Engineering Mathematics

with Applications to
Fire Engineering
http://taylorandfrancis.com
Engineering Mathematics
with Applications to
Fire Engineering

Khalid Khan
Tony Lee Graham
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2018 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed on acid-free paper

International Standard Book Number-13: 978-1-138-09884-8 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to
publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materi-
als or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material
reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained.
If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in
any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, micro-
filming, and recording, or in any information storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright​.com (http://www​
.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-
8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that
have been granted a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identifi-
cation and explanation without intent to infringe.

Library of Congress Cataloging‑in‑Publication Data

Names: Khan, Khalid Mahmood, 1963- author. | Graham, Tony Lee, author.
Title: Engineering mathematics with applications to fire engineering / Khalid Khan
and Tony Lee Graham.
Description: Boca Raton : Taylor & Francis, 2018. | Includes bibliographical references
and index.
Identifiers: LCCN 2018001261| ISBN 9781138098848 (hardback : acid-free paper) | ISBN
9781315104270 (ebook)
Subjects: LCSH: Engineering mathematics. | Engineering mathematics--Problems,
exercises, etc. | Fire protection engineering--Mathematics.
Classification: LCC TA332.5 .K43 2018 | DDC 620.001/51--dc23
LC record available at https://lccn.loc.gov/2018001261

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com
Contents

Preface xiii

Authors xv

1 Review of Basic Concepts 1

1.1 Degrees of Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Rounding Numbers (Common Method) . . . . . . . . . . . . . . . . 1
1.1.2 Round-to-Even Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.3 Decimal Places . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.4 Significant Places . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Scientific Notation (Standard Form) . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.1 How Does Scientific Notation Work? . . . . . . . . . . . . . . . . . . 7
1.2.2 Addition and Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.3 Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.4 Division . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Basic Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.1 Algebraic Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.2 Evaluating Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4 Linear Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.1 Solving Linear Equations . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.2 Transposing Formulae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5 Linear Simultaneous Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5.1 Elimination Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5.2 Substitution Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6 Quadratic Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6.1 Solving Quadratic Equations Using the Formula . . . . . . . . 21
1.7 Trigonometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.7.1 Right-Angled Triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.7.2 Scalene Triangles (Sine and Cosine Rules) . . . . . . . . . . . . . 25
1.7.2.1 Sine Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.7.2.2 Cosine Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

v
vi Contents

1.7.3 Resultant Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.7.3.1 Adding Two Forces . . . . . . . . . . . . . . . . . . . . . . . 28
1.7.4 Basic Trigonometric Identities . . . . . . . . . . . . . . . . . . . . . . 29
1.7.5 Radian Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1.7.5.1 Radians on the Calculator . . . . . . . . . . . . . . . . . . 31
1.8 Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.8.2 Measures of Averages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.8.2.1 Data in a Frequency Table . . . . . . . . . . . . . . . . . . 34
1.8.2.2 Grouped Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
1.8.3 Measures of Spread . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.8.3.1 Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.8.3.2 Interquartile Range . . . . . . . . . . . . . . . . . . . . . . . 36
1.8.3.3 Standard Deviation (σ) . . . . . . . . . . . . . . . . . . . . . 37
1.8.3.4 Sample Standard Deviation (s) . . . . . . . . . . . . . . 38
1.8.4 Change of Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
1.9 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

2 Introduction to Probability Theory 45

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.1.1 Mutually Exclusive Events . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.1.2 Independent Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.1.3 Conditional Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.1.4 Bayes’ Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.1.4.1 Generalization of Bayes’ Theorem . . . . . . . . . . . 50
2.1.5 Tree Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.2 Discrete Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.2.1 Discrete Probability Distribution . . . . . . . . . . . . . . . . . . . . 52
2.2.2 Expectation Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.2.3 Variance and Standard Deviation . . . . . . . . . . . . . . . . . . . . 54
2.3 Continuous Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.3.1 Probability Density Function (pdf) . . . . . . . . . . . . . . . . . . . 56
2.3.2 Cumulative Distribution Function (cdf) . . . . . . . . . . . . . . . 56
2.3.3 Expectation of a Continuous Random Variable . . . . . . . . . 58
2.3.4 Variance and Standard Deviation of a Continuous
Random Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

3 Vectors and Geometrical Applications 65

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.1.1 Magnitude and Unit Vectors . . . . . . . . . . . . . . . . . . . . . . . . 66
3.1.1.1 Unit Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.1.2 Addition and Subtraction of Vectors . . . . . . . . . . . . . . . . . . 68
3.1.3 Scalar and Vector Products . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.1.3.1 Scalar Product (Dot Product) . . . . . . . . . . . . . . . . 69
3.1.3.2 Vector Product (Cross Product) . . . . . . . . . . . . . . 70
3.1.3.3 How to Calculate a × b . . . . . . . . . . . . . . . . . . . . 71
3.1.4 Projection of Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
 Contents vii

3.2 Vector Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

3.2.1 Vector Equation of a Line . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.2.1.1 Intersection of Lines . . . . . . . . . . . . . . . . . . . . . . 75
3.2.2 Vector Equation of Planes . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.2.2.1 Generalizing for Any Plane in Space . . . . . . . . . 77
3.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

4 Determinants and Matrices 81

4.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.2 Introduction to Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.2.1 2 × 2 Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.2.2 Properties of Determinants . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.2.2.1 Multiplying a Determinant by a Number . . . . . . 84
4.2.3 3 × 3 Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.3 Introduction to Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.3.1 Order of a Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.3.2 Addition and Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.3.3 Matrix Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.3.3.1 Multiplying a Matrix by a Scalar . . . . . . . . . . . . . 88
4.3.3.2 Multiplying Two Matrices . . . . . . . . . . . . . . . . . . 89
4.3.3.3 How to Multiply Two Matrices . . . . . . . . . . . . . . 89
4.3.4 Special Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.3.5 Powers of Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.3.6 Inverse of a Square Matrix . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.3.7 Eigenvalues and Eigenvectors . . . . . . . . . . . . . . . . . . . . . . . 96
4.3.8 Diagonal Factorization of Matrices . . . . . . . . . . . . . . . . . . 100
4.4 Solving Systems of Linear Equations . . . . . . . . . . . . . . . . . . . . . . . 103
4.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.4.2 Gaussian Elimination Method . . . . . . . . . . . . . . . . . . . . . . 104
4.4.3 Matrix Inversion Method . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.4.3.1 Matrix Method for Solving Simultaneous
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5 Complex Numbers 117

5.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.2 Introduction and the Imaginary j . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.2.1 Some Properties of j . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.2.2 Complex Numbers: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.3 Arithmetic Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.3.1 Addition and Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . .118
5.3.2 Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.3.2.1 Conjugate Numbers . . . . . . . . . . . . . . . . . . . . . . 119
5.3.3 Division . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.4 Argand Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.4.1 Drawing a Diagram of Complex Numbers . . . . . . . . . . . . 120
viii Contents

5.5 Polar and Exponential Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

5.5.1 Polar Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.5.1.1 Multiplying and Dividing Complex Numbers
in Polar Form . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.5.2 Exponential Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.5.3 Powers of Complex Numbers . . . . . . . . . . . . . . . . . . . . . . 122
5.5.4 De Moivre’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.6 Roots of Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.7 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

6 Introduction to Calculus 131

6.1 Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.1.1 Definition of a Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.1.1.1 Differentiating Fractional and Negative
Powers of x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.1.2 Stationary Points (Maxima and Minima) . . . . . . . . . . . . . 136
6.1.2.1 Practical Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.1.2.2 Second Derivative Method . . . . . . . . . . . . . . . . 138
6.1.3 Differentiating Products and Quotients . . . . . . . . . . . . . . 139
6.1.3.1 Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6.1.3.2 Quotients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.1.4 Standard Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.1.5 Function of a Function (Chain Rule) . . . . . . . . . . . . . . . . . 143
6.2 Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6.2.1 Introduction and the Riemann Sum . . . . . . . . . . . . . . . . . 144
6.2.2 Fundamental Theorem of Calculus (Optional Section) . . . 146
b

6.2.2.1 How to Compute the Integral

∫ f (x) dx . . . . . . . 146
a
6.2.3 Standard Integrals and Areas under Curves . . . . . . . . . . . 149
6.2.3.1 Finding the Area under a Curve . . . . . . . . . . . . 150
6.2.4 Improper Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
6.3 Integration Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6.3.1 Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6.3.2 Partial Fractions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
6.3.2.1 Type 1: Different Linear Factors . . . . . . . . . . . . 158
6.3.2.2 Type 2: Denominator with a Repeated Factor . . . 159
6.3.2.3 Type 3: Denominator with a Quadratic Factor . . . 160
6.3.2.4 Performing the Final Integration . . . . . . . . . . . . 161
6.3.3 Integration by Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167

7 Ordinary Linear Differential Equations 171

7.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
7.2 Types of Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
7.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
7.2.2 Order of a Differential Equation . . . . . . . . . . . . . . . . . . . . 173
7.2.3 Degree of a Differential Equation . . . . . . . . . . . . . . . . . . . 173
 Contents ix

7.2.4 Linearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173

7.2.5 What Is Meant by Solving Differential Equations? . . . . . 174
7.3 First-Order Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . 174
7.3.1 Simplest Situation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
7.3.2 Separating Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
7.3.3 Integrating Factor Technique . . . . . . . . . . . . . . . . . . . . . . 179
7.4 Second-Order Differential Equations . . . . . . . . . . . . . . . . . . . . . . . 182
7.4.1 Complementary Function (CF) . . . . . . . . . . . . . . . . . . . . . 183
7.4.1.1 General Solution for the Complementary
Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.4.1.2 How to Find the Complementary Function . . . . 184
7.4.2 Types of Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
7.4.2.1 Case 1: Real and Distinct Roots m1 and m2 . . . . 185
7.4.2.2 Case 2: Real and Repeated Roots . . . . . . . . . . . 186
7.4.2.3 Case 3: Complex Conjugate Roots . . . . . . . . . . 187
7.4.3 Particular Integral (P.I.) . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
7.4.3.1 How to Find the Particular Integral . . . . . . . . . . 188
7.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

8 Laplace Transforms 199

8.1 Why Do We Need the Laplace Transform? . . . . . . . . . . . . . . . . . . 199
8.2 Derivation from a Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
8.3 Introduction and Standard Transforms . . . . . . . . . . . . . . . . . . . . . . 201
8.3.1 Schematic Representation of Laplace Transforms . . . . . . 202
8.3.2 Standard Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
8.3.3 Linearity of Laplace Transforms . . . . . . . . . . . . . . . . . . . . 205
8.3.4 Basic Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
8.4 Inverse Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
8.5 Discontinuous Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
8.5.1 Heaviside Unit Step Function . . . . . . . . . . . . . . . . . . . . . . 210
8.5.1.1 Calculating the Laplace Transform of H(t–c) . . . 210
8.5.1.2 Unit Step at Origin . . . . . . . . . . . . . . . . . . . . . . . 211
8.5.2 The Delta Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
8.5.2.1 The Delta Function at the Origin . . . . . . . . . . . . 212
8.6 Shift Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
8.7 Method for Solving Linear Differential Equations . . . . . . . . . . . . 214
8.8 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

9 Fourier Series and Fourier Transforms 229

9.1 Periodic Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
9.2 Fourier Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
9.2.1 Periodic Functions of Period T . . . . . . . . . . . . . . . . . . . . . 230
9.2.2 General Properties and Orthogonal Functions . . . . . . . . . 231
9.2.3 Fourier Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
9.3 Complex Form of the Fourier Series . . . . . . . . . . . . . . . . . . . . . . . 238
9.4 Fourier Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
9.4.1 Nonperiodic Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
9.4.2 Fourier Transform Pair . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
x Contents

9.4.3 What Does the Fourier Transform Represent? . . . . . . . . . 245

9.4.4 Properties of the Fourier Transform . . . . . . . . . . . . . . . . . 247
9.4.4.1 Linearity Property . . . . . . . . . . . . . . . . . . . . . . . 247
9.4.5 Convolution of Two Functions . . . . . . . . . . . . . . . . . . . . . 248
9.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

10 Multivariable Calculus 255

10.1 Partial Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
10.1.1 Introduction and Definition . . . . . . . . . . . . . . . . . . . . . . . . 255
10.1.1.1 Partial Derivatives Defined . . . . . . . . . . . . . . . . 256
10.1.1.2 Instantaneous Rate of Change . . . . . . . . . . . . . . 257
10.1.2 Higher Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
10.1.2.1 Clairaut’s Theorem . . . . . . . . . . . . . . . . . . . . . . 259
10.1.2.2 Antiderivatives When There Are
Multiple Variables . . . . . . . . . . . . . . . . . . . . . . . 259
10.1.3 Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
10.1.3.1 Chain Rule with One Variable . . . . . . . . . . . . . . 260
10.1.3.2 Chain Rule with Multivariables . . . . . . . . . . . . . 261
10.1.4 Directional Derivatives and Gradients . . . . . . . . . . . . . . . 263
10.1.5 Stationary Points (Maxima, Minima, and Saddle Points) . . . 266
10.1.5.1 Summary to Find Maximum or Minimum
Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
10.2 Higher-Order Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
10.2.1 Double Integrals and Fubini’s Theorem . . . . . . . . . . . . . . 269
10.2.1.1 An Application of Double Integration . . . . . . . . 272
10.2.2 Double Integration Using Polar Coordinates . . . . . . . . . . 273
10.2.2.1 Using Polar Coordinates to Calculate Double
Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
10.2.3 General Regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
10.2.4 Triple Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
10.2.4.1 Cartesian Coordinates . . . . . . . . . . . . . . . . . . . . 281
10.2.5 3-D Coordinate Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 283
10.2.5.1 Integrals in the New Coordinate Systems . . . . . 284
10.2.6 General Change of Coordinate Systems . . . . . . . . . . . . . . 286
10.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
10.3.1 Application of Double Integration . . . . . . . . . . . . . . . . . . . 289
10.3.2 Application of Triple Integration (Center of Mass) . . . . . . 292
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295

11 Vector Calculus 297

11.1 Differentiation and Integration of Vectors . . . . . . . . . . . . . . . . . . . 297
11.1.1 Derivatives of Vector Functions . . . . . . . . . . . . . . . . . . . . 297
11.1.2 Integrating Vector Functions . . . . . . . . . . . . . . . . . . . . . . . 299
11.2 Vector Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
11.3 Line Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
11.3.1 The ds-Type Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
11.3.2 The dr −Type Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
11.3.3 Summary of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
 Contents xi

11.4 Gradient Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308

11.4.1 Conservative Vector Fields . . . . . . . . . . . . . . . . . . . . . . . . 310
11.4.2 Testing for Conservativeness . . . . . . . . . . . . . . . . . . . . . . . 313
11.5 Green’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
11.5.1 Properties of Green’s Theorem . . . . . . . . . . . . . . . . . . . . . 319
11.6 Divergence and Curl of Vector Fields . . . . . . . . . . . . . . . . . . . . . . 321
11.6.1 2-Dimensional Definitions . . . . . . . . . . . . . . . . . . . . . . . . 321
11.6.2 Alternative Forms of Green’s Theorem . . . . . . . . . . . . . . 323
11.6.2.1 Curl Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
11.6.2.2 Divergence Form . . . . . . . . . . . . . . . . . . . . . . . . 324
11.6.3 3-Dimensional Definitions . . . . . . . . . . . . . . . . . . . . . . . . 325
11.7 Surface Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
11.7.1 Parametric Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
11.7.1.1 Summary of the Main Types of Surfaces . . . . . 330
11.7.2 Surface Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
11.7.3 Tangent Planes and Normal Vectors . . . . . . . . . . . . . . . . . 333
11.7.3.1 Normal Vector to the Tangent Plane . . . . . . . . . 334
11.7.3.2 General Formula for the Normal Vector
to a Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
11.7.4 Normal Vectors to Surfaces . . . . . . . . . . . . . . . . . . . . . . . 336
11.7.5 Applications of Surface Integrals . . . . . . . . . . . . . . . . . . . 339
11.8 Stokes’ Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
11.9 Divergence Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
11.10 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353

Answers 357

Index 363
http://taylorandfrancis.com
Preface

E ngineering is the application of scientific knowledge to solving problems in

the real world. There are many branches and fields within the different engi-
neering disciplines. All disciplines, whether they be mechanical, civil, chemical,
electrical, or fire, have common skills that rely heavily on rational thinking and
logical decision making. A very important component required in all engineering
fields is a good understanding of applied mathematical concepts.
This book was written for undergraduate degree–level students studying a first
course in engineering. Its primary aim is students studying fire engineering and
related courses, although there are many other real-world applications given that
are relevant to other areas of engineering. The motivation for writing such a book
was that although there are many engineering mathematics textbooks available
for fire engineering students, there was not a book that showed how the math-
ematical concepts studied were applied to their particular discipline. This book
not only gives the foundation mathematics needed for fire engineering degree
courses but also gives real-world fire applications showing how these concepts
are used in practice.
This book is based on a combination of lectures that have been delivered over
a twenty-year period at international educational institutes and at the University
of Central Lancashire in the United Kingdom across the various levels of under-
graduate study. The Division of Fire and Safety Engineering at the University of
Central Lancashire has developed a leading position in the United Kingdom and
overseas with respect to the delivery of undergraduate and postgraduate courses
in fire safety engineering, management, and leadership studies. The step-by-step
methodology used together with plenty of practical applications in the real world
make this book an essential aid in the understanding of mathematical concepts
for most engineering disciplines. Other engineering mathematics textbooks gen-
erally assume a certain level of mathematical understanding and therefore some-
times certain key steps are omitted in the discourse; this is not the case with the
current book.
Each chapter of the book has a similar format starting with the development
of the basic mathematical theory, which is then followed by real-world applica-
tions especially to fire engineering, and then concluding with a problems section.

xiii
xiv Preface

We consider the presentation in this text as unique in that there are real-world fire
engineering applications given alongside the mathematical content that underpins
those applications.
The overall structure of the book is that it begins with a review of the key
elementary mathematical concepts and focuses on important concepts such as
transposition of formulae, as these form an essential part of many engineering
solutions. An introduction to probability theory with discrete and continuous ran-
dom variables are important concepts used in fault tree analysis in risk assessment
and reliability theory, respectively. Determinants and matrices lead to solving a
system of linear equations using different techniques such as Gaussian elimina-
tion and the matrix inversion method. Vectors and normal vectors to surfaces are
considered and form the basis of concepts in surface integrals. Use of complex
numbers and their applications in electrical circuit theory and the role they play
in the solutions to differential equations are covered. There is an introduction to
one-variable calculus with the fundamentals of differentiation, integration tech-
niques, and importance of integration as a summing process for a multitude of
engineering applications. Methods of solving ordinary linear differential equa-
tions are introduced with emphasis on the Laplace transform method as a valuable
tool in finding solutions to problems. Higher-dimensional multivariable calculus
dealing with partial derivatives, double and triple integrals, and general change
of coordinate systems are covered. Finally, in the last chapter on vector calculus,
vector fields representing physical phenomenon are considered along with con-
cepts of divergence and curl, and applications of Green’s, Stokes’, and divergence
theorems are given.
We would like to thank the staff at CRC/Taylor & Francis who have contrib-
uted to the production of this book and to Dr. Alan Burns for his valuable com-
ments on the draft version of the book. Finally, we wish to express our gratitude
to the University of Central Lancashire for providing a conductive academic envi-
ronment that allowed us to complete this project.
Authors

Khalid Khan, BSc (Hons), MSc, PhD, received his BSc (Hons) in mathemat-
ical physics, and MSc and PhD in control systems all from the University of
Manchester Institute of Science and Technology in the United Kingdom. Dr Khan
then spent two years as a consultant engineer working on safety, reliability and
risk assessment problems in the energy industry. Subsequently, he moved abroad
and after spending some period of time as an assistant professor of mathemat-
ics at Etisalat University in the United Arab Emirates he returned to the United
Kingdom and took up a position at the University of Central Lancashire.
Dr Khan is currently a senior lecturer in engineering mathematics in the
School of Engineering at the University of Central Lancashire. He teaches on a
range of mathematics modules within the fire degree programs and contributes
to other mathematics teaching within the school and college. He is currently the
course leader for the Foundation Degree in Fire Safety Engineering. Dr Khan as
part of the fire team has been involved in the development of a range of courses
in fire safety engineering and management that are currently running at one of
University of Central Lancashire’s international partnerships in Qatar. Dr Khan is
also a senior fellow of the Higher Education Academy (HEA), which is a British
professional institution promoting excellence in higher education.
Dr Khan’s research interests are in the area of mathematical modeling of sys-
tem behavior in a range of applications. His current work focuses on fire suppres-
sion using sprinkler systems and on mathematical models of collective motion of
self-propelled particles in homogeneous and heterogeneous mediums. Dr Khan
currently has over thirty publications and is also a member of two journal review
panel boards.

Tony Graham, BSc (Hons), PhD, is a senior lecturer and course leader at the
University of Central Lancashire, United Kingdom. He is best known for teaching
fire safety engineering to thousands of students over twenty-four years in differ-
ent countries and for papers on compartment fire dynamics and the phenomenon
of flashover fire. His teaching also includes risk engineering and engineering
analysis. He has taught courses at International College of Engineering and
Management in Sultanate of Oman and also at City University of Hong Kong.

xv
xvi Authors

He is a senior fellow of Higher Education Academy and Secretary of the

Combustion, Explosion and Fire Engineering research group in the School
of Engineering and has served on both the Academic Board, and Academic
Standards and Quality Assurance Committee. His research was funded by
Nuffield Foundation and is ongoing. Dr Graham earned his BSc (Hons) in theo-
retical physics at Lancaster University, and subsequently was awarded his PhD
in fire engineering in 1998 from the University of Central Lancashire. For the
future, Dr Graham is looking at two books on fire engineering science and risk
engineering. Dr Graham, along with his beloved wife Anna and daughter Natalia
now prosper in Preston, Lancashire.
 1 Review of Basic
Concepts

1.1 Degrees of Accuracy
1.1.1 Rounding Numbers (Common Method)
In the real world, when dealing with numbers a degree of accuracy is needed. For
example, if a piece of wood of a certain length was required, asking for a length of
123.732461 centimeters would not be sensible as such accurate measurements are
not possible. What is more usual is some form of rounding. The method of round-
ing is commonly used in mathematical applications in science and engineering.
It is the one generally taught in mathematics classes in high school. The method
is also known as round-half-up. It works as follows:

• Decide which is the last digit to keep.

• Increase it by 1 if the next digit is 5 or more (this is called rounding up).

• Leave it the same if the next digit is 4 or less (this is called rounding
down).

Example 1.1
3.044 rounded to hundredths is 3.04 (because the next digit, 4, is less
than 5).
3.045 rounded to hundredths is 3.05 (because the next digit, 5, is 5 or
more).
3.0447 rounded to hundredths is 3.04 (because the next digit, 4, is less
than 5).

For negative numbers, one rounds the absolute value and reapplies the sign
afterward.

1
2 Review of Basic Concepts

Example 1.2
−2.1349 rounded to hundredths is −2.13.
−2.1350 rounded to hundredths is −2.14.

1.1.2 Round-to-Even Method
The round-to-even method method, also known as unbiased rounding or Gaussian
rounding, exactly replicates the common method of rounding except when the
digit(s) following the rounding digit starts with a 5 and has no nonzero digits
after it. The new algorithm becomes

• Decide which is the last digit to keep.

• Increase it by 1 if the next digit is 6 or more, or a 5 followed by one or

more nonzero digits.

• Leave it the same if the next digit is 4 or less.

• Otherwise, if all that follows the last digit is a 5 and possibly trailing
zeros, then increase the rounded digit if it is currently odd; else, if it is
already even, leave it alone.

All rounding schemes have two possible outcomes: increasing the rounding
digit by one or leaving it alone. With traditional rounding, if the number has a
value less than the halfway mark between the possible outcomes, it is rounded
down; if the number has a value exactly halfway or greater than halfway between
the possible outcomes, it is rounded up. The round-to-even method is the same
except that numbers exactly halfway between the possible outcomes are some-
times rounded up, sometimes down.
Despite the custom of rounding the number 4.5 up to 5, 4.5 is no nearer to 5
than it is to 4 (it is 0.5 away from both). When dealing with large sets of scientific
or statistical data, where trends are important, traditional rounding on average
biases the data upward slightly. Over a large set of data, or when many subsequent
rounding operations are performed as in digital signal processing, the round-to-
even rule tends to reduce the total rounding error, with (on average) an equal por-
tion of numbers rounding up as rounding down. This generally reduces upward
skewing of the result.

Examples 1.3
3.016 rounded to hundredths is 3.02 (because the next digit, 6, is 6 or
more).
3.013 rounded to hundredths is 3.01 (because the next digit, 3, is 4 or
less).
3.015 rounded to hundredths is 3.02 (because the next digit is 5, and the
hundredths digit, 1, is odd).
3.045 rounded to hundredths is 3.04 (because the next digit is 5, and the
hundredths digit, 4, is even).
3.04501 rounded to hundredths is 3.05 (because the next digit is 5, but it
is followed by nonzero digits).
 1.1 Degrees of Accuracy 3

Table 1.1 Showing Accuracy of the “Round-to-Even” Method

Original Number “Old” Rounding Method “Round-to-Even” Method
3.55 3.6 3.6
3.65 3.7 3.6
3.75 3.8 3.8
3.85 3.9 3.8
3.95 4.0 4.0
Mean = 3.75 Mean = 3.8 Mean = 3.76

Example 1.4
Table 1.1 shows how the round-to-even system compares with the old sys-
tem of rounding. There are five original data points that are rounded to the
tenths and then their average is calculated. It can be seen that the round-to-
even method is much more accurate than the old method of rounding.

1.1.3 Decimal Places
When rounding a number, one is usually told how to round it. It is simplest when
one is told how many places to round to, but one should also know how to round
to a named place, such as to the nearest thousand or to the ten-thousandths place.
Also, it may be required to know how to round to a certain number of significant
digits; this is dealt with later.
Using the first few digits of the decimal expansion of pi = π = 3.14159265... in
the following examples.

Example 1.5
Round pi to five decimal places. First, count out the five decimal places,
then look at the sixth place:

3.14159 265

A little line separating the fifth place from the sixth place has been drawn.
This can be a good way of keeping your place, especially if dealing with
lots of digits.
The fifth place has a 9 in it. Looking at the sixth place, it has a 2 in it.
Since 2 is less than 5, the 9 will not be rounded up; that is, just leave the
9 as it is. In addition, delete the digits after the 9. Then pi, rounded to five
decimal places, is given as 3.14159.

Example 1.6
Round pi to four decimal places. First, go back to the original number:
3.14159265. Count off four places, and look at the number in the fifth place:

3.1415 9265

The number in the fifth place is a 9, which is greater than 5, so round

up in the fourth place, truncating the expansion at four decimal places.
4 Review of Basic Concepts

That is, the 5 becomes a 6, the 9265… part disappears, and pi, rounded to
four decimal places, is given as 3.1416.
This rounding works the same way when rounding to a certain named
place, such as the hundredths place. The only difference being a bit more
careful in counting off the places needed. Just remember that the decimal
places count off to the right in the same order as the counting numbers
count off to the left. That is, for regular numbers, the place values are

(ten-thousands) (thousands) (hundreds) (tens) (ones)

For decimal places, a “oneths” is not there, but all the other fractions are

(decimal point) (tenths) (hundredths) (thousandtths) (ten-thousandths)

Example 1.7
Round pi to the nearest thousandth. The “nearest thousandth” means that
one needs to count off three decimal places (tenths, hundredths, thou-
sandths), and then round:

3.141 59265

Then pi, rounded to the nearest thousandth, is 3.142.

Example 1.8
Round 18.796 to the hundredths place. The hundredths place is two decimal
places, so count off two decimal places, and round according to the third
decimal place:

18.79 6

Since the third decimal place contains a 6, which is greater than 5, one has
to round up. But rounding up a 9 gives a 10. In this case, round the 79 up to
an 80 as 18.80.
One might be tempted to write this as 18.8, but, since rounded to the
hundredths place (to two decimal places), one should write both decimal
places. Otherwise, it looks like rounding to one decimal place, or to the
tenths place, and the answer could be counted off as being incorrect.

1.1.4 Significant Places
Rounding can also be carried out to an appropriate number of significant digits.
What are significant digits? Well, they are sort of the “interesting” or “important”
digits. For example:

3.14159 has six significant digits (all the numbers give you useful information).

1000 has one significant digit (only the 1 is interesting; you do not know
anything for sure about the hundreds, tens, or units places; the zeroes
 1.1 Degrees of Accuracy 5

may just be placeholders; they may have rounded something off to get
this value).

1000.0 has five significant digits (the “.0” tells us something interesting
about the presumed accuracy of the measurement being made: that the
measurement is accurate to the tenths place, but that there happens to be
zero tenths).

0.00035 has two significant digits (only the 3 and 5 tell us something; the
other zeroes are placeholders, only providing information about relative
size).

0.000350 has three significant digits (that last zero tells us that the measure-
ment was made accurate to that last digit, which just happened to have a
value of zero).

1006 has four significant digits (the 1 and 6 are interesting, and the zeros
have to be counted, because they are between the two interesting
numbers).

560 has two significant digits (the last zero is just a placeholder).

560. (notice the point after the zero) has three significant digits (the decimal
point tells us that the measurement was made to the nearest unit, so the
zero is not just a placeholder).

560.0 has four significant digits (the zero in the tenths place means that
the measurement was made accurate to the tenths place, and that there
just happen to be zero tenths; the 5 and 6 give useful information, and
the other zero is between significant digits, and must therefore also be
counted).

Here are the basic rules for significant digits:

1. All nonzero digits are significant.

2. All zeros between significant digits are significant.

3. All zeros that are both to the right of the decimal point and to the right of
all nonzero significant digits are themselves significant.

Following are some rounding examples; each number is rounded to four, three,
and two significant digits.

Example 1.9
Round 742,396 to four, three, and two significant digits:

742,400 (four significant digits)

742,000 (three significant digits)
740,000 (two significant digits)
6 Review of Basic Concepts

Example 1.10
Round 0.07284 to four, three, and two significant digits:

0.07284 (four significant digits)

0.0728 (three significant digits)
0.073 (two significant digits)

Example 1.11
Round 231.45 to four, three, and two significant digits:

231.4 (four significant digits)

231 (three significant digits)
230 (two significant digits)

1.2 Scientific Notation (Standard Form)

Scientific notation, also sometimes known as standard form or as exponential
notation, is a way of writing numbers that accommodates values too large or
small to be conveniently written in standard decimal notation. Scientific notation
has a number of useful properties and is often favored by engineers, scientists,
and mathematicians, who work with such numbers.
Look at the following numbers:

300,000,000 m/sec, the speed of light

0.000 000 000 753 kg, mass of a dust particle

Scientists have developed a shorter method to express very large and very small
numbers. This method is called scientific notation. Scientific notation is based on
powers of the base number 10.

Example 1.12
Our galaxy to which the sun belongs is called the Milky Way. It con-
tains at least 100,000,000,000 stars. Now let’s look at this number:
100,000,000,000. It can be written as 1.0 × 100,000,000,000. It is the large
number 100,000,000,000 that causes the problem. But that is just a multiple
of 10. In fact, it is 10 times itself 11 times:

10 × 10 × 10 × 10 × 10 × 10 × 10 × 10 × 10 × 10 × 10 = 100, 000, 000, 000

A more convenient way of writing 100,000,000,000 is 1011. The small num-

ber to the right of the 10 is called the exponent, or the power of ten. It rep-
resents the number of zeros that follow the 1.

So, one would write 100,000,000,000 in scientific notation as 1.0 × 1011.

This number is read as follows: one point zero times ten to the eleventh
power.
 1.2 Scientific Notation (Standard Form) 7

So generally, any number can be written in scientific form as

A × 10 N (1.1)

where 1 ≤ A < 10 and N is any positive or negative integer.

1.2.1 How Does Scientific Notation Work?

As stated earlier, the exponent refers to the number of zeros that follow the 1. So

101 = 10
10 2 = 100
10 3 = 1, 000

and so on. Similarly, 100 = 1, since the zero exponent means that no zeros follow
the 1.
Negative exponents indicate negative powers of 10, which are expressed as
fractions with 1 in the numerator (on top) and the power of 10 in the denominator
(on the bottom). So

10 −1 = 1/10
10 −2 = 1/100
10 −3 = 1/1, 000

and so on. This allows one to express other small numbers this way. For example,

2.5 × 10 −3 = 2.5 × 1/1, 000 = 0.0025

Every number can be expressed in scientific notation. In Example 1.12,

100,000,000,000 should be written as 1.0 × 1011.
This illustrates another way to think about scientific notation: The exponent
will tell you how the decimal point moves; a positive exponent moves the decimal
point to the right, and a negative one moves it to the left. So, for example,

4.0 × 10 2 = 400 (2 places to the right of 4)

4.0 × 10 −2 = 0.04 (2 places to the left of 4)

Note that scientific notation is also sometimes expressed as E (for exponent), as

in 4 E 2 (meaning 4.0 × 10 raised to 2). Similarly, 4 E –2 means 4 times 10 raised
to –2, or = 4 × 10 –2 = 0.04. This method of expression makes it easier to type in
scientific notation.

1.2.2 Addition and Subtraction

The key to adding or subtracting numbers in scientific notation is to make sure the
exponents are the same.
8 Review of Basic Concepts

Example 1.13

(2.0 × 10 2 ) + (3.0 × 10 3 )

can be rewritten as

(0.2 × 10 3 ) + (3.0 × 10 3 )

Just add 0.2 + 3 and keep the 103 intact. Your answer is 3.2 × 103, or 3,200.
This can be checked by converting the numbers first to the more familiar
form. So,

2 × 10 2 + 3.0 × 10 3 = 200 + 3, 000 = 3, 200 = 3.2 × 10 3

Example 1.14

(2.0 × 10 7 ) − (6.3 × 10 5 )

The problem needs to be rewritten so that the exponents are the same.
So this can be written as

(200 × 10 5 ) − (6.3 × 10 5 ) = 193.7 × 10 5

which in scientific notation would be written as 1.937 × 107.

1.2.3 Multiplication
When multiplying numbers expressed in scientific notation, the exponents can
simply be added together. This is because the exponent represents the number of
zeros following the one. So,

101 × 10 2 = 10 × 100 = 1, 000 = 10 3

Checking that it is seen 101 × 102 = 101+2 = 103.

Similarly,

101 × 10 −3 = 101−3 = 10 −2 = .01

Again, when checking, it is seen that 10 × 1/1000 = 1/100 = .01.

Example 1.15
Multiply the following numbers: (4.0 × 105) × (3.0 × 10 –1). The 4 and the
3 are multiplied, giving 12, but the exponents 5 and –1 are added, so the
answer is 12 × 104, or 1.2 × 105.
Checking: (4 × 105) × (3 × 10 –1) = 400,000 × 0.3 = 120,000 = 1.2 × 105.
 1.3 Basic Algebra 9

1.2.4 Division
Example 1.16

(6.0 × 108 )
(3.0 × 10 5 )

To solve this problem, first divide the 6 by the 3, to get 2. The exponent
in the denominator is then moved to the numerator, reversing its sign (this
will be explained further when dealing with indices). So, move the 105 to
the numerator with a negative exponent, which then looks like this: 2 ×
108 × 10 –5. All that is left now is to solve this as a multiplication problem,
remembering that all that needs to be done for the 108 × 10 –5 part is to add
the exponents. So, the answer is 2.0 × 103 or 2,000.

Note: Usually the risk of pedestrian dying in a transport accident is quoted

as 1 in 47,773. This then gives the risk of dying as 2.1 × 10 –5 (correct to two
significant digits). The risk of dying from a fire is quoted as six in a million
and can be w­ ritten as 6 × 10 –6.

1.3 Basic Algebra
1.3.1 Algebraic Notation
Algebraic notation describes how algebra is written. It follows certain rules and
conventions, and has its own terminology.
For example, the expression in Figure 1.1 has the following components to it:
1: exponent (power); 2: coefficient; 3: terms; 4: operators; 5: constant, and with x,
y: variables.
A coefficient is a numerical value or letter representing a numerical constant
that multiplies a variable (the operator is omitted).
A group of coefficients, variables, constants and exponents that may be sepa-
rated from the other terms by the plus and minus operators. Letters represent
variables and constants.
By convention, letters at the beginning of the alphabet (e.g., a, b, and c) are
typically used to represent constants, and those toward the end of the alphabet
(e.g., w, x, and y) are used to represent variables.
Algebraic operations work in the same way as arithmetic operations, such
as addition, subtraction, multiplication, division, and exponentiation, and are
applied to algebraic variables and terms.

2 1 2

7 x11 + 3xy – 8

3 4 3 4 5

Figure 1.1 Composition of an algebraic expression.

10 Review of Basic Concepts

Multiplication symbols are usually omitted, and implied when there is no

space between two variables or terms, or when a coefficient is used. For example,
5 × x2 is written as 5x2.

1.3.2 Evaluating Expressions
Algebraic expressions may be evaluated and simplified, based on the basic prop-
erties of arithmetic operations (addition, subtraction, multiplication, division, and
exponentiation).
Added terms are simplified using coefficients. For example, x + x + x + x can
be simplified as 4x (where 4 is a numerical coefficient).
Multiplied terms are simplified using exponents. For example, x × x × x × x is
represented as x4.
Like terms are added together, for example, 5x + 4p + 1 − 2x + 6p + 4 is w­ ritten
as 3x + 10p + 5.
Brackets can be multiplied out, using the distributive property. For example,
4(x + 3) can be written as 4 × x + 4 × 3, which can be written as 4x + 12.
This idea can be extended to multiply out two brackets as (x + 4) (x + 3). Here
in the first bracket the x term multiplies the (x + 3) and then +4 term multiplies
the (x + 3) as follows:

( x + 4)( x + 3) = x ( x + 3) + 4( x + 3) = x 2 + 3x + 4 x + 12 = x 2 + 7 x + 12.

