𝑘
𝜌𝑔

𝑇(𝑡) Dependent variable

Worked Example 5

Assume
𝑉 ∝ 𝐸 𝛼 𝜌 𝛽 𝑔𝛾
Consider the dimensions

LHS - 𝐿3 RHS - 𝑀𝐿2 𝑇 −2 𝛼

𝑀𝐿−3 𝛽
𝐿𝑇 −2 𝛾
= 𝑀 𝛼+𝛽
𝐿 2𝛼−3𝛽+𝛾
𝑇 −2𝛼−2𝛾

By equating dimensions of LHS and RHS

3 3 3
𝛼= , 𝛽=− , 𝛾=−
4 4 4
𝑇(𝑡)
Then Dependent variable
3 3
𝐸 4 𝐸 4
𝑉∝ so that 𝑉 = 𝑘
𝜌𝑔 𝜌𝑔
Buckingham’s Pi Theorem
Buckingham’s Pi Theorem
Theorem:

If a physical problem involves 𝑛 variables and these variables contain 𝑘

independent fundamental dimensions (such as mass, length, time), then the
relationship among the variables can be reduced to:
𝑛 − 𝑘 dimensionless groups (called 𝜋 terms)
Buckingham’s Pi Theorem
In mathematical form

If
𝑓 𝑥1 , 𝑥2 , … , 𝑥𝑛 = 0
Then we can write
𝑔 Π1 , Π2 , … , Π𝑛−𝑘 = 0

or
Π𝑖 = 𝑔 Π1 , Π2 , … , Π𝑖−1 , Π𝑖+1 , Π𝑛−𝑘
Buckingham’s Pi Theorem
Procedure:

• Clearly define the problem and identify the important variables

𝑓 𝑥1 , 𝑥2 , … , 𝑥𝑛 = 0

• Express each of n variable in terms of its fundamental dimensions

(eg: M L² T⁻², M L² T⁻³ I⁻² )

• Determine the number of Pi groups, 𝑗 = 𝑛 − 𝑘

• From j dimensionless Pi groups, check that they are all indeed dimensionless

• Express the result in form

𝑔 𝛱1 , 𝛱2 , … , 𝛱𝑛−𝑘 = 0 or 𝛱𝑖 = 𝑔 𝛱1 , 𝛱2 , … , 𝛱𝑖−1 , 𝛱𝑖+1 , 𝛱𝑛−𝑘

Worked Example 5: Simple Pendulum Problem
The variable are
𝑚 − 𝑚𝑎𝑠𝑠 , 𝑙 − 𝑙𝑒𝑛𝑔𝑡ℎ , 𝑡 − 𝑡𝑖𝑚𝑒,
𝑔 − 𝑔𝑟𝑎𝑣𝑖𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑓𝑜𝑟𝑐𝑒

Number of variables 𝑛 = 4
Number of dimensions 𝑘 = 3 (M, 𝐿 and 𝑇)

By Buckingham’s Pi theorem, there are 4 −

3 = 1, Π groups
Worked Example 5: Simple Pendulum Problem
Π = 𝑚𝛼 𝑙 𝛽 𝑡 𝛾 𝑔𝛿 𝛼=0
𝛽=1
Let
Consider the dimensions 𝛾 = −2
𝑚 − 𝑀, 𝑙 − 𝐿, 𝑡 − 𝑇 , 𝑔− 𝐿𝑇 −2 𝛿 = −1
So that Then
𝛼 𝛽+𝛿 𝛾−2𝛿
Π=𝑀 𝐿 𝑡
Then 𝑙 𝛽 𝑡 𝛾 𝑔𝛿 = 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝛼=0
𝛽+𝛿 =0 𝑙1 𝑡 −2 𝑔−1 = 𝑘(= 𝐶 2 )
𝛾 − 2𝛿 = 0 𝑙
𝑡=𝐶
𝑔
Worked Example 6:
Use Buckingham's pi theorem to derive a formula for the gravitational force
𝐹 between two masses 𝑚₁ and 𝑚₂ separated by distance 𝑅.
Include Newton's gravitational constant 𝐺 since purely dimensional reasoning alone
requires it to balance units.