Expressions can be factored. For example, 6x + 24x3, by taking a factor of 6x from

both terms can be written as 6x(1 + 4x2).

1.4 Linear Equations
1.4.1 Solving Linear Equations
In engineering, the physical modeling of system behavior is done using the language
of mathematics. Different types of equations are derived that model the systems and
the simplest type of equations are called linear. An equation of the form ax + b = 0,
where a and b are constants, is said to be a linear equation with variable x.

Note: For an equation to be linear, the power the variable, in this case x, has to
be raised to is one.

The method of solving linear equations is to collect all the terms involving x
on one side of the equation and everything else on the other side. The idea is to
isolate the variable x to be on its own. The way this is achieved is through using
certain operations like addition, subtraction, multiplication, division, and others
(i.e., squaring and square rooting) to manipulate the equation so as to keep both
sides of the equation the same. This is illustrated in the following examples.

Example 1.17
Solve the following equation to find x:  x + 3 = 7.
Here it is easy to see what the answer should be for x, x = 4. But how can
this be arrived at systematically since for more complicated equations
 1.4 Linear Equations 11

the answer will not be obvious. The approach is to consider the equa-
tion as

Left Hand Side (LHS) = Right Hand Side (RHS)

So, to keep this balanced, any operation carried out on the LHS must
also be carried out on the RHS. Starting with the equation given earlier:

x+3=7 (−3 from both sides of the equation)

x +3−3= 7−3 ( tidying up both sides)
x=4 (solved for x )

Example 1.18
Solve the following equation to find x:  x – 5 = 4.
Starting with the equation:

x−5=4 (+5 from both sides of the equation)

x −5+5= 4+5 ( tidying up both sides)
x=9 (solved for x )

Example 1.19
Solve the following equation to find x:  3x = 12.
Starting with the equation:

3 x = 12 (divide by 3 on both sides of the equation))

3 x 12
= (tidying up both sides)
3 3
x=4 (solved for x )

Example 1.20
x
Solve the following equation to find x:   = 6
Starting with the equation, 5

x
=6 (multiply by 5 on both sides of the equationn)
5
x
×5=6×5 (tidying up both sides)
5
x = 30 (solved for x )

These basic rules can be applied to more complex equations as follows.

12 Review of Basic Concepts

Example 1.21
Solve the following equation to find x:  5x − 3 = 2x + 15
Starting with the equation:

5 x − 3 = 2 x + 15 (subtract 2 x from both sides)

5 x − 3 − 2 x = 2 x + 15 − 2 x (tidy up both sides)
3 x − 3 = 15 ( + 3 to both sides)
3 x − 3 + 3 = 15 + 3 (tidy up both sides)
3 x = 18 (divide by 3 both sides)
3 x 18
= (tidy up both sides)
3 3
x=6 (solved for x )

Example 1.22
Solve the following equation to find x:  5(x + 3) + 4 (2x − 3) = 2(2x + 15)
Starting with the equation:

5( x + 3) + 4 (2 x − 3) = 2(2 x + 15) (expand out the brackets))

5x + 15 + 8 x − 12 = 4 x + 30 (tidy up both sides)
13 x + 3 = 4 x + 30 (subtract 4 x from both sides)
13 x + 3 − 4 x = 4 x + 30 − 4 x (tidy up both sides)
9 x + 3 = 30 ( – 3 from both sides)
9 x + 3 − 3 = 30 − 3 (tidy up both sides)
9 x = 27 (divide by 9 both sides)
9 x 27
= (tidy up both sides)
9 9
x=3 (solved for x )

Note: Sometimes the linear equations are disguised because they involve
fractions. A good strategy is to multiply every term by the lowest common
multiple (LCM) of the denominators, as seen in the next example.

Example 1.23
x−5 4−x
Solve the following equation to find x:   − =5
4 3
 1.4 Linear Equations 13

Starting with the equation:

x−5 4−x (multiply by 12 both sides as this thhe

− =5
4 3 LCM of 4 and 3)
12( x − 5) 12(4 − x )
− = 5 × 12 (tidy up both sides)
4 3
3( x − 5) − 4(4 − x ) = 60 (expand out brackets)
3 x − 15 − 16 + 4 x = 60 (tidy up both sides)
7 x − 31 = 60 ( + 31 on both sides)
7 x − 31 + 31 = 60 + 31 (tidy up both sides)
7 x = 91 (divide by 7 both sides)
7 x 91
= (tidy up both sides)
7 7
x = 13 (solved for x )

Note: These basic ideas of solving linear equations are very important in
transposing equations.

1.4.2 Transposing Formulae
In science and engineering, formulae are used to relate physical quantities to each
other. It is found in electrical circuit theory that the power P is related to the cur-
rent I and resistance R by the following equation:

P = I 2R

Here, P is called the subject of the formula.

One may have several sets of corresponding values of I and P and want to find
the corresponding values of R. Much time and effort will be saved if the formula
was expressed with R as the subject because one then only needs to substitute the
given values of I and P in the rearranged formula.
The process of rearranging a formula so that one of the other symbols becomes
the subject is called transposing the formula. The rules used in the transposition
of formulae are essentially the same as those used in solving equations, as seen
in the previous section.

Example 1.24: Symbols Connected by a Plus Sign

Transpose T = t + 10   to make t the subject.
Rewrite this as t + 10 = T.
Now, subtracting 10 from both sides gives t = T – 10.
14 Review of Basic Concepts

Example 1.25: Symbols Connected by a Minus Sign

Transpose P – 5V = F   to make P the subject.
Adding 5V from both sides gives P = F + 5V.

Example 1.26: Symbols Connected as Products

Transpose W = IV   to make V the subject.
Divide both sides by I:

W IV
=
I I
or
W W
=V ⇒ V=
I I

Example 1.27: Symbols Connected as a Quotient

V
Transpose the formula R =    to make V the subject.
Multiply both sides by I: I

V
R×I= ×I
I
or
RI = V ⇒ V = RI

Example 1.28: Symbols Connected by Multiple Operations

Transpose V = E + IR    to make I the subject.

Subtract E from both sides:

V − E = IR
Divide both sides by R:

V−E V−E
=I ⇒ I=
R R

Example 1.29: Formulae Containing Brackets

Transpose R = P(1 + 5t)   to make t the subject.
Remove the bracket on the RHS:

R = P + 5Pt

Subtract P from both sides:

R − P = 5Pt
 1.4 Linear Equations 15

Dividing both sides by 5P:

R − P 5Pt R−P
= ⇒ t=
5P 5P 5P

Example 1.30: Formulae Containing Roots and Powers

Transpose J = I 2Rt    to make I the subject.
Divide both sides by Rt:

J
I2 =
Rt

Take the square root of both sides:

I
I=
Rt

Example 1.31

Transpose x = z − y   to make z the subject.

Square both sides:

x2 = z − y

Add y to both sides:

x2 + y = z ⇒ z = x2 + y

Example 1.32: More Difficult Problem

x+2
If y = , transpose to make x the subject.
x −1
Multiple by x − 1 both sides:

( x − 1) y = x + 2

Remove the brackets:

xy − y = x + 2

Collecting like terms:

xy − x = y + 2

Take x as a common factor:

x ( y − 1) = y + 2
16 Review of Basic Concepts

Divide by (y − 1):

y+2
x=
y −1

Example 1.33
Alpert’s equation for the ceiling jet velocity when the jet is far away from
the fire is given as

  13 1 2 
Q H
U = 0.195  
 5 
r 6 

where velocity, U, is in meters per second (m/s); total energy release rate,
 is in kilowatts (kW); and the ceiling height and radial position (r and
Q,
H) are in meters (m).
Transpose this equation to make Q the subject. First, rewrite to make Q
on the LHS:

  13 1 2 
Q H
0.195   =U
 5 
r 6

Divide by 0.195 both sides:

  13 1 2 
Q H U
 =
 5  0 . 195
r 6 

5
Multiply both sides by r 6 :

1 1  U  56
Q 3 H = r
 0.195 
2

1
Divide both sides by H 2 :

5
1  U  r 6
Q =
 0.195  1 2
3

H
 1.5 Linear Simultaneous Equations 17

Raise both sides to the power 3

3
 
5   
3 5
 U r 6 U r 2
Q =    = 
 0.195  H 21
  0.195  H 2 3

Substituting Values into a Formula

In most practical situations, one knows the value of certain variables and
constants and is required to calculate the value of another variable. In this
case, it may be needed to substitute values into the equation (remembering
to use correct units at all times) and calculate the unknown variable.

Example 1.34
9
If F = C + 32, then find F if C = 20.
5

Solution: Substituting in C = 20 gives

9
F= (20) + 32 ⇒ F = 9 × 4 + 32
5
F = 68

Now, if one needed to find the value of C given F, then it would be easier
to rearrange the equation first to make C the subject and then substitute in
the given value of F (see next example).

Example 1.35
9
Given that F = C + 32, find the value of C if F = 86.
5

Solution: Rearranging

9
C = F − 32
5
5
C = (F − 32)
9

So putting in F = 86 yields

5
C= (86 − 32) ⇒ C = 30
9

1.5 Linear Simultaneous Equations

The concern here is primarily with two equations in two variables such as x and
y, and this idea can be used to find the intersection of straight lines. There are
different methods of solving linear simultaneous equations: elimination method,
substitution method, and graphical method. The graphical method is generally
18 Review of Basic Concepts

not as accurate as the others and is not considered here. The substitution method
can sometimes involve awkward fractions and so the elimination method is gener-
ally the most preferred method.

1.5.1 Elimination Method
The method of elimination makes the coefficients of one of the variables equal,
and then the two equations are either added or subtracted in order to eliminate
that variable.

Note: In the equation 3x − 5y = 7, the coefficients of x and y are 3 and –5, respectively.

Example 1.36
Solve the simultaneous equations

x+y=6 (1)

x−y=4 (2)

Solution: This is the simplest case, and since the coefficients of both the x
and y are the same it is easy. Just add Equations 1 and 2 to eliminate the
y variable since the y coefficients are of opposite signs. If, however, the x
variable was to be eliminated first, then Equations 1 and 2 would need to be
subtracted since the coefficients are of the same signs.
Adding Equation 1 to 2 gives

2 x = 10 ⇒ x = 5

Now substituting this x = 5 into Equation 1 gives

x + y = 6 ⇒ 5+ y = 6 ⇒ y =1

Finally, the solutions are x = 5, y = 1.

Note: Remember to check the answers by seeing if they fit both the original
Equations 1 and 2.

Example 1.37
Solve the following simultaneous equations:

x + 2y = 5 (1)

3x − y = 1 (2)
 1.5 Linear Simultaneous Equations 19

Now both the coefficients of x and y are different, so the task is first to
make one of the variable coefficients the same. Here there is no preferred
choice and so try to eliminate the y variable.
To make the coefficients of y the same, multiply Equation 2 by 2, which
gives

x + 2y = 5 (3)

6x − 2 y = 2 (4)

So now adding Equations 3 and 4 gives

7x = 7 ⇒ x =1

and substituting this x = 1 back into Equation 1 gives

x + 2y = 5 ⇒ 1 + 2y = 5 ⇒ 2y = 4 ⇒ y = 2

Final solutions are x = 1, y = 2.

Finally, the most difficult situation for simultaneous equations follows.

Example 1.38
Solve the simultaneous equations

2 x + 3 y = 13 (1)

7 x − 5 y = −1 (2)

Solution: Here there is no simple choice to eliminate any of the variables.

To eliminate the y variable, Equation 1 gets multiplied by 5 and Equation 2
gets multiplied by 3 and this then has both the y coefficients equal to 15, that
is, the lowest common multiple of 3 and 5 as follows:

10 x + 15 y = 65 (3)

21x − 15 y = −3 (4)

Adding Equations 3 and 4 together gives

31x = 62 ⇒ x = 2

substitute this x = 2 into Equation 1 gives,

2 x + 3 y = 13 ⇒ 4 + 3 y = 13 ⇒ 3 y = 9 ⇒ y = 3

The final solution is x = 2, y = 3.

20 Review of Basic Concepts

1.5.2 Substitution Method
The second method of solving simultaneous linear equations involves rearrang-
ing one of the equations and substituting into the other. This technique is called
the method of substitution. It requires making one of the variables the subject and
then using this value into the second equation.

Example 1.39
Solve the following simultaneous equations using the method of substitution:

7 x + 2 y = 11 (1)

4x + y = 7 (2)

Solution: Here it is easier (by avoiding fractions) to rearrange Equation 2 to

make y the subject to give

y = 7 − 4x

This value for y can now be substituted into Equation 1 to give

7 x + 2(7 − 4 x ) = 11
7 x + 14 − 8 x = 11
− x = −3
x=3

Using this value for x into y = 7 − 4x gives y = 7 − 12 ⇒ y = −5.

The final solution is then x = 3, y = −5.

1.6 Quadratic Equations
We have seen that equations like 2x + 1 = 0 and x + 2 = 0 are examples of linear
equations. Now, suppose one has an equation like this: (x + 2)(2x − 1) = 0. This is
not a linear equation; it is called a quadratic equation because if the two sets of
brackets are expanded, it gives x2 + 3x − 2 = 0.

Note: Here the highest power of the variable x is now 2, and the word quadratic
is derived from the Latin word quadratus meaning “square.”

A quadratic equation has not one solution like the linear equations already seen,
but two solutions. What are they? Start by looking at the shape of this equation:

If (one number ) × (another number ) = 0

The only way that this can happen is if one of the numbers in the brackets is
zero. In other words, either x + 2 = 0 or 2x − 1 = 0. Now, it has already been shown
how to solve simple linear equations like these.
 1.6 Quadratic Equations 21

1
If x + 2 = 0, then x = −2. And if 2x − 1 = 0, then 2 x = 1 ⇒ x = .
2
So the answers to the equation (x + 2)(2x − 1) = 0 are x = −2 or x = 0.5.

Example 1.40
One can write the answers to the following quadratic equations fairly easily.

1
( x − 4)(3 x + 1) = 0 Solution: x = 4 or x = −
3
1 2
(10 x + 1)(5 x − 2) = 0 Solution: x = − or x =
10 5
1
x (2 x − 1) = 0 Solution: x = 0 or x =
2
3
( x − 7)(8 x − 3) = 0 Solution: x = 7 or x =
8

These equations are quadratic equations that have already been factor-
ized, that is, written as

(one expression) × (another expression) = 0

that is, (x − a)(x − b) = 0, where a and b are called the “roots” of the equation.

Leaving the topic of factorizing quadratics aside, let’s turn to the situation
when the quadratic equation doesn’t factorize. For example, x2 + 8x + 7 = 0
­factorizes and can quickly be solved as shown earlier. But changing the
final number to say 8, that is, x2 + 8x + 8 = 0, means that this cannot be
done by the factorizing method. Also in realistic engineering problems it is
highly unlikely that the quadratic equation generated will factorize. What
then? In all general cases, a quadratic formula can be used to solve the
problem. It is better to consider this method as it can be used in all cases
encountered.

1.6.1 Solving Quadratic Equations Using the Formula

Given the general quadratic equation,

ax 2 + bx + c = 0 (1.2)

where a, b, and c are numbers, then the solutions to this equation are given by the
formula known as the general solution to a quadratic equation as

− b ± b 2 − 4 ac
x= (1.3)
2a

This term underneath the square root sign is given a special name called the
discriminant and the symbol Δ, where, Δ = b2 − 4ac.
22 Review of Basic Concepts

Notes:
• If the number under the square root sign is positive, there are two real
and distinct solutions.
• If the number under the square root sign is zero, there is one real and
repeated solution.
• If the number under the square root sign is negative, there are no real
solutions. This topic is dealt with in Chapter 5, where it will be shown that
in this third case the answers can in fact be written as complex solutions.

Example 1.41
Solve the equation x2 + 8x + 8 = 0 correct to four decimal places.

Solution: First compare the given quadratic equation with the general qua-
dratic ax2 + bx + c = 0. This then determines the values of the coefficients
a, b, and c to be used in the formula as a = 1, b = 8, and c = 8. Substituting
these values into the general formula given by Equation 1.3 gives

− b ± b 2 − 4 ac
x=
2a

−8 ± 82 − 4(1)(8) −8 ± 32 −8 ± 4 2
x= = = = −4 ± 2 2
2(1) 2 2

Thus x = −1.1716 or x = −6.8284.

Example 1.42
Solve the equation 4x2 − 3x − 11 = 0 correct to three decimal places.

Solution: Again, first compare the given quadratic equation with the general
quadratic ax2 + bx + c = 0. Determining the values of the coefficients a, b,
and c as follows: a = 4, b = –3, and c = –11. Substituting these values into
the general formula given by Equation 1.3 gives

− b ± b 2 − 4 ac
x=
2a

−(−3) ± (−3)2 − 4(4)(−11) 3 ± 185 3 ± 13.60

x= = =
2(4) 8 8

Thus, x = 2.075 or x = −1.325.

Note: All quadratic equations can are solved in the same way with just dif-
ferent coefficients for the values of a, b, and c.
 1.7 Trigonometry 23

1.7 Trigonometry
Many problems in engineering especially mechanics involve forces and these can
be represented as either right angled triangles or general scalene triangles. The
need to solve these problems requires solutions to triangular problems. First, the
methods of solution to right angled triangles is considered.

1.7.1 Right-Angled Triangles
A right-angled triangle is one in which one of the angles is 90 degrees. Consider
the general right-angled triangle shown in Figure 1.2. First, the labeling of the
sides is very important. The longest side of the right-angled triangle is always
called h the hypotenuse, the side facing opposite the angle is o the opposite, and
the side next to the angle is called a the adjacent side, as shown in Figure 1.2.

Hypotenuse
h
Opposite
o

Adjacent
a

Figure 1.2 A general right-angled triangle.

When dealing with just the sides of the triangle, Pythagorean theorem can be
used to relate the different sides as follows:
h 2 = a 2 + o2 (1.4)

Hence, given any two sides of a right-angled triangle, the third side can be calcu-
lated using Equation 1.4.

Example 1.43
Find the missing length AB as shown in the triangle given in Figure 1.3.

h
6 cm

A C
8 cm

Figure 1.3 A right angled triangle with a missing side.

24 Review of Basic Concepts

Solution: Using the labeling notation of Figure 1.2 gives a = 8 cm and b =

6 cm. Then using Equation 1.4, side h can be calculated as

h 2 = a 2 + o2 = 82 + 62 = 64 + 36 = 100

giving h = 100 = 10 cm.

When dealing with angles and sides, for the right-angled triangle there are
three trigonometric ratios sin θ, cos θ, and tan θ and these are defined as follows:

o
sin θ = (1.5)
h

a
cos θ = (1.6)
h

o
tan θ = (1.7)
a

These three formulae can more easily be remembered by using the acronym:

SOH CAH TOA (1.8)

o
Note: Given that sin θ = then to find the angle θ, the inverse sine (i.e., sin –1)
h  o
function would have been used on both sides to give θ = sin− 1   . Similarly,
 h
inverse function expressions exist for the cosine and tangent functions.

When solving any triangle problem first make sure all sides are labeled cor-
rectly, and then identify which one of the ratios is required to solve the problem.

Example 1.44
Find the missing sides o and a in the triangle shown in Figure 1.4.

h = 10 cm
o

30°
A C
a

Figure 1.4 Right-angled triangle with unknown sides.

 1.7 Trigonometry 25

Solution: To find the side o, use a trigonometry identity involving the sides
o
o and side h. Using, SOH CAH TOA, that is, sin θ = needs to be used and
substituting in values for θ and h gives h

o
sin 30° = ⇒ o = 10 sin 30° ⇒ o = 5 cm
10

Example 1.45
Find the angle θ in the triangle given in Figure 1.5.

o= 3

θ
A C
a=1

Figure 1.5 Right-angled triangle with missing angle.

Solution: Having the sides o and a, use a trigonometric identity involving

o
sides o and a. SOH CAH TOA leads to the need to use tan θ = . Substituting
in values for a and o gives a

3  3
∴ tan θ = ⇒ θ = tan −1  ,
1  1 

that is, θ = 60°.

1.7.2 Scalene Triangles (Sine and Cosine Rules)

Many of the problems encountered will form a general triangle and this is called
a scalene triangle. In such cases, two new formulae are necessary to solve these
triangular problems. The first rule is called the sine rule and the second rule is
called the cosine rule.

1.7.2.1 Sine Rule
Given a general triangle as shown in Figure 1.6, the labeling is again very important
as the formulae depend on it. Here, the angles are denoted by the capital letters and
the side opposite that angle is denoted by the corresponding lowercase letter.
The sine rule relates the sine of an angle with its opposite side as follows:

a b c
= = (1.9)
sin A sin B sin C
26 Review of Basic Concepts

c
a

A
b C

Figure 1.6 General scalene triangle.

Note: In Equation 1.9 any two of the relationships are used as required,
that is,
a b a c b c
= or = or =
sin A sin B sin A sin C sin B sin C

Example 1.46
In the triangle given in Figure 1.7, find the length of side AC.

Solution: First find the missing angle B, angle B is given by 180° − (83° +
62°) = 35°
a b c
So now using the sine rule, = = , since the side AC (i.e.,
sin A sin B sin C
side b) is required and side a and angle A are also known, then using

a b 10 b 10 × sin 35
= ⇒ = ⇒ b= = 5.78 cm
sin A sin B sin 83 sin 35 sin 83

Note: It is not always possible to make use of the sine rule directly, so
sometimes there is a need for the second rule.

c 83° b

B 62°
a = 10 C

Figure 1.7 A general triangle to find missing sides and angles.

 1.7 Trigonometry 27

c
a

A
b C

Figure 1.8 General scalene triangle.

1.7.2.2 Cosine Rule
The cosine rule is needed when there are two given sides and only the angle
between them is given. Consider the general triangle in Figure 1.8.
The Cosine rule is stated as

a 2 = b 2 + c 2 − 2bc cos A (1.10)

And by symmetry,

b 2 = a 2 + c 2 − 2ac cos B (1.11)

c 2 = a 2 + b 2 − 2ab cos C (1.12)

To find the angles when all three sides are known, Equation 1.10 can be rear-
ranged to give

b2 + c2 − a2
cos A = (1.13)
2bc

Example 1.47
Find the length AC given in Figure 1.9.

Solution: Using the cosine rule given by Equation 1.11, b2 = a2 + c2 − 2ac cos B
gives

b 2 = 10 2 + 82 − 2 × 10 × 8 × cos 60° ⇒ b 2 = 84 ⇒ b = 9.17 cm

c=8 b

62°

B
a = 10 C

Figure 1.9 A general triangle using the cosine rule.

28 Review of Basic Concepts

1.7.3 Resultant Forces
1.7.3.1 Adding Two Forces
Consider two forces of magnitude, F1 and F2 , that act upon a particle. If these
forces are placed end to end, it can be seen that they have the same effect as
a single force of magnitude F, as in Figure 1.10. This force is known as the
resultant force. The resultant force will form the third side in a triangle of
forces. To calculate this resultant force and its direction, use of the cosine and
sine rule are made.

F2 Resultant
R

F2
F1
F1

Figure 1.10 Resultant forces forming general triangles.

Example 1.48
Two forces of magnitude, 6 N and 5 N, act on a particle. The angle
between the forces is 40°. Find the magnitude and direction of the resul-
tant force.

Solution: Drawing the resultant diagram as shown in Figure 1.11.

First, using the cosine rule in the triangle of forces to calculate the size of
the resultant force gives

F 2 = 62 + 52 − 2 × 6 × 5 cos 140° = 106.96

F = 10.34 N

To find the direction of the resultant force with respect to the 6 N force
means finding the angle θ in Figure 1.11. This can be done using the sine
rule as

5 F 5 × sin 140
= ⇒ sin θ = ⇒ θ = 18.1°
sin θ sin 140 10.34

So, the resultant force has magnitude 10.34 N and is at an angle 18.1° to the
6 N force.

Resultant
5N R
40° 140°
θ 5N

6N 6N

Figure 1.11 Diagram showing resultant force forming a general triangle of forces.
 1.7 Trigonometry 29

1.7.4 Basic Trigonometric Identities

Having already seen the trigonometric ratios for sin θ, cos θ, and tan θ, that is,

o a o
sin θ = , cos θ = , tan θ =
h h a

There are important trigonometric identities that can be derived and are very
useful in applications later on. Two basic identities are

cos2 θ + sin 2 θ = 1 (1.14)

sin θ
tan θ = (1.15)
cos θ

Note: These can be proved from the aforementioned trigonometric ratios and
can be used to prove other trigonometric identities as well as solve trigonometric
equations.

Example 1.49
3 4
Find the value of tan θ when sin θ = and cos θ = − .
5 5
Solution:

sin θ 3  4  3  5  3
tan θ = = ÷ − = × − =−
cos θ 5  5  5  4  4

Example 1.50

1 − sin θ 1
Show that ≡ − tan θ .
cos θ cos θ

Note: When asked to “show that” or “prove that” then consider starting
with one side of the identity and, step by step, reduce it to the same form as
the other side of the identity.

Solution:

1 − sin θ 1 sin θ 1
LHS: ≡ − ≡ − tan θ
cos θ cos θ cos θ cos θ

Example 1.51

2 − cos2 θ
Show that ≡ 1.
1 + sin 2 θ
30 Review of Basic Concepts

Solution: Since cos2 θ + sin 2 θ = 1 ⇒ cos2 θ = 1 − sin 2 θ

So,

2 − cos2 θ 2 − 1(1 − sin 2 θ ) 1 + sin 2 θ

LHS: ≡ ≡ ≡1
1 + sin 2 θ 1 + sin 2 θ 1 + sin 2 θ

There are other useful compound formulae for the sine, cosine, and tangent as
follows:
cos( A ± B) = cos A cos B  sin A sin B (1.16)

sin( A ± B) = sin A cos B ± cos A sin B (1.17)

tan A ± tan B
tan( A ± B) = (1.18)
1 tan A tan B

Other important trigonometric formulae can be derived from Equations 1.16

and 1.17 by adding and subtracting these to give the following:

1
cos( A) cos( B) = ( cos( A + B) + cos( A − B)) (1.19)
2

1
sin( A)sin( B) = ( cos( A − B) − cos( A + B)) (1.20)
2

1
sin( A) cos( B) = (sin( A + B) + sin( A − B)) (1.21)
2

The preceding formulae are very important whenever the need arises to trans-
form the product of sines and cosines into sums and are a very useful in the
techniques of integration and applications in areas such as Fourier series, seen
later in Chapter 9.

1.7.5 Radian Measure
When it comes to measuring angles generally the measurement that has been used
is degrees. A whole circle is 360°, a straight line is 180°, and a right angle is 90°,
for example. However, when it comes to finding gradients of curves, or rates of
change (i.e., to differentiate trigonometric functions), a different unit of measure-
ment is used called the radian.
Radian is short for the radius angle, and it means the angle given at the center
of a circle by an arc of one radius as shown in Figure 1.12.
Now the relationship between radians and degrees can be determined as fol-
lows: There are 360° in a complete revolution of the circle, the circumference of a
circle is 2πr, and so the number of radiuses r around the circumference is given by
2π r
= 2π radians. Therefore, this gives an important formula relating degrees
r
to radians:
360° ≡ 2π radians (1.22)
 1.8 Statistics 31

1 radian r

Figure 1.12 Definition of the radian angle.

Table 1.2 Relationships between Degrees and Radians

Angle in Degrees Angle in Radians
360° 2π
180° π
π
90°
2
π
60°
3
π
45°
4

The standard relationships that exist between degrees and radians is shown in
Table 1.2.
When talking about an angle in degrees, one should write the degrees sign
as 45.2°. When talking about an angle in radians, a sign is not used. Thus
π
cos(60°) = cos   = 0.5 and sin (25°) = 0.4226, but sin (25) = −0.1324 because
 3
this means sin (25 radians).

1.7.5.1 Radians on the Calculator

Trigonometric problems can be solved in either degrees or radians. Probably the
need to use radians will arise when doing certain types of problems involving
calculus. The calculator can be set to work in radians instead of degrees using the
Mode button.

1.8 Statistics
1.8.1 Introduction
Statistics deals with all aspects of data: collecting the data, pictorial representa-
tion of the data, and numerical analysis and drawing final conclusions on findings.
There are vast areas where statistics are used ranging from opinion pollsters (test-
ing public opinions on issues), governmental national statistics, and in science and
engineering testing theories.
32 Review of Basic Concepts

Two types of variables

Qualatative Quantative
(no numerical value,
(having numerical value)
colour of eyes etc.)

Discrete variable Continuous variable

(countable, number of (measured height, weight
people etc.) etc.)

Figure 1.13 Different types of variables in statistics.

There are two types of variables, as shown on Figure 1.13, and any quantity
that varies is called a variable.
There are some different aspects to statistical data that need defining and fur-
ther explanation:

Primary data—This is data collected for a specific investigation. This could

involve carrying out actual experiments, sending out questionnaires, and
so forth.

Secondary data—Organizations and governments publish a vast amount

data on a wide variety of subjects that appear in different publications.
They provide useful data for investigations but they were not collected
specifically for the investigation.

Population—This is all the possible data for the research question. If the
question was to determine the average height of male adults in the United
Kingdom, then the population would need to include measuring every
single adult male.

Sample—This is just part of the data set. In some practical situations, it

makes no sense to consider the whole population, as this may be too large
(i.e., in millions) and so a smaller sample is considered more appropriate.
Sampling is very important and useful; it reduces the amount of data that
needs to be collected.

Sampling without bias—A bias is anything that makes the sample unrepresenta-
tive (e.g., asking only certain members of a community their views a topic).

Random sampling—A random sample of size n is a sample selected so that

all possible samples of size n have an equal chance of being selected.

1.8.2 Measures of Averages
There are three main measures of average of a set of numerical data:

Mode—The one that occurs most frequently (or often).

 1.8 Statistics 33

Median—Arrange the data in order of magnitude (size), then find the cen-
tral value. If there is an even number of data values, then use halfway
between the two middle values.

Mean—Add all the data values and divide by the total number of data points.

Note: For most purposes the mean is considered the most useful measure of aver-
age since it uses all data.

The symbol μ (pronounced myü) is used for the mean of the whole population
and the symbol x (x bar) is used for the mean of the sample. The definition of the
mean can be more concisely written in formula form as

x=
∑x i
(1.23)
n

Note: Here the symbol ∑ x means to sum or add together all the x values.
i i

Equation 1.23 is used for a simple set of data (i.e., a small number of data
points) and where n is the number of data points.

Example 1.52
The number of fires reported in 20 consecutive weeks is given by the data
set as:

4 7 12 13 0 5 21 13 10 6
6 8 15 9 6 0 14 12 6 8

Find the modal, median, and mean number of fires per week.

Solution: It is first easier to put the data set in order as

0 0 4 5 6 6 6 6 7 8 8 9 10 12 12 13 13 14 15 21

The modal number is, 6 as this occurs more times than all others.
For the median, since there is an even number of data points (i.e., 20 data
points) look at the 10th and 11th data values. In this case they are both 8 and
8, which gives the middle of these as

8+8
Median = =8
2
34 Review of Basic Concepts

For the mean, use the formula given by Equation 1.23:

x=
∑x i
=
175
= 8.75.
n 20

1.8.2.1 Data in a Frequency Table

Sometimes when large data sets are involved, it is no longer easy or practical to
write out the data points as a list especially if there are hundreds or thousands
of data points. In this situation, it is more convenient to use a frequency table to
represent the data values. The frequency is the number of times an observation
occurs. The formula for the mean now modifies as

x=
∑x f i i
(1.24)
∑f i

where fi is just the frequency associated with the ith data point.

Example 1.53
A survey was carried out to see the distribution of the ages (in years) of fire
engines in a particular region of the country. Twenty-seven vehicles were
surveyed and the results are shown in Table 1.3. Find the mode, median, and
mean age of the fire engines.

Solution: The first thing to do is redraw the table with an extra column for
the product term xi fi, as shown in Table 1.4. The columns can be filled in
and the products computed. At the bottom of the columns, the sums of the
columns can also be calculated.

Mode: 0 (i.e., highest frequency). These will be new fire engines.

27 + 1
Median: There are 27 data points. To find the middle, 2
= 14th ⇒
data point ⇒ 2 years.

Mean: x =
∑x f =
i i 45
= 1.67 years.
∑f i
27

Table 1.3 A Frequency Table

Showing Ages of Fire Engines
Ages of Fire Engine, Frequency,
xi fi
0 9
1 4
2 6
3 5
4 2
5 0
6 1
 1.8 Statistics 35

Table 1.4 Constructing the Extended Frequency Table

Age of Fire Engine, Frequency, Product,
xi fi xi fi
0 9 0×9=0
1 4 1×4=4
2 6 2 × 6 = 12
3 5 3 × 5 = 15
4 2 4×2=8
5 0 5×0=0
6 1 6×1=6

∑ f = 27
i ∑ x f = 45
i i

1.8.2.2 Grouped Data
Sometimes the individual data points are not provided but only the grouped fre-
quency table is given. In this case it is not possible to give the exact values of the
three averages but only estimates.
For the mean, the method is the same as for Example 1.48, but first the mid-
point of the grouped data set must be determined. This is a good approximation
since some of the data values will be higher than this and some will be lower than
this middle value. The next example shows how this method works.

Example 1.54
Consider the marks for 61 students who took a test as given in Table 1.5.
Find the modal class, median, and the mean percentage mark for this class.

Solution:

Modal class: 51%–60% (this is the most frequently occurring).

Median: Middle value, 27 + 1 = 31st data value, so it’s the 6th data value
2
in the class interval 51–60 and so needs calculating as follows:

6
51 + × (class width = 10) = 53.73%
22

Table 1.5 A Frequency Table

Showing Grouped Data
Marks%, Frequency,
xi fi
1–10 1
11–20 4
21–30 3
31–40 7
41–50 10
51–60 22
61–70 8
71–80 3
81–90 2
91–100 1
36 Review of Basic Concepts

Table 1.6 Calculating an Estimate of the Mean for a Grouped

Data Set
Marks%, Frequency, Midpoint%, Product,
xi fi xm xmfi
1–10 1 5.5 5.5
11–20 4 15.5 62
21–30 3 25.5 76.5
31–40 7 35.5 248.5
41–50 10 45.5 455
51–60 22 55.5 1221
61–70 8 65.5 524
71–80 3 75.5 226.5
81–90 2 85.5 171
91–100 1 95.5 95.5

∑ f = 61i ∑x f = 3085.5
m i

The mean is obtained as before, but now the midpoint of the class width is
needed since the exact data values are not known, as shown in Table 1.6.
So the estimated mean is

∑ x f = 50.58%
m i

∑f i

1.8.3 Measures of Spread
In the previous section it was seen that for a data set an average value could be
calculated to indicate something about the data points. However, this does not
say anything about how the data points are spread out. There are three common
measures of spread: The range (R), the interquartile range (IQR), and the standard
deviation (σ) are defined next.

1.8.3.1 Range
The range for a set of data is the difference between the highest value and the
lowest value (i.e., it considers the extreme values of the data set).

Example 1.55
See the following two data sets A and B:

A: 4, 5, 5, 6, 7, 9
B: 1, 3, 3, 5, 6, 8, 10, 12

The range for set A is R = 9 – 4 = 5 and for data set B is R = 12 – 1 =

11. So it can be seen that even though the two data sets have the same mean
(i.e., 6), the set B has more variation in the data points than set A.

1.8.3.2 Interquartile Range
When a set of n data is written in order of magnitude (size), the median is given
(n + 1)
by th item of data.
2
 1.8 Statistics 37

The quartiles are found in a similar way:

• The lower quartile is the median of the lower half of the data.

• The upper quartile is the median of the upper half of the data.

• Then the Interquartile range (IQR) = Upper quartile – Lower quartile.

Example 1.56
Find the interquartile range (IQR) for the following set of data:

24, 24, 25, 26, 26, 26, 27, 27, 30, 33, 33, 35, 35, 36, 43

Solution: There are 15 data values, so the median is the eighth data value.
Thus, median = 27.

Lower quartile has 7 data values ⇒ median of this is the fourth value = 26.
Upper quartile has 7 data values ⇒ median of this is 35.

∴ IQR = UQ − LQ = 35 − 26 = 9

1.8.3.3 Standard Deviation (σ)

The range and IQR do not use all the data to measure the spread. A more useful
indicator of the average spread of data about the mean is the standard deviation
and is used along with the mean widely in science and engineering. The formulae
for standard deviation for a population and for a sample are as follows.

1.8.3.3.1 Population Standard Deviation (σ)

Standard deviation of a population for a simple data set x1, x2, x3, …, xn, is denoted
by σ and is given by

σ=
∑ (x − µ)i
2

(1.25)
n

where μ is the population mean and n is the number of data points.

For a frequency distribution

σ=
∑ (x − µ) f
i
2
i
(1.26)
∑f i

where fi are the individual frequencies associated with the xi.

Note: The units of σ are the same as the units of x and x .

38 Review of Basic Concepts

1.8.3.4 Sample Standard Deviation (s)

If the data is from a small sample of a population, the population mean μ is not
calculated but rather the sample mean x . Then the estimated standard deviation
(s) is given by

s=
∑ (x − x ) i
2

(1.27)
n −1

Here, division is by n − 1 instead of n.

Again, for a frequency distribution

s=
∑ (x − x ) f
i
2
i
(1.28)
∑ f −1 i

Note: Generally, it is much easier to calculate the mean and standard deviation
using a calculator (in statistics mode).

Example 1.57
Look at the following set of data and find the mean and standard deviation:

xi : 1, 2, 3, 4, 5

Solution:

Mean: µ =
∑x i
=
15
=3
n 5
Now use Table 1.7 to find the standard deviation σ as

∴σ=
∑ (x − µ)
i
2

=
10
= 2 = 1.414
n 5

Table 1.7 Calculating the Standard

Deviation of a Simple Data Set
xi (xi − μ) (xi − μ)2
1 –2 4
2 –1 1
3 0 0
4 1 1
5 2 4

∑ (x − µ)i
2
= 10
 1.8 Statistics 39

Example 1.58
The number of fires over 52 weeks was recorded at a particular fire station.
The results are shown as follows:

No. of fires, xi 4 5 6 7 8 9 10
Frequency, fi 3 10 18 6 5 6 4

Calculate the mean and standard deviation of the number of fires per
week.

Solution: The calculation are shown in Table 1.8 as

µ=
∑x f i i
=
346
= 6.65
∑f i
52

σ=
∑ (x − µ) f
i
2
i
=
143.77
= 2.765 = 1.66
∑f i
52

Finally, the square of the standard deviation (σ2) is given a special name
as the variance.
Table 1.8 Calculating Mean and Standard Deviation for Data
in a Frequency Table
xi fi xi fi (xi − μ) (xi − μ)2 fi
4 3 12 –2.65 21.0675
5 10 50 –1.65 27.225
6 18 108 –0.65 7.605
7 6 42 0.35 0.735
8 5 40 1.35 9.1125
9 6 54 2.35 33.135
10 4 40 3.35 44.89

∑ f = 52 ∑ x f = 346
i i i ∑ (x − µ)
i
2
fi = 143.77

1.8.4 Change of Scale
The weekly wages of employees in a small company have a mean of £290 and stan-
dard deviation of £42. If there is a pay rise of £15 for each employee, what happens
to the mean and standard deviation? Since each wage is increased by £15, the mean
wage will be increased by £15 to £305. However, the standard deviation measures
variability and this is unchanged since all wages have increased by the same amount.
If instead of a flat rate increase an increase of 10% is given to each employee, the
variability would increase because the higher paid employees would get a larger rise.
In this case both the mean and standard deviation would increase by 10%. In sum-
mary, if a variable is increased by a constant amount, its average will be increased
by this amount, but the spread will be unchanged. If the variable is multiplied by a
constant amount, both its average and spread will be multiplied by this amount.
40 Review of Basic Concepts

1.9 Applications
Example 1.59: Relationship between Sizes of Pool Fires to Flame Heights
A schematic diagram of a pool fire with its flame height is shown in
Figure 1.14. From pool fire experiments, a power law fit was determined
relating the heat release rate Q (kW) and the diameter of the pool fire D (m)
to the flame height L (m) using Heskestad’s equation as follows:

2
L = 0.235 Q 5 − 1.02 D (1.29)

Knowing the heat release rate for a fire and its diameter size, then using
Equation 1.29 can predict the height of the flame.
Consider a 500 kW diesel pool fire with a diameter of 1.5 m, the flame
height is given by substituting in the data values into Equation 1.29 as

2
L = 0.235(500) 5 − 1.02(1.5) = 1.29 m

D = 1.5 m

Figure 1.14 Diesel pool fire with its associated flame height.

Example 1.60: Enclosure Fire with a Ceiling

Figure 1.15 shows the development of a fire within an enclosure of height
H. When the fire plume interacts with the ceiling it cannot travel fur-
ther upward, so it creates a horizontal ceiling jet. The temperature of the

Layer of hot gas and smoke

Fuel array

Figure 1.15 Smoke layer development with an enclosure ceiling.

 1.9 Applications 41

smoke layer measured radially far away from the fire is given by Alpert’s
equation as

2
(Q /r ) 3
Tjet − T0 = 5.38 (1.30)
H

where Tjet is the smoke layer temperature, T0 is the ambient room tempera-
ture, Q is the heat release rate, r is the radial distance of the smoke, and H
is the room height.
If a sprinkler was located 5m radially from the fire with a ceiling height
of 3m and the fire size is 1200 kW with an ambient temperature of 20°C
then what is the expected gas temperature at the sprinkler?
Equation 1.30 can be transposed to make Tjet the subject of the formula
to give

2
(Q /r ) 3
Tjet = T0 + 5.38 (1.31)
H

Substituting in the values of the parameters into Equation 1.31 gives

2
(240) 3
Tjet = 20 + 5.38 = 89°C
3

Note: If the temperature to activate a sprinkler or smoke detector is known,

then the equation can be transposed to solve for Q.  This can be useful for
design purposes in fire calculations to determine the fuel size required to
activate a sprinkler or smoke detector.

Example 1.61: Direction of Smoke Flow with Wind Effects

A smoking fire in a room is located at a point A as shown in Figure 1.16.
Smoke flows vertically upward with a velocity Vs = 1.2 ms−1. If a window is
open and the wind with velocity Vw = 0.6 ms−1 is flowing due east, calculate

Vw

Vs

Figure 1.16 Smoke movement with wind effects.

42 Review of Basic Concepts

the resultant velocity VR of the smoke flow and the direction it flows with
reference to the horizontal floor.
First constructing a triangle of velocities gives Figure 1.17.

Vw = 0.6

Vs = 1.2
VR
α

Figure 1.17 Resultant velocity of the smoke flow.

It can be seen from the right-angled triangle that

V = 1.22 + 0.62 = 18 ∴ VR = 1.34 ms−1.
2
R
The direction of the smoke flow is given by the angle α as
1.2
tan α = ⇒ α = 63.4°.
0.6

Example 1.62: Change in Mean and Standard Deviation

A sprinkler, designed to extinguish house fires, is activated at high tem-
peratures. A batch of sprinklers is tested and found to be activated at
a mean temperature of 72°C with a standard deviation of 3°C. Find the
mean and standard deviation of the temperature in degrees Fahrenheit.
The formula relating degrees centigrade to degrees Fahrenheit is
F = 1.8C + 32. So the mean would be multiplied by 1.8 and add to it 32
to give

Mean = (72 × 1.8) + 32 = 161.6°F

But for the standard deviation, only carry out the multiplication since the
addition of 32 will have no effect on the variability giving

Standard deviation = 3 × 1.8 = 5.4°F

Problems
1.1 Transpose the following formulae for the variable given:

3.6V
a. L = for A
A

b. F = G + 7H for H

k
c. α = for k
ρc

d. v = h − gt for t
 Problems 43

d2
e. t = for d
4α

f. I = εσ AT 4 for T

g H2
g. τ = t for t
H S

1 2
h. E = mv for v
2

h+3
i. g = for h
h−5

1
 Q  3
j. U = 0.96   for Q
 H

1.2 The equation for the smoke jet Tj is given by Equation 1.31 as

Tj = T0 + 5.38
(Q /r ) 3

a. By transposing, find an expression for the heat release rate Q .

b. Given that H = 15m, r = 8m, T0 = 20°C, and Tj = 100°C determine


the value of Q.

1.3 Solve the following equations:

a. a + b = 2 and 5a + b = 14

b. 2x + 3y = 4.5 and 5x − 2y = 6.5

c. x2 + 5x − 3 = 0

d. 3x2 − 6x − 11 = 0

1.4 A fireboat in Hong Kong starting at a point O is crossing a river that

has two parallel banks. The width of the river is 200 m. The water
in the river is flowing at a speed of V ms−1. Point A is directly oppo-
site point O on the other bank. The velocity of the boat relative to
the water is 20 ms−1 at angle of 70° to the bank. The boat lands at a
point B were a building is burning, which is 30 m from A. The angle
between the actual path of the boat and the bank is θ°. Figure 1.18
shows the corresponding river and the velocities of the boat and the
water.
44 Review of Basic Concepts

30 m
(Burning building)

A B

V ms–1

20 ms–1 200 m

70° θ°

O
Fire boat station

Figure 1.18 Fireboat crossing a river bank.

a. Find the time it takes the fireboat to cross the river and the actual
direction of the boat θ.

b. Determine the velocity of the water flow.

1.5 The number of defective components reported to a company per week

is given as follows:

No. of defective components, xi 5 7 10 13 16 19

Frequency, fi 2 3 8 5 3 1

Calculate the mean number of defective components reported per

week and the standard deviation.
 2 Introduction to
Probability Theory

2.1 Introduction
Most people deal with uncertain events each day in their lives but rarely pause to
calculate chances or probabilities. Yet whole professions, such as the insurance
industry, pensions, investment advisors, and bookmakers are founded upon prob-
ability. In engineering, the ideas of probability are very important when consider-
ing the overall reliability of systems and in the areas of risk assessment.
How is the probability of something happening defined? The usual definition
supposes that one repeats an “experiment” many times and records the outcomes.
For example, roll a fair dice many times and record how often the result is a six.
The probability of a six occurring is given by

Number of times a six occurs

P(6) =
Total number of throws

1
If the die was fair and rolled lots of times, the expected probability would be 6 .

Therefore, generally the probability of an event A is defined as

Number of ways A can occur

P( A) = (2.1)
Total number of possible outcomes

Note: If A denotes the event A happening, then A′ or (Ac) is called the complement
of A and denotes the event A not happening.

45
46 Introduction to Probability Theory

There are several consequences of this definition. A probability must lie

between zero and 1. Also, adding the probabilities of all the outcomes, the total
must be 1. This gives the following: P(A) + P(A′) = 1 and so

P( A′) = 1 − P( A) (2.2)

For example, for a fair dice P(1) = P(2) = P(3) = P(4) = P(5) = P(6). And of
course 1 + 1 + 1 + 1 + 1 + 1 = 1. But it can also be concluded, for example, since
6 6 6 6 6 6

P(6) = 1
, then P(not a 6) = 1 − 1 = 5 .
6 6 6
It is not always possible to assign a probability using equally likely outcomes.
If someone wanted to know the probability of going to the bus stop and having
to wait for a bus for more than 5 minutes, then some trials would have to be done
to find the waiting times. The relative frequency of an event is the proportion of
times it has been observed to happen. If somebody went for a bus on 40 weekday
mornings and on 16 of these they had to wait more than 5 minutes, one could
16
assign a probability of = 0.4 to the event of having to wait more than 5 minutes.
40

2.1.1 Mutually Exclusive Events

Mutually exclusive events are those that cannot both happen, for example, scoring
a 3 and scoring a 4 when throwing a dice. One can find the probability that either
one event or the other happens by adding the probabilities of the two events.

Example 2.1
When rolling a dice, the

1 1 1
P(either getting a 3 or 4) = + =
6 6 3

Example 2.2
When choosing one card from a pack, the probability of getting a ten card
or a jack card is

4 4 8 2
P(a ten or a jack ) = P( ten) + P( jack ) = + = =
52 52 52 13

Generally, for mutually exclusive events the relationship is

P( A or B) = P( A) + P( B) (2.3)

This can be extended for any number of mutually exclusive events as

( )
P A1 or A2 or A3 ……or An = P( A1 ) + P( A2 ) + P( A3 ) +……. + P( An )
 2.1 Introduction 47

Or in concise form as,

 n  n


P  Ai  =
 i =1 
∑ P( A )
i =1
i

If the events are not mutually exclusive these formulae are modified. For
example Equation 2.3 now becomes

P ( A or B) = P( A) + P( B) − P( A and B) (2.4)

This is because these events can happen together and the probability of
B is included in the probability of A and vice versa, so this means that the
probability has been included twice and so one of these must be subtracted
from the answer.

2.1.2 Independent Events
Independent events are those that have no influence on each other. When tossing
a coin twice, the result of the first throw has no effect on the second.
In such a case, the probability of both events happening is found by multiply-
ing the separate probabilities.

Example 2.3

1 1 1
P (head on both tosses of a coin) = P (head) × P (head) = × =
2 2 4
1 1 1
P (rolling two sixes with a dice) = P (6) × P (6) = × =
6 6 36

Generally, for independent events:

P( A and B) = P( A) × P( B) (2.5)

This can be extended for any number of independent events as

( )
P A1 and A2 and A3 ……and An = P( A1 ) P( A2 ) P( A3 )…….P( An )

or in concise form as,

 n  n


P  Ai  =
 i =1 
∏ P( A )
i =1
i

If the events are not independent, then one has conditional probabilities
and these formulae are modified (see later).
48 Introduction to Probability Theory

Example 2.4
The probability that telephone calls to a railway timetable inquiry service
are answered is 0.7. If three calls are made find the probability that all three
are answered and exactly two are answered.

Solution: If A is the event of a call being answered, P(A) = 0.7, then A′ is the
probability of a call not being answered, and P(A′) = 1 – 0.7 = 0.3.
Using the multiplication rule the probability of AAA = 0.7 × 0.7 ×
0.7 = 0.343.
If one call is unanswered it could be the first, second, or the third call:

A′AA with probability 0.3 × 0.7 × 0.7 = 0.147

AA′A with probability 0.7 × 0.3 × 0.7 = 0.147
AAA′ with probability 0.7 × 0.7 × 0.3 = 0.147

These three outcomes are mutually exclusive and so one can apply the
addition law and find the probability of exactly two calls being answered to
be 0.147 + 0.147 + 0.147 = 0.441.

2.1.3 Conditional Probability
Sometimes the events A and B are not independent, and so the probability of the
event A happening depends on the event B happening. This is represented as follows:

P(A|B) denotes the probability that event A happens given that event B
happens.

Two events A and B are independent if P(A) = P(A|B).

So the multiplication law given by Equation 2.5 becomes more general as

P( A and B) = P( A) × P ( B A ) (2.6)

Example 2.5
James buys ten apparently identical oranges. Unknown to him the flesh of
two of these oranges is rotten. He selects two of the ten oranges at random
and gives them to his friend. Find the probability that

a. Both oranges are rotten.

b. Exactly one of the oranges is rotten.

Solution: Sometimes in probability type problems it is easier to see what is

happening using a tree diagram (discussed later in Section 2.1.5) as shown
in Figure 2.1.

2
a. The probability of both oranges being rotten is = 0.022.
90
16 16
b. The probability of exactly one orange being rotten is + = 0.35.
90 90
 2.1 Introduction 49

1st orange 2nd orange Outcome Probability

2 1 2
1 × =
Rotten Rotten, rotten 10 9 90
9

2 2 8 16
Rotten Rotten, o.k. × =
10 10 9 90
8
O.k.
9

2
Rotten 8 2 16
9 O.k., rotten × =
8 10 9 90
O.k.
10

7 8 7 56
O.k. O.k., o.k. × =
9 10 9 90

Figure 2.1 Tree diagram showing possible outcomes and probabilities.

2.1.4 Bayes’ Theorem
Founded by the Rev. Thomas Bayes (1701–1761), who apart from being a minister
was a statistician and philosopher, Bayes’ theorem is a fact about probabilities
and has a lot of real-world applications. It gives a way of working out what the
conditional probabilities should be.
A notation that is used for the following case is that the probability of a hypoth-
esis, H, given that a new piece of evidence E is written as P(H\E).
Bayes’ theorem states

P( E \H ) P( H )
P( H \E ) = (2.7)
P(E )

Proof: Generally, we have from the multiplication laws of probabilities:

P( H and E ) = P( H ) P( E \H )

P( E and H ) = P( E ) P( H \E )

Now these two are equal, so

P( E ) P( H \E ) = P( E \H ) P( H )

P( E \H ) P( H )
P( H \E ) =
P(E )

which is Bayes’ theorem.

50 Introduction to Probability Theory

Example 2.6
One day a person who does not feel well decides to go onto the Internet to
find out what might be wrong. Let’s say the person found an illness called
hypothesitis, H. So the probability of the symptoms given hypothesitis is
P(E\H) = 0.95.
But Bayes’ theorem can be used to find the probability that the person
will have hypothesitis given the symptoms as

P( E \H ) P( H )
P( H \E ) =
P(E )

The following information is still needed. The prior probability of

hypothesitis P(H), which is found to be a rare disease with probability
P(H) = 0.00001. Also, the kind of symptoms does the person have, P(E)
(e.g., headache and cold) is P(E) = 0.01. So now P(H\E) is given by, using
Equation 2.7, as

P( E \H ) P( H ) 0.95 × 0.0001
P( H \E ) = = = 0.00095
P(E ) 0.01

Very small indeed!

2.1.4.1 Generalization of Bayes’ Theorem

Often, for some partition {Aj} of the sample space, the event space is given by or
conceptualized in terms of P(Aj) and P(B\Aj). It is then useful to compute P(B)
using the total law of probability as

P ( B) = ∑ P(B \A ) P( A )
j
j j

which then gives a generalized Bayes’ theorem as

P( B \Ai ) P( Ai )
P( Ai \B) = (2.8)
∑ P(B \A ) P( A )
j
j j

Example 2.7
The output from a factory is produced on three machines: A1, A 2 , and
A 3. The three machines account for 20%, 30%, and 50% of the out-
put, respectively. The fraction of defective items produced is 5% for
A1, 3% for A 2 , and 1% for A 3. If an item is chosen at random and the
output is found to be defective, what is the probability it was produced
by machine A 3?

Solution: Let Ai denote the event that a randomly chosen item was made by
the ith machine (i = 1, 2, 3). Let B denote the event that a randomly chosen
item is defective. So we know P(A1) = 0.2, P(A2) = 0.3, and P(A3) = 0.5.
Also, P(B\A1) = 0.05, P(B\A2) = 0.03, and P(B\A3) = 0.01.
 2.1 Introduction 51

So using Bayes’ theorem

P( B \A3 ) P( A3 )
P( A3 \B) =
∑
3
P( B \A j ) P( A j )
j =1

0.01 × 0.5 5
P( A3 \B) = = = 0.208
0.05 × 0.2 + 0.03 × 0.3 + 0.01 × 0.5 24

2.1.5 Tree Diagrams
An alternative approach to solving probability problems involving a series of
events is with a tree diagram. Consider the case of telephone calls made to an
engineering supply company. The probability that a call is answered is given as
0.7. A tree diagram showing the possible outcomes is given in Figure 2.2. Each
branch shows the possible outcomes of each call and their probabilities. Here, A =
call answered and A′ = call not answered. The outcome of the three calls is found
by reading along the branches leading to it, and the probability of this outcome is
found by multiplying the individual probabilities along these branches.

Note: This is an important concept in assessing the risk of certain events occur-
ring. See examples in applications section.

From the tree diagram the probabilities of different outcomes can be calcu-
lated. The probability of all three calls being answered AAA can be seen to be
0.343. The probability of exactly two calls being answered is the sum of the
probabilities of the three outcomes: AAA′, AA′A, and A′AA = 0.147 + 0.147 +
0.147 = 0.441.

1st call 2nd call 3rd call Outcome Probability

A 0.7 AAA 0.7 × 0.7 × 0.7 = 0.343

A 0.7 0.7 × 0.7 × 0.3 = 0.147

A’ 0.3 A A A’

A 0.7 A A’ A 0.7 × 0.3 × 0.7 = 0.147

A 0.7 A’ 0.3
A’ 0.3 A A’ A’ 0.7 × 0.3 × 0.3 = 0.063

A’ A A 0.3 × 0.7 × 0.7 = 0.147

A 0.7
A 0.7 0.3 × 0.7 × 0.3 = 0.063
A’ 0.3 A’ 0.3 A’ A A’

A’ 0.3 A 0.7 A’ A’ A 0.3 × 0.3 × 0.7 = 0.063

A’ 0.3 A’ A’ A’ 0.3 × 0.3 × 0.3 = 0.027

Figure 2.2 Tree diagram showing possible outcomes and probabilities.

52 Introduction to Probability Theory

2.2 Discrete Random Variables

2.2.1 Discrete Probability Distribution
Example 2.8
Consider a board game where a turn consists of throwing a die and then
moving a number of squares equal to the score on the die. “The number of
squares moved in a turn” is a variable because it can take different values,
namely, 1, 2, 3, 4, 5, and 6.
However, the value taken at any one turn cannot be predicted but depends
on chance. For these reasons “the number of squares moved in a turn” is
called a random variable. Although the result of the next throw of the die
cannot be predicted, it is known that, if the die is fair, the probability of
getting each value is 1 .
6

Note: A random variable is a quantity whose value depends on chance.

A convenient way of expressing this information is to let X stand for the

1
number of squares moved in a turn. Then, for example, P( X = 3) = means
1 6
the probability that X takes the value 3 is . Generalizing, P(X = x) means
6
the probability that the variable X takes the value x.

Note: The capital letter stands for the variable itself and the small letter
stands for the value the variable takes.

This notation can be used in Table 2.1 below to give the possible values
for the number of squares moved and the probability of each value. This
table is called the probability distribution of X.

Note: The probability distribution of a random variable is just a list-

ing of the possible values of the variable and the corresponding
probabilities.

Table 2.1 Probability Distribution of X, the

Number of Squares Moved in a Turn for a Single
Throw of a Die
x 1 2 3 4 5 6 Total
1 1 1 1 1 1
P(X = x) 1
6 6 6 6 6 6
 2.2 Discrete Random Variables 53

Example 2.9
A bag contains two red and three blue marbles. Two marbles are selected at
random without replacement and the number, X, of blue marbles is counted.
Find the probability distribution of X.
The tree diagram illustrating this situation is shown in Figure 2.3, where
R1 denotes the event that the first marble is red and R2 the event that the
second marble is red. Similarly, B1 and B2 stand for the events that the first
and second marbles, respectively, are blue. X can take the values 0, 1, and 2.

P( X = 0) = P( R1 and R2 ) = P( R1 ) × P R2 R1 = ( ) 2 1
× =
2 1
=
5 4 10 10

P( X = 1) = P( B1 and R2 ) + P( R1 and B2 )
( )
= P( B1 ) × P R2 B1 + P( R1 ) × P B2 B1 ( )
3 2 2 3 12 3
= × + × = =
5 4 5 4 20 5

P( X = 2) = P( B1 and B2 ) = P( B1 ) × P B2 B1 = ( ) 3 2
× =
6
=
5 4 20 10
3

The probability distribution of X is shown in Table 2.2.

Note: For any random variable, X, the sum of the probabilities is 1 and
this is given as

∑ P( X = x ) = 1 (2.9)

1
P (R2|R1)= R1 and R2
4

2
P (R1)=−
5 R1 and B2
3
P (B2|R1) =
4

2
P (R2|B1)= B1 and R2
4
3
P(B1) =
5

2
P (B2|B1)= B1 and B2
4

Figure 2.3 Tree diagram showing possible outcomes and probabilities.

Table 2.2 Probability Distribution

of the Number of Blue Marbles X
x 0 1 2 Total
1 6 3
P(X = x) 1
10 10 10
54 Introduction to Probability Theory

2.2.2 Expectation Values
The expected value or the mean of a probability distribution is denoted by μ. The
new symbol is used in order to distinguish the mean of a probability distribution
from x, the mean of a data set. μ is often called the expectation or expected value
of X and is denoted by E(X).
The expectation of a random variable X is defined by

E(X ) = µ = ∑x p i i (2.10)

Example 2.10
Find the expected value of the variable X, which has the probability distri-
bution shown in Table 2.3.

Solution:

     
E(X ) = ∑ x p =  1 × 16  +  2 × 16  +  3 × 16 
i i

 1  1  1
+  4 ×  +  5 ×  +  6 ×  = 3.5
 6  6  6

Table 2.3 Probability Distribution and Associated

Probabilities
x 1 2 3 4 5 6 Total
1 1 1 1 1 1
P(X = x) 1
6 6 6 6 6 6

Example 2.11
Find the expected value of the variable Y, which has the probability distri-
bution shown in Table 2.4.

Solution: Again using the formula gives

E (Y ) = ∑ y p = 4 121
i i

Table 2.4 Probability Distribution and Associated Probabilities

y 1 2 3 4 5 7 8 9 10 11 12 Total
1 1 1 1 1 1 1 1 1 1 1
P(Y = y) 1
6 6 6 6 6 36 36 36 36 36 36

2.2.3 Variance and Standard Deviation

Just as the spread in a data set can be measured by the standard deviation or vari-
ance, so it is possible to define a corresponding measure of a random variable. The
symbol used for the standard deviation of a random variable is σ (read as “sigma”)
and its square, σ2, is the variance of a random variable denoted by Var(X).
 2.3 Continuous Random Variables 55

The variance of a random variable X is defined as

σ 2 = Var ( X ) = ∑ ( x − µ ) p =∑ x p − µ
i
2
i
2
i i
2
(2.11)

The standard deviation of a random variable is σ, the square root of Var(X). It

is in practice simpler to calculate Var(X) using ∑x p − µ .
2
i i
2

Example 2.12
Calculate the standard deviation of the random variable X given in
Example 2.10.
Solution: First calculate ∑x p :2
i i

       1
∑ x p =  1 × 16  +  2 × 16  +  3 × 16  +  4
2
i i
2 2 2 2
× 
6
 1  1  91 1
+  52 ×  +  6 2 ×  = = 15
 6  6 6 6

From the previous Example 2.10, μ = E(X) = 3.5. Using definition of

variance

Var ( X ) = ∑x p − µ
2
i i
2 1
= 15 − 3.52 =
6
35
12
= 2.92

Then calculate the standard deviation, which is just the square root of
the variance:

35
σ= = 1.71 (correct to three significant figurees)
12

2.3 Continuous Random Variables

Note: In this section, finding the integral of functions is required and so the
topic of integration is important. If one is not familiar with integrating func-
tions, then it is advised to study Chapter 6 first, then return to this section
afterward.

It has already been shown that for a discrete random variable X, it is possible to
allocate probabilities to each discrete value, x, that X can take. When considering
a continuous random variable this is not the case.
For a continuous random variable X, probabilities are allocated to each of the
range of values that the variable can take. This is done by defining a function f(x)
called the probability density function (pdf).
56 Introduction to Probability Theory

2.3.1 Probability Density Function (pdf)

The probability density function f(x) allocates probabilities to each of the range of
values that the continuous random variable can take and is defined such that for
f(x) ≥ 0, for all values of x, then

∫ f (x) dx = 1
all x
(2.12)

Then, the probability that the random variable X takes a value in the range
a ≤ x ≤ b is given by the integral
x =b

P (a ≤ X ≤ b) =
∫ f (x) dx
x =a
(2.13)

Example 2.13
A continuous random variable X has the pdf defined by

 3
 (1 − x 2 ) −1 ≤ x ≤ 1
f (x) =  4
 0 otherwise


Calculate P(0.2 ≤ X ≤ 0.5).

Solution:
0.5
3
P(0.2 ≤ X ≤ 0.5) =
∫ 4 (1 − x ) dx
0.2
2

0.5
3 x  3
3  0.53   0.23  
 x −  =  0.5 −  −  0.2 − 
4 3  0.2 4  3 3 
= 0.196 (correct to three significant digits)

2.3.2 Cumulative Distribution Function (cdf)

The cumulative distribution function, F(x), for a continuous random variable, X,
having a pdf f(x) is given by the formula

F ( x ) = P( X ≤ x ) =
∫ f (x) dx
−∞
(2.14)

dF ( x )
where = f ( x ).
dx

Note: If f(x) is defined only on the range of values a ≤ x ≤ b, then this becomes
 2.3 Continuous Random Variables 57

F ( x ) = P( X ≤ x ) =
∫ f (x) dx
a
(2.15)

Example 2.14
The continuous random variable X has the following cumulative distribu-
tion function:

 0 x≤0
 3
 x
F (x) =  0<x≤4
 64
 1 x>4

a. Find P(X ≤ 3).

b. Sketch the graph of f(x).

Solution:
33 27
a. P( X ≤ 3) = F (3) = = .
64 64

dF ( x )
b. Using, f ( x ) =
dx

3/4

x
0 4

Figure 2.4 Graph of the probability density function f(x).

The pdf is shown in Figure 2.4 and defined as

 3x 2
 0≤x≤4
f ( x ) =  64
 0 otherwise

58 Introduction to Probability Theory

2.3.3 Expectation of a Continuous Random Variable

The expected value of a continuous random variable X having a pdf f(x) is
denoted by E(X), where

E(X ) =
∫ x f (x) dx
allx
(2.16)

Example 2.15
A random variable T has the pdf given by

 0 02 0 < x < 10
f (t ) =  . t
 0 otherwise

Find E(T).

Solution: Using the formula given by Equation 2.16 gives

10

E (T ) =
∫ t f (t) dt
0
10

=
∫ 0.02t dt
0
2

10
 0.02t 3  2
=  =6
 3 0 3

Note: In general, the following results can be shown to be true.

E (aX ) = aE ( X ) (2.17)

E (aX + b) = aE ( X ) + b (2.18)

where a and b are constants.

2.3.4 Variance and Standard Deviation

of a Continuous Random Variable
The variance of a continuous random variable X is defined as

2
Var ( X ) = E ( X 2 ) −  E ( X )  (2.19)

Usually the variance is Var (X) = σ2.

The standard deviation of a continuous random variable X is denoted by σ and
is given by

σ = Var ( X ) (2.20)
 2.3 Continuous Random Variables 59

Example 2.16
A random variable T has the pdf given by

 0 02 0 < x < 10
f (t ) =  . t
 0 otherwise

Find the variance Var(T) and the standard deviation σ.

2
From the previous Example 2.15 it was found that E (T ) = 6 .
3
Solution:

2
Var (T ) = E (T 2 ) −  E (T ) 

So

10

E (T ) =
2
∫ t f (t) dt
0
2

10

=
∫ 0.02t dt
0
3

10
 0.02t 4 
=  = 50
 4 0

Therefore,

2
 2 5
Var (T ) = 50 − 6  = 5
 3 9

And the standard deviation is

5
σ= 5 = 2.36
9

Note: In general, the following results can be shown to be true for the vari-
ance of a continuous random variable X:

Var (a) = 0 (2.21)

60 Introduction to Probability Theory

Var (aX ) = a 2 Var ( X ) (2.22)

Var (aX + b) = a 2 Var ( X ) (2.23)

where a and b are constants.

2.4 Applications
Example 2.17: Event Tree Analysis (ETA) Showing
Probabilities of Outcomes
Failure of a complicated engineering system can lead to different damage
scenarios. The consequence of a particular failure event may depend on a
sequence of events following the failure. The means for systematic identi-
fication of the possible event sequences is the so-called event tree. This is
a visual representation, indicating all events that can lead to different sce-
narios. In the following example, first identify the events.
Consider the initiation event A, fire ignition reported to a fire squad.
After the squad has been alerted and done its duty at the place of accident,
a form is completed where a lot of information about the fire can be found:
type of alarm, type of building, number of staff involved, and much more.
Here the focus is on the condition of the fire at the arrival of the fire brigade.
This is described as

E1: Smoke production without flames

and the complement as

E1c : A fire with flames (not merely smoke producction)

The place where the fire was extinguished is described by the event

E2 : Fire was extinguished in the item where it started

and the complement as

E2c : Fire was extinguished outside the item

Figure 2.5 shows the events and corresponding number of cases.

Consider the case, where there was a fire with flames at the arrival and
that the fire was extinguished outside the place where it started. The prob-
abilities can be calculated from the event tree or using the conditional prob-
ability formula Equation 2.6.
From the event tree this is

65 30 30
E1c E2c = × = = 0.3
100 65 100
 2.4 Applications 61

or using Equation 2.6 gives

( ) ( ) (
P E1c and E2c = P E1c P E2c E1c = ) 65 30 30
× =
100 65 100
= 0.3

(32)
E2
(35)

E1
E2C (3)

(100)

(65) E2 (35)

E1C

E2C (30)

Figure 2.5 The event tree with the numbers within the parentheses indicating the
number of cases observed after 100 fire ignitions.

Example 2.18: Fault Tree Analysis (FTA)

to Calculate Risk Assessment
The fault tree analysis method can be used as an essential element for
risk assessment studies and accident investigations. The method starts
with a schematic diagram representing the system components, which
have an associated probability of failure attached to them and uses a
bottom-up method to analyze the system failure (i.e., the top event). The
diagram consists of the system components combined with the logic
operators the “And” and “Or” gates. The “And” gate implies that for
two components the system will only fail if the first component and the
second fails. For an “Or” gate, the system will fail if either the first or
second component fails.
As an example, when tackling fire incidents firefighters need a constant
supply of water to deal with the fire. The top event here can be that there is
no water output for dealing with the incident. The water supply comes via
an electrical pump system and a diesel pump system. The electrical pump
system can fail due to the pump or the electrical supply, and similarly, the
diesel pump system can fail due to the pump or diesel supply. Each system
component has an associated probability of failure and working. From the
bottom up a total probability failure can be calculated for the top event
using the rules of probabilities for “And” and “Or” systems. The preced-
ing system can be represented by a qualitative and quantitative fault tree as
shown Figure 2.6.
Starting with the bottom probabilities and using the rules for “And”
and “Or” the higher probabilities can be calculated so that the top event
failure rate can be determined. If the fire company is not happy with
the failure rate for the top event of 0.01206 or 1/83 failures, then fur-
ther action can be taken to increase the overall reliability of the system.
62 Introduction to Probability Theory

Water fails
(0.01206)

AND

Electric pump Diesel pump

system E system D
(0.06) (0.201)

OR OR

Pump E Electric supply Pump D Diesel supply

(0.05) (0.01) (0.2) (0.001)

= Basic event = Undeveloped event

Figure 2.6 Fault tree diagram showing the probabilities of failure in parentheses.

Here, the diesel pump D is the most unreliable component and so mak-
ing this more reliable (i.e., reducing its failure rate to 0.1) would mean
the top event failure rate now becomes 0.00606 or 1/165 failures, which
is a good improvement. Even further analysis could be carried out by
the company to reduce the risk of failure by introducing a second (i.e.,
a spare) diesel pump and seeing what the effect of this would have on
the overall failure rate. It turns out that introducing “redundancy” is a
more effective method for increasing the overall reliability of a system
(see Problem 2.8).

Example 2.19: Application of Bayes’ Theorem

for Conditional Probabilities
Three companies X, Y, and Z supply sprinkler systems to a university. The
percentage of sprinklers supplied and the probability of these being defec-
tive is shown in Table 2.5.
Given that a sprinkler is found to be defective, what is the probability of
it being supplied by company Y?

Table 2.5 Percentage of Sprinkler’s Supplied and Probability

of Being Defective
Percentage of Sprinklers Probability of Being
Company Supplied Defective
X 60% 0.01
Y 30% 0.02
Z 10% 0.03
 Problems 63

Using Bayes’ theorem, that is, Equation 2.8, gives

P(Y ) P( D \Y )
P(Y \D) =
P( X ) P( D \X ) + P(Y ) P( D \Y ) + P( Z ) P( D \Z )

0.3 × 0.02 0.0066

P(Y \D) = = = 0.4
0.6 × 0.01 + 0.3 × 0.02 + 0.1 × 0.03 0.015

that is, there is a 40% chance that the defective sprinkler was supplied by
company Y.

Problems
2.1 A card is picked at random from a shuffled pack. What is the probability
of getting

a. A diamond

b. An ace or a king

c. An ace or a red card

2.2 A group of firefighters is called to a fire incident. Seven are full-time,

five are part-time, and two are reserves. If a firefighter is chosen at ran-
dom to drive the fire engine, what is the probability it will be a part-time
firefighter?

2.3 Two used fire engines were bought by a fire station. The probability
that fire engine A is working in a year’s time is 0.9 and the probability
that the second fire engine B is working in a year’s time is 0.7. Find the
probability that

a. Both fire engines are working in a year’s time.

b. At least one of the fire engines is still working in a year’s time.

2.4 A fair coin is tossed three times. Find the probability that the number of
tails is 0, 1, 2, or 3.

2.5 A high-rise building decides, retrospectively, to install smoke alarms

in all its apartments. It buys the smoke alarms from two companies:
company A supplies 65% and company B supplies 35% of the alarms.
The probability that a smoke alarm is defective from company A is 0.01
and from company B is 0.03. After routine testing, it was found that a
smoke alarm was defective. What is the probability it was supplied by
company B?

2.6 Calculate the expected value and the standard deviation of the variable
X, which has the probability distribution given in Table 2.6.
64 Introduction to Probability Theory

Table 2.6 Probability Distribution

and Associated Probabilities
x 1 2 3 Total
1 3 2
P(X = x) 6 6 6 1

2.7 Given that a system has a failure probability density function f(t), where
time is measured in years as follows,

 0 t<0

 1
f (t )  t 0≤t ≤4
 8
 0 t>4

Calculate the mean time to failure (MTTF) and the standard deviation
for the system.

2.8 Consider the system given in Example 2.18. It was found that reducing
the diesel pump failure rate by half to 0.1 reduced the overall failure
rate from 1/83 to 1/165. To further improve this system, the company
decides to install a second pump (i.e., a spare) with the same failure rate
of 0.2 rather than improving the reliability of the first pump. Calculate
the new system failure rate for the top event “water fails” and comment
on your findings.
 3 Vectors and
Geometrical
Applications

3.1 Introduction
In engineering, there are different mathematical quantities that are used to
describe physical phenomena. These quantities can be divided into two catego-
ries: scalar or vector. These two quantities are defined as follows:

Scalar quantity—Something that has only size/magnitude. There are many

examples of these, such as time, temperature, volume, mass, and speed
(usually a number and units).

Vector quantity—Something that has both size/magnitude and direction.

Examples of these are force, velocity, and acceleration.

Note: A vector is usually distinguished from a scalar by a line either on top of

or below the symbol, that is, as a or a. Both notations are used in this book.

To represent vectors in two- or three-dimensional space, coordinate axes are

used with unit vectors defined along each axis as follows:
A unit vector is any vector that has size (or magnitude) equal to one. In the
x, y, z direction of space these unit vectors are defined as follows: i , j are unit
vectors in the x and y directions in 2-D, respectively, and so vector a = 3 i + 4 j is
given by three units along the x-axis and four units along the y-axis and is shown
in Figure 3.1. Similarly, i , j , and k are unit vectors in the x, y, and z directions in
3-D, respectively, and so a vector b = 2 i + 3 j + 7 k is shown in Figure 3.2.

65
66 Vectors and Geometrical Applications

y
2-D

4 a = 3i + 4j

x
3

Figure 3.1 Vector represented in 2-D space.

3-D
z

7 b = 2i + 3j + 7k

2
3
y

Figure 3.2 Vector represented in 3-D space.