𝑚2 Variable Symbol Dimension

𝐹 Force F MLT⁻²
𝐹 Mass 1 m₁ M
𝑅 Mass 2 m₂ M
Separation R L
𝑚1
Gravitational constant G M⁻¹L³T⁻²
Worked Example 6:
Dimensions – M, L , T

𝑘=3

Variables
𝐹, 𝑚1 , 𝑚2 , 𝑅, 𝐺

𝑛=5
So that
𝑗 =5−3=2
Worked Example 6:
Choose repeat variables
The Rule
• The number of repeating variables must equal the number of fundamental
dimensions, k.
• Here, 𝑘 = 3 (𝑀, 𝐿, 𝑇), so we need exactly 3 repeating variables — which is why
𝑚₁, 𝑅, and 𝐺 were chosen.

• Why This Rule Exists

• The repeating variables form the dimensional basis of the problem. They need to:
Span all fundamental dimensions — together they must cover 𝑀, 𝐿, and 𝑇 completely
• Be dimensionally independent — no repeating variable's dimensions can be formed by
combining the others (analogous to linearly independent vectors in a basis)
• Not include the variable you're solving for — 𝐹 and 𝑚₂ were excluded for this reason
Form 1st Π term Let
Chosen 3 variables combine with 𝐹 𝛼=1
Π1 = 𝐹 𝛼 𝑚1 𝛽 𝑅𝛾 𝐺 𝛿
Then
𝛿 = −1
= 𝑀𝐿𝑇 −2 𝛼
𝑀 𝛽
𝐿 𝛾
𝑀−1 𝐿3 𝑇 −2 𝛿
𝛽 = −2
𝛾=2
= 𝑀𝛼+𝛽−𝛿 𝐿𝛼+𝛾+3𝛿 𝑇 −2𝛼−2𝛿
Then
𝛼+𝛽−𝛿 = 0 So that
𝐹𝑚1 −2 𝑅 2 𝐺 −1 = Π1
𝛼 + 𝛾 + 3𝛿 = 0
−2𝛼 − 2𝛿 = 0 𝐹𝑅 2
2 = Π1
𝑚1 𝐺
Form 2nd Π term Let
Chosen 3 variables combine 𝛼=1
with 𝑚2 Then
Π1 = 𝑚2 𝛼 𝑚1 𝛽 𝑅 𝛾 𝐺 𝛿 𝛿=0
𝛽 = −1
= 𝑀 𝛼 𝑀 𝛽 𝐿 𝛾 𝑀−1 𝐿3 𝑇 −2 𝛿 𝛾=0

= 𝑀𝛼+𝛽−𝛿 𝐿𝛾−3𝛿 𝑇 2𝛿 So that

Then 𝑚2 𝑚1 −1 = Π2
𝛼+𝛽−𝛿 =0
𝑚2
𝛾 − 3𝛿 = 0 = Π2
𝑚1
2𝛿 = 0
Worked Example 6:
Then we can write
𝐹𝑅 2 𝑚2
= 𝑓
𝑚1 2 𝐺 𝑚1