3.1.1 Magnitude and Unit Vectors

Giving a vector r , the size or magnitude of the vector r is denoted by r and shown
in Figure 3.3. The magnitude is calculated geometrically using the Pythagorean
theorem in 2-D and 3-D space. Generally, for the vector r = a i + bj , the mag-
nitude r is given by r = a 2 + b 2 , that is, the square root of the sum of the
components squared.
 b
In 2-D, the angle of the vector , that is, the argument, is given by θ = tan −1   ,
 a
but more generally if the vector is in a different quadrant, then trigonometry is
used to calculate θ.
Given a general vector in 3-D space, r = x i + y i + zk , as shown in Figure 3.4.
Generally, for the vector r = x i + y i + zk the magnitude r is given as
r = x 2 + y 2 + z 2 , again the square root of the sum of the components squared.
 3.1 Introduction 67

r = ai + bj

θ
x
0

Figure 3.3 General vector with its magnitude in 2-D space.

r = xi + yj + zk
r

Figure 3.4 General vector with its magnitude in 3-D space.

Example 3.1
In 2-D space:
If r = 3 i + 4 j , then the magnitude of r is given as r = 32 + 4 2 , r = 25 ,
so r = 5.

Example 3.2
In 3-D space:
Given the vector r = i + j − 4 k , find the magnitude r .

r = (1)2 + (1)2 + (−4)2 , r = 18 , so r = 4.24.

3.1.1.1 Unit Vectors
r
Given a vector r , a unit vector in the direction of r is denoted by r̂, where r̂ =.
r
This means that to find a unit vector, take the vector and divide its components by
the magnitude of that vector.
68 Vectors and Geometrical Applications

Example 3.3
For the vector given by r = 2 i − 4 j + 3k to find the unit vector in the direc-
r
tion of r , using r̂ = gives r = (2)2 + (−4)2 + (3)2 , r = 4 + 16 + 9 and
r
so r = 29. This then gives

1
rˆ = ⋅ (2 i − 4 j + 3 k )
29

As will be seen in later chapters vectors have very important applica-

tions in engineering in areas such as vector fields. An example of a basic
vector field is shown next.

Example 3.4
What does the vector field defined in 2-D space given by F ( x , y) = x i look
like? To see this, take different points in space and see what the vector field
becomes.
So for the point (1,0), F (1, 0) = 1 i and for any point (1, y), F (1, y) = 1 i .
Similarly, for the point (2, y), F (2, y) = 2 i . This vector field is shown in
Figure 3.5.

x
–4 –3 –2 –1 0 1 2 3 4

Figure 3.5 Vector field defined by F ( x , y) = x i .

3.1.2 Addition and Subtraction of Vectors

Addition of vectors is carried out with simple adding of the corresponding i , j ,
and k components. Figure 3.6 shows how to add vectors pictorially.
To calculate the vector c = a + b , just add the corresponding components as fol-
lows: Given a = 3 i + 2 j − 5k and b = i − 3 j + 7 k , then c = a + b = 4 i − j + 2 k .
Subtraction of vectors is carried out with subtracting of corresponding i , j ,
and k components. Figure 3.7 shows how to subtract vectors pictorially.
 3.1 Introduction 69

b c
b

a a

Figure 3.6 Addition of two vectors a and b.

a
b

–b
d
a

Figure 3.7 Subtraction of two vectors a and b.

To calculate the vector d = a − b, just subtract the corresponding components.

Given a = 3 i + 2 j − 5k and b = i − 3 j + 7 k , then d = a − b = 2 i + 5 j − 12 k .

3.1.3 Scalar and Vector Products

First, consider what is meant by multiplication of a vector by a scalar k. In the
following equation, a = k b, this means that the vector a is k times as long as the
vector b and in the same direction. If k happens to be negative, then they are in
the opposite direction.
When considering the product of two vectors, there is some freedom to what
is meant by this. Generally, there are two different products that are defined: one
called the scalar product since the result is a scalar quantity and the other is the
vector product as this produces a vector after the multiplication.

3.1.3.1 Scalar Product (Dot Product)

Consider two vectors a and b and an angle θ between them as shown in Figure 3.8.
The definition of the scalar product is given by

a ⋅ b = a b cos θ = b ⋅ a (3.1)

This can be thought of as the length of a multiplied by the component of b

along a or vice versa.

Note: The scalar product is always written with a dot between the two vectors
being multiplied.

Figure 3.8 Two vectors with an angle between them in space.

70 Vectors and Geometrical Applications

The formula given by Equation 3.1 is useful for the following situations: (1) to
find the angle between two vectors, and (2) to prove if two vectors are perpen-
dicular to each other.
Consider the arbitrary vectors a = a1 i + a2 j + a3 k and b = b1 i + b2 j + b3 k .
To calculate the a ⋅ b term, this is given as

a ⋅ b = a1b1 + a2b2 + a3b3 (3.2)

This is because the terms i . i = j . j = k . k = 1 and i . j = j . k = i . k = 0.

Example 3.5
Given a = 3 i + 2 j + k and b = i + 5 j − 3k , find the angle between the vec-
tors. Using Equation 3.2 first gives

a ⋅ b = (3)(1) + (2)(5) + (1)(−3)

= 3 + 10 − 3 = 10

Now, to find the angle θ between the vectors using Equation 3.1 gives

a ⋅ b = a b cos θ

10 = 32 + 22 + 12 ⋅ 12 + 52 + (−3)2 ⋅ cos θ

a⋅b 10 10
cos θ = , cos θ = = = 0.452
a⋅b 14 ⋅ 35 7

 10 
θ = cos−1  = 63.14°
 7 

Note: If a.b = 0, then the two vectors are perpendicular, since a and
b ≠ 0, so cos θ = 0, which implies that the angle θ = 90°.

3.1.3.2 Vector Product (Cross Product)

The definition of the vector product is given as

a × b = a b sin θ nˆ (3.3)

where n̂ is a unit normal vector to the plane containing both a and b. There are
two directions perpendicular to any plane, here the direction is such that a, b, and
n̂ form a right-handed set of axes.

Note: The vector product can have a “×” or a “∧” notation between the vectors,
Figure 3.9 shows the arrangement of the vectors.
 3.1 Introduction 71

n
b

Figure 3.9 The vector product of a and b lies along the z axis.

Equation 3.3 for the vector product is useful for producing a vector that is per-
pendicular to both the vectors a and b.

3.1.3.3 How to Calculate a × b
If two vectors are given as a = 3 i + 2 j + k and b = i + 2 j − 3k , then the vec-
tor product is given as a × b = (3 i + 2 j + k ) × ( i + 2 j − 3k ). Now this can
be slightly complicated to work out by expanding of the brackets and using
i × i = j × j = k × k = 0, while the cross products follow the right-handed axes
rules. This process can be seen using the cyclic diagram as shown in Figure 3.10.
However, the vector product is more easily calculated using a 3 × 3 determi-
nant. Since what is required is the vector c = a × b, in the determinant the first
row is just the i , j , and k components. Then on the second row will be the com-
ponents of the vector a and finally on the third row the components of the vector
b as follows:

i j k
c =a×b= 3 2 1 = +i 2 1 − j 3 1 +k 3 2
2 −3 1 −3 1 2
1 2 −3

c = −8 i + 10 j + 4 k

Now, if this is a vector perpendicular to both a and b, then this can be checked
using the scalar product of c with a and b. The result should be zero.
Check: a ⋅ c = 0 −33 + 20 + 13 = 0
Also, b⋅c = 0 − 8 + 20 − 12 = 0
This proves that both vectors are perpendicular to this vector c = a × b.

i
i × j = k
j × k= i
k× i = j
etc.
k j

Figure 3.10 Cyclic rotation showing vector products for unit vectors i , j , and, k.
72 Vectors and Geometrical Applications

3.1.4 Projection of Vectors
When talking about the projection of a vector onto another vector, this generally
means the projection of vector a on to the line created by the vector b. The fol-
lowing cases could arise as shown in Figure 3.11.
a or
a

b b
Projb a Projb a

Figure 3.11 Projection of vector a onto b.

This idea is similar to a sundial casting a shadow on a tabletop.

How do you find the projection of a onto b? Consider the situation in Figure 3.12.

Projb a

Figure 3.12 Calculating the projection of the vector a onto b.

First find the length of the projection of a on to b. This is given by using trigonom-
etry. The length of projection onto b is = a cos θ . This is in terms of the angle θ.
But in terms of the vectors a and b, make use of the dot product as follows:

a ⋅ b = a b cos θ

gives

a⋅b
cos θ =
a b

Therefore, replacing for cos θ, the length of the projection of a on to b is now

a⋅b a⋅b
a =
a b b

a⋅b
So the length of the projection vector is , but its direction is that of the
b
vector b. Therefore, multiplying the length by a unit vector b̂ will give the vector
projection of a onto b as follows:

a⋅b ˆ
Projb a = b (3.4)
b
 3.1 Introduction 73

b
Now, = bˆ can be written out again in a neater form as
b
( )
Projb a = a . bˆ bˆ (3.5)

Note: This can be useful when imagining a fire hose spraying foam from the
ground toward a fire on an oil tank. This gives a way to combine the horizontal
wind speed with the form jet in that direction.

Example 3.6
Find the distance from the point (5, 5) to the line y = 2x shown in Figure 3.13.
In this problem, the use of projections can help to reach a solution.

Solution: To find the perpendicular distance d of the point (5, 5) to the line,
first project the vector (5, 5) onto the line. A vector that describes the line
y = 2x is given as the vector 〈1, 2〉 . You need to find

 〈1, 2〉   〈1, 2〉 
Proj〈1,2 〉 〈5, 5〉 =  〈5, 5〉.
 5   5 

15
= 〈1, 2〉 = 3〈1, 2〉 = 〈3, 6〉
5

So the projection of 〈5, 5〉 onto the line is at the point P (3, 6). This is the
closest point to (5, 5) that exists on the line y = 2x.
So the distance required is the distance between two points (5, 5) and
(3, 6), which is just given by the distance between two points formula,
d = ( x1 − x 2 )2 + ( y1 − y2 )2 , that is, d = (5 − 3)2 + (5 − 6)2 = 5 .

y y = 2x

P
d
(5, 5)

Figure 3.13 Perpendicular distance from a point to the line.

74 Vectors and Geometrical Applications

3.2 Vector Geometry
3.2.1 Vector Equation of a Line
To find the vector equation of the straight line shown in Figure 3.14, suppose P is
an arbitrary point on a straight line and A and B are given points on the same line
and O is the origin. P has a position vector r, the line is parallel (in the direction)
of AB, and A and B are both points on the line.

B Straight-line

a
b P

Figure 3.14 Vector equation of a line.

Now using vector addition gives the relationship

OP = OA + AP

where OP = r (variable), OA = a (fixed), and AP = t AB (t is a real parameter).

Therefore, the vector equation of a straight line can be written as

r = a + t AB (3.6)

Note: The vector equation of a straight line through two fixed points with position
vectors a and b is given by

r = a + t (b − a) (3.7)

(Since AB = OA + OB from the diagram.)

Example 3.7
Find the vector equation of the straight line parallel to the vector i + 2 j − 5k
going through the point with position vector 2 i − 3 j + k .

Solution: Using the vector equation of a line is given by Equation 3.6

r = a + t AB:

r = 2 i − 3 j + k + t ( i + 2 j − 5k )
 3.2 Vector Geometry 75

3.2.1.1 Intersection of Lines
In 2-D space two lines intersect or are parallel. In 3-D space two lines can inter-
sect or they may be either parallel or skew (i.e., missing each other).
Given two lines in space r1 = a1 + t b1 and r2 = a2 + s b2, then for them to inter-
sect means r 1 = r2 (i.e., a1 + t b1 = a2 + s b2 ), where the unique values of t and s
can be found.

Note: If unique values of t and s cannot be found, then the lines do not intersect
and are said to be skew.

Example 3.8
Find the point of intersection of the lines:

r1 = 2 i + 3 j + t (− i + 2 j ) and r2 = − i + j + s(3 i − 2 j )

Solution: Lines intersect when r1 = r2

2 i + 3 j + t (− i + 2 j ) = − i + j + s(3 i − 2 j )

(2 − t ) i + (3 + 2t ) j = (−1 + 3s) i + (1 − 2s) j

Equating the coefficients of the i and j components gives

2 − t = −1 + 3s (i.e. 3s + t = 3)

and

3 + 2t = 1 − 2s (i.e. t + s = −1)

Solving these simultaneously gives t = −3 and s = 2.

Therefore, the lines do meet and the point of intersection is given
by using either the t = −3 or s = 2 back into the equations of the lines
r1 or r2, respectively. Using t = −3 into r1 = 2 i + 3 j + t (− i + 2 j ) gives
2 i + 3 j − 3(− i + 2 j ) = 5 i − 3 j as the point of intersection of the two lines.

Note: If the lines are in 3-D space, follow the same method except now
there will be three simultaneous equations involving s and t. Solve for s
and t using any two of the three equations and test to see if these fit into the
third equation. If they fit into the third equation, then the lines do intersect.
If they do not fit the third equation, then the lines are said to be skew.
76 Vectors and Geometrical Applications

Example 3.9
Two lines r1 and r2 are given by the equations:

r1 = 3 i + j − 4 k + s(−2 i − 3 j + 6 k )

r2 = 2 i − 3 j + k + t ( i − j − k )

where s and t are parameter values.

Show that these lines intersect and find the position of the vector of P,
the point of intersection.

Solution: For intersection of the lines r1 = r2, this gives

(3 − 2s) i + (1 − 3s) j + (−4 + 6s) k = (2 + t ) i + (−3 − t ) j + (1 − t ) k

Equating the coefficients of the i , j , and k components gives

3 − 2s = 2 + s , 1 − 3s = −3 − t , − 4 + 6s = 1 − t

Solving the first two equations gives s = 1 and t = –1, and these fit the
third equation so therefore the lines do intersect.
The point of intersection is given by using s = 1 into equation for r1 and
the point becomes P( i − 2 j + 2 k ).

3.2.2 Vector Equation of Planes

A plane in space is completely specified by knowing one point in it, together with
a vector that is perpendicular to the plane.

Example 3.10
Find the equation of the plane with normal vector n = 〈−1, 2, 3〉 and goes
through the point Q (1, 3, 1) as shown in Figure 3.15.

R –1, 2, 3

P (x, y, z)

Q (1, 3, 1)

Figure 3.15 Equation of a plane, normal to a vector and passing through a point.
 3.2 Vector Geometry 77

To find this plane, for a given point (x, y, z) the vector PQ and QR must
be perpendicular to each other. So, using the properties of the scalar prod-
uct, then the dot product of QR with PQ must be zero.
Therefore, 〈−1, 2, 3〉 ⋅ 〈 x − 1, y − 3, z − 1〉 = 0

−( x − 1) + 2( y − 3) + 3( z − 1) = 0

the equation of the plane is −x + 2y + 3z = 8.

3.2.2.1 Generalizing for Any Plane in Space

First, one needs a single point that the plane goes through and a normal vector
n to the plane. Let the point in the plane be (x0, y0, z0) and the normal vector be
n = a, b, c as shown in Figure 3.16.
z

a, b, c

(x, y, z)

(x0, y0, z0)

Figure 3.16 General plane in space.

So this now gives

〈a, b, c〉 ⋅ 〈 x − x0 , y − y0 , z − z0 〉 = 0

a( x − x 0 ) + b( y − y0 ) + c( z − z0 ) = 0

or collecting all the constants onto one side gives the general equation of a plane
as

ax + by + cz = d (3.8)

Example 3.11
Find the equation of a plane that acts as a “mirror” for the points (1,3,–1)
and (1,–1,1) as shown in Figure 3.17.
The plane is perpendicular to the line joining the two points.

n = (1, −1,1) − (1, 3, −1) = 〈0, −4, 2〉

78 Vectors and Geometrical Applications

(1, –1, 1)

M
y
(1, 3, –1)

Figure 3.17 Plane with a mirror image of a point.

 x +x y +y z +z 
To find the midpoint M, use the midpoint formula  1 2 , 1 2 , 1 2  ,
 2 2 2 
which gives

 1 + 1 3 − 1 −1 + 1 
 , ,  = (1,1, 0)
2 2 2 

Having a point and the normal vector, the equation of the plane is given
by using Equation 3.8 as

0( x − 1) − 4( y − 1) + 2( z − 0) = 0

4 y − 2z = 4

3.3 Applications
Example 3.12: Resultant Smoke Flow
A small fire in a corridor has smoke flowing vertically at a speed of Vs =
1.3 ms–1. An open window blows wind due east at a speed of Vw = 2.5 ms–1
as shown in Figure 3.18.
The resultant speed VR and direction θ the smoke moves can be obtained
using vectors as follows: VR2 = 1.32 + 2.52 = 7.94 and VR = 2.8 ms–1.
The direction of smoke flow relative to the horizontal is
 1.3 
θ = tan −1  = 27.5°.
 2.5 

Vw = 2.5

Vs = 1.3
VR
θ

Figure 3.18 Wind effect on fire smoke.

 Problems 79

Example 3.13: Applicability of the Scalar Product

1. If the force F acts on a body that moves through a distance d, the
work done W is given by the formula

W = F . d = F d cos θ (3.9)

Here, the cos θ factor in the scalar product allows for the fact that
it is only the component of F along the direction of motion that does
the work.
2. Given two unit vectors û1 and û 2, then the angle between these two
vectors is given by

cos θ = uˆ 1. uˆ 2 (3.10)

Example 3.14: Force on a Charge in a Magnetic Field

In a region of magnetic induction given by B, a particle moves with velocity
v and charge q as shown in Figure 3.19. The particle experiences a force F
whose size or magnitude is given by qvB sin θ and whose direction happens
to be perpendicular to the directions of both the vectors v and B. So, it turns
out that F = qv × B gives the right magnitude and direction for the force and
as a consequence the particle gets deflected.

θ B

Figure 3.19 Particle moving in a magnetic field.

Problems
3.1 Find a unit vector û in the direction of the vector AB, where the points
A and B are given by A( i + 3 j + 5k ) and B(3 i + 2 j + k ).

3.2 Find the angle between the vectors a = 〈1, 2, −4 〉 and b = 〈4, 2, 7〉.

3.3 Find the value of the constant α, which makes the vectors a = 〈1, 2, α 〉
and b = 〈2, 3, 4 〉 perpendicular to each other.

3.4 Figure 3.20 shows a rhombus OACB, with sides OA and OB being given
by the vectors a and b, respectively. Given the diagonals OC = a + b
and BA = a − b, show that the diagonals are perpendicular to each other.
80 Vectors and Geometrical Applications

B
C

a–b
b

a+b

0 A
a

Figure 3.20 A rhombus with sides as vectors a and b.

3.5 Find a vector perpendicular to both the vectors a = 3 i + j − 2 k and

b = i − 2 j + k.

3.6 Show that the two lines given by

r1 = i + j + 3k + s(3 i + 8 j + 2 k )

r2 = 2i + 5 j − k + t ( i + 2 j + 3k )

intersect and find their point of intersection.

3.7 The three points A(2 i + j − k ), B( i − 2 j + 5k ), and C (3 i + 2 j − 2 k ) lie

in a plane. Find the equation of the plane passing through the three
points.

3.8 The 2-D rotational flow of a fluid can be represented by the vector field

〈− y, x 〉
F ( x , y) =
x 2 + y2

which is not defined at (0, 0).

a. Determine that the magnitude of the vector field.

b. Hence, sketch the vector field.

3.9 Smoke flows vertically upward at a speed of 0.8 ms–1 when it is blown
by wind traveling at 0.6 ms–1 in a southeasterly direction. Find the resul-
tant speed of the smoke flow and its direction relative to the vertical.
 4 Determinants
and Matrices

4.1 Background
Matrices with different dimensions are a standard method for solving a wide
range of problems in many disciplines. In science and engineering, when consid-
ering particular problems, the derivation of the solutions to these problems can
come down to solving a system of linear equations. In this chapter, different meth-
ods for solving a linear system of equations will be discussed and these methods
involve the concepts of determinants and matrices.

4.2 Introduction to Determinants
Determinants arise naturally in the solution of a set of linear equations. Also
they will help solve linear equations using the matrix inversion method (see
Section 4.3.6). Starting with a general set of two linear simultaneous equations
as follows:

ax + by = e (4.1)

cx + dy = f (4.2)

where a, b, c, d, e, and f are constants. How do you find the values of x and y?
Since the values of a, b, c, and d are not known, all that can be done is try to
eliminate either x or y.
If one decides to get rid of y, then to make the coefficients of y the same, first
multiply Equation 4.1 by d and Equation 4.2 by b:

adx + bdy = de (4.3)

bcx + bdy = bf (4.4)

81
82 Determinants and Matrices

Then subtracting Equation 4.4 from Equation 4.3:

de − bf
adx − bcx = de − bf ∴ x (ad − bc) = de − bf ∴ x=
ad − bc

A similar approach, getting rid of x, would give the result as

af − ce
y=
ad − bc

Now provided that the denominator term ad − bc ≠ 0, then a symbol can be

used to define a 2-by-2 determinant using parallel lines (∣ ∣), as shown next:

a b  ad − bc
(4.5)
c d

Note: The symbol ≜ means “defined as.”

This definition given by Equation 4.5 is just the difference in product of the
diagonals.
Notice also how these answers for x and y contain lots of similar expressions.
Using the definition of a determinant given earlier, they can be written in the fol-
lowing form:

e b a e
f d c f
x= , y=
a b a b
c d c d

or alternatively as

x y 1
= = (4.6)
e b a e a b
f d c f c d

This is called Cramer’s rule. Equation 4.6 is a nice and neat way to represent
the solutions to Equations 4.1 and 4.2.
 4.2 Introduction to Determinants 83

4.2.1 2 × 2 Determinants
Now using the definition of a determinant given by Equation 4.5

a b  ad − bc
c d

Example 4.1

Calculate the following determinant:   6 2

3 4
Solution:

6 2 = 6 × 4 − 2 × 3 = 24 − 6 = 18
3 4

Example 4.2

Calculate the following determinant:   −2 −5

3 4
Solution:

−2 −5 = −2 × 4 − (−5) × 3 = −8 + 15 = 7
3 4

4.2.2 Properties of Determinants
Looking at these answers in the above examples, one can notice some facts about
determinants.

• If rows are swapped for columns, the value of the determinant is

unchanged.

a b = a c
c d b d
84 Determinants and Matrices

• If one row (or column) is equal to the other, the value of the determinant
is zero.

a b = 0 = a a
a b c c

• If one row (or column) is a multiple of the other, the value of the deter-
minant is zero.

a b = 0 = a ka
ka kb c kc

This last property shows that, for example,

100 500 = 0
1 5

4.2.2.1 Multiplying a Determinant by a Number

Look what happens if every element of a 2 × 2 determinant is multiplied by 3:

3 p 3r p r
= 9 ps − 9qr = 9( ps − pr ) = 9
3q 3s q s

In other words, the value of the determinant is multiplied by 9.

Multiplying just one row (or column) of a determinant by a number has the
effect of multiplying the value of the determinant by that number. For example,

m n 5m n m n
5 = = = 5(mp − on)
o p 5o p 5o 5 p

It can be seen why the last property mentioned earlier comes about:

100 500 = 100 1 5 = 100(5 − 5) = 100(0) = 0

1 5 15

In fact, there is a much “stronger” property that can be easily proved. First, a
numerical example to show this:

102 504 = 2 4 = − 4 = 6
10
1 5 1 5
 4.2 Introduction to Determinants 85

Here the determinant is made much easier to work out by subtracting 100 and 500
from the top row.
This can be made a general rule as follows:

a + kc b + kd = (a + kc)d − (b + kd )c = ad + kcd − bc − kcd

c d
a b
= ad − bc =
c d

Or the value of a determinant does not change if one adds to one row a multiple
of another row.
This might not seem important for 2 × 2 determinants, but all the preceding
rules apply to determinants of any size, and it is with larger ones that they come
in useful.

4.2.3 3 × 3 Determinants
The basic building block of a determinant is the 2 × 2 one already discussed. All
larger ones are broken down into combinations of 2 × 2 determinants, according
to the rule of signs given as

+ − +…
− + −
+ − +


Then going along any row or column, multiplying each element by

1. The sign in that place

2. The 2 × 2 determinant revealed when covering up the row and column

with that element in

This following example shows how this is carried in practice.

Example 4.3
Take the determinant

3 4 2
0 1 5
1 6 8

Suppose this is expanded along the top row. The element 3 is multiplied

by “+,” and by the 2 × 2 determinant 1 5 , since that is what is revealed

68
if one covers up the row and column containing the element 3. Next move
86 Determinants and Matrices

along to the element 4. This is multiplied by “–” and by 0 5 . Finally, take

1 8

the element 2. This is multiplied by “+” and by 0 1 . Now adding up the

1 6
three parts gives the value of the determinant as

3 4 2
1 5 −4 0 5 +2 0 1
0 1 5 = +3
6 8 1 8 1 6
1 6 8
= 3(−22) − 4(−5) + 2(−1) = −48

Suppose trying to expand this using a different row; let’s say row 2. Then
the working is similar, but remember that now the expansion starts with a
minus sign this time giving

3 4 2
4 2 +1 3 2 −5 3 4
0 1 5 = −0
6 8 1 8 1 6
1 6 8
= 0(not necessary) + 1(22) − 5(14) = −48

The same result is obtained expanding along any row or column.

Now what about using a column instead of a row? From the first prop-
erty, the value of the determinant is unchanged if the rows and columns are
swapped. Therefore, a column can be regarded as equivalent to a row, and the
calculation can be done in a similar way, for example, using the last column:

3 4 2
0 1 −5 3 4 +8 3 4
0 1 5 = +2
1 6 1 6 0 1
1 6 8
= 2(−1) − 5(14) + 8(3) = −48

Why all the fuss about being able to use any row or column to evaluate a deter-
minant? Because, as seen, it is very handy if some of the elements are zero. Thus,
using the first column or the second row would be quicker ways of evaluating the
above determinant.

Example 4.4
Evaluate the following determinants:

3 6 2
1. 5 2 1   Expanding along row 3 would speed the working.
0 4 0
 4.3 Introduction to Matrices 87

−4 2 7
2. 3 5 0    Expanding along column 3 would help.
1 8 0

Example 4.5
Find the values of k for which

k 3 −5
0 k −1 4 = 0
0 0 k +5

Solution: Expanding the determinant along first column gives

k 3 −5
k −1 4
0 k −1 4 = k = k ( k − 1)( k + 5) = 0
0 k +5
0 0 k +5

This implies k = 0, k = 1, or k = –5 as the solutions.

4.3 Introduction to Matrices
In English, one thinks of the word matrix as meaning a “grid” or an array.” In
mathematics, it has a special meaning: An “object” made up of a rectangular
arrangement of “things,” or elements, which behaves according to certain rules.
These elements might be numbers, letters, complex numbers, or other things, but
the rules that govern matrix arithmetic are always the same.
When a matrix is written, it is always enclosed in brackets (either curly or
square). This is very important, because there are other rectangular arrays, as
seen in the previous section, called determinants that are written differently, and
the different types must not be confused!

Example 4.6
All the following are matrices

 a 
1    2+ j −5 − 3 j 
0 ,  b  ,
   
3 2  c   j 1 − j 
 d 

Note: Matrices that have a single column (as the second one above) are
called column vectors and those that have a single row are called row
vectors.
88 Determinants and Matrices

4.3.1 Order of a Matrix
Clearly, the preceding matrices have different sizes or “order.” The order of a
matrix is the number of rows by the number of columns it has and is written gen-
erally as m × n, where m is the number of rows and n is the number of columns.

Example 4.7
In Example 4.6, the order of the matrices are as follows: 2 × 2, 4 × 1, and
2 × 2.
If two matrices are equal, clearly they must be the same size and shape;
also corresponding elements must be identical. So if, for example, the two
   
matrices  a b  and  3 −5  are equal, then there are four equations
 c d  2 1 
that must be true: a = 3, b = −5, c = 2, and d = 1.

4.3.2 Addition and Subtraction

If it is required to add or subtract two matrices, to be able to do this, they first have
to have the same size and shape, that is, have the same order. Then it is a matter
of simply adding or subtracting the corresponding elements as shown in the next
example.

Example 4.8

 2 1 8   6 1 8   8 2 16 
  +  = 
 1 5 3   0 2 1  1 7 4 
 0 6 4   5 3 0   5 9 4 

or

 2 1 8   6 1 8   −4 0 0 
  −  = 
 1 5 3   0 2 1   1 3 2
 0 6 4   5 3 0   −5 3 4 

4.3.3 Matrix Multiplication
4.3.3.1 Multiplying a Matrix by a Scalar
Adding a matrix A to itself, it makes sense to call the result 2A. The effect will be
to multiply each element by 2 as follows:

 3 6   3 6   6 12   3 6
  +  =  = 2 
 4 5   4 5   8 10   4 5
 4.3 Introduction to Matrices 89

This idea can be extended to apply to multiplying a matrix by any num-

ber (scalar): Simply multiply each element by that number as shown next:

   
k  a b  =  ka kb 
 c d   kc kd 

Note: This is different than multiplying a determinant by a constant.

4.3.3.2 Multiplying Two Matrices

In order to discuss this, it is better to define the way to describe the size and shape
(or order) of a matrix. It takes too long to say: “A matrix with three rows and four
columns.” Instead this can be referred to as a “3 × 4 matrix.” When looking at a
matrix, first go

1 2 −3 7
3
down the left-hand side: 3 5 1 −2 This is a 3 × 4 matrix
0 2 6 8

4
then along the bottom

4.3.3.3 How to Multiply Two Matrices

Consider a matrix A order m × n and a second matrix B of order p × q, then the
multiplication of the two matrices, AB, is only defined if the number of columns
of A, given by (n), is equal to the number of rows of B, given by (p), that is (n = p
the inner numbers). Then the product exists and is of order given by m × q (the
outer numbers). This can be represented as follows:

 2 6
 1 −2 7 
If A =   and B =  1 8  , then how to determine the product AB,
 2 5 4  
 3 5
2×3 3× 2

Now since inner numbers are the same 3 = 3, that is, n = p, for these two matri-
ces, then this product exists. The answer will be a matrix of order given by the
outer numbers, which is 2 × 2. This can be represented at the moment generally as

 
 1 −2 7   2 6   R1C1 R1C2 
  1 8  =  
 2 5 4  RC RC 
 3 5  2 1 2 2
2×3 3× 2 2×2
90 Determinants and Matrices

This product matrix has four elements in it. These elements are calculated by
seeing the location of each element in the product matrix and carrying out the
corresponding multiplication of the row and column from the original two matri-
ces and adding all the multiplications together.
In the above example, the first element in the product matrix, R1C1 is given by
row 1 of the first matrix multiplying column 1 of the second matrix as follows:

 2
R1C1 = (1 − 2 7)  1  = (1) × (2) + (−2) × (1) + (7) × (3) = 2 − 2 + 21 = 21
 
 3

Note: Here the multiplication is carried out with corresponding elements, that
is, the first element of the row with the first element of the column, and so on.

Similarly, for the other three elements of the product matrix:

 6
R1C2 = (1 − 2 7)  8  = (1) × (6) + (−2) × (8) + (7) × (5) = 6 − 16 + 35 = 25
 
 5

 2
R2C1 = (2 5 4)  1  = (2) × (2) + (5) × (1) + (4) × (3) = 4 + 5 + 12 = 21
 
 3

 6
R2C2 = (2 5 4)  8  = (2) × (6) + (5) × (8) + (4) × (5) = 12 + 40 + 20 = 72
 
 5

Now these four elements can be put into the result of the product as

 
 1 −2 7   2 6   21 25 
  1 8  = 
 2 5 4   21 72 
 3 5

Finally,
 
AB =  21 25 
 21 72 
 4.3 Introduction to Matrices 91

Example 4.9
Given the following matrices,

 1.5 
 50 100 150 50   
 2.0  , calculate the product AB.
A=  and B =
 60 80 25 50   3.5 
 4.0 

Solution: Matrix A has order 2 × 4 and matrix B has order 4 × 1. Since the
inner numbers match, then the product exists and the result will be a matrix
of order 2 × 1, that is, the outer numbers as follows:

 1.5 
 50 100 150 50   2.0   R1C1   1000 
   =  = 
 60 80 25 50   3.5   R2C1   537.5 
 4.0 

This is the essence of matrix multiplication. Note that in this last exam-
ple the orders of the matrices were

(2 × 4) (4 × 1) giving a (2 × 1) matrix

order of first matrix order of second matrix

and that the multiplication could only be done because the “4’s matched”:
the number of columns in the first one matched the number of rows in the
second one.
Could these particular matrices have been multiplied the other way
around, that is, does the product BA exist? By considering the structure
this gives

1 .5
2 .0 50 100 150 50 Because these numbers
=?
3 .5 60 80 25 50 don’t match up, implies can’t
do that matrix multiplication
4 .0
this way round: the product
doesn’t exist.

(4 × 1) (2 × 4)
92 Determinants and Matrices

Example 4.10

 3 5   1 0   23 10 
   = 
 2 1  4 2   6 2 

Example 4.11

 2 3   0 4 2   6 23 7 
   = 
 1 4   2 5 1   8 24 6 

Example 4.12

 3  18 6 24 
  (6 2 8) =  36 12 48 
 6  
 1  6 2 8 

What is observed is even if both matrix products AB and BA exist, they

may not be equal. In general, for matrices, AB ≠ BA.

Note: That this is quite different behaviour from ordinary numbers!

4.3.4 Special Matrices
As with normal algebra, the number 1 has the property that 1 × a = a × 1 = a.
With matrices, there is a similar situation using the identity matrix I as follows:

 1 0 0 
 1 0   etc.
I2 =   and I 3 =  0 1 0
 0 1 
 0 0 1 

These are called the identity matrix, I, which have the special property that for
any other matrix A gives A I = I A = A. The identity matrix is always square, but
can be any size, such as 2 × 2, 3 × 3, 4 × 4, and so forth.
 4.3 Introduction to Matrices 93

Finally, this section will conclude with other special types of matrices:

• A square matrix could have all its elements equal to zero: this is called
the null matrix.

• If all the elements except those on the main diagonal are zero, then it is
called a diagonal matrix.

• As noted above, the identity matrix, I, is a diagonal matrix with all its
nonzero elements equal to 1.

• The transpose of a matrix is a new matrix obtained by swopping rows

into columns and vice versa. To transpose a matrix, row 1 will become
column 1, and row 2 will become column 2, and so forth. The transpose
of matrix A is written as AT. For example, if

 
A= 1 0 5
 4 2 1

then

 1 4
A = 0 2 
T
 
 5 1

and it can be seen that the transpose of a 2 × 3 matrix must be a 3 × 2 matrix.

The transpose of a square matrix will also be a square matrix, of the same
order. What do you notice about these two matrices and their transposes?

 3 1 8   3 1 8 
B= 1 2 4  BT =  1 2 4 
   
8 4 9  8 4 9 

 0 −1 −2   0 1 2
C= 1 0 5  CT =  −1 0 −5 
   
 2 −5 0   −2 5 0 

B is an example of a symmetric matrix: B = BT.

C is an example of a skew-symmetric matrix: C = −CT.
94 Determinants and Matrices

4.3.5 Powers of Matrices
As with normal algebraic expressions, powers of matrices are calculated as
follows:

A2 = A × A
A3 = A × A × A
A4 = A × A × A × A


Note: A3 = A2 × A, A4 = A2 × A2 , and so forth.

Example 4.13

 
Given a square matrix A =  3 5  , calculate A2.
 2 1
Solution:

     
A2 = A × A =  3 5  ×  3 5  =  19 20 
 2 1   2 1   8 11 

4.3.6 Inverse of a Square Matrix

The inverse of a matrix A is denoted by A−1 such that AA−1 = I, where I is the
identity matrix.
To calculate the inverse of a square matrix A use the following formula:

1 T
A−1 = Ac (4.7)
A

where |A| is the determinant of the matrix A, and AcT is the matrix of cofactors of
A transposed. The cofactors of the matrix A are given by first taking into account
an appropriate sign for each element in the matrix using the following structure:

 + − + …
 
 − + − 
 + − + 
  

Then each element is made up of the sign with the determinant of elements that
are left after removing the row and column containing that element.
Generally, if A is a square matrix, then the “minor” of entry aij is denoted by
Mij and is defined to be the determinant of the submatrix that remains after the ith
row and jth column are deleted from A.
 4.3 Introduction to Matrices 95

The number (−1)i+j Mij is denoted by Cij and is called the cofactor of entry aij.
The next examples show how this works for a 2 × 2 and 3 × 3 matrix.

Example 4.14
The determinant of a general 2 × 2 matrix

 
A= a b 
 c d

is given as |A| = ad − bc and the matrix of cofactors of

   
Ac =  d −c  and AcT =  d − b 
 −b a   −c a 

This gives the inverse of A as

1  d −b 
A−1 = (4.8)
ad − bc  −c a 

Example 4.15

 
Calculate the inverse of the matrix A =  2 3  .
 1 5
Solution:

A = 10 − 3 = 7

1  5 −3 
A−1 = .
7  −1 2 

Check

1 7 0   1 0 
AA−1 = = =I
7  0 7   0 1 

Example 4.16
Following is an example of the inverse of a 3 × 3 matrix using the formula
1 T
A−1 = Ac .
A
96 Determinants and Matrices

Find the inverse of the matrix

 0 0 1 
A =  2 −1 3 
 
 1 1 4 

Solution: Expanding along row 1 gives the determinant as

A = 1 2 −1 = 3.
1 1

Matrix of cofactors is

 
 + −1 3 − 2 3 + 2 −1 
 14 1 4 1 1 
 
Ac =  − 0 1 + 0 1 − 0 0 
1 4 1 4 1 1 
 
 0 1 − 0 1 + 0 0 
 + 
 −1 3 2 3 2 −1 

 −7 −5 3 
Ac =  1 −1 0 
 
 1 2 0 

 −7 1 1 
A =  −5 −1 2
T 
c
 
 3 0 0 

 −7 1 1 
−1 1 
So therefore, A = −5 −1 2  .
3
 3 0 0
Check to see that AA = I.
−1

Note: In general, for higher-order matrices, for example, 3 × 3 and 4 × 4,

finding inverses can be difficult and as such it is much easier to use a
matrix calculator to work them out.

4.3.7 Eigenvalues and Eigenvectors

If a column matrix is regarded as representing a vector, then premultiplying this
column matrix by a square matrix will result in another column matrix or another
vector, for example,
 4.3 Introduction to Matrices 97

 2 1   6   15 
 −2 5   3  =  3 
    

   
One could say the vector  6  has been transformed into the vector  15 
 3  3 
 
by the square matrix  2 1  .
 −2 5 
Both the direction and the length, or magnitude (see Chapter 3, Section 3.1.1),
of the vector have changed. But will this be true for every vector that gets trans-
formed with this matrix?
Clearly, all the matrix multiplications cannot be done by hand. An Excel
workbook could easily be created that shows both vectors and changes the direc-
tion of the “input” vector and the modulus. It is noticed that there are two direc-
tions for which the vectors seem to be “in line with one another,” in other words
the modulus may have changed as a result of the transformation but not the
direction.
 
If the input vector is any vector along the direction  1  , let’s call it
 1
 k
  , where 0 ≠ k ∈ , then the “output” vector is also along the same direction.
 k
Moreover, for any such vector, the output vector is always 3 times as long as the
input, as follows:

 2 1   k   3k   k
 −2 5    =  3  = 3  
  k   k   k

 
Also, if the input vector is any vector along the direction  1  , let’s call it
 2
 k 
  , where 0 ≠ k ∈ , then the output vector is also along the same direction.
 2k 
Moreover, for such vectors, the output vector is always 4 times as long as the
input, for

 2 1   k   4k   k 
 −2 5   2  =  8  = 4  2 
  k   k   k

These two “multiplying factors,” 3 and 4, are called the eigenvalues of the
 
matrix  2 1  .
 −2 5 
Each eigenvalue is associated with a direction; this direction is called the eigen-
vector associated with that eigenvalue. (Eigen is German for “characteristic.”)
98 Determinants and Matrices

 
Hence the matrix  2 1  has an eigenvalue 3, with associated eigenvec-
 −2 5 
 k  
tor   , 0 ≠ k ∈ ; and an eigenvalue 4 with associated eigenvector  k  ,
 k  2k 

0 ≠ k ∈ .

Eigenvalues and eigenvectors are important properties of matrices, with many

applications in engineering as well as in pure and applied mathematics.
One can determine whether a square matrix has eigenvalues and eigenvectors,
and if so, what they are. The deciding property seems to be that the output vec-
tor is a multiple of the input vector. Taking the matrix from earlier, this can be
written as

 2 1  x   x
 −2 5   y  = λ  y 
    

where λ ∈ is the multiple or

AX = λ X
∴ λ X − AX = 0
∴ (λ I − A)X = 0

Note: This could also have been written as (A – λI)X = 0.

Here, in matrix form, is a set of two homogeneous equations in two variables.

There will always be the trivial solution, x = y = 0, but for a nontrivial solution
to exist,

det (λ I − A) = 0,

In the preceding example,

   
det  λ 0  −  2 1   = λ 2 1 = (λ 2 − 7λ + 12) = 0
−
 0 λ   −2 5   2 λ−5

∴ (λ − 3)(λ − 4) = 0

so λ = 3 or 4.
For any square matrix A, the expression det (λI – A) is called the characteris-
tic polynomial of A. The characteristic polynomial is sometimes written simply
 4.3 Introduction to Matrices 99

as h(λ). The equation det (λI – A) = 0, is called the characteristic equation of A.

Then any λ with the property, A X = λ X is called an eigenvalue of A and then X
is called the eigenvector of A corresponding to λ.

Example 4.17
Find the eigenvalues and associated eigenvectors of the matrix

 1 0 −1 
A= 1 2 1 
 
 2 2 3

Solution: The characteristic equation is given by det (λI – A) = 0. In this

case

λ −1 0 1
−1 λ − 2 −1 = 0
−2 −2 λ − 3

Evaluating along the top row

(λ − 1){(λ − 2)(λ − 3) − 2} + {2 + 2(λ − 2)} = 0

∴ (λ − 1){λ 2 − 5λ + 4} + 2(λ − 1) = 0
∴ (λ − 1){λ 2 − 5λ + 6} = 0
∴ (λ − 1)(λ − 2)(λ − 3) = 0
∴ λ = 1, 2, or 3

There are three eigenvalues, namely 1, 2, and 3.

Take each in turn and find an associated eigenvector.

1 0 −1   x  x
using λ = 1 in AX = λ X :     = 1 
1 2 1  y  y
2 2 3  z   z 

∴ x − z = x, so z=0
Also, x + 2y + z = y so x = −y
 k
 
So, an eigenvector could be,  − k  , for any real (non-zero) number k.
 0 
100 Determinants and Matrices

Sometimes the “scaling factor:” k is omitted, and the eigenvector is

 1
 
given as  −1  , where it is understood that any multiple of this vector
 0 
will also work.
Similarly,

1 0 −1   x  x
using λ = 2 in AX = λ X :     = 2 
1 2 1  y  y
2 2 3  z   z 

∴ x − z = 2x, so x = −2
also, x + 2 y + z = 2 y, so 2y = 2y
and 2 x + 2 y + 3z = 2 z , so 2 y = − z − 2 x = − z + 2z = z

 −2 k 
 
So an eigenvector could be,  k  , for any real (nonzero) number k.
And finally,  2k 

1 0 −1   x  x
using λ = 3 in AX = λ X :     = 3 
1 2 1  y  y
2 2 3  z   z 

 k 
 
So an eigenvector could be,  − k  , for any real (nonzero) number k.
Therefore, the matrix  −2 k 

1 0 −1
A= 1 2

1

2 2 3

 1   −2 
has eigenvalues 1, 2, and 3 with corresponding eigenvectors  −1  ,  1  ,
   
 1  0  2
 
and  −1 .
 −2 

4.3.8 Diagonal Factorization of Matrices

For an eigenvalue equation AX = λX, suppose there are n independent eigenvec-
tors of the matrix A. If these eigenvectors are put in a column P, that is,
 4.3 Introduction to Matrices 101

    
 
P =  X1 X 2   X n 
    
 

then

AP = A X1 X 2 Xn = 1 X1 2 X2 n Xn

1 0
AP = X1 X 2 Xn
0 λn

Diagonal eigenvalue
matrix D

Therefore, AP = PD and P−1 AP = D provided P is invertible, requiring n inde-

pendent eigenvectors.
Therefore, giving the following results,

D = P −1 AP (4.9)

A = PDP −1 (4.10)

How do you make use of this information? Since

A = PDP −1
A2 = PDP −1PDP −1 = PD 2 P −1
A3 = PDP −1PDP −1PDP −1 = PD 3 P −1

and so on, this is generalized to power k:


Ak = PDP −1PDP −1PDP −1 … PDP −1 = PD k P −1

This gives an important result:

Ak = PD k P −1 (4.11)

So, eigenvalues and eigenvectors give an effective way of finding powers of

matrices.
The diagonal matrix D and its powers are just the powers of the diagonal ele-
ments and so any power of a matrix can be raised by just three matrices multiplied
together.
102 Determinants and Matrices

The matrix A is sure to have n independent eigenvectors and therefore be diag-

onalizable if

• All the λ’s are different or distinct, that is, no repeated eigenvalues.

• Having repeated eigenvalues means there may or may not have been
independent eigenvectors.

Example 4.18
Determine the diagonal factorization of a 2 × 2 matrix given by

 
A= 4 1 
 −8 −5 

Solution: A can be diagonalized if A has two independent eigenvectors.

First, finding the eigenvalues and eigenvectors of A:

AX = λ X

( A − λI )X = 0

det ( A − λ I ) = 0

 1 =0
det  4 − λ 
 −8 −5 − λ 

λ 2 + λ − 12 = 0

(λ + 4)(λ − 3) = 0

gives λ1 = −4 and λ2 = 3.
The corresponding eigenvectors are for

 
λ1 = 3, the eigenvector is X1 =  1  .
 −1 

 
λ2 = −4, the eigenvector is X 2 =  1  .
 −8 
 4.4 Solving Systems of Linear Equations 103

So,

 
P= 1 1 
 −1 −8 

which gives

1 
P −1 = −  −8 −1 
7 1 1

Now

 
D= 3 0 
 0 −4 

A = PDP−1 (factorization as a product of three matrices)

Ak = PD k P −1

Now, powers of the matrix A can be calculated as,

A4 = PD 4 P −1
1   
A4 = −  1 1   81 0   −8 −1 
7  −1 −8   0 256   1 1 
 −25 
A4 =  56 
 200 281 

4.4 Solving Systems of Linear Equations

4.4.1 Introduction
It was shown previously that there are different ways of solving a set of linear
equations. One method was to make use of determinants, that is, using Cramer’s
rule. Given two equations in two unknowns

ax + by = e

cx + dy = f

This has solutions

x y 1
= =
e b a e a b
f d c f c d

provided that they exist.

104 Determinants and Matrices

The advantage of this method is that only one of the answers needs to be cal-
culated, if that is all that is needed.
Another option is to make use of the inverse matrix method (see Section 4.4.3).
Writing the set of equations in matrix form as AX = B, the solution, provided it
exists, is given by X = A–1B. The advantage of this method is that results can be
obtained for lots of inputs by changing the right-hand set of values and using the
same inverse matrix A–1 each time.
Each method has its advantages of course, but they also have the disadvantage
that they do not give us much information about situations when there is not a
unique solution. As engineers, it is very important to be able to tell what happens
in extreme cases! Also, there may be a need to know when there might be more
than one solution for a set of equations.

4.4.2 Gaussian Elimination Method

The Gaussian elimination method for solving sets of linear gives more informa-
tion about no solutions or lots of solutions, as well as finding “the solution” when
it exists. It’s probably easiest to learn this method from a worked example.

Example 4.19
Solve the set of equations:

− x + y + 2z = 2
3x − y + z = 6
− x + 3y + 4z = 4

Step 1: Write the augmented matrix:

 −1 1 2 2
 3 −1 1 6 

 −1 3 4 4

Step 2: Perform elementary row operations on the augmented matrix

until it is in echelon form. Echelon form means that the first nonzero ele-
ment in any row lies to the right of the first nonzero element in the row
above. (Note: Some other definitions of echelon form exist, including hav-
ing a one as the first nonzero element in each row. This definition is not used
here, although it is easy to divide each row by a suitable number to get it in
this form.) The matrix will have the following structure:

 a b c d
 
 0 e f  g
 0 0 h  i 

where a, b, c, d, e, f, g, h, and i are constants. The basically idea is to aim for

a “triangle of zeros” in the bottom left-hand corner of the matrix.
 4.4 Solving Systems of Linear Equations 105

Elementary row operations can be

• Multiplying any row by a number

• Interchanging any two rows
• Adding to any row a multiple of another row

Generally, all this is easier to do than to explain! It is always a good idea

to note what has been done at each step.

−1 1 2 2
Let’s get two zeros in the first column of the matrix 3 − 1 1 6 by
−1 3 4 4
Row 2 = Row 2 + 3 Row 1,
(R2: R2 + 3R1) −1 1 2 2
0 2 7 12
and Row 3 = Row 3 – Row 1, 0 2 2 2
(R3: R3 – R1)

Now let’s get a zero in the second −1 1 2 2

column by 0 2 7 12
Row 3 = Row 3 – Row 2, 0 0 − 5 − 10
(R3: R3 – R2)

Note: When putting a zero in the second column, one must always use row 2 with
a row 3 otherwise using row 1 with row 3 will change the zero value already cre-
ated in the first element of row 3 and so undoing what has already been achieved.

The objective has been achieved, since the nonzero elements start one
place to the right as we go down the rows. The augmented matrix is in
echelon form.
Step 3: The solution can now be found by back-substitution. Imagine the
equations returned to their original form. They would now read

− x + y + 2z = 2
2 y + 7 z = 12
−5z = −10

From the last equation it can be seen that dividing each side by –5 gives
z = 2. Then, looking at the middle equation and using the value for z gives
2y + 14 = 12, so 2y = −2 and y = –1. Finally, moving to the first equation and
using the two values that are known already gives −x + (−1) + 2(2) = 2, from
which x = 1. Putting all the results together gives the solutions as

x = 1, y = −1, z=2.

This is an example where there was a unique solution to the set of

equations.
106 Determinants and Matrices

Example 4.20: If No Solution Exists

Consider the equations

3x + 2 y + z = 3
2x + y + z = 0
6 x + 2 y + 4 z = −4

Let’s try the method of Gaussian elimination on this set of equations.

First, write the augmented matrix:

 3 2 1  3 
 
 2 1 1  0 
 6 2 4  6 

Let’s get two zeros in the first column

by
Row 2 = Row 2 – 2/3 Row 1 3 2 1 3
(R2: R2 – 2/3 R1)
0 −1 1 −2
3 3
and Row 3 = Row 3 – 2 Row 1 0 −2 2 0
(R3: R3 – 2R1)

Now let’s get a zero in the second 3 2 1 3

column by
0 −1 1 −2
Row 3 = Row 3 – 6 Row 2 3 3
(R3 = R3 – 6R2) 0 0 0 12

It can be seen that the last row contains a contradiction. By writing it out
as an equation it becomes 0 = 12? This tells one that the system of equations
has no solution. It is said that the system is inconsistent.

Example 4.21: If Infinitely Many Solutions Exist

Consider the equations

3x + 2 y + z = 3
2x + y + z = 0
−x − z = 3

Let’s try the method of Gaussian elimination on this set of equations.

First, write the augmented matrix:

 3 2 1  3
 
 2 1 1  0
 −1 0 −1  3
 4.4 Solving Systems of Linear Equations 107

Let’s get two zeros in the first column by

3 2 1 3
Row 2 = Row 2 – 2/3 Row 1
(R2: R2 − 2/3 R1) 0 −1 1 −2
3 3
0 2 −2 4
and Row 3 = Row 3 + 1/3 Row 1 3 3
(R3: R3 + 1/3 R1)

Now let’s get a zero in the second column 3 2 1 3

by
Row 3 = Row 3 + 2 Row 2 0 −1 1 −2
3 3
(R3: R3 + 2R2) 0 0 0 0

The last row gives nothing useful, since 0 = 0. The second row gives

1 1
− y + z = −2
3 3
or, multiplying each side by 3,

− y + z = −6

z is arbitrary; in other words, it can take any value. Suppose, letting a

parameter such as t represent any number for z, that is, z = t. Then y can be
found from y = z + 6 = t + 6. Then, from the first equation, 3x + 2y + z = 3,
gives 3x + 2(t + 6) + t = 3 or 3x = −3t − 9. So x = −t − 3.
So the “solution” can be given as, x = −t − 3, y = t + 6, z = t.

Any set of numbers that satisfied these equations would also satisfy the
original set of simultaneous equations. The parameter t can be replaced with
any real number. Hence, there are infinitely many solutions to this system.
The three situations be generalized as (1) a unique solution, (2) no real
solutions, and (3) infinitely many solutions. It’s definitely got something to
do with the last row (or rows) of the echelon form matrix being zero.
The rank of a matrix is the number of nonzero rows when it has been
reduced to echelon form. The three preceding situations finished up looking
like the following:

Unique solution: rank(A) = rank(A,d) = 3

0 0 c
non-zero

Infinite number of solutions: rank(A) = rank(A,d) = 2

0 0 0 0

No solutions: rank(A) = 2, rank(A,d) = 3

0 0 0 c
non-zero

Similar rules can be applied to any number of equations.

108 Determinants and Matrices

4.4.3 Matrix Inversion Method

Two simultaneous linear equations can be written out in matrix form. Consider
the following simultaneous equations:

x + 2y = 4
3x − 5 y = 1

These can be written in matrix form as

 1 2  x  4
 3 −5   =
  y   1 

Denoting

   x  
A =  1 2 , X =   , B =  4 
 3 −5   y  1

This gives the following matrix equation AX = B. This is the matrix of the simul-
taneous equations. Here the unknown is the matrix X, since A and B are already
known. A is called the matrix of coefficients.

4.4.3.1 Matrix Method for Solving Simultaneous Equations

Given the system of equations in matrix form

AX = B (4.12)

Multiplying both sides of Equation 4.12 by the inverse of A, provided it exists,

gives

A−1 AX = A−1B (4.13)

But A−1 A = I, the identity matrix. Also, IX = X, so this leaves

X = A−1 B (4.14)

This result given by Equation 4.14 gives a method for solving simultaneous
equations. First write the equations in matrix form, calculate the inverse of the
matrix of coefficients A−1, and finally perform a matrix multiplication.

Note: When det(A) = 0 there are no solutions and so the matrix is not invertible
because of division-by-zero problems.
 4.4 Solving Systems of Linear Equations 109

Example 4.22
Solve the simultaneous equations

x + 2y = 4
3x − 5 y = 1

Solution: Putting these in matrix form gives

 1 2  x  4
 3 −5   =
  y   1 

AX = B

 
Now calculate the inverse of A =  1 2  :
 3 −5 

1  −5 −2 
A−1 = −
11  −3 1


Then X is given by using

1  −5 −2   4 
X = A−1 B = −
11  −3 1   1 
1  −22 
=−
11  −11 
 
=  2
1

Hence x = 2, y = 1 are the solutions.

The same method can now be used to solve a system of three simultane-
ous equations as shown in the next example.

Example 4.23
Solve the simultaneous equations

− x − 2 y + 2z = 1
2x + y + z = 7
3 x + 4 y + 5z = 26
110 Determinants and Matrices

Solution: These can be put into a matrix equation as

 −1 −2 2   x   1 
 2 1 1  y =  7 
    
 3 4 5   z   26 

AX=B

The inverse of A is calculated as

 1 18 −4 
−11  
A = −7 −11 5 
23 
 5 − 2 3

To find the solution

 1 18 −4   1   23 
1    1  
X = A−1 B = −7 −11 5   7  = 46
23  23  
 5 − 2 3   26   69 

 1
X =  2
 
 3

that is, x = 1, y = 2, and z = 3.

4.5 Applications
Example 4.24: Fire Modeling Using Markov Chains
A Markov chain is a process that satisfies the Markov property, meaning
one can make predictions for the future of the process based solely on its
present state just as well as one could knowing the process’s full history.
The process involves the probabilities of being in states and can be mod-
eled using matrices. The use of matrices and matrix operations can provide
solutions to problems and is useful to consider them here.
Consider the four different states of a fire as follows: O, fire is out;
S, smoke development; F, flashover; and B, full burning fire. The time steps
are in minutes and Figure 4.1 shows the transition diagram modeling the
fire with transition probabilities between the different states. It is required
to find how long it will take the fire in the different states to go out, that is,
to reach the absorbing state O.
 4.5 Applications 111

0.9 0.6 0.1

B F

0.3 0.4
0.1

S
1.0
O

0.6

Figure 4.1 Transition diagram modeling a fire.

For this setup, a transition matrix P can be derived from the transition
diagram for the different states as follows,

O S F B
O 1 0 0 0
S 0.6 0 0.4 0
P=
F 0 0.3 0.1 0.6
B 0 0.1 0 0.9

Within this transition matrix P, there is a smaller submatrix that can be

formed by removing the absorbing state O and is defined as Q, where

 0 0.4 0 
Q = 0.3 0.1 0.6 

 
 0.1 0 0.9 

It turns out from the theory that an important fundamental matrix FM

gives the times from the different states to the absorbing state and can be
formed using

FM = ( I − Q)−1

where identity matrix I = I3 in this case and the power –1 indicates the
inverse of the (I − Q) matrix.
112 Determinants and Matrices

Constructing

 1 0 0   0 0.4 0   1 −0.4 0 
( I − Q) =  0 1 0  −  = 
   0.3 0.1 0.6   −0.3 0.9 −0.6 
 0 0 1   0.1 0 0.9   −0.1 0 0.1 

and calculating the inverse of this 3 × 3 matrix gives

S  1.67 0.74 4.4 

FM = ( I − Q)−1 = F  1.67 1.85 11.1 
 
B  1.67 0.74 14..4 

So, to obtain the average time to get from the different fire states to the
absorbing state (fire out) is given by adding along each row as follows:

Smoking state, S: 1.67 + 0.74 + 4.4 = 6.8 minutes

Flashover state, F: 1.67 + 1.85 + 11.1 = 14.6 minutes
Full burn state, B: 1.67 + 0.74 + 14.4 = 16.8 minutes

Example 4.25: Diagonal Factorization in Difference Equations

Given a first-order difference equation uk+1 = Auk governing some system
behavior with given start vector u 0, this equation can be solved as follows:

u1 = Au0
u2 = Au1 = A2u0
u3 = Au2 = AA2u0 = A3u0


uk = Aku 0 is the general solution to the equation for any k.

Now, to find what will be the state after, say, 100 iteration or time inter-
vals, letting k = 100 gives

u100 = A100u0

To calculate A100 using just matrix multiplication is very difficult, but if

one can diagonalize A using its eigenvalues and eigenvectors as A = PDP−1,
then it is much easier to find A100.

Note: This type of difference equation uk+1 = Auk is seen to appear in many
practical applications such as Markov Chains in the area of probabilis-
tic risk. Also, uk could be used to represent the position in an evacuation
model on a floor grid during a fire.
 4.5 Applications 113

Example 4.26: Dimensional Analysis in Fluid Mechanics

Consider the problem of how the drag force is affected on a smooth sphere
in some uniform fluid flow. The important parameters that can affect the
drag force F might be D the diameter of the sphere, v the velocity of the
fluid, ρ the density of fluid, and μ the viscosity of the fluid. This can be rep-
resented as F = f (D, v, ρ, μ), that is, F is some function of these four param-
eters that needs to be determined. If experiments were to be carried out to
see this dependence, say, 10 values of varying the diameter D while fixing
the other three parameters could be carried out. In the same way, varying
the other three parameters would have to be conducted as well. This will
require a lot of experiments and the time taken to do all these would be very
large and in reality impractical to do.
What is needed is a more efficient method and dimensional analysis is a
way that can reduce the problem to a more realistic situation in which only a
few experiments need to be carried out. The method used to solve this par-
ticular drag force problem uses the Buckingham Pi theorem, which is not
discussed here. However, within this method the property of dimensional
homogeneity is used. All equations will have the same units on both sides
of the defining equation and it is this property that then requires solving a
set of linear simultaneous equations.
To illustrate this idea of dimensional homogeneity further consider a sim-
pler problem. In mechanics, the horizontal range R traveled by a projectile
can depend on the parameters horizontal velocity Vx, the vertical velocity
Vy, and the gravitational acceleration g, that is, can say that R = f (Vx, Vy, g).
To determine this actual relationship, the units of the parameters are first
determined with velocity (meters/second) and acceleration (meters/seconds
squared). Using L = meter, T = time, L x = length in horizontal direction, and
Ly = length in the vertical direction, then the range R can be represented as

R ∝ Vxa Vyb g c (4.15)

where the parameters a, b, and c are powers to be determined using dimen-

sional homogeneity. Putting in the units for both sides of Equation 4.15
gives

1 a b c
 Lx  =  Lx T −1   L y T −1   L y T −2 

Considering the different dimensions and balancing the powers on both

sides gives the following:

L x : 1 = a, L y : 0 = b + c, T : 0 = − a − b − 2c

Now these form a system of three equations in three unknowns that need
to be solved to find a, b, and c, and hence give the form of relationship for R.
These can be solved by different methods but here the Gaussian elimination
method of Section 4.4.2 is used as follows:
114 Determinants and Matrices

a + b + 2c = 0
b+c= 0
a =1

First, the augmented matrix:

 1 1 2 0 
 
 0 1 1 0 
 1 0 0 1 

R3: R3 – R1 gives

 1 1 2 0 
 
 0 1 1 0 
 0 −1 −2  1 

R3: R3 + R2 gives

 1 1 2 0 
 
 0 1 1 0 
 0 0 −1 1 

The equations now read as,

a + b + 2c = 0
b+c= 0
−c = 1

From this it is seen that c = −1, b = 1, and a = 1, taking these values

and substituting them back into Equation 4.15 gives the relationship for the
range as

Vx Vy
R∝
g

Problems
4.1 Calculate the following determinants.

4 1 3
5 1
a. b. −1 2 7
3 −2
1 5 0
 Problems 115

1 k 2
4.2 Find the value of the constant k for which   k −1 0 1 = 0
1 2 0

 2 1 5 4 2 1
4.3 Let A =   and B =   , calculate the following.
 1 3 −2   3 −1 7 

a. A + B

b. A − B

4.4 Given

 4 1 
 
A =  3 2  and B =  2 −3 
 5 0   1 4 

calculate the following.

a. AB       b. B3        c. AAT

4.5 Solve the following set of simultaneous equations using the Gaussian
elimination method.

a. x + 2y + z = 7        b. x+ y−z = 2
3x + y + 4 z = 5 x + 2y = 5
2 x + 3 y − z = 14 2x + 3y − z = 7

4.6 An electrical network has currents i1, i2, and i3 given by the following set
of equations:

i1 + 3i2 + i3 = 9
2i1 − i2 + 5i3 = 7
4i1 + 2i2 − i3 = 11

Using the matrix inversion method, find the currents in the network.

4.7 For the matrix

2 1 3 
A=1 2 3 
 
 3 3 20 

a. Show that the characteristic equation given by det (λI − A) = 0 is

λ3 − 24λ2 + 65λ − 42 = 0.
116 Determinants and Matrices

b. Hence, find the eigenvalues and eigenvectors of the matrix A.

 
4.8 Given the 2 × 2 matrix A =  2 1  , determine the following.
 9 2
a. Diagonal Factorization of the matrix A.

b. Hence, calculate the matrix A10.

4.9 Consider again Example 4.24 with the four states of fire: O, fire is out;
S, smoke development; F, flashover; and B, full burning fire. The time
steps are given in minutes and Figure 4.2 shows the transition diagram
modeling the fire with transition probabilities between the different
states. Calculate how long it will take the fire in the different states to
reach the absorbing state O (i.e., to go out).

0.9 0.75 0.05

B F

0.2 0.5
0.1

S
1.0
O

0.5

Figure 4.2 Transition diagram modeling a fire with the probabilities between
states.
 5 Complex Numbers

5.1 Background
The origins of complex numbers came about when mathematicians were trying to
solve certain types of equations. Equations of the form x2 = 9, which had two real
solutions x = 3 and x = −3, were of no problem. But when the equation was of the
form x2 = −9, this has no real solutions since the square of a real number cannot
be negative. This was a problem. The idea to overcome this was to extend the real
numbers with the imaginary unit called j = −1, where j2 = −1, so that solutions
to equations such as x2 = −9 could now be found.
Originally, the symbol i was used to represent −1, but in many engineering
fields i was the symbol for electrical current and so the symbol j is used in engi-
neering instead.

5.2 Introduction and the Imaginary j

When solving a quadratic equation using the formula such as the general qua-
dratic equation ax2 + bx + c = 0, where a, b, and c are numbers. Then the solutions
are given by the formula

− b ± b 2 − 4 ac
x= (5.1)
2a

Notes:

• If the number under the square root sign is positive, there are two real and
distinct solutions.

• If the number under the square root sign is zero, there is one repeated solution.

• If the number under the square root sign is negative, there are no real solutions.

117
118 Complex Numbers

In the next section, it will be seen that in the third case the solutions can in fact
be written as complex numbers.
Now, what if the quadratic equation is given as x2 + 6x + 10 = 0? Using the
formula given by Equation 5.1 with a = 1, b = 6, and c = 10 gives

−6 ± 36 − 40 −6 ± −4
x= =
2 2

Unfortunately, the calculator cannot work out −4 . In fact, with “normal” (or
real) numbers, there is not any number that when multiplied by itself can give a
negative result.
But suppose there was an invented “number,” call it j, such that j2 = −1. What
then? It can be seen that

(2 j)(2 j) = 4 j 2 = 4(−1) = −4

and

(−2 j)(−2 j) = 4 j 2 = 4(−1) = −4

So −4 can be written as 2j or −2j and the solutions to the quadratic equation

can now be written as

−6 ± 2 j
x= = −3 + j or − 3 − j
2

These are complex solutions to the quadratic equation.

5.2.1 Some Properties of j
If j2 = −1, then j3 = −j, j4 = 1, j5 = j, and so on. You could also write
−4 = ±2 j, −9 = ±3 j, −16 = ±4 j, and so on.

5.2.2 Complex numbers:
A complex number is a number like 2 + 3j. The real part of 2 + 3j is 2. The imagi-
nary part of 2 + 3j is 3. (Note: The imaginary part doesn’t include j.)
In any equation with complex numbers, the real part of one side has to equal
the real part of the other side; similarly for the imaginary parts. There are two
equations rolled up as one. If it is given that x + yj = 3 − 4j, then it can be seen
straight away that x = 3 and y = −4.

5.3 Arithmetic Operations
5.3.1 Addition and Subtraction
It is easy to add and subtract complex numbers. It goes just like ordinary algebra;
add the real parts together and add the imaginary parts together.
 5.3 Arithmetic Operations 119

Example 5.1
1. (3 + j) + (1 + 2j) = 4 + 3j
2. (2 − 3j) − (1 + j) = 1 − 4j
3. (1 + 4j) + 2(5 − j) = 11 + 2j

5.3.2 Multiplication
Multiplying complex numbers is also simple, as long as you remember that j2 = −1.

Example 5.2
1. j(3 + j) = 3j + j2 = −1 + 3j
2. (2 + j)(4 + 5j) = 2(4 + 5j) + j(4 + 5j) = 8 + 10j + 4j + 5j2 = 3 + 14j

5.3.2.1 Conjugate Numbers
When multiplying complex pairs such as (2 − j)(2 + j) = 5 and (1 + 6j)(1 − 6j) =
37 notice here that each time that the result is a real number. The “partner” of
the complex number is called its conjugate. So the conjugate of 2 − j is 2 + j. The
conjugate of 2 + j is 2 − j. The conjugate of −5 − 7j is −5 + 7j. The conjugate of
100 j is −100 j. The conjugate of 27 is 27. The conjugate of a + bj is a − bj. Also
notice that the result of multiplying a number by its conjugate is always real and
positive. Generally, the following is found:

(a + bj)(a − bj) = a 2 + b 2

The conjugate of the complex number z is written either as z* or as z . So if

z = 2 + 3j, then z* = 2 − 3j and zz* = 13. And if z = 1 + 5j, then z* = 1 − 5j, and
zz* = 26.

5.3.3 Division
When it comes to division of complex numbers, it makes no sense to be dividing
by imaginary numbers. This idea of the complex conjugate helps out here as an
equivalent division can be done such that the denominator is now a real number
as seen in the next example.

Example 5.3

1 1(2 − j) 2− j 2− j 2 1
= = = = − j = 0.4 − 0.2 j
2 + j (2 + j)(2 − j) 22 + 12 5 5 5

Here, the conjugate of the denominator (2 + j) is (2 − j), and this is mul-

tiplied top and bottom such that the denominator becomes a real number,
which is 5 in this case.
120 Complex Numbers

Im

2 + 3j

Re

2 – 3j

Figure 5.1 Argand diagram showing complex number representation.

5.4 Argand Diagram
5.4.1 Drawing a Diagram of Complex Numbers
It is often helpful to draw the complex plane, or Argand diagram, and represent
the complex number 2 + 3j, say, as a vector starting at the origin and finishing at
the point (2,3) as shown in Figure 5.1.

Note: Adding and subtracting complex numbers follows the rules of vector addi-
tion and subtraction.

5.5 Polar and Exponential Form

5.5.1 Polar Form
Figure 5.2 shows the diagram of the complex plane showing the number 4 + 3j.
The complex number 4 + 3j can also be written as a distance from the origin
to the point and an angle measured anticlockwise from the real axis as (5 cos
36.9°) + j(5 sin 36.9°) or 5(cos 36.9° + j sin 36.9°). This is called the polar form
of the number. (4 + 3j is called the rectangular form). Sometimes the polar form

Im

4 + 3j

5
3
36.9°
Re
4

Figure 5.2 Different representations of the complex number.

 5.5 Polar and Exponential Form 121

is written in shorthand to 5 ∠ 36.9°. (The length is called the modulus. The angle
is called the argument.)

Note: The angle can be expressed in several ways, for example, –30° could be
written as 330°; 200° could be written as –160°.

Generally, a complex number in rectangular form is a + bj, as shown in

Figure 5.3.
Im

a + bj

r
b
θ°
Re
a

Figure 5.3 General point a + bj in the Argand diagram.

It can be seen that the modulus r is given by the Pythagorean theorem as

r = a 2 + b 2 and the angle θ is obtained using basic trigonometry and in this case
b
as θ = tan −1 .
a
An important property of the polar form is that this angle θ is not unique and
it is the case that θ + 360° n (where n = 0, 1, 2,…) would also be at the same posi-
tion in the complex plane. This will be very important when calculating roots of
equations in Section 5.6.
Also, the following relationship exists to give the polar form as

z = a + bj = r cosθ + r sin θ j = r (cosθ + j sin θ )

Note: Most calculators can easily swap a number from one form of a complex
number to the other using the Pol and Rec function buttons.

5.5.1.1 Multiplying and Dividing Complex Numbers in Polar Form

The polar form makes it very easy to multiply or divide complex numbers.
Because of the way sines and cosines work

(5 ∠ 25°)(2 ∠ 60°) = (5 × 2) ∠ (25° + 60°) = 10 ∠ 85°

Or when multiplying two numbers in polar form, just multiply their lengths
and add their angles.
122 Complex Numbers

Similarly,

5 ∠25°  5 
=   ∠(25° − 60°) = 2.5 ∠ (−35°)
6 ∠60°  2 

Or when dividing two numbers in polar form, just divide their lengths and
subtract their angles.

5.5.2 Exponential Form
Using Euler’s identity, ejθ = cos θ + j sin θ. Then, z = r (cos θ + j sin θ) = r ejθ. In
this form r is the modulus and θ is the argument in radians. This is known as the
exponential form of a complex number.

Note: The proof of Euler’s identity uses the Taylor series expansions of the expo-
nential, cosine, and sine functions, and is omitted here.

5.5.3 Powers of Complex Numbers

Multiplying a complex number by itself gives a power of that number:

(5 ∠25°)(5 ∠25°)(5 ∠25°)(5 ∠25°) = (5 ∠25°)4

Because of the way multiplication works with complex numbers in polar form,
this can be worked out easily as

(5 × 5 × 5 × 5)∠(25° + 25° + 25° + 25°) = 54 ∠ (4 × 25°).

This shows an important result that

(5 ∠25°)4 ≡ 54 ∠ (4 × 25°)

This result is known as De Moivre’s theorem and is stated more generally in

the next section.

5.5.4 De Moivre’s Theorem

De Moivre’s theorem relates to raising powers of complex numbers in polar form
as follows:

{r (cos θ + j sin θ )}n = r n (cos nθ + j sin nθ ) (5.2a)

Or

{r ∠θ}n = r n ∠ nθ (in shortened form) (5.2b)

 5.6 Roots of Equations 123

The result works for both positive and negative whole number powers and is
used for finding the roots of equations in the next section.

5.6 Roots of Equations
De Moivre’s theorem can also be used with fractional powers to find square roots,
cube roots, and so forth of complex numbers. There’s a difference, however:
This time, there should be more than one answer. To see why, try calculating
z3 if z = 2 ∠60°.
From De Moivre,

z 3 = {2 ∠60°}3 = 8 ∠180°

and this can be expressed in rectangular form as –8. If it is drawn on the complex
plane, it will look like Figure 5.4.
But suppose one tries the same thing with the complex number w = 2 ∠300°.
Then z3 = 8 ∠900°, and turning this answer into rectangular form, it’s the same
as before: –8. Are there any other numbers that would give the answer –8 when
cubed? Without knowing anything about complex numbers, one can clearly say
that (−2)3 = −8. Therefore, –2 can be written as a complex number in polar form
as 2 ∠180°.
So there appears to be three different complex numbers that, when cubed, give
the answer –8. These are as follows: z1 = 2 ∠60°, z2 = 2 ∠180°, z3 = 2 ∠300°. These
1
are the three cube roots of –8, or the three values of (−8) 3 . If these are drawn on
the complex plane, they will look like Figure 5.5. Notice, how they are equally
spaced (120°) around a circle.
The procedure to find all the different values for the roots is very straightfor-
1
ward. Suppose one needs to find z 4 , if z is the complex number 81 ∠160°. This
can be done by completing the following steps:
Im

Re
–8 0

Figure 5.4 Number –8 on the Argand diagram.

124 Complex Numbers

Im

Re
0

Figure 5.5 The three cube roots of –8 on the Argand diagram.

1 1
Step 1: z 4 = {81 ∠160°}4 Make sure the numbers are in polar form.
1
Step 2: = {81 ∠(160° + 360° n)}4 Add 360°n to the angle.

1
1
Step 3: = 814 ∠ (160° + 360° n) Using De Moivre.
4
= 3 ∠(40° + 90° n) Tidying up.

 3 ∠40°

 3 ∠130°
Step 4: = Taking the values of n = 0, 1, 2, 3…
 3 ∠220°
 3 ∠310°

To draw these, they would be equally spaced (90°) around a circle.

Notes:

• Starting with a number that is not in polar form, the first task is to turn it
into polar form.

• The answers are arrived at by taking n = 0, 1, 2, 3, … and go on until they

start to repeat. In the earlier example, the next one would be 3 ∠400°,
which is the same as 3 ∠40°, which is the first solution again.