The simplest function can be written as

𝑚2 𝑚2
𝑓 =𝐶
𝑚1 𝑚1
Then
𝐹𝑅 2 𝑚2
=𝐶
𝑚1 2 𝐺 𝑚1
Experimental results show that 𝐶 = 1.
Then
𝐺𝑚1 𝑚2
𝐹=
𝑅2
Worked Example 7:
Using Buckingham Pi Theorem, determine the dimensionless
parameters governing the drag force acting on a sphere.
Drag force 𝐹 depends on Diamensions involved
• Fluid density 𝜌 • 𝐹 − 𝑀𝐿𝑇 −2
• Velocity of the sphere 𝑣 • 𝜌 − 𝑀𝐿−3
• Sphere Diameter 𝐷 • 𝑣 − 𝐿𝑇 −1
• Viscosity 𝜇 • 𝐷 −𝐿
• 𝜇 − 𝑀𝐿−1 𝑇 −1
Then there is an equation of
5 variables 𝐹, 𝜌, 𝑣, 𝐷 and 𝜇 Dimensions are 𝑀, 𝐿, 𝑇
So that 𝑛 = 5. 𝑘=3
𝐹 = 𝑓 𝜌, 𝑣, 𝐷, 𝜇

No of Pi terms; 5 − 3 = 2
Worked Example 7:
Now since there are 3 dimensions, we choose 3 independent
variables say;
𝐷−𝑀
𝑣 − 𝐿𝑇 −1
𝜌 − 𝑀𝐿−3

Together they contain all 𝑀, 𝐿, 𝑇

Form 1st Π term Let
Chosen 3 variables combine with 𝛼=1
𝐹 Then
Π1 = 𝐹 𝛼 𝜌𝛽 𝑣 𝛾 𝐷𝛿 𝛽 = −1
𝛾 = −2
= 𝑀𝐿𝑇 −2 𝛼
𝑀𝐿−3 𝛽
𝐿𝑇 −1 𝛾 𝐿𝛿 𝛿 = −2

= 𝑀𝛼+𝛽 𝐿𝛼−3𝛽+𝛾+𝛿 𝑇 −2𝛼−𝛾

So that
Then 𝐹𝜌−1 𝑣 −2 𝐷−2 = Π1
𝛼+𝛽 =0
𝛼 − 3𝛽 + 𝛾 + 𝛿 = 0 𝐹
= Π1
𝜌𝑣 2 𝐷2
−2𝛼 − 𝛾 = 0
Form 2nd Π term Let
Chosen 3 variables combine with 𝛼=1
𝐹 Then
Π2 = 𝜇 𝛼 𝜌 𝛽 𝑣 𝛾 𝐷 𝛿 𝛽 = −1
𝛾 = −1
= 𝑀𝐿−1 𝑇 −1 𝛼
𝑀𝐿−3 𝛽
𝐿𝑇 −1 𝛾 𝐿𝛿 𝛿 = −1

= 𝑀𝛼+𝛽 𝐿−𝛼−3𝛽+𝛾+𝛿 𝑇 −𝛼−𝛾

So that
Then 𝜇𝛽−1 𝑣 −1 𝐷−1 = Π2
𝛼+𝛽 =0
𝜇
−𝛼 − 3𝛽 + 𝛾 + 𝛿 = 0 = Π2
𝜌𝑣𝐷
−𝛼 − 𝛾 = 0
Worked Example 7:
Then we can write
𝐹 𝜇
2 2
=𝑓
𝜌𝑣 𝐷 𝜌𝑣𝐷
Worked Example 7:
Find a mathematical model determining the terminal velocity of a raindrop falling
from a motionless cloud, using the Buckingham's pi theorem

Variable Symbol Dimensions

Terminal velocity V LT⁻¹

Raindrop diameter d L
Air density ρ ML⁻³
Dynamic viscosity μ ML⁻¹T⁻¹
Gravitational acceleration g LT⁻²
 𝑛 = 5, 𝑘 = 3 Let
Then 𝑗 = 5 − 3 = 2 𝛼=1
The 3 independent
Then
𝑑, 𝜌, 𝜇 𝛽=1
Which contain all 3 dimensions 𝑀, 𝐿, 𝑇 𝛾=1
Then 𝛿 = −1