• The number of solutions to expect is given by looking at the bottom num-

1
ber of the fractional power; for example, z 5 will have five values equally
1
spaced around a circle and z 7 will have seven values equally spaced
around a circle. So, for the mth root there are m values.
 5.7 Applications 125

• Having found the set of answers in polar form, these can easily be turned
into rectangular form if required.

• The principal root is defined as the one closest to the positive real axis.

Example 5.4
Determine all solutions of the equation z3 − 125 j = 0, giving your answers
in rectangular form.

Solution:

z3 − 125 j = 0
∴ z3 = 125 j Rearranging into standard equation form.
1
1
∴ z = (125 j) 3
Taking both sides to the power .
1
3
= {125 ∠90°}3 Change the number into polar form.
1
= {125 ∠(90° + 360° n)}3 Add 360° n to the angle.
1
1
= 125 ∠ (90° + 360° n)
3
Using De Moivre’s theorem.
3
= 5 ∠(30° + 120° n) Tidying up.
 5 ∠30°

=  5 ∠150° Taking enough values of n = 0, 1, 2, to get all
 5 ∠270°
 solutions.
 4.33 + 2.5 j

=  −4.33 + 2.5 j Change into rectangular form if required.
 −5 j


5.7 Applications
The use of complex numbers occurs in many areas of science and engineering,
and one such application is in determining conditions for flashover fires and disas-
ters to occur. Another important area is in electrical circuit theory with alternat-
ing currents, resistor, inductor, and capacitor circuits.

Example 5.5: Conditions for Flashover Fires to Occur

In a room on fire with a single door and a single fire bed, fire modeling can
lead to a simplified equation of the temperature T against time t:

dT
K = a2T 2 − a1T + 1 (5.3)
dt

where K is a constant depending on mass and specific heat capacity. The

right-hand side of Equation 5.3 is a quadratic equation in T with defining
126 Complex Numbers

parameters a1 and a2. From theory it turns out that if the roots of this qua-
dratic equation are complex in nature, then a flashover fire will occur.
As an example if a1 = 0.01 and a2 = 0.2 then the quadratic equation on
the right-hand side (RHS) becomes

0.2T 2 − 0.01T + 1 = 0

Solving this using Equation 5.1 gives

0.01 ± −0.7999
T=
0.4

which gives the two solutions as T = 0.025 ± j 2.236, which are two complex
conjugate roots. Hence for these parameter values the fire would develop
into a flashover state and have possible disastrous consequences.

Example 5.6: Representing Phases on the Complex Plane

Consider the RCL electrical circuit in Figure 5.6. From some basic electri-
cal circuit theory, the resistor has effective resistance R (ohms), the inductor
L (henrys) has effective resistance (known as reactance) X L , and the capaci-
tor C (farads) has effective resistance (known as reactance) XC.
From experiment, these are found to be as follows:

1
XC =
wC

X L = wL

where w = 2π f and f is the frequency of the source.

Now, to consider the total resistance of the series circuit one needs to
consider all three components, that is, R, L, and C. Again, from experiment

V
AC
i(t) L

Figure 5.6 An electrical RCL circuit.

 5.7 Applications 127

Im
j

XL

+90°

Re
0
–90° R

XC

Figure 5.7 The different phases for the component on the complex plane.

it is found that the current through the different components is either in

phase or out of phase with the voltage source V.
A phase diagram for the three components showing these phase shifts
can be drawn. For the resistor there is no phase shift of the current to the
voltage. For the inductor the voltage leads the current by +90°. For the
capacitor the voltage lags the current by −90°. To keep the resistances and
reactances separate due to the different phases, these can be put onto a dia-
gram using the real and imaginary axis for the corresponding phases. The
phasor diagram is shown in Figure 5.7.
So, knowing the values of R, L, C, and f for a given circuit, then the resis-
tances and reactances for the components can be calculated.

Example 5.7: Impedances and Currents in Electrical Circuits

Consider the circuit shown in Figure 5.8. To calculate the total impedance
for the circuit usually, called Z, add all the real and imaginary components
together as follows:

Z = 4 + j10 + 3 + (− j6)

Z = 7 + j4

Now to find the magnitude and phase angle of the total impedance Z, this
complex impedance can be converted into polar form to give Z = 8.06 ∠ 29.7°,
which gives Z = 8.06 Ω.

Note: Other quantities like the current in the circuit can be found using
V
complex number division, that is, using the relationship = .
Z
128 Complex Numbers

4Ω j 10 Ω

AC i(t)
100 V 3Ω

–j6Ω

Figure 5.8 An electrical series circuit with resistors, inductor and capacitor.

Example 5.8: Application of complex Numbers to Forces

Consider the force diagram in Figure 5.9. To find the resultant force and
the angle at which the force is acting, the forces can first be represented as
complex numbers in polar form and then in rectangular form.
The resultant force is given by FR = F1 + F2 + F3. Using the polar form of
a complex number as z = r cos θ + r sin θ j, each of the three forces can be
converted first into polar form then rectangular form as

F1 = 50 cos 45 + 50 sin 45 j = 35.36 + 35.36 j

F2 = 60 cos150 + 60 sin 150 j = −51.96 + 30 j

F3 = 40 cos 255 + 40 sin 255 j = −10.35 − 38.64 j

Im

F2 = 60 N F1 = 50 N
45°
60°

Re

15°

F3 = 40 N

Figure 5.9 Forces acting on a body.

 Problems 129

Im

FR = 37.95 N
26.72

44.75°
Re
–26.95

Figure 5.10 Resultant force FR on the Argand diagram.

So, the result force is FR = (35.36 + 35.36 j) + (−51.96 + 30 j) + (−10.35 −

38.64 j). Now, just add all the complex numbers to give, FR = −26.95 + 26.72 j.
This can be put onto an Argand diagram as shown in Figure 5.10. So,
the resultant force has a magnitude FR = 37.95 N and acts at an angle of
135.25° in an anticlockwise direction from the real axis.

Problems
5.1 Solve the following quadratic equations:

a. x2 + 4x + 5 = 0

b. 2x2 − x + 7 = 0

c. 3x2 + 6x + 11 = 0

5.2 Given that z1 = 3 + 5j and z2 = 1 + 2j, find the following:

a. z1 + z2

b. z1 − 3z2

c. z1z2

z1
d.
z2

5.3 Solve the following equation to find the two pairs of values of a and b:
(a + 3j) (4 − bj) = 37 + 9j.

5.4 Find the roots of the following equations, giving your answers in rect-
angular form.

a. z3 = 5 + 4j

b. z 5 − 2 + 3 j = 0
130 Complex Numbers

c. z 2 = 18 3 − 18 j

d. z3 + 27 j = 0

5.5 Consider the electrical circuit in Figure 5.11.

a. Find the total impedance Z for the circuit.

b. Hence, find the size of the current i in the circuit, giving the answer
in milliamps.
R = 32 Ω

XC = 15.92 Ω

V(t) =12 Vrms

i(t)
AC

XL = 62.83 Ω

Figure 5.11 An electrical RCL series circuit.

5.6 For a room fire the equation for the temperature T against time t is given
by Equation 5.3 as

dT
K = a2T 2 − a1T + 1.
dt

Determine the relationship between the parameters a1 and a2 for flash-

over to occur.
 6 Introduction
to Calculus

6.1 Differentiation
6.1.1 Definition of a Limit
For a straight line, to find the slope (or gradient) of the line, which is the same
across the whole line, it is the change in y against the change in x (Figure 6.1).

y
y = mx + c

Δy

Δx

Figure 6.1 The slope of a line given as a change in y against a change in x.

The formula for the slope is given as m, where

change in y ∆y
m= = (6.1)
change in x ∆ x

What happens if we have some general curve as shown in Figure 6.2. What is
meant by the slope of this curve? Clearly, the slope is changing at different points.

131
132 Introduction to Calculus

y = f (x)

Figure 6.2 Finding slopes of a general curve.

So, the slope at a given point P can be found by saying that this is also the slope
of the tangent line at that point, as shown in Figure 6.3.

y
y = f (x)

Tangent line

Figure 6.3 The slope of a curve as the slope of the tangent line.

At different points the slope would be different because of the different tangent
lines at those points. To find the slope at the different points derivatives and lim-
its are used. Consider the curve given in Figure 6.4. To find the slope at point P,
another point Q near to P is considered. The slope of the line joining P to Q, that
is, the secant line, is an approximation to the slope at P. So, the closer Q gets to
P the better the approximation of the slope at P. The slope of the line PQ is given
by using Equation 6.1 as

f ( x + h) − f ( x ) f ( x + h) − f ( x )
Slope of PQ = = (6.2)
x+h − x h

Now by letting h → 0, the line PQ becomes the tangent line at P to the curve,
which is the just slope of the curve at the point P. Therefore, the slope at the point
P is defined as
 6.1 Differentiation 133

y
y = f (x)

Q(x + h, f(x + h))

P(x, y)
x
x x+h

Figure 6.4 Finding the slope of a general curve at a point P.

dy f ( x + h) − f ( x )
Slope = = lim (6.3)
dx h→ 0 h

This is the definition of the derivative, that is, the slope of a curve at a particular
point P. This definition can be used to find slopes of different curves and to derive
a general formula.

Example 6.1
Review the curve y = x2 in Figure 6.5. Using the formula given by Equation
6.3, the slope of PQ is given as

dy ( x + h) 2 − x 2 2 xh + h 2
Slope = = lim = lim = lim (2 x + h)
dx h→0 h h→0 h h→0

y = x2

Q(x + h, (x + h)2)

P(x, x2)
x
x x+h

Figure 6.5 Slope of the curve y = x2.

134 Introduction to Calculus

Therefore, as h → 0,

dy
Slope = = 2x
dx
for the curve y = x2.
This process can be generalized for any function of the form y = axn.
Similar analysis gives

dy a( x + h)n − ax n
Slope = = lim
dx h→0 h

Using the binomial expansion of (x + h)n as

n(n − 1) n− 2 2
( x + h)n = x n + nx n−1h + x h + O(h 3 + higher )
2!

 n(n − 1) n− 2 2 
a  x n + nx n−1h + x h + O(h 3 + higher )  − ax n
dy
Slope = = lim  2 ! 
dx h→0 h

So,

dy  n(n − 1) n− 2 
= lim a nx n−1 + x h + O(h 2 + higher ) 
dx h→0  2! 

Now taking the limit as h → 0, the second and higher terms become zero
and this gives for any curve given of the form y = axn the derivative as

dy
= anx n−1 (6.4)
dx

This is now a general formula for finding the slopes of curves for powers of x.

Example 6.2
Given

dy dy
y = x7 , find . Solution: = 7x 6
dx dx
dy dy
y = 3x 5 , find . Soolution: = 15 x 4
dx dx
dy dy
y = 2 x −3 , find . Solution: = −6 x −4
dx dx
dy dy
y = x, find . Solution: =1
dx dx
dy dy
y =1 find . Solution: =0
dx dx
 6.1 Differentiation 135

Example 6.3
dy
If y = x5, find , then the gradient at the point (2, 32) on this curve.
dx
Solution:

dy
= 5x 4
dx

dy
At the point (2, 32), the gradient is = 5 × 24 = 80.
dx

dy
Note: Finding is called differentiating y with respect to x.
dx

What about adding or subtracting powers of x? See the following example.

Example 6.4
dy
If y = x6 + 3x4 − x2 + x, find.
dx
When differentiating, simply add and subtract the separate differentiated
dy
terms. So, in this case, becomes
dx

dy
= 6 x 5 + 12 x 3 − 2 x + 1
dx

6.1.1.1 Differentiating Fractional and Negative Powers of x

The same rule applies as with integer powers of x, that is, using the formula given
by Equation 6.4
dy
= anx n−1
dx

applies when the power n is a fraction or a negative number as well.

Example 6.5
1
If y = 2 x 2 − 6 x −4
then the slope becomes
1
dy −
= x 2 + 24 x −5.
dx

1
When the function is written in terms of x or
, then to begin with it is
x2
probably best to rewrite it with powers that are fractions or negative num-
bers, then use the rule given by Equation 6.4.
136 Introduction to Calculus

As a reminder, the following are useful when changing to fractional or

negative powers:

1 1
= x −2 x = x2
x2

1 1 −
1
= x −3 =x 2
x3 x

1 1 −
m
= x −n =x n
xn n
m
x

6.1.2 Stationary Points (Maxima and Minima)

Look at the graph shown in Figure 6.6 given by y = f(x). Notice that the highest
dy
point on the graph of y against x is reached when is zero. This is because the
dx
gradient at this point on the graph is zero. So, for such points the x values of the
stationary points can be found by putting the gradient function equal to zero, and
solving this equation. Then the y value can be found from the original equation
y = f(x).

dy
=0
dx

Figure 6.6 Graph of y = f(x) showing stationary points.

Example 6.6
Find the highest point on the graph of y = 5 + 2x − x2.

Solution: First find the gradient function

dy
= 2 − 2x
dx
 6.1 Differentiation 137

y These ‘highest’ and ‘lowest’ points are

called ‘maxima’ (single ‘maximum’)
and ‘minima’ (single ‘minimum’)

x
0 4

Figure 6.7 Graph showing a minimum point.

At the highest point, the gradient is zero. So,

dy
= 2 − 2x = 0
dx

Solving gives x = 1.
To determine the y coordinate, substitute the x value into y = 5 + 2x − x2
to give y = 6. So, the highest point on this graph is (1, 6).

The gradient will also be zero if our graph has a lowest point, like the
graph of y = x2 − 4x as sketched in Figure 6.7.
At a maximum or minimum point,

dy
=0 (6.5)
dx

Is it possible to tell from the equations whether it is a maximum or a mini-

mum point? Clearly, this is important in practical situations; sometimes one
wants to maximize profit or minimize wastage and so on.
There are several ways to check out the nature of a stationary point
(so-called because y is moving neither up nor down at such a point, hence
“stationary”). One way is the practical test and the other is the second
derivative test.

6.1.2.1 Practical Test
The practical test looks at what’s happening to the gradient on either side of the
dy
point. Taking the previous example, y = 5 + 2x − x2 with = 2 − 2 x , you can
dy dx
construct Table 6.1. Here the slope is calculated on either side of the station-
dx
ary point x = 1. By looking at the shape of the slope formed, this indicates that the
point (1, 6) is a maximum point.
138 Introduction to Calculus

Table 6.1 Considering the Slope on Either Side of the

Stationary Point
Value of x L (x = 0) (x = 1) R (x = 2)

dy
Sign of 2 0 –2
dx

6.1.2.2 Second Derivative Method

The second derivative method can sometimes be much quicker than the practical
test. It involves differentiating twice. Consider what’s happening to the gradient
as one moves from left to right past a maximum point as shown in Figure 6.8.
So, this can be summarized as

d  dy  d 2 y d2y
= is negative at a maximum,
, that is, <0 (6.6)
dx  dx  dx 2 dx 2

On the other hand, consider a minimum point shown in Figure 6.9. This can
be summarized as

At a local maximum, the gradient

is changing from positive to
negative, in other words it’s
decreasing. So the ‘gradient of
the gradient’ is negative.

Figure 6.8 How the gradient function changes at a maximum point.

At a local minimum, the gradient

is changing from negative to
positive, in other words it’s
increasing. So the ‘gradient of
the gradient’ is positive.

Figure 6.9 How the gradient function changes at a minimum point.

 6.1 Differentiation 139

Figure 6.10 Second derivative is zero, showing points of inflection.

d  dy  d 2 y d2y
  = 2 is positive at a minimum, that is, >0 (6.7)
dx dx dx dx 2

What happens if the second derivative is zero? Well, there might be a point of
inflection, that is, neither a maximum or minimum point, but still the slope is zero
as shown in Figure 6.10.
The easiest way to check it out is to make a table, as in the first method of the
practical test.

6.1.3 Differentiating Products and Quotients

So far, differentiating powers of x and finding maxima and minima have been
considered. But functions such as y = (1 + 3x)(2x4 + 6x) have not been tackled, at
least not without multiplying out the brackets. Next, the derivatives of functions
that are products and quotients are considered.

6.1.3.1 Products
This type of function consists of one part multiplied by another; it is a product of
the two parts. The way we do it is to differentiate only one part at a time.
If y = (1 + 3x)(2x4 + 6x), then

dy
= (1 + 3x)(8x3 + 6) + (3)(2x 4 + 6x)
dx

plus
put down the first part differentiate the first part

differentiate the second put down the second

This idea can be expressed mathematically by the product rule formula.

If y = u.v, where u and v are functions of x then

dy dv du
=u +v (6.8)
dx dx dx
140 Introduction to Calculus

Example 6.7
dy
If y = (x2 − 2)(x2 + 1), find , simplifying your result.
dx
Solution:

dy
= ( x 2 − 2)(2 x ) + (2 x )( x 2 + 1) = 4 x 3 − 2 x
dx

6.1.3.2 Quotients
Sometimes there may be functions like

x2 + 1
y=
x2 − 1

dy
How do you find in this case?
dx
Well, there’s a formula that can be used similar to the one for the product rule.
It is a little more complicated than the one for products, but quite simple to use.
u
It is called the quotient rule. Expressed mathematically, if y = , where u and v
are functions of x, then v

du dv
dy v dx − u dx
= (6.9)
dx v2

Expressed in words this is,

minus

(put down the bottom part)(differentiate the top) — (put down the top)(differentiate the bottom)
(put down the bottom)2

Note: Start with the bottom part; remember it’s a minus sign linking the terms on
the top. The squared term on the bottom is not usually multiplied out.

Example 6.8

5x + 1 dy
If y = ,   find . Simplify your results (top only).
5x + 3 dx
 6.1 Differentiation 141

Solution:

dy (5 x + 3)(5) − (5 x + 1)(5)
=
dx (5 x + 3)2

dy 10
=
dx (5 x + 3)2

6.1.4 Standard Functions

So far, the slope dy when y is some power of x, a product and a quotient have been
dx
considered. It is now time to consider how to differentiate such functions as e3x,
sin x, and cos x etc.
The following results can all be proved from first principles, but for the
moment it is better to just make use of them. There is no need to learn these,
but there will be a need to be able to look them up and use them, both with and
without substituting numbers. Table 6.2 shows some standard functions that are
commonly used.

Table 6.2 Derivatives of Some Standard Functions

dy
y
dx
xn nxn−1
ekx kekx
ln x 1
x
sin kx k cos kx (provided x is in radians)
cos kx −k sin kx (provided x is in radians)
tan kx k sec2 kx (provided x is in radians)

Examples 6.9
dy
1. If y = x2 + 3ex, find when x = 2.
dx
Solution:

dy
= 2 x + 3e x = 26.2
dx

dy
2. If y = 4x + sin x, find when x = 1. (Be careful!
dx
Remember to
Solution: use radians
when working
dy these out.)
= 4 + cos x = 4.54
dx
142 Introduction to Calculus

Clearly, it’s not difficult to differentiate such functions. What about using
the rules for product and quotient? They apply just as well to these functions
as they did to powers of x. But do make sure the correct rule is being used.
To recap again, there are two arrangements to watch out for they are the
product and quotient.

dy dv du
Product: y = u.v then use =u +v
dx dx dx
du dv
v −u
u dy dx dx
Quotient: y = then use =
v dx v2
In the following, more examples using the product and quotient rule are
given.

Example 6.10
If y = x2 e3x, then

dy
= ( x 2 )(3e3 x ) + (2 x )(e3 x ) = 3 x 2e3 x + 2 xe3 x
dx

Example 6.11

ln x
If y = , then
x

 1
( x )   − (1)(ln x )
dy  x 1 − ln x
= =
dx x2 x2
Next is a more complex problem to show how to make use of these
derivatives.

Example 6.12
2x
Find any maximum or minimum points on the graph of y = .
ex
dy
Solution: Remember, the procedure is to first find , then put this equal to
dx
zero and solve it to find x. Next find the y-values that correspond with the
x-value(s), and last decide which sort of stationary point it is (maximum,
minimum, or point of inflection).
(This function is a quotient, so use the quotient rule when differentiating.)

2x dy
Starting with y = , find .
ex dx
dy (e x )(2) − (e x )(2 x ) 2(1 − x )
= =
dx (e x )2 ex
 6.1 Differentiation 143

dy 2
At the stationary point, = 0 and so 1 − x = 0 giving x = 1 and y = .
dx e
 
So, the stationary point is at  1, 2  .
 e
To check what kind of point it is, you could use the second derivative
test, but here it is simpler to use the practical test about the point x = 1 as
shown in Table 6.3.
Table 6.3 Slope of the Curve about the Stationary Point
Value of x L (x = 0) (x = 1) R (x = 2)

dy
Sign of 2 0 –0.27
dx

 2
So, the point  1,  is a local maximum point and the graph of this func-
e
tion is shown in Figure 6.11.

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5 4
x

2x
Figure 6.11 Graph of the function y = .
ex

6.1.5 Function of a Function (Chain Rule)

Consider a function to differentiate that looks like the following: y = (x2 + 5x + 1)9.
This combines two ideas that can be tackled: y = (x2 + 5x + 1) and y = x9
Think of a function as a box or machine that does something to an input. Then
the combined function can be pictured as shown in Figure 6.12.
Clearly, the rate at which the second function changes depends on how fast
dy
the first one is changing. To get for the combined function, just combine the
dx
derivatives from each one in a simple way.
144 Introduction to Calculus

x2 + 5x + 1 x9

Figure 6.12 Treating the function as a combined function process.

If y = (x2 + 5x + 1)9, This rule is also

called the ‘chain
dy rule’:
then = 9 (x2 + 5x + 1)8 × (2x + 5)
dx
dy dy du
= ×
dx du dx
differentiate the inside
where, in this case,
don’t change
the inside of the bracket y = u9 and
u = x2 + 5x + 1

Note: The two parts are multiplied together.

Example 6.13
dy
Find for the following functions.
dx

1. y = (4x − 5)10

dy
Solution: = 10(4 x − 5)9 (4) = 40(4 x − 5)9
dx
2. y = sin (x2)

Solution: dy
= cos( x 2 )(2 x ) = 2 x cos( x 2 )
dx

6.2 Integration
6.2.1 Introduction and the Riemann Sum
b

What does
∫ f (x) dx represent? An incomplete answer is to say it is the area under
a
the curve shown in Figure 6.13. Yes, it does give the area under the curve, but the
 6.2 Integration 145

y = f (x)

x
a b

Figure 6.13 Area under the curve y = f(x).

area under the curve is just an application of integration. There are many appli-
cations of integration, as will be shown in the later chapters. A more complete
b

answer to the question is as follows:

∫ f (x) dx is the “limit of approximation by
a
breaking things into pieces.” This can be explained further with the diagram in
Figure 6.14.

Δx

a xp* b

Figure 6.14 An interval on the x axis.

Consider the interval from a to b. Take this interval and chop it up into pieces of
varying lengths. Then for each piece pick a point. Then evaluate the function f at
these points and then sum over all the pieces multiplied by the width of the piece.
So,
b

∫ f (x) dx  lim
∆x → 0 ∑
all pieces
f ( x*p )∆ x (6.10)
a

This idea can be used in many applications.

Example 6.14
Find the area under the curve y = f(x) from a to b as shown in Figure 6.15.

Area ≈ Area of the rectangle ≈ ∑

all pieces
f ( x*p )∆ x
146 Introduction to Calculus

y = f (x)

Total Area ≈ A1 + A2 + A3 + A4
A1 A2 A3 A4

x
a Δx b

Figure 6.15 Finding the area under the curve in an interval.

By chopping up the interval into smaller and smaller rectangles as Δx → 0,

the actual area under the curve is obtained.
Therefore,

lim
∆ x→0 ∑
all pieces
f ( x*p )∆ x =
∫ f (x) dx = Area
a

The curly symbol ò is understood to mean a limiting summation.

6.2.2 Fundamental Theorem of Calculus (Optional Section)

b

6.2.2.1 How to Compute the Integral

∫ f(x)dx
a
You will need to make use of the fundamental theorem of calculus, some function
x

f that contains the interval [a, b]. Now define F ( x ) =

interval [a, b] as shown in Figure 6.16.
∫ f (t) dt , where x is in the
a
Taking the derivative of F(x) as F(x)′ gives

 F (x + ∆ x ) − F (x) 
F ( x )′ = lim  
∆ x→0
 ∆x 

F(x)
y = f (t)

t
a x b

Figure 6.16 Defining an area in a region.

 6.2 Integration 147

This is from the definition of the limit.

This can be written as

 x+∆ x x 
F ( x )′ = lim 

∫ a
f (t ) dt −
∫ a
f (t ) dt 

∆ x→0  ∆x 
 

The area between x and x + Δx is shown in Figure 6.17.

So,

x+∆ x x x+∆ x

∫
a
f (t ) dt −
∫ f (t) dt = Shaded area = ∫
a x
f (t ) dt

Therefore,

x+∆ x 
1  
F ( x )′ = lim
∆ x→0 ∆ x

 x
f (t
∫) dt 


The mean value theorem of definite integrals expresses that there exists a c
where (x ≤ c ≤ x + Δx) such that

x+∆ x

f (c)∆ x = area under the curve =

∫
x
f (t ) dt

So,

x+∆ x
1
f (c) =
∆x ∫ x
f (t ) dt

F(x) y = f (t)

f (c)

t
a x c x + Δx b

Figure 6.17 Area in the region x and x + Δx.

148 Introduction to Calculus

Or there exists a c in [x, x + Δx] where

F ( x )′ = lim { f (c)}
∆ x→0

But x ≤ c ≤ x + Δx, so as Δx → 0, c → x and f(c) → f(x) as Δx → 0. Therefore,

F′(x) = f(x)
Why is this important?
For some function f(x), then there exists a function F(x), thus the derivative of this
function is equal to the function f(x). Or, there is an antiderivative function F(x)
any continuous function f(x). This is also a connection between differentiation
and integration.
b

Now, having this, we can compute ∫ f (x) dx. This is now given by
a

∫ f (x) dx = F (b) − F (a)

a
(6.11)

where F′(x) = f(x).

Since integration is an antiderivative process it then follows that the rule for
differentiating any power of x is reversed when for integrating any power of x.

If y = xn, then

x n+1
∫ x n dx =
n +1
+C (6.12)

provided that n ≠ −1 and C is an arbitrary constant.

Note: In words, this particular rule becomes “add one to the power and divide
by the new power.”

Example 6.15

x5
1.
∫ x 4 dx =
5
+ C

3x 6 x6
2.
∫ 3 x 5 dx =
6
+ C =
2
+ C

x1.5
3.
∫ x 0.5 dx =
1.5
+ C

x −1 1
4.
∫ x −2 dx =
−1
+ C = − + C
x
 6.2 Integration 149

These are called indefinite integrals as the value of the constant C is unknown.
For definite integrals, when there are limits of integration, use the for-
mula given by Equation 6.11.
b

∫ f (x) dx = F (b) − F (a)

a

Example 6.16

2 x dx = [ x 2 + C ]0 = (12 + C ) − (0 2 + C ) = 1 − 0 = 1
∫
1

Note: The C has canceled out. This will always happen with a definite inte-
gral, and so there is no need to write it.

Hence, the preceding calculation would normally be written as

2 x dx = [ x 2 ]0 = 12 − 0 2 = 1 − 0 = 1
∫
1

6.2.3 Standard Integrals and Areas under Curves

All the integrating so far has been dealing with powers of x.
Just as with differentiating, finding integrals of other functions may be
necessary. These are summarized in the Table 6.4.
Table 6.4 Standard Integrals for Some Common Functions
y
∫ y dx
xn x n+1
+ C , n ≠ −1
n +1
1 ln x + C
x
ekx e kx
+C
k
sin kx cos kx
− +C
k
cos kx sin kx
+C
k
sec2 kx tan kx
+C
k
tan x −ln(cos x) + C
1 1 −1 x
tan +C
x 2 + a2 a a

Note: x must be in radians when using the trigonometric functions.

150 Introduction to Calculus

The shaded area is

given by,
5
y dx
2

x
2 5

Figure 6.18 Calculating areas under curves.

6.2.3.1 Finding the Area under a Curve

Provided that the whole curve is above the x-axis, the definite integral gives the
area under a curve as shown in Figure 6.18.

Example 6.17
1. Find the area under the curve y = 10x − x2 between the x-values 2
and 8.

Solution: The area will be

8 8
 x 2 x 3   8 2 83   2 2 23 
∫
2
(10 x − x ) dx = 10 −  = 10 −  − 10
2
 2 3 2  2 3   2
−
3
 = 132

So, the required area is 132 square units.

2. Find the area under the curve y = sin x between the x-values 0
and π.

Solution: The area will be

∫ (sin x) dx = [ − cos x ] = {− cos π} − {− cos 0} = 1 + 1 = 2

π
0
0

So, the required area is 2 square units.

Note: One, or both, of the limits may be negative, as long as the graph stays
above the x-axis.

3. Find the area under the curve y = x3 + 4x2 + 3x + 10 between the

x-values –3 and –1.
 6.2 Integration 151

Solution: The area will be

−1 −1
 x4 x3 x2 
∫
−3
( x + 4 x + 3 x + 10) dx = 
3 2
 4
+4
3
+ 3 + 10 x 
2  −3
 7  1 2
= −9  − −32  = 22
 12   4 3

So, the required area is 22.6667 square units (to four decimal places).
This area is shown in Figure 6.19.
20
18
16 It is found that the
area under the
14
graph, between
12 the two vertical
y 10 lines, is 22.67
8 square units.
6
4
2
0
–4 –2 2 x 4

Figure 6.19 Graph of the function y = x3 + 4x2 + 3x + 10.

What happens if the graph goes below the x-axis? Everything below the
axis results in a negative contribution to the integral, so if there are
parts above and parts below the axis, then there is a need to work out
the upper and lower bits separately. Figure 6.20 shows a cubic function.

4
2
–3 –2 –1 1 2 x 3 4 5
0
–2
–4
–6
y –8
–10
–12
–14

Figure 6.20 Graph of the cubic function y = x3 − 3x2 − 4x.

4. Find the total area enclosed by the graph and the x-axis between
x = −1 and x = 4.

Solution: First, work out the area above the x-axis between x = −1 and x = 0:
0 0
 x4 x3 x2   3
∫
−1
( x − 3 x − 4 x ) dx = 
3 2
 4
− 3 − 4  = {0} − −  = 0.75
3 2  −1  4

So, this area is 0.75 square units.

152 Introduction to Calculus

Now work out the area below the x-axis between x = 0 and x = 4:
4 4
 x4 x3 x2 
∫
0
( x 3 − 3 x 2 − 4 x ) dx = 
 4
− 3 − 4  = {−32} − {0} = −32
3 2 0

The integral is negative, but the area is positive; so this area is 32 square
units.
Adding together these two areas, the total area enclosed by the graph and
the x-axis over the interval −1 ≤ x ≤ 4 is 32.75 square units.

6.2.4 Improper Integrals
b

bounded and finite in the interval [a, b]. a

∫
Recall that for the integral to exist and call it definite, f ( x ) dx , then f(x) must be

But sometimes there can be the situation as shown in Figure 6.21. f(x) is a
b

bounded function but not finite, so

∫ f (x) dx is an improper integral. In Figure 6.22,
a
the interval is finite but f(x) is not bounded. Again, this type of integral shown in
Figure 6.22 is called an improper integral.

(i) f (x)

f (x) is bounded but it is not finite [a, ∞]

x
a ∞

Figure 6.21 f(x) is a bounded function but not finite.

f(x)
(ii)

b
∫a f (x) dx–interval is finite but
the function is not bounded

x
a b

Figure 6.22 The interval is finite but f(x) is not bounded.

 6.2 Integration 153

f (x)

1
y=
x2

x
1 ∞

1
Figure 6.23 The function y = is finite but unbounded.
x2

Example 6.18
1
Consider the function given by the curve y = , as shown in Figure 6.23.
x2
Find the area under the curve from x = 1 to x = ∞.

Solution: This is expressed as

∞
1
A=
∫x 1
2
dx

The following describes how to deal with these types of integrals (one of
the limits is infinite ∞).
Let x go up to some value x = t.

∞ t
1 1
∫
1
x2
dx = lim
t → ∞ ∫x
1
2
dx

Do the integral as normal and at the end use the limit as t → ∞.

t t t
1  x −1 
= lim
t → ∞ ∫
1
x2
dx = lim
t → ∞ ∫
1
x −2 dx = lim 
t → ∞  −1 1


t
 −1   −1 −1   1
= lim   = lim  −  = lim 1 − 
x t 1 t →∞  t
t → ∞
 1 t → ∞
 

∞
1 1
So, as t → ∞,
t
→ 0 . Therefore,
∫x
1
2
dx = 1

Since this number is finite, the integral

154 Introduction to Calculus

∞
1
∫x
1
2
dx

is convergent. Otherwise, the integral is said to be divergent.

Example 6.19
+∞

Calculate
∫x e
−∞
2 − x3
dx .

This can be split into two equivalent integrals:

+∞ 0 +∞

∫x e
−∞
2 − x3
dx =
∫x e
−∞
2 − x3
dx +
∫x e
0
2 − x3
dx

0 v

lim
t → −∞ ∫x e
t
2 − x3
dx + lim
v → ∞ ∫x e
0
2 − x3
dx

Start with

0 0
 −e − x 3   −1 1 − t 3 
lim
t → −∞ ∫x e
t
2 − x3
dx = lim 
t → −∞  3 t
 = lim
t →
m 
−∞
+ e 
 3 3 

− t3
As t → −∞, e → ∞, so

∫x e
−∞
2 − x3
dx → ∞

The first part is divergent and so the original integral is divergent,

that is,

+∞

∫x e
−∞
2 − x3
dx → ∞
 6.3 Integration Techniques 155

Example 6.20
There is an infinite discontinuity in the following integral. There is a prob-
lem when x = 0 in the integrand.
9
1
Calculate
∫
0
x
dx.

Here, replace the zero with the symbol a and let this a → 0.

9 9
1  1
∫x
−
lim 2 dx = lim +  2 x 2  a = lim+  2 9 − 2 a 
a → 0+ a → 0 a→ 0
a

as a → 0, a → 0.
Therefore,

9
1
∫
0
x
dx = 6

6.3 Integration Techniques
6.3.1 Substitution
In this method, the idea is to replace the complicated part of the integral by letting
it equal a new variable, say, u. This then involves changing the integral over to a
u variable problem with respect to du.

2
Take the first example,
∫ x + 3 dx. Write a new letter, say, u, to stand for the
expression x + 3. Now, one must be very careful to keep track of the three vari-
ables x, y and u. The working will look like this after the substitution for u and
replacing for dx.

u = x+3
2 2
∫ ∫
du
dx = du = 2 ln(u) = 2 ln( x + 3) + C ∴ =1
x+3 u dx
∴ du = dx

The reader could check back by differentiating this result and the result
2
is indeed .
x+3
156 Introduction to Calculus

Example 6.21

Calculate
∫ 4 cos(100πx) dx .
First, make the substitution. Let u = 100πx and also replacing for dx
gives

du 4
∫ 4 cos(100πx) dx = ∫ 4 cos(u) 100π = ∫ 100π cos(u) du u = 100 πx
du
4 ∴ = 100π
= ( − sin(u)) + C dx
100 π
du
4 ∴ = dx
=− sin(100 πx ) + C 100π
100 π

In all of these examples, the connection between du and dx has involved

a number. But there might be a situation where one wants to replace x2 + 3,
say, with the letter u.
du du
In this case, = 2 x and so dx will have to be replaced by .
dx 2x

Example 6.22

6x
∫x 2
+3
dx

Make the substitution. Let u = x2 + 3 and also replacing for dx gives

6x 6 x du 3 u = x2 + 3
∫ x2 + 3
dx =
∫ .
u 2x
=
∫ u
du
∴
du
= 2x
= 3 ln(u) + C dx
du
= 3 ln( x 2 + 3) + C ∴ = dx
2x

Finally, definite integrals (those with limits) can be done by the sub-
stitution method. Of course, one can do it just as earlier, using the limits
at the end to calculate the answer. But it is much quicker to shortcut a
little by changing the limits so that one does not return to the variable
x at all.

Example 6.23
Consider the integral with limits as

5
2x
∫x
1
2
+7
dx

Now, recall that the limits 1 and 5 mean x = 1 and x = 5.

 6.3 Integration Techniques 157

When changing the variable of integration to u, at the same time the

limits of u-values can be changed too. If this is done, there is no need to
return to the variable x at all.
But one will have to calculate the two u limits; this can be done in the
du
margin with the ‘ ’ calculation.
dx
So, the whole solution will look like this,

5 32
2x 2 x du u = x2 + 7
∫1
x +7
2
dx =
∫
8
.
u 2x
∴
du
= 2x
32 dx
du
∫ u = [ ln u ]
32
= du
8 ∴ = dx
8
2x
= ln(32) − ln(8) when x = 1, u = 8
= ln(4) when x = 5, u = 32

6.3.2 Partial Fractions
If one is faced with an integral that contains a fraction, first check to see if the top
is the derivative of the bottom of the fraction. For example:

2x 3x 2 + 4 x x
∫ x +3
2
dx ,
∫ x3 + 2x 2 − 5
dx ,
∫x 2
+4
dx

Note: In the case of the last example, the exact of the derivative on the top does
not exist, but it can be made to do so as follows,

x 1 2x
∫x 2
+4
dx =
2 ∫x 2
+4
dx

and now this can be done like the others using the method of substitution.
If, however, the top does not contain the derivative of the bottom, and if
the bottom part can be factorized, then try using integration using partial
fractions.
The different types of forms of partial fractions splits are summarized as
follows:

1. If the factors on the bottom are different and linear:

3x + 1 A B C
= + +
( x + 1)( x + 2)( x − 3) x + 1 x + 2 x − 3
158 Introduction to Calculus

2. If there is a repeated factor on the bottom:

1 A B C
= + +
( x + 2)( x − 1)2 x + 2 x − 1 ( x − 1)2

3. If a factor has higher powers than just x:

3x + 1 A Bx + C
= + 2
( x − 1)( x + 1) x − 1 x + 1
2

There may be a combination of these types. A, B, and C are constants

that need to be worked out before doing the integral.
First, let’s try to work out some partial fractions, leaving the integration
part for later.

6.3.2.1 Type 1: Different Linear Factors

Example 6.24
5x − 1
Express in partial fractions.
( x + 1)( x − 2)
1. Write the fraction as separate parts, putting A, B, C, … for the
unknown numerators of the partial fractions. To use the example
above, write as follows:

5x − 1 A B
= +
( x + 1)( x − 2) x + 1 x − 2

2. Make the “little fractions” back up into one “big fraction.”

5x − 1 A B A( x − 2) + B( x + 1)
= + =
( x + 1)( x − 2) x + 1 x − 2 ( x + 1)( x − 2)

3. Put the tops of the right- and left-hand sides equal. This follows, since
the bottoms are equal.

5 x − 1 ≡ A( x − 2) + B( x + 1)

4. Substitute values for x that make each bracket equal to zero. This will
give the values for A, B, and so on.

In this case, the first bracket can be made zero if x = 2, and the second
bracket is zero if x = −1.
Letting x = 2:

5(2) − 1 = A(0) + B(2 + 1)

∴ 9 = 3B
∴B = 3
 6.3 Integration Techniques 159

Letting x = –1:

5(−1) − 1 = A(−1 − 2) + B(0)

∴ −6 = −3 A
∴A= 2

Now, the partial fraction split has finished. Next, write the partial frac-
tion out as

5x − 1 2 3
= +
( x + 1)( x − 2) x + 1 x − 2

6.3.2.2 Type 2: Denominator with a Repeated Factor

Example 6.25
1
Express in partial fractions.
( x + 2)( x − 1)2
First, write the form of the partial fraction, referring to different forms:

1 A B C
= + +
( x + 2)( x − 1) 2
x + 2 x − 1 ( x − 1)2

Now proceed as before. First, make the right-hand side into one big fraction:

1 A B C A( x − 1)2 + B( x + 2)( x − 1) + C ( x + 2)
= + + =
( x + 2)( x − 1)2 x + 2 x − 1 ( x − 1)2 ( x + 2)( x − 1)2

Then make the top parts equal, since the bottom parts are the same:

1 = A( x − 1)2 + B( x + 2)( x − 1) + C ( x + 2)

And now need to think of which numbers we could substitute for x that
will make the brackets equal to zero. If x = 1, then

1 = A(0)2 + B(2)(0) + C (3)

1
So, 1 = 3C, which gives C = .
If x = −2, then 3

1 = A(−3)2 + B(0)(−3) + C (0)

1
So, 1 = 9A, which gives A = .
9
Now, to find B, either substitute any other number for x and make use of
the values of A and C, or look at the powers of on each side of the equation
to equate coefficients of the powers of x.
160 Introduction to Calculus

Start with

1 = A( x − 1)2 + B( x + 2)( x − 1) + C ( x + 2)

Looking at the coefficients of

1
[ x 2 ]: 0 = A+ B = +B
9
1
∴ B=−
9

The work is done; just write out the partial fraction in full:

1 1 1
−
1
= 9 + 9 + 3
( x + 2)( x − 1)2 x + 2 x − 1 ( x − 1)2

This is more useful written with the constants at the front as follows:

1 1 1  1 1  1 1 
=  − +
( x + 2)( x − 1)2 9  x + 2  9  x − 1  3  ( x − 1)2 

6.3.2.3 Type 3: Denominator with a Quadratic Factor

Example 6.26
3x + 1
Express in partial fractions.
( x − 1)( x 2 + 1)
Start by writing the correct form for the partial fraction, referring to the
list of types:

3x + 1 A Bx + C
= + 2
( x − 1)( x + 1) x − 1 x + 1
2

Now proceed as before, first making the right-hand side into one big
fraction:

3x + 1 A Bx + C A( x 2 + 1) + ( Bx + C )( x − 1)
= + =
( x − 1)( x 2 + 1) x − 1 x 2 + 1 ( x − 1)( x 2 + 1)

Now putting the top parts equal, since the bottom parts are the same:

3 x + 1 = A( x 2 + 1) + ( Bx + C )( x − 1)
 6.3 Integration Techniques 161

This time, there will only be one number that can be substituted for x to
obtain A.
If x = 1, then

3(1) + 1 = A(2) + ( Bx + C )(0)

∴4 = 2A
∴A= 2

Check the constant terms by putting x = 0, 1 = A + C(−1), so C = 1.

Finally, to get a value for B, look at the coefficients of x2:

[ x 2 ]: 0 = A+ B
∴ B = −2

All three numbers A, B, and C are found and now the partial fraction can
be written out in full as

3x + 1 2 −2 x + 1
= + 2
( x − 1)( x + 1) x − 1 x + 1
2

6.3.2.4 Performing the Final Integration

Usually, will end up with three different types of function to integrate as follows:

3
1. ∫ x + 1 dx
2
2. ∫ ( x + 4) 2
dx

2x + 5
3. ∫x 2
+1
dx

The first two can be done by substitution, although the results are different.
The last integral has to be tackled in a different way. Another standard integral
must be used, which again can be looked up in a standard table of integrals as
shown in Table 6.2 and is reproduced here.

1 1 x
∫x 2
+a 2
dx = tan −1 + C
a a
(6.13)

Again looking at the last integral from earlier,

2x + 5
∫x 2
+1
dx
162 Introduction to Calculus

what must be done here is split the top of the fraction and make two parts out of
the expression as follows:

2x + 5 2x 5
∫x 2
+1
dx =
∫x 2
+1
dx +
∫x 2
+1
dx

The first part can be done by substitution again; the second part can be
done using the standard integral given by Equation 6.13, where a2 = 1, and
so a = 1. Putting all this together gives the solution as

2x + 5 2x 5
∫x 2
+1
dx =
∫x 2
+1
dx +
∫x 2
+1
dx = ln( x 2 + 1)) + 5 tan −1 x + C

6.3.3 Integration by Parts
Sometimes there is a product of two functions that needs integrating as follows:

∫xe x
dx

There is a formula that is derived from the product rule and can be used to find
the integral of a product of two functions, that is,

dv du
∫ u. dx dx = u.v − ∫ v. dx dx (6.14)

dv
To use this formula, the term, which is , must be able to find its inte-
dx
gral easily. Now, if both terms are easy to integrate, then any powers of x is
chosen to be the u term.

Example 6.27

Evaluate
∫xe x
dx
dv
Since both functions are easy to integrate let u = x and then = ex.
dx
du
For the right-hand side of Equation 6.14, you need the and the v terms.
dx
du
= 1 and v = e x
dx

so,

∫ xe dx = xe − ∫ e dx
x x x
 6.3 Integration Techniques 163

This gives the final answer as

∫ xe dx = xe
x x
− e x + C = ( x − 1)e x + C

Example 6.28

Evaluate
∫ x ln(x) dx
Now clearly the ln(x) is not easy to integrate so it must be the u term, that
dv
is, u = ln(x) and then = x.
dx

du 1 x2
= and v =
dx x 2

Using the formula given by Equation 6.14 gives the solution as

x2 x2
∫ x ln( x ) dx =
2
ln( x ) +
4
+C

Remember, the use of integration to find areas ‘under’ curves, and, as an

extension of this, areas ‘between two graphs’.
In general, it’s easy to find the area enclosed between two graphs by
using the formula,

Area =
∫ {(Upper graph) − ( Lower graph)} dx
a
(6.15)

Example 6.29
Review Figure 6.24. To find the bounded area between the curves as shown
in Figure 6.24, simply find the points of intersection of the two curves by
equating the y values as follows:

4 − x 2 = x 2 − 2x

x2 − x − 2 = 0

so x = −1 and x = 2.
Performing the integration as given by Equation 6.15 gives

2 2

Area =
∫ (4 − x ) − ( x
−1
2 2
− 2 x ) dx =
∫ (−2x
−1
2
+ 2 x + 4) dx = 9 square units.
164 Introduction to Calculus

8
This shows the graphs of
6
y = 4 – x2
y = x2 – 2x
4

–2 –1 0 1 2 3
x
–2

–4

Figure 6.24 Area bounded between two curves.

6.4 Applications
Example 6.30: Practical Optimization Problem
In some real practical situations, there may be some maximization or mini-
mization problems. Consider the following example of constructing an
open cylindrical tank using minimum material.
An open tank that has vertical sides and a circular base is to be constructed
from metal so as to use the minimum amount of material. If the capacity of
the tank is to be 8 m3, find the dimensions of the tank shown in Figure 6.25.

Solution: The volume of the tank gives πr2h = 8. The surface area of the
tank is given as S = πr2 + 2πrh. You need to find the value of r (and h),
which makes S a minimum.
From the first equation, h can be defined as

8
h=
πr 2

Base radius r, height h

h

Figure 6.25 Open tank with radius r and height h.

 6.4 Applications 165

Substituting this into the equation for S gives

16
S = πr 2 + .
r

dS
Now the problem is one of minimizing S with respect to r. Find = 0.
Differentiating S with respect to r gives dr

dS 16
= 2πr − 2 = 0
dr r

Solving for r gives

8
r3 =
π

which gives

1
 83
r =   = 1.37
 π

This then gives h as

8
h=
π (1.37)2

which gives h = 1.37 m.

To check this gives a minimum value of S, not a maximum. Using the
second derivative method gives

d 2S 32
= 2π − 16(−2)r −3 = 2π + 3
dx 2 r

When r = 1.37 m, this is a positive number = 18.7. So S is a minimum as

required.

Example 6.31: Heat Released during a Fire

An application of how integration can be used in fire engineering is the
development of an initial fire growth curve where Q (t ) is the heat release
rate and t is the time in seconds, as shown in Figure 6.26. This represents
the initial fire growth.
166 Introduction to Calculus

Q̇

˙ = 0.01 t2
Q

t
0 300

Figure 6.26 Heat release rate Q (t ) against time t.

Solution: The heat-released Q in the time interval from t = 0 to t = a seconds

is given by the formula

a
dQ
Q=
∫ dt dt
0
(6.16)

To calculate the heart released in the first 300 seconds of a fire with the
heat-release rate given as a t-squared function, Q (t ) = 0.01t 2, using the for-
mula given in Equation 6.16 gives

300 300
 0.01 t 3  (0.01)(300)3 270000
Q=
∫
0
0.01 t dt = 
2
 3 0
 =
3
=
3
= 90000 J = 90 KJ

Example 6.32: Applications in Reliability Theory

System or component failure can have a devastating impact on a human
level and also on a financial level, so it is important to be able to predict
when things are going to fail. The reliability of a component (or system) at
time t, say, R(t), is defined in terms of probability as R(t) = P(T > t), where
T is the lifetime of the component. This is interpreted as the component is
still functioning at time t. R(t) is called the reliability function.
If f(t) is the probability density function (pdf) of the failure function T,
then the probability of failure up to a time T ≤ t can be written as

P(T ≤ t ) =
∫ f (s) ds,
0
 Problems 167

Now, the reliability of a component means the probability that the sys-
tem has not failed in the interval [0, t]. So,

R(t ) = 1 − P(T ≤ t )

and in terms of calculations, the reliability R(t) becomes

t ∞

R(t ) = P(T > t ) = 1 −

∫
0
f (s) ds =
∫ f (s) ds
t

It can be seen that integrals can play an important role in calculating physi-
cal quantities that can be used to predict system behavior.
Another important application in the reliability of systems is that given
the failure probability density function f(t), then the expected time to failure
E(T) can be calculated using the formula

E (T ) =
∫ t f (t) dt
0

Now, since f(t) is some function of time and so is t, this integral will be a
product of two functions of time. For certain functions of f(t), this will need
to be done using the integration by parts method discussed in the earlier
sections.
The expected value is a measure of when something will typically fail.
This is an important measure as this allows for maintenance of systems to
take place in advance of this time, which will prevent a system from failing
and can also save lives and money.

Problems
dy
6.1 Find for each of the following:
dx
a. y = 4x3 − 5x2 + 7x − 11

b. y = 8 x ( x + 5)

c. y = x2 e3x

sin x
d. y =
x

e. y = (x3 + 11)5

6.2 Find the coordinates of the stationary points for the curve with equation

16
y = 8 − x2 −
x2
168 Introduction to Calculus

Determine the nature of the stationary points.

6.3 A glass window consists of a rectangle with sides of length 2r cm by

h cm and a semicircle of radius r cm as shown in Figure 6.27. The total
area of one surface of the glass is 500 cm2.

a. Show that the perimeter P of the window is given by

 π 500
P = 2+  r +
 2 r

b. Determine the value of r for which P has a stationary value and

hence determine its nature.

2r

Figure 6.27 Glass window made from a rectangle and a semicircle.

6.4 Evaluate each of the following:

a.
∫ (4 x 3
+ 9 x 2 − 10 x + 4) dx b.
∫ x (3 x + 5) dx

x5 + 1
c. ∫ x2
dx

d.
∫ 6x x 2 + 5 dx

6
2x + 1
e.
∫ (x − 4)(x + 2) dx
5
f.
∫ xe 4x
dx

g.
∫ e sin x dx
x
h.
∫x n
ln x dx

6.5 Given that the initial fire growth is governed by the heat release rate
Q (t ) = 0.3t 2 , determine the heat released by the fire during the first 10
seconds of burning.
 Problems 169

6.6 Given the failure probability density function f(t), the time t being mea-
sured in operator work years as follows:

 1
 (t − 1), 1≤ t ≤ 9
f (t ) =  32
 0, otherwise


Calculate the expected time to failure E(t).

http://taylorandfrancis.com
 7 Ordinary Linear
Differential
Equations

7.1 Background
In many physical situations, equations arise that involve differential coefficients
dy d 2 y dy
such as , , etc. These equations are called differential equations since
dx dx 2 dt
they contain differential coefficients. Differential equations arise naturally in the
modeling of real phenomena in engineering. The following example shows some
different areas in which they can occur.

Example 7.1
1. If a body, for example, a sprinkler droplet, falls freely under gravity
g, the distance traveled s is given by using Newton’s second law of
motion as

d 2s
=g (7.1)
dt 2

where s is the distance fallen in time t.

2. Consider the electrical RL circuit shown in Figure 7.1. An electrical

circuit with a resistor R and an inductor L in series, the current i flow-
ing in the circuit is given by using Kirchhoff’s voltage law as

di
L + Ri = E (7.2)
dt

where i is the current at time t.

171
172 Ordinary Linear Differential Equations

E(t)
AC
i(t) L

Figure 7.1 A simple RL electrical circuit.

3. When a mass oscillates on the end of a spring and is subject to a fric-

tional resistance proportional to its speed, the equation of motion may
be written as

d 2x dx
m 2
+r + sx = 0 (7.3)
dt dt

where x is the displacement from the equilibrium position at time t;

and m, r, and s are constants.

Note: In all of these cases the problem is to find the dependent variable in terms
of the independent one. For example, from Equations 7.1 to 7.3, respectively, find
s in terms of t, find i in terms of t, or find x in terms of t.

The formulation of a mathematical equation to represent a physical situation is

referred to as mathematical modeling. Differential equations being a big subject
area, it is important to define some of the important terminology associated with
them in the next section.

7.2 Types of Differential Equations

7.2.1 Introduction
An equation in which at least one term is a differential coefficient is called a
differential equation. Ordinary differential equations (ODEs) involve only one
independent variable x and a dependent variable y, and one or more differential
coefficients. Some further examples of these are given next.

Example 7.2

dy
x = 3y (7.4)
dx
 7.2 Types of Differential Equations 173

Example 7.3

d2y dy
+ 2 + y = sin x (7.5)
dx 2 dx

Differential equations represent dynamical relationships, that is, quantities that

change and so are found to occur in many scientific and engineering problems,
and are essential to the study of transient and nonsteady system behavior.

7.2.2 Order of a Differential Equation

The order of a differential equation is given by the highest derivative found in the
equation.
What is the order of the differential equations given in Example 7.2 and
Example 7.3? Example 7.2 is an example of a first-order differential equation
dy
because the highest derivative is . Example 7.3 is an example of a second-order
dx d2y
differential equation because the highest derivative is 2 .
dx
What is the order of the differential equations given in the following examples?

Example 7.4

d3y d2y dy
3
+ 8 y2 2 + 2 + y = 3 (7.6)
dx dx dx
d3y
The order is 3 because the highest derivative is the third derivative 3
term. dx

Example 7.5
3
 dy 
 dx  + y = x
2
(7.7)

dy
The order is 1 because the highest derivative is the first derivative term.
dx
7.2.3 Degree of a Differential Equation
The degree of a differential equation is the power to which the highest derivative
is raised.
In Example 7.5, Equation 7.7 has order 1 but the degree is 3, since the highest
dy
derivative is being raised to the power 3.
dx
7.2.4 Linearity
A differential equation is linear if it is linear in the dependent variable y and its
derivatives. The differential equations in Examples 7.2 and 7.3,

dy d2y dy
x = 3y and 2
+2 + y = sin x
dx dx dx

are both linear.

174 Ordinary Linear Differential Equations

But the differential equations given by Example 7.4 and 7.5,

3
d3y d2y dy  dy 
3
+ 8 y2 2 + 2 +y=3 and  dx  + y = x
2

dx dx dx

3
d2y  dy 
are both nonlinear because of the y 2 and   terms being present.
dx 2  dx 
Consider the differential equation given in the following example.

Example 7.6

dy dy
x3 +4 + y = x5 (7.8)
dx dx

This equation is still linear since it is linear in y and its derivatives the x
powers are not a problem, because x is the independent variable not the
dependent variable.

7.2.5 What Is Meant by Solving Differential Equations?

A differential equation represents a relationship between two variables. The same
relationship can often be expressed in a form that does not contain the differential
coefficient. The following example illustrates this idea.

Example 7.7

dy
If = 2x ⇒ y = x 2 + C
dx
These expressions are the same relationship in different forms.
Converting a differential equation into a direct equation between y and x is
called solving the differential equation.

Note: When solving a first-order differential equation, the solution will

contain only one arbitrary constant. A second-order differential equation
will produce a solution with two arbitrary constants, and so on.

7.3 First-Order Differential Equations

When dealing with first-order differential equations, the task is to see what type
of method is most appropriate to deal with the problem. The next sections con-
sider three methods that can be used to solve certain first-order differential equa-
tions, starting with the simplest case, which is using direct integration.
 7.3 First-Order Differential Equations 175

7.3.1 Simplest Situation
dy
If the differential equation can be arranged in the form = f ( x ), then this type
dx
of equation can be solved by direct integration and there is generally no compli-
cations involved. The following example shows this simplest case.

Example 7.8
Solve
dy
x = 6x3 + 7 (7.9)
dx

Solution: Dividing Equation 7.9 by x gives

dy 7
= 6x 2 + (7.10)
dx x

Integrating both sides gives

dy  7
∫ dx dx = ∫  6x 2
+  dx
x
(7.11)

y = 2 x 3 + 7 ln x + C (7.12)

Note: Equation 7.12 is called a general solution to the differential equation

because of the arbitrary constant C present in the solution.

To obtain the constant C, extra information is needed relating x and y.

Once the value of C is known, then the particular solution is said to be
obtained.

7.3.2 Separating Variables
dy
If the differential equation is not of the simple type, that is, = f ( x ), then it
dx
dy
may be the case that the differential equation is of the form = f ( x , y). Then the
dx
variable y on the right-hand side prevents solving by direct integration.
In this case, the function f(x,y) can be split into two separate functions as
follows:

• F(x), a function containing just x terms

• G(y), a function containing just y terms

176 Ordinary Linear Differential Equations

It may be the case that f(x,y) can be made into a separable form as follows:

dy
= f ( x , y) = F ( x ).G ( y) (7.13)
dx

or

dy F (x)
= f ( x , y) = (7.14)
dx G ( y)

If the differential equation can be written as either Equation 7.13 or 7.14, then
it is possible to separate the variables x and y, and then rearrange the equations to
integrate separately as shown in the following examples.

Example 7.9
Solve

dy 2x
= (7.15)
dx y + 1

This equation is of the form

dy F (x)
= f ( x , y) = (7.16)
dx G ( y)

Solution: Multiplying Equation 7.15 by the (y + 1) term gives

dy
( y + 1) = 2x (7.17)
dx

Equation 7.17 is now of the form in which the variables are separated.
Now integrating both sides of Equation 7.17 with respect to x gives

dy
∫ ( y + 1) dx dx = ∫ 2x dx (7.18)

dy
∫ ( y + 1) dx dx =
∫ 2x dx (7.19)

∫ ( y + 1) dy = ∫ 2x dx (7.20)

Equation 7.20 can now be integrated separately on each side to give

y2
+ y = x2 + C (7.21)
2
 7.3 First-Order Differential Equations 177

This then the general solution to the differential equation given by

Equation 7.15.

Example 7.10
Suppose we require the equation of a curve that satisfies the following dif-
ferential equation:

dy cos x
2 = (7.22)
dx y

and passes through the point (0, 2).

Solution: Separating the variables by multiply Equation 7.22 by y and inte-

grating both sides gives

dy
∫ 2y dx dx = ∫ cos x dx (7.23)

dy
∫ 2y dx dx =
∫ cos x dx (7.24)

∫ 2y dy = ∫ cos x dx (7.25)

y 2 = sin x + C (7.26)

Now at the point (0, 2) substituting in x = 0 and y = 2 into Equation 7.26

gives

4 sin 0 + C ⇒ C = 4

So the particular solution becomes

y 2 = sin x + 4 (7.27)

Example 7.11
Solve

dy
= (1 + x )(1 + y) (7.28)
dx

Solution: Dividing Equation 7.28 by (1 + y) separates the variables then

integrating both sides gives

1 dy
∫ (1 + y) dx dx = ∫ (1 + x) dx (7.29)
178 Ordinary Linear Differential Equations

1 dy
∫ (1 + y) dx dx =
∫ (1 + x) dx (7.30)

1
∫ (1 + y) dy = ∫ (1 + x) dx (7.31)

Integrating both sides of Equation 7.31 gives the solution as

x2
ln(1 + y) = x + +C (7.32)
2

This is the general solution to the differential Equation 7.28.

Example 7.12
Solve

dy y 2 + xy 2
= (7.33)
dx x 2 y − x 2

Separating the variables requires more work in this problem. First a com-
mon factor of y2 on the numerator and x2 on the denominator of Equation
7.33 can be factored out to give

dy y 2 (1 + x )
= (7.34)
dx x 2 ( y − 1)

( y − 1)
To separate the variables, multiply Equation 7.34 by on both sides
y2
and then integrating gives

( y − 1) dy (1 + x )
∫ y 2 dx
dx =
∫ x2
dx (7.35)

( y − 1) dy 1+ x
∫ y 2 dx
dx =
x2
dx
∫ (7.36)

Dividing out on both sides of Equation 7.36 by the denominators gives

 y −2   −2 x 
∫  y 2 − y  dy = ∫  x + 2  dx
x
(7.37)
 7.3 First-Order Differential Equations 179

1   1
∫  y − y −2
 dy = ∫  x −2
+  dx
x
(7.38)

Integrating both sides of Equation 7.38 gives

ln y + y −1 = − y −1 + ln x + C (7.39)

Tidying up both sides gives the solution of the differential Equation 7.33 as

1 1
ln y + = ln x − + C (7.40)
y x

Note: When trying to solve first-order differential equations, it may be

the case that the differential equation cannot be solved by separating the
variables. In this case another approach is needed to solve the differential
equation. The next section considers the method of using the integrating
factor technique.

7.3.3 Integrating Factor Technique

If the differential equations have either of the following structures:

dy
= f (x) (7.41)
dx

This type can be solved using direct integration. Or of the form,

dy
= F ( x ).G ( y) (7.42)
dx

This type can be solved using separating variables method.

However, if the differential equation is not of the form given by Equation 7.41
or 7.42, then it may be possible to see if the differential equation is of the follow-
ing form:

dy
+ P ( x ) y = Q( x ) (7.43)
dx

where P and Q are functions of x.

Equation 7.43 cannot generally be solved using the method of separating vari-
ables. Here the aim is to try to make the left-hand side of Equation 7.43 into a
complete differential coefficient from which we can integrate easily. That is to see
if Equation 7.43 can be written as
180 Ordinary Linear Differential Equations

d
(……) = Q
dx

To do this, Equation 7.43 has to be multiplied by a factor called an integrating

factor (I.F.), which turns out to be

I.F. = e ∫
P ( x ) dx
(7.44)

If Equation 7.42 is multiplied by the I.F., then it becomes

dy ∫ P ( x ) dx
+ Pye ∫ = Qe ∫
P ( x ) dx P ( x ) dx
e (7.45)
dx

Now the left-hand side of Equation 7.45 is the differential coefficient of

ye ∫ P ( x ) dx
, so Equation 7.45 can now be written as

d
dx
(
ye ∫
P ( x ) dx
= Qe ∫)P ( x ) dx
(7.46)

Integrating both sides of Equation 7.46 gives

ye ∫
∫
= Qe ∫
P ( x ) dx P ( x ) dx
dx (7.47)

The solution to the differential of the form given by Equation 7.43 can be writ-
ten in a form that is easier to remember as follows:

∫
y( I.F.) = Q( I.F.) dx (7.48)

where

I.F. = e ∫
P ( x ) dx
(7.49)

Example 7.13
Solve

dy
+ 5 y = e2 x (7.50)
dx

Solution: Comparing with the standard form equation,

dy
+ P ( x ) y = Q( x ) (7.51)
dx
 7.3 First-Order Differential Equations 181

implies P(x) = 5 and Q(x) = e2x.

I.F. = e ∫ = e∫
P ( x ) dx 5 dx
= e5 x

This gives the I.F. as

I.F. = e5 x (7.52)

Using Equation 7.47 gives the solution as

ye5 x =
∫e 2x 5x
e dx (7.53)

Simplifying the exponents gives

ye5 x =
∫e 7x
dx (7.54)

e7 x
ye5 x = +C (7.55)
7

Now dividing both sides of Equation 7.55 by e5x gives the final solution as

e2 x
y= + Ce −5 x (7.56)
7

Example 7.14
Solve

dy
( x + 1) + y = ( x + 1)2 (7.57)
dx

Solution: Compare with the standard form

dy
+ P ( x ) y = Q( x ) (7.58)
dx

Equation 7.57 is not in standard form and needs dividing by (x + 1)

throughout to give

dy 1
+ y = ( x + 1) (7.59)
dx x + 1
182 Ordinary Linear Differential Equations

1
This implies that P( x ) = and Q(x) = x + 1.
x +1

1
∫ dx
I.F. = e ∫
P ( x )dx
= e x +1 = e ln(x +1) = x + 1
∴ I.F. = x + 1 (7.60)

Using Equation 7.47 gives the solution as

y( x + 1) =
∫ (x + 1) dx
2
(7.61)

Integrating the right-hand side of Equation 7.61 gives

( x + 1)3
y( x + 1) = +C (7.62)
3

Dividing both sides of Equation 7.62 by (x + 1) gives the solution as

( x + 1)2 C
y= + (7.63)
3 x +1

7.4 Second-Order Differential Equations

In the previous section, the differential equations considered had in them the
highest derivative of order 1. Now more complicated differential equations are
considered where the highest derivative is of order 2. These are called second-
order differential equations. The general second-order linear differential equation
is of the following form:

d2y dy
a( x ) 2
+ b ( x ) + c( x ) y = f ( x ) (7.64)
dx dx

where a(x), b(x), c(x), and f(x) are generally functions of x.

Alternatively, a shorthand notation is used to represent the derivative functions as

dy d2y
y′ = and y′′ =
dx dx 2

Then Equation 7.64 can be written in shorthand notation as

a( x ) y′′ + b( x ) y′ + c( x ) y = f ( x ) (7.65)

If the right-hand side of Equation 7.65 is identically zero (i.e., f(x) = 0), then

a( x ) y′′ + b( x ) y′ + c( x ) y = 0 (7.66)
 7.4 Second-Order Differential Equations 183

This is called the homogeneous equation as it only contains terms in y and its
derivatives.
Linear constant coefficient second-order differential equations are a spe-
cial case of Equation 7.65 where a(x), b(x) and c(x) are all constants (i.e., only
numbers). So, the differential equations to be solved in this section are of the
following type:

ay′′ + by′ + cy = f ( x ) (7.67)

where a, b, and c are constants.

The general solution of Equation 7.67 is made up of two parts:

General solution = Complementary function + Particular integral

y(t ) = yCF + yPI (7.68)

The task is to find these two functions yCF and yPI separately and then add
them together to get the final solution. The next section considers how to com-
pute the complementary function yCF.

7.4.1 Complementary Function (CF)

The complementary function is obtained by solving Equation 7.67 with f(x) = 0,
that is, the homogenous equation:

ay′′ + by′ + cy = 0 (7.69)

Note: A second-order linear differential equation always has two and only two
linearly independent complementary solutions.

7.4.1.1 General Solution for the Complementary Function

If y1(x) and y2(x) are two linearly independent solutions to the homogeneous
equations:

ay′′ + by′ + cy = 0 (7.70)

Then y1(x) and y2(x) are both solutions to Equation 7.70, that is,

ay1′′+ by1′ + cy1 = 0 (7.71)

ay2′′ + by2′ + cy2 = 0 (7.72)

Then the linear combination of y1(x) and y2(x)

yCF (t ) = α y1 ( x ) + β y2 ( x ) (Principle of superposition) (7.73)

184 Ordinary Linear Differential Equations

where α and β are constants, is also a solution to Equation 7.70.

Note: Equation 7.73 is called the general solution for the complementary
function.

7.4.1.2 How to Find the Complementary Function

The complementary function is given as the solution to the homogeneous Equation
7.70, that is, with f(x) = 0:

ay′′ + by′ + cy = 0 (7.74)

It is obtained by assuming that Equation 7.70 has solutions of the exponential

form

y = e mx (7.75)

where m is a constant (real or complex) that needs to be determined.

For y = emx to be a solution to Equation 7.70, it must satisfy Equation 7.70.
Starting with y and calculating its first and second derivatives y′ and y″ gives

y′ = me mx , y′′ = m 2e mx (7.76)

Substituting y, y′, and y″ back into Equation 7.70 gives

am 2e mx + bme mx + ce mx = 0 (7.77)

e mx (am 2 + bm + c) = 0 (7.78)

Since emx does not equal zero implies for Equation 7.78 to be equal to zero, then
the quadratic equation must be equal to zero, that is,

am 2 + bm + c = 0 (7.79)

Equation 7.79 is given a special name called the characteristic or auxiliary

equation.
Next, an example is shown of how to calculate the complementary function
yCF.

Example 7.15
Solve the following second-order linear constant coefficient differential
equation:

y′′ + 3 y′ + 2 y = 0 (7.80)
 7.4 Second-Order Differential Equations 185

Solution: Assume solution of the form y = emx, then y′ = memx and y″ =

m2 emx.
Substituting for y, y′, and y″ back into Equation 7.80 gives

m 2e mx + 3me mx + 2e mx = 0 (7.81)

e mx (m 2 + 3m + 2) = 0 (7.82)

Now, the term emx ≠ 0.

This means that the quadratic equation must be equal to zero, that is,

m 2 + 3m + 2 = 0 (7.83)

This factorizes as (m + 2)(m + 1) = 0, which give the values for m as

m = −2, m = −1. Substituting these values of m back into y = emx, gives the
two linearly independent solutions as

y1 = e −2 x and y2 = e − x (7.84)

The general complementary solution to Equation 7.80 is now given as a

general combination of y1 and y2, that is,

y = α e −2 x + β e − x (7.85)

where α and β are arbitrary constants to be determined from initial

conditions.

7.4.2 Types of Solutions
It has already been shown that by assuming an exponential solution of the form
y = emx to the constant coefficient equation ay″ + by′ + cy = 0 gives rise to the
characteristic or auxiliary equation am2 + bm + c = 0, which is solved for m
in order to fully determine the general complementary solution to the equation
ay″ + by′ + cy = 0.
The general solution to the characteristic equation am2 + bm + c = 0 is given by

− b ± b 2 − 4 ac
m= (7.86)
2a

and the roots depend on the sign of the discriminant Δ = b2 − 4ac.

There are three cases to consider:

7.4.2.1 Case 1: Real and Distinct Roots m1 and m2

If b2 − 4ac > 0, then the solutions of the characteristic Equation 7.79 are real and
different as follows:

m = m1 and m = m2
186 Ordinary Linear Differential Equations

where

− b + b 2 − 4 ac − b − b 2 − 4 ac
m1 = and m1 =
2a 2a

The general solution to the complementary function is given as

yCF = α e m1x + β e m2 x (7.87)

Example 7.16
Solve

y′′ + 5 y′ + 6 y = 0 (7.88)

Solution: Assume the answer in the form y = emx.

The characteristic equation is

m 2 + 5m + 6 = 0

⇒ (m + 2)(m + 3) = 0

m = −2 and m = −3

yCF = α e −2 x + β e −3 x

7.4.2.2 Case 2: Real and Repeated Roots

If b2 − 4ac = 0, then the solutions of the characteristic Equation 7.79 are repeated
roots given as

b
m1 = −
2a

There is one linearly independent solution, which is given by y1 = e m1 x. To con-

struct the general solution to the complementary function, there needs to be a
second linearly independent solution.
It can be shown that if y1 = e m1 x is one solution to Equation 7.74, then the second
independent solution is y2 = xe m1 x.

Note: If one solution was q1 = e m1t , then the other would be q2 = te m1t.

The general solution for the complementary function is now given as

yCF = α e m1x + β xe m1x (7.89)

 7.4 Second-Order Differential Equations 187

Example 7.17
Solve
y′′ + 6 y′ + 9 y = 0 (7.90)

Solution: Assume the answer is of the form y = emx.

The characteristic equation is m 2 + 6m + 9 = 0, which implies
(m + 3) (m + 3) = 0, m = −3 is a repeated root. So one solution is y1 = e –3x
and the second solution is y 2 = xe−3x.
The general solution for the complementary function is now

yCF = α e −3 x + β xe −3 x

which can also be written as yCF = (α + β x)e−3x by taking e−3x as a common

factor.

7.4.2.3 Case 3: Complex Conjugate Roots

If b2 − 4ac < 0, then the solutions to the characteristic equation are m = m1 and
m = m2, where m1 and m2 are complex conjugate solutions.

m1 = p + jq and m2 = p − jq

b 4 ac − b 2
with p = − and q = .
2a 2a
Then the solution to the complementary function is

yCF = α e( p+ jq ) x + β e( p− jq ) x (7.91)

Taking out the epx as a common factor gives

(
yCF = e px α e jqx + β e − jqx ) (7.92)

Since this contains imaginary exponentials, these can be replaced by using

Euler’s identity:

e jθ = cos θ + j sin θ

This then gives the following:

e jqx = cos qx + j sin qx and e − jqx = cos qx − j sin qx

Therefore, the general solution for the complementary function can now be
written as

yCF = e px ( A cos qx + B sin qx ) (7.93)

where A and B are constants.

188 Ordinary Linear Differential Equations

Example 7.18
Solve

y′′ + 4 y′ + 9 y = 0 (7.94)

Solution: Assume the answer in the form y = emx.

The characteristic equation is m2 + 4m + 9 = 0. This can be solved using
the quadratic formula to give the solutions as a complex conjugate pair:

m = −2 ± j 5

So, here p = −2 and q = 5, giving the general solution to the comple-

mentary function as

yCF = e −2 x ( A cos 5 x + B sin 5 x ) (7.95)

7.4.3 Particular Integral (P.I.)

So far, consideration has been given to second-order differential equations of the
form

ay′′ + by′ + cy = 0 (7.96)

that is, where f(x) = 0 has been made. Now considering the full differential
equation:

ay′′ + by′ + cy = f ( x ) (7.97)

Substituting in the complementary function would make the left-hand side of

Equation 7.97 equal to zero and not f(x), so there must be a further term in the
solution that will make the left-hand side equal to f(x) and not equal to zero.
This extra function is called the particular integral (P.I.). If the solutions to the
complementary function are real and different solutions, let the particular integral
be called X(x). Then the complete solution y(x) will be of the form

y = α e m1x + β xe m1x + X ( x ) (7.98)

X(x) is an extra function known as the particular integral (yet to be determined).

7.4.3.1 How to Find the Particular Integral

To find the particular integral, one assumes the most general form of the function
on the right-hand side of Equation 7.97.

ay′′ + by′ + cy = f ( x )

that is, general form of the function f(x),

 7.4 Second-Order Differential Equations 189

Table 7.1 Particular Integral yPI for Different Functions f(x)

f(x) Try yPI as
k (constant) yPI = C
kx yPI = Cx + D
kx2 yPI = Cx2 + Dx + E
k sin wx or k cos wx yPI = C sin wx + D cos wx
αekx yPI = Cekx

Table 7.1 shows which functions for the particular integral to try for the differ-
ent functions of f(x).

Note: The constants C, D, E, and so on are determined by substituting yPI and

its derivatives back into Equation 7.97 and equating coefficients on both sides.

Example 7.19
If f(x) = sin 4x, then try the particular integral as yPI = C sin 4x + D cos 4x.

If f(x) = f(x) = x + 2ex, then try the particular integral as yPI = Cx + D + Eex.

Finally, putting all the theory together for the complementary function
and the particular integral, the next example shows how to solve a general
second-order differential equation.

Example 7.20
Solve
y′′ − 5 y′ + 6 y = 2 sin 4 x (7.99)

subject to initial conditions, x = 0, y =

27
, and y′ = 117 .
25 50
Solution: The general solution is given by y(t) = yCF + yPI.

First Finding the Complementary Function (  yCF)

Making the right-hand side of Equation 7.99 equal to zero, f(x) = 0, gives

y′′ − 5 y′ + 6 y = 0 (7.100)

Try a solution of the form y = emx.