= 𝐿𝑇 −1 𝛼
𝐿 𝛽
𝑀𝐿−3 𝛾
𝑀𝐿−1 𝑇 −1 𝛿
So that

Then 𝑣𝑑𝜌
𝑀: 𝛾+𝛿 = 0 = Π1
𝜇
𝐿: 𝛼 + 𝛽 − 3𝛾 − 𝛿 = 0
𝑇: −𝛼 − 𝛿 = 0
Now Let
Π2 = 𝑔 𝛼 𝑑 𝛽 𝜌𝛾 𝜇 𝛿 𝛼=1
= 𝐿𝑇 −2 𝛼
𝐿 𝛽
𝑀𝐿−3 𝛾
𝑀𝐿−1 𝑇 −1 𝛿 Then
𝛽=3
𝛾=2
Then
𝛿 = −2
𝑀: 𝛾+𝛿 =0
𝐿: 𝛼 + 𝛽 − 3𝛾 − 𝛿 = 0
So that
𝑇: −2𝛼 − 𝛿 = 0
𝑔𝑑 3 𝜌2
= Π2
𝜇2
Worked Example 7:
Then we can write
𝑣𝑑𝜌 𝑔𝑑 3 𝜌2
=𝑓
𝜇 𝜇2

Assume a power law (standard assumption in dimensional analysis)

𝑛
𝑣𝑑𝜌 𝑔𝑑 3 𝜌2
=𝐶
𝜇 𝜇2

3 2 𝑛
𝜇 𝑔𝑑 𝜌
𝑣=𝐶
𝑑𝜌 𝜇2
 Exercise :
Redo the above problem by adding the “raindrop density ρ_d”

Variable Symbol Dimensions

Terminal velocity V LT⁻¹

Raindrop diameter d L

Fluid (air) density ρ ML⁻³

Dynamic viscosity of air μ ML⁻¹T⁻¹

Gravitational acceleration g LT⁻²

Raindrop density ρ_d ML⁻³

Answer:
𝜇 𝑔𝑑 3 𝜌2 𝜌𝑑
𝑉= ⋅𝑓
𝜌𝑑 𝜇2 𝜌
Past Paper Questions
July 2025
Q1. c)

A train operator for a particular line is buying some new rolling stock. The trains supplied
by one company accelerate in such a way that their speed (𝑉1 ) in m/s, varies with time (t)
in seconds from standing start according to the expression:

𝑡2
𝑉1 𝑡 = 35
1 + 𝑡2

However, the expression for the (𝑉2 ) of trains from a competitor supplier is
𝑉2 𝑡 = 100[1 − 𝑒 −0.02𝑡 ]

1. How far will each train travels in 30 seconds from a standing start.?

2. Which train would you recommend

 Distance Speed at 30s Maximum
m/s speed (m/s)

Train 1 996 35 35
Train 2 774 45 100
July 2025
The sound power(𝑊) from a turbulent jet through a nozzle is
believed to depend on the jet centerline velocity ( 𝑉 ) nozzle
diameter (𝐷 ) , speed sound 𝐶 and fluid density (𝜌). The
temperature and composition of the fluid in the jet is the same as
the ambient fluid. Using the Buckingham Pi theorem, find a relation
between the sound power and other parameters of the system
Quantity Symbol Dimensions SI unit Quantity Symbol Dimensions SI unit

Velocity v L T⁻¹ m/s Force F M L T⁻² N

Acceleration a L T⁻² m/s² Pressure/Stress σ M L⁻¹ T⁻² Pa

Energy E M L² T⁻² J Power P M L² T⁻³ W

Density 𝜌 M L⁻³ kg/m³ Damping coeff. c M T⁻¹ kg/s

Voltage V M L² T⁻³ I⁻¹ V Resistance R M L² T⁻³ I⁻² Ω

Bending M M L² T⁻² N·m 2nd moment area I L⁴ m⁴

moment
Shear stress 𝜏 M L⁻¹ T⁻² N/m² Viscosity 𝜇 M L⁻¹ T⁻¹ Kg/ms
(Pressure/Area)