The characteristic equation is

m 2 − 5m + 6 = 0

⇒ (m − 2)(m − 3) = 0

m = 2 and m = 3 (this is case 1 for different types of solutions) implies the solu-
tion is

yCF = α e 2 x + β e3 x
190 Ordinary Linear Differential Equations

For the Particular Integral (  yPI)

Look at the right-hand side of Equation 7.99. The function is f(x) = 2 sin 4x.
Try

yPI = A cos 4 x + B sin 4 x (7.101)

Differentiating Equation 7.101 twice gives

′ = −4 A sin 4 x + 4 B cos 4 x
yPI

′′ = −16 A cos 4 x − 16 B sin 4 x

yPI

Now by substituting yPI, yPI′ and yPI″ back into Equation 7.99 gives

−16 A cos 4 x − 16 B sin 4 x − 5(−4 A sin 4 x + 4 B sin 4 x ) + 6( A cos 4 x + B sin 4 x ) =

2 sin 4 x

Collecting all the cos 4x and sin 4x terms together on the left-hand side gives

(−16 A + 6 A − 20 B) cos 4 x + (−16 B + 20 A + 6 B)sin 4 x = 2 sin 4 x + 0 cos 4x

(−10 A − 20 B) cos 4 x + (−10 B + 20 A)sin 4 x = 2 sin 4 x + 0 cos 4 x

Here, the right-hand side has a 0 multiplying the cos 4x, so like terms can be
compared on both sides.
By equating the coefficients of the cosine terms on both sides gives

−10 A − 20 B = 0 ⇒ − 10 A = 20 B

⇒ A = −2 B (7.102)

By equating coefficients of the sine terms on both sides gives,

−10 B + 20 A = 2 ⇒ − 10 B = 2 − 20 A

⇒ − 5 B = 1 − 10 A (7.103)

Solving Equations 7.102 and 7.103 simultaneously for A and B gives

2 1
A= and B=− (7.104)
25 25

Therefore the particular integral can be obtained by using these values of A

and B back into Equation 7.101 to give

2 1 1
yPI = cos 4 x − sin 4 x = (2 cos 4 x − sin 4 x ) (7.105)
25 25 25
 7.5 Applications 191

The general solution is given by

y(t ) = yCF + yPI

1
y(t ) = α e 2 x + β e3 x + (2 cos 4 x − sin 4 x ) (7.106)
25

Now to finally finish off the complete solution, the constants α and β need to be
27
found using the given initial conditions. The initial conditions are x = 0, y = ,
25
117
and y′ = 50
.
27
Using x = 0 and y = in Equation 7.106 gives
25

α +β =1 (7.107)

Differentiating Equation 7.106 gives

1
y′(t ) = 2α e 2 x + 3β e3 x − (8 cos 4 x + 4 sin 4 x ) (7.108)
25

117
Using x = 0 and y ′ = in Equation 7.108 gives
50

2α + 3β = 2.5 (7.109)

Solving Equations 7.107 and 7.109 simultaneously gives

α = 0.5 and β = 0.5 (7.110)

The complete solution to the original differential Equation 7.99 is given as

1
y(t ) = 0.5e 2 x + 0.5e3 x + (2 cos 4 x − sin 4 x ) (7.111)
25

The next section is of applications that show how differential equations arise
in real-world engineering problems and how using the different methods studied
so far can help solve these real problems.

7.5 Applications
Example 7.21: Calculating the Time for the Smoke Layer to Develop
In this problem, as the fire progresses, the dangerous smoke layer develops
as shown in Figure 7.2. The time for the smoke layer to reach a certain
height can be calculated by solving a differential equation. In Figure 7.2,
room height is H and lower layer height is z. The fire is treated as a point

source of heat Q. The mass flow rate of the lower layer to the upper layer is
given by m p, the plume mass flow rate. The plume is considered only as a
means of transporting mass from the lower layer to the upper layer.
192 Ordinary Linear Differential Equations

Layer of hot gas and smoke

H
z

Fuel array

Figure 7.2 Two-zone model for the smoke layer.

Applying the conservation of mass and energy to the lower layer gives
a simplified differential equation for smoke filling (in dimensionless form)
as follows:
1 5
+ 0.21(Q *) 3 y 3 = 0
dy
(7.112)
dτ

given y = 1 when τ = 0.
z
The dimensionless height given by y = varies from 0 to 1 and gives the
H
fraction of room height below the smoke layer. The dimensionless time is
H2 g
given by τ = t , where t is time in seconds, g equals 9.81 ms–2, H is
S H
the room height, and S the floor area.
Equation 7.112 is an example of a differential equation that can be solved
using the method of separating variables as follows.
1 5
Subtracting 0.21(Q *) 3 y 3 from both sides of Equation 7.112 gives,

1 5
= − 0.21(Q *) 3 y 3
dy
(7.113)
dτ
5
Dividing by y 3 and integrating both sides of Equation 7.113 with respect
to dτ gives

5 1
dy = − 0.21(Q *) 3 d τ
∫ ∫
−
y 3 (7.114)

2 1
y = − 0.21(Q *) 3 τ + C
3 −3
− (7.115)
2

using the initial conditions y = 1 when τ = 0 gives C = − 3 .

2

2 1
y = − 0.21(Q *) 3 τ −
3 −3 3
− (7.116)
2 2
 7.5 Applications 193

2 1
= 0.14 (Q *) 3 τ + 1
−
y 3 (7.117)

So now we have an equation relating τ and y as follows:

1 2
0.14 (Q *) 3 τ = y 3 − 1
−
(7.118)

This Equation 7.118 can now be used to calculate the time t taken for a
room to fill with smoke to any height.
If a pool of kerosene is ignited releasing 186 kW in a room with floor
area 5.62 m by 5.62 m and a height of 5.95 m, the time until the smoke layer
has filled half the room can be calculated as follows: First, calculate Q*
using the following formula:

Q 186
Q * = 5
= 5
= 0.002 (7.119)
1100 H 2 1100(5.95) 2

For half the room, y = 0.5, and so Equation 7.118 now gives τ as

1 2
−
0.14(0.002) 3 τ = (0.5) 3 −1

Rearranging this to find τ yields τ = 33.3 seconds.

Now since

H2 g
τ =t = 1.44t
S H

this implies that the time is t ≈ 23 seconds. It therefore takes less than half
a minute for a 186 kW fire to fill half the room with smoke.

Note: A 186 kW fire is approximately a wastepaper bin sized fire.

Example 7.22: Probabilistic Modeling Using Continuous

Markov Processes
Using a probabilistic model of system behavior for a continuous time
Markov process, the resulting model can be described by a set of differ-
ential equations known as the Kolmogorov forward equations. Consider a
simple two-state problem in which a system is either working (R) or not
working (F), as shown in Figure 7.3.
For a two-state dynamical Markov process represented by Figure 7.3,
λ is the failure rate and μ is the repair rate, which are both constants.
194 Ordinary Linear Differential Equations

R F

Figure 7.3 Two-state dynamical process. The R state represents the system work-
ing and the F state represents not working.

The rate of change of the probability of being in state R is decreased by λ

and increased by μ like a probability flow and are given by the Kolmogorov
forward equations:

dR
= −λ R + µF (7.120)
dt

where λR is the probability of the system operating at time t and not failing
in the time step dt, and μF is the probability of the system being in the failed
state at time t and being repaired in the time step dt.
Similarly, for the rate of change of the system being in state F is given by

dF
= −µF + λ R (7.121)
dt

Equations 7.120 and 7.121 are first-order differential equations that need
to be solved to find the probabilities of being in each state.

Solution: Starting with the equation for the R state (Equation 7.120) and
having the initial condition that R(0) = R0 (i.e., the initial probability of the
system working)

dR
= −λ R + µF (7.122)
dt

Replacing for F = 1 −R, into Equation 7.121 gives

dR
= − λ R + µ (1 − R) = − λ R + µ − µ R (7.123)
dt

This tidies up as

dR
= µ − (λ + µ ) R (7.124)
dt

which can be written as

dR
+ (λ + µ ) R = µ with initial condition R(0) = R0 (7.125)
dt
 7.5 Applications 195

Equation 7.125 is an example of a differential equation that can be solved

using the integration factor method.
Comparing Equation 7.125 with the standard form equation

dy
+ P(t ) y = Q(t ) (7.126)
dt

gives P(t) = (λ + μ) and Q(t) = μ.

The integrating factor (I.F.) then becomes

I.F. = e ∫
( λ + µ ) dt
= e( λ + µ )t (7.127)

The solution is given by Equation 7.47 as

R(t )e( λ + µ )t =
∫ µe ( λ + µ )t
dt (7.128)

µ
R(t )e( λ + µ )t = e( λ + µ )t + C (7.129)
(λ + µ )

µ
R(t ) = + Ce − ( λ + µ )t (7.130)
(λ + µ )

Using the initial condition t = 0, R = R0 gives

 µ 
C =  R0 −
 (λ + µ ) 

So finally, the solution for R(t) is

µ  µ  − ( λ + µ )t
R(t ) = +  R0 − e (7.131)
(λ + µ )  (λ + µ ) 

A similar relationship can be found for F(t) using F(t) = 1 – R(t) as

λ  λ  − ( λ + µ )t
F (t ) = + F0 − e (7.132)
(λ + µ )  (λ + µ ) 

These solutions then give the probabilities for the system to be in a work-
ing or not working state.

Example 7.23: Finding the Current in a RL Circuit

Consider the practical electrical RL circuit shown in Figure 7.4, The prob-
lem is to determine the current i(t) in the circuit.
196 Ordinary Linear Differential Equations

E0 i(t) L

Figure 7.4 An electrical RL series circuit.

Using Kirchhoff’s voltage law gives the differential equation for the cir-
cuit as

di
L + Ri = E0 (7.133)
dt

subject to the initial conditions i = 0 when t = 0.

Notes:
• The values of L the inductance, R the resistance, and E 0 the voltage
source are constants.
• Equation 7.133 is a type of differential equation in which there is a
choice of methods that can be used to solve it, either by separating
variables or the integrating factor method. The solution shows the
method of separating variables.

Solution: Starting with Equation 7.133 and subtracting the term Ri from
both sides yields

di
L = E0 − Ri (7.134)
dt

Dividing both sides of Equation 7.134 by (E 0 − Ri) gives

L di
=1 (7.135)
( E0 − Ri) dt

Integrating both sides of Equation 7.135 with respect to dt gives

 L 
∫  E ∫
 dt = 1 dt
0 − Ri 
(7.136)
 Problems 197

Integrating both sides of Equation 7.136 gives

L
− ln( E0 − Ri) = t + C (7.137)
R

Now using the initial conditions i = 0 when t = 0 gives C as

L
C=− ln E0 (7.138)
R

Substituting this value of C back into Equation 7.137 and rearranging

gives

 E0  R
ln  = t (7.139)
 E0 − Ri  L

Now to find i(t), Equation 7.139 is raised to the power e giving

R
E0 t
= eL (7.140)
E0 − Ri

Inverting both sides of Equation 7.140 gives

R
E0 − Ri − t
=e L (7.141)
E0

Rearranging this to make i(t) the subject gives the final solution for the
current in the circuit as

i(t ) =
E0
R
( − t
1− e L
R
) (7.142)

Problems
Solve the following differential equations using the appropriate method.

dy x
7.1 = with y(0) = 1
dx y

dy
7.2 = ( y − 3)( x + 5)
dx

dy 1
7.3 =
dx xy + x
198 Ordinary Linear Differential Equations

dy
7.4 x − y = x3
dx

dy
7.5 (1 − x 2 ) − 2 xy = x 4 with y(0) = 1
dx

d 2 y dy
7.6 − − 2y = x + 2
dx 2 dx

d2y dy
7.7 2
− 10 + 25 y = 10
dx dx

d2y dy
7.8 2
+4 + 5 y = 13e3 x with y(0) = 0 and y′(0) = 1.5
dx dx

dT
7.9 = 1 + 5T with T(0) = 1
dt

7.10 Consider the following RC circuit in Figure 7.5. Kirchhoff’s voltage

law for the circuit gives the following equation.

1
Ri +
C ∫ i dt = E 0


where R, C, and E 0 are all constants. By differentiating the above
equation, derive a differential equation for the current i(t) in the circuit
t
−
and hence show that i(t ) = ke RC , where k is a constant.

E0 i(t) C

Figure 7.5 An electrical RC series circuit.

 8 Laplace Transforms

The Laplace Transform is an integral transform named after its founder Pierre-
Simon Laplace. It takes a function of continuous time t (t > 0) to a function of a
complex variable s (frequency). As discussed earlier in Chapter 7, when modeling
real world problems, the formulation of differential equations naturally arises in
many different fields of engineering.

8.1 Why Do We Need the Laplace Transform?

The Laplace transform is a very important tool in engineering disciplines as it
enables the following:

1. It helps to solve linear differential equations with given initial conditions,

for systems that can be described by the following types of equations:

d2y dy
a 2
+ b + cy = E (8.1)
dx dx

2. The method is also particularly useful if the inputs to the differential

equations, that is, the E term in Equation 8.1 are discontinuous inputs like
the unit step function.

3. Incorporates the initial conditions at the start of the solution to the

problem.

4. In systems engineering, the system is broken down into components

as blocks. Each block can be represented in the s-domain and then
manipulated.

199
200 Laplace Transforms

8.2 Derivation from a Power Series

Most mathematics textbooks will start with the formula for the Laplace transform
without any reference to how it comes about mathematically. Here, it is more
appropriate to first consider how the Laplace transform is derived.
Starting with a discrete power series, this can be written as follows:

∑a x
0
n
n
= a0 + a1x + a2 x 2 +…

And for some an this series can be written in closed form as, say, A(x):

∑a x 0
n
n
= A( x )

Using a slightly different notation for the coefficients an = a(n) this becomes

∑ a(n) x
0
n
= A( x ) (8.2)

So, different values of a(n) can produce a different closed form sum A(x). Some
examples of using specific a(n)’s are given next.

Example 8.1
If a(n) = 1, then the series just becomes

∑x
0
n
= 1 + x + x 2 +…

This is just the geometric series with first term a = 1 and common ratio
r = x.
This has a sum to infinity given by

a
S∞ =
1− r

So, the series in closed form becomes

1
A( x ) =
1− x

provided ∣x∣ < 1

 8.3 Introduction and Standard Transforms 201

Example 8.2
1
If a(n) = , then the series is
n!
∞

∑ 0
xn
n!
x
= 1+ +
x2
1! 2!
+…

This is just the series for the exponential function of x. Therefore,

A(x) = ex in closed form.
So far, the power series has been one using a discrete summation. Now
considering the continuous analog of the above case one obtains the cor-
responding Laplace transform as follows.
∞ ∞

Replace the summation

0
∑
with the integral for continuous time.
0
∫
Replace the discrete integers n = 0, 1, 2, 3,… by continuous time t such that
t goes between 0 < t < ∞ for all values of t. Therefore, Equation 8.2 becomes

∫ a(t)x
0
t
dt = A( x ) (8.3)

Equation 8.3 could be left in this form, but for integration purposes it is
not good to have x as the base. It is usually better to have e as the function.
This can be done by letting x = eln(x), so xt = (eln(t))t
Now for this integral to converge, as t → ∞, x has be less than 1 and x > 0
to avoid any imaginary numbers appearing. So, 0 < x < 1 implies that ln(x) <
0 in this range for x.
Let –s = ln(x) and replace a(t) = f(t) in Equation 8.3:

∫ f (t)e
0
− st
dt = F (s) (8.4)

This is called the Laplace transform of the function f(t). This is just the
continuous analog of the summation of a power series.

8.3 Introduction and Standard Transforms

If f(t) is a piecewise continuous function for t ≥ 0, then the Laplace trans-
form of f(t) is defined as

∞
ℒ  f (t )  =
∫
0
f (t )e − st dt = F (s) (8.5)

Note: s = σ + jw (that is in general a complex variable).

202 Laplace Transforms

The Laplace transform pair can be denoted as follows:

L
f (t) F (s)

L–1
t-domain s-domain

The operator ℒ is used to represent the Laplace transform and the operator
ℒ –1 is used to represent the inverse Laplace transform.

8.3.1 Schematic Representation of Laplace Transforms

The Laplace transform method for solving differential equations can be more
easily understood using the schematic diagram in Figure 8.1 showing the process
being applied. From the diagram, it can be seen that the differential equation to
be solved is first Laplace transformed. Second, inputting the initial conditions
then results in an algebraic expression for the variable in the s-domain. Since this
part is all algebraic, it is fairly easy to manipulate to determine an expression for
Y(s). Finally, use the inverse Laplace transforms to generate the solution back in
the time domain, y(t), and hence, obtaining the solution to the original problem.

Laplace transform process

Ordinary
differential Algebraic Inverse
Laplace
laplace Solution
equations transform equation transform
with initial
conditions Y (s) y (t)

t-Domain s-Domain t-Domain

Figure 8.1 Schematic representation of the Laplace transform process.

8.3.2 Standard Transforms
In the aforementioned process, the differential equation needs to be Laplace
transformed and so this requires Laplace transforming different functions of time
dy d2y
as well as derivative terms like and 2 etc.
dt dt
Most of the Laplace transforms and subsequently the inverse Laplace trans-
forms are generally taken from a standard table of results. Therefore, it is use-
ful to first show some of the more common Laplace Transforms of functions in
a table format and then to see how they are derived from first principles using
the formula definition. Table 8.1 gives some functions and their corresponding
Laplace transforms. All the functions in the table can be proved using the basic
definition given by Equation 8.5 and some further manipulation. Next, a few of
the results are proved to show how they are derived using the formula definition.
 8.3 Introduction and Standard Transforms 203

Table 8.1 Standard Laplace Transforms of

Some Common Functions
f(t) F(s)
1 1
s
t 1
s2
eat 1
s−a
sin (at) a
s2 + a2
cos (at) s
s2 + a2
tn n!
s n +1
tn e-at n!
(s + a)n +1

Example 8.3
Laplace transform f(t) = 1.
∞

If f(t) = 1, then using formula F (s) =

∫ f (t)e
0
− st
dt gives

∞ ∞
 e − st   1 1
∫
= 1e
0
− st
dt = 
 − s 0
 = (0) −   =
−s s
1
F (s ) =
s

Example 8.4
Laplace transform f(t) = t.
∞

If f(t) = t, then F (s) =

∫t e
0
− st
dt . Here use integration by parts (Chapter 6,

Section 6.3.3), that is, using the formula

dv du
∫ u. dt dt = u.v − ∫ v. dt dt
dv du e − st
Let u = t = e − st gives =1 and v=
dt dt −s
This gives

∞ ∞ ∞ ∞
 e − st  e − st e − st  e − st  1
F (s ) =  t
 − s 0
 − ∫
0
−s
dt = 0 +
∫
0
s
dt =  2  = 2
 − s 0 s
1
F (s ) = 2
s
204 Laplace Transforms

Example 8.5
Laplace transform f(t) = eat.
∞

If f(t) = e , then F (s) =

at
∫e e
0
at − st
dt

∞ ∞
 e − t( s − a )  1
F (s ) =
∫e
0
− t( s − a )
dt = 
 −(s − a)  0
 = 0−
− (s − a)
1
F (s ) =
(s − a)

Note: Other Laplace transforms such as sin at and cos at can be done with-
out the need of integration by parts in a much more convenient manner
using Euler’s formula and complex numbers.

Euler’s formula is ejθ = cosθ + j sinθ, so it implies that ejat = cos at + j sin at.
Since this formula contains a real part cos at and imaginary part sin at, then
if one Laplace transforms ejat the real part will be the Laplace transform of
cos at while the imaginary part will be the Laplace transform of sin at. This
is shown in Example 8.6.

Example 8.6
∞

f (t) = e , then F (s) =

jat
∫e
0
jat − st
e dt

∫e (
− s − ja )t
F (s ) = dt
0
∞
 e − ( s− ja )t 
Therefore, F (s ) =  
 −(s − ja)  0
  1 
F (s ) =  0 −  − 
  s − ja  
1
F (s ) = ,
s − ja

Now this needs to be written as a real and imaginary part, that is, a + jb.
One can multiply the top and bottom by the complex conjugate of the bot-
tom (s – ja), that is, by (s + ja), to give

(s + ja) s + ja s a
F (s ) = = 2 = 2 +j 2
(s − ja)(s + ja) s + a 2
s +a 2
s + a2
 8.3 Introduction and Standard Transforms 205

So this gives

ℒ [ e jat ] = ℒ  cos at + j sin at  =

s a
+j 2
s2 + a2 s + a2

from which it can be seen that

s
ℒ  cos at  =
s + a2
2

and

a
ℒ sin at  =
s2 + a2

as required.

8.3.3 Linearity of Laplace Transforms

From the definition of integration, it follows that if we have two functions f(t) and
g(t), and these are both of exponential order, then

ℒ  af (t ) + bg(t )  = aℒ  f (t )  + bℒ  g(t )  = aF (s) + bG (s) (8.6)

This follows from the property that the integral of the sum of two functions is
equal to the sum of the two separate integrals. This property of linearity allows
the Laplace transforms of sums of functions to be found easily. The next example
shows how this is can be applied.

Example 8.7
Given f(t) = t and g(t) = e3t, then using Equation 8.6 gives

ℒ [ 5t − 11e3t ] = 5ℒ[t ] − 11ℒ [ e3t ] =

5 11
−
s2 s − 3

and, conversely, it follows for the inverse Laplace transforms:

5 11  −1  1  −1  1 
5t − 11e3t = ℒ−1  2 −  = 5ℒ  2  − 11ℒ  s − 3 
 s s − 3  s
   

This property will be very useful when solving differential equations

later on.

8.3.4 Basic Relations
There are some basic relations that can be used when solving differential equation
problems of which the most important ones are usually the Laplace transforms
of the first and second derivative of functions. These are properties 6 and 7 in
the Table 8.2 where a is an arbitrary constant. These relationships can be proved
206 Laplace Transforms

Table 8.2 Basic Properties of Laplace Transforms

Operation Time Domain s-Domain
1. Time shifting f(t–a) e–asF(s)

2. Time scaling f(at) 1  s

F 
a  a
3. Multiplying by an exponential in t-domain eatf(t) F(s–a)
d
4. Multiplying by (t) tf (t) −  F (s) 
ds
∞
5. Dividing by (t) 1
t
f (t )
∫ F (s) ds
s

6. First derivative f ′(t) sF(s) – f(0)

7. Second-order derivative f ″(t) s2F − sf(0) – f ′(0)
t

∫ f (t) dt
1
8. Integration F (s )
s
0

using the basic definition given by Equation 8.5. The following examples show
how properties 1 and 6 of Table 8.2 are derived.

Example 8.8

Show that the ℒ[ f (t − a)] = e − as F (s) (Property 1).

∞

Using the formula given by Equation 8.5, ℒ  f (t )  =

∫ f (t)e
0
− st
dt,

ℒ  f (t − a)  =
∫ f (t − a)e
0
− st
dt (8.7)

by replacing f(t) with f(t − a) and making a change of variable by letting

u = t − a. This gives du = dt and a is a constant and t = u + a. The limits
of integration remain the same and making the above substitutions into
Equation 8.7 gives

∞ ∞

f (u)e ( ) du =
∫ ∫ f (u)e
− s u+ a − su − as
ℒ  f (t − a)  = e du
0 0

The e−as term can be taken out in front of the integral since it does not
depend on u.
∞

ℒ  f (t − a)  = e − as
∫ f (u)e
0
− su
du

But
∫ f (u)e
0
− su
du = F (s) , so the following result is obtained:

ℒ  f (t − a)  = e − as F (s)
 8.4 Inverse Transforms 207

Example 8.9
Show that the ℒ[ f ′(t )] = sF (s) − f (0) (Property 6).
∞

Using the definition gives ℒ  f ′(t )  =

∫ f ′(t)e
0
− st
dt. Now this integral has

to done by using integration by parts. Letting

dv
u = e − st and = f ′(t )
dt

gives

du
= − se − st and v = f (t )
dt

Therefore,

∞ ∞

∫ f (t)(−se ∫ f (t)e
∞
ℒ  f ′(t )  = e − st f (t )  0 − − st
) dt =  0 − f (0)  + s − st
dt
0 0

= − f (0) + s F (s )

giving the result that

ℒ  f ′(t )  = s F (s) − f (0)

Note: This result together with the second derivative transform (property 7)
are used extensively when solving differential equations.

8.4 Inverse Transforms
In the final part of the process of solving differential equations (see Figure 8.1),
one needs to obtain the original function back in the time domain. This will
require once again using the standard table of transforms and inverses. Usually,
the expression that is required to be inverse Laplace transformed cannot be read-
ily obtained from the table as it stands. However, with the use of partial fraction
decomposition (see Chapter 6, Section 6.3.2) expressions can be obtained that can
be inverse Laplace transformed easily.
Next, examples are given on how to find the inverse Laplace transform of some
given functions.
208 Laplace Transforms

Example 8.10
Find

 2 
ℒ−1  2 2 .
 s (s + 4) 

This is not readily available as it stands in the standard tables. The frac-
2
tion 2 2 has to be split into simpler fractions using partial fraction
s (s + 4)
decomposition:

2 A B Cs + D
= + + (8.8)
s (s + 4)
2 2
s s2 s2 + 4

Multiplying throughout by the s2(s2 + 4) term gives the following:

2 = A(s)(s 2 + 4) + B(s 2 + 4) + (Cs + D)s 2 (8.9)

Using easy values of s (i.e., s = 0) and then equating coefficients on both

sides of Equation 8.8 gives

1 1
A = 0, B = , C = 0, and D = −
2 2

Therefore, Equation 8.8 can now be written as

2 1 1
= −
s (s + 4)
2 2
2s 2
2(s + 4)
2

 2  −1  1  −1  1 1 
ℒ −1  2 2  = ℒ  2 −ℒ 2 2 
 s ( s + 4 )   2 s   ( s + 4) 
 2  1 −1  1  1 −1  1 
ℒ −1  2 2 = ℒ  2 − 2ℒ  2 
 s ( s + 4 )  2 s   (s + 4) 

Finally, the inverse transforms are obtained from using the table of stan-
dard transforms:

 2  1 1
ℒ −1  2 2  = 2 t − 4 sin(2t )
 s ( s + 4 ) 
 8.4 Inverse Transforms 209

Example 8.11
s+4
If F (s) = , find f(t).
(s + 3)(3s − 2)
Solution: First using the partial fractions method gives

s+4 A B
= + (8.10)
(s + 3)(3s − 2) s + 3 3s − 2

Multiplying throughout by the (s + 3) (3s − 2) term gives

s + 4 = A(3s − 2) + B(s + 3)

2
Using easy values of s, that is, s = –3 and s = , gives
3

1 14
A=− and B =
11 11

Therefore, Equation 8.10 becomes by writing the constants at the front as

s+4 −1 1 14 1
= +
(s + 3)(3s − 2) 11 s + 3 11 3s − 2

and so

f (t ) = ℒ −1 [ F (s) ]
 s+4   −1 1   14 1 
f (t ) = ℒ −1   = ℒ −1   + ℒ −1  
 (s + 3)(3s − 2)   11 (s + 3)   11 (3s − 2) 
 1 
−1 −1  1  14 −1  3 
f (t ) = ℒ   + ℒ  
11  (s + 3)  11  s − 2 

  
3  

Here in the term (3s – 2) the 3 has been taken out of the bracket first and
then using the standard table of transforms gives

2
−1 −3t 14 3 t
f (t ) = e + e
11 33
210 Laplace Transforms

So far, it has been shown how to Laplace transform and inverse Laplace
transform certain continuous time functions. However, in some engineering
problems, the input to the system is discontinuous in time and hence con-
sideration is now given on how to Laplace transform some of these kinds
of input functions.

8.5 Discontinuous Functions
8.5.1 Heaviside Unit Step Function
The unit step function has the effect of switching on or switching off at some pre-
described value of the time t. The function is shown in Figure 8.2. This function
is the Heaviside unit step and is denoted by f(t) = H(t – c), where

f (t ) = 0 :t <c
f (t ) = 1 :t ≥c

8.5.1.1 Calculating the Laplace Transform of H(t – c)

ℒ  H (t − c)  =
∫ H (t − c)e
0
− st
dt

But H (t − c)e − st = 0 t <c

− st
=e t ≥c
Therefore,
∞ ∞ ∞
 − st  e − sc e − sc
ℒ  H (t − c)  =
∫
0
H (t − c)e − st dt =
∫
c
e − st dt =  e  = 0 −
 − s c −s
=
s

Generally,

e − sc
ℒ  H (t − c)  = (8.11)
s

This is the Laplace transform of unit step function operating at t = c.

f (t)

c t

Figure 8.2 Heaviside unit step function operating at t = c.

 8.5 Discontinuous Functions 211

8.5.1.2 Unit Step at Origin

If the unit step occurs at the origin, then putting c = 0 in H(t – c) gives f(t) = H(t)
and this is shown in Figure 8.3.
Also, the Laplace transform of the unit step function at the origin becomes

e− s 0 1
ℒ  H (t )  = =
s s

using c = 0 in Equation 8.11.

More generally, special kinds of discontinuous input functions can be described
using the Heaviside unit step functions operating at different points. An example
is the square pulse operating from t = a to t = b as given in Figure 8.4.
Here, we can represent f(t) by subtracting two Heaviside unit step functions as

f (t ) = H (t − a) − H (t − b)

Hence, the Laplace transform of the square pulse becomes

e − sa e − sb
ℒ  f (t )  = L  H (t − a) − H (t − b)  = −
s s

using Equation 8.11.

Next, another discontinuous function known as the delta (or impulse) function
is considered.

f (t)

0 t

Figure 8.3 Heaviside unit step function operating at t = 0.

f (t)

0 a b t

Figure 8.4 Square pulse operating between t = a and t = b.

212 Laplace Transforms

8.5.2 The Delta Function

Graphically the delta function is a rectangular pulse of zero width, infinite height,
and can be represented by a single vertical line with an arrowhead as shown in
Figure 8.5.
The definition of the delta function operating at t = a is

δ (t − a) = 0, t≠a
δ (t − a) = ∞, t=a

and the area of the pulse is 1.

The Laplace transform of δ(t – a) is then given by

∫
ℒ δ (t − a)  = δ (t − a)e − st dt
0
∞

=e − sa
∫ δ (t − a) dt
0

But
∫ δ (t − a) dt = 1 (just the area of the delta function).
0
Therefore, the Laplace of the delta function operating at t = a is

ℒ δ (t − a)  = e − sa (8.12)

f (t) δ (t–a)

0 a t

Figure 8.5 Delta function operating at t = a.

8.5.2.1 The Delta Function at the Origin

The delta function at the origin as shown in Figure 8.6 is defined as

δ (t ) = 0, t≠0
δ (t ) = ∞, t=0
 8.6 Shift Theorems 213

δ (t)

0 t

Figure 8.6 Delta function operating at t = 0.

It follows that the Laplace transform of the delta function at the origin with
a = 0 is

ℒ δ(t )  = 1

using Equation 8.12.

Note: Use of these types of discontinuous input functions can arise in engineering
problems. For example, in fire these can be used to determine what happens if a
window breaks or a suppression system activates. Hence, knowing their Laplace
Transforms enables solutions to problems as is shown later in the applications
section (Section 8.8).

8.6 Shift Theorems
When working with differential equations with discontinuous input functions, the
final part of inverse Laplace transforming sometimes requires the use of the first
and second shift theorems and so these are stated next:

• First shift theorem (s-shifting)

f (t ) ↔ F (s), then e at f (t ) ↔ F (s − a)

• Second shift theorem (t-shifting)

If F (s) = ℒ  f (t )  , then e − cs F (s) = ℒ  f (t − c) H (t − c)  (8.13)

Again, these shift theorems can be derived using the definition formula of the
Laplace transform. These are useful in some cases for finding inverse Laplace
transforms, that is, taking the inverse Laplace transform of Equation 8.13 gives
the following important result:

ℒ −1 e − cs F (s)  = f (t − c) H (t − c) (8.14)

214 Laplace Transforms

The following examples show how to make use of the second shift theorem to
find inverses.

Example 8.12
e −4 s
Find a function f(t) whose transform is F (s) = .
s2
Solution: The numerator e–4s corresponds to e–cs (i.e., c = 4 ) in Equation
8.14. Therefore, this indicates that there is a H (t – 4) term present and as

1
F (s ) =
s2

implies that f (t) = t, which is the inverse Laplace transform of t.

 e −4 s 
ℒ −1  2  = (t − 4) H (t − 4)
 s 

using Equation 8.14.

A systematic method for solving ordinary linear differential equations with

a range of different inputs can now be outlined and followed as given next.

8.7 Method for Solving Linear Differential Equations

To solve linear differential equations by the Laplace Transform method, the fol-
lowing steps can be applied:

Step 1: Rewrite the differential equation in terms of the Laplace transform

in the s-domain (i.e., Laplace transform every term).

Step 2: Insert the initial conditions.

Step 3: Rearrange the equation algebraically to give the Laplace transform

of the solution.

Step 4: Finally, determine the inverse Laplace transform to obtain the solution
(with the help of partial fractions and the table of standard transforms).

The following are key relationships that are used frequently in step 1:

ℒ  x (t )  = X (s)
 dx 
ℒ   = sX (s) − x (0)
 dt 
 d 2x 
ℒ  2  = s 2 X (s) − sx (0) − x (0)
 dt 

The next few examples show how to apply the preceding methodology to a range
of different differential equations.
 8.7 Method for Solving Linear Differential Equations 215

Example 8.13
Find the solution to the first-order differential equation

dx
+ 3 x = e −2t (8.15)
dt

with given initial conditions that x = 2 when t = 0.

Solution:

Step 1: Laplace transforming Equation 8.15:

 dx 
ℒ   + ℒ[3 x ] = ℒ [ e −2t ]
 dt 

1
sX (s) − x (0) + 3 X (s) =
s+2

Step 2: Insert initial conditions, that is, x(0) = 2:

1
sX (s) − 2 + 3 X (s) =
s+2

Step 3: Find X(s) using algebraic manipulation

1
(s + 3) X (s) = +2
s+2
2s + 5
X (s ) =
(s + 2)(s + 3)

Step 4: Find the inverse Laplace transform

 2s + 5 
x (t ) = ℒ −1  X (s)  = ℒ −1  
 (s + 2)(s + 3) 

This is done by using partial fraction splitting of the fraction

2s + 5
(s + 2)(s + 3)

 1 1  −1  1  −1  1 
x (t ) = ℒ −1  X (s)  = ℒ −1  +  = ℒ  (s + 2)  + ℒ  (s + 3) 
 ( s + 2) ( s + 3)     

Therefore, x(t) = e–2t + e–3t using standard tables.

216 Laplace Transforms

Example 8.14
Solve the following second-order differential equation:

d2y dy
2
+ 5 + 4 y = 3δ(t − 2) (8.16)
dt dt

with initial conditions y(0) = 2 and y′(0) = –2.

Solution:

Step 1: Laplace transforming Equation 8.16:

 d2y   dy 
ℒ  2  + 5ℒ   + 4ℒ[ y] = 3 ℒ δ(t − 2) 
 dt   dt 

s 2Y (s) − sy(0) − y (0) + 5 ( sY (s) − y(0) ) + 4Y (s) = 3e −2 s

Step 2: Insert initial conditions:

s 2Y (s) − 2s + 2 + 5(sY (s) − 2) + 4Y (s) = 3e −2 s

Step 3: Find Y(s) using algebraic manipulation

(s 2 + 5s + 4)Y (s) − 2s − 8 = 3e −2 s
3 e −2 s + 2s + 8
Y (s ) =
(s 2 + 5s + 4)
3 e −2 s + 2(s + 4)
Y (s ) =
(s + 1)(s + 4)
3 e −2 s 2
Y (s ) = +
(s + 1)(s + 4) (s + 1)

Step 4: Find the inverse Laplace transform. Splitting

3
(s + 1)(s + 4)

using partial fractions gives

 1 1  −2 s 2
Y (s ) =  −  e +
 (s + 1) (s + 4)  (s + 1)

and hence
 8.7 Method for Solving Linear Differential Equations 217

y(t ) = ℒ −1[Y (s)]

 1 1  −2 s −1  2 
y(t ) = ℒ −1  −  e + 2 ℒ  (s + 1) 
 ( s + 1) ( s + 4 )   

Therefore, using the second shift theorem and standard table gives the
solution as

y(t ) = [ e − (t − 2) − e −4 (t − 2) ] H (t − 2) + 2e − t

Example 8.15
This is a more complicated example in which there are two-coupled second-
order differential equations to solve.
Solve the following second-order simultaneous equations:

d 2x
+ 2x − y = 0 (8.17)
dt 2

d2y
+ 2y − x = 0 (8.18)
dt 2

with initial conditions

x (0) = 4 and x (0) = 0; y( 0 ) = 2 and y (0) = 0

Solution:

Step 1: Laplace transforming Equations 8.17 and 8.18:

 d 2x 
ℒ  2  + 2ℒ[ x ] − ℒ[ y] = L[0]
 dt 

 d2y 
ℒ  2  + 2ℒ[ y] − ℒ[ x ] = L[0]
 dt 

s 2 X (s) − s x (0) − x (0) + 2 X (s) − Y (s) = 0

s 2Y (s) − s y(0) − y (0) + 2Y (s) − X (s) = 0

Step 2: Put in initial conditions

s 2 X (s ) − 4 s + 2 X (s ) − Y (s ) = 0
s 2Y (s) − 2s + 2 Y(s) − X(s) = 0
218 Laplace Transforms

Step 3: Simplifying algebraically

(s 2 + 2) X (s) − Y (s) = 4 s (8.19)

(s 2 + 2)Y (s) − X (s) = 2s (8.20)

These are two simultaneous equations in X(s) and Y(s) and we can
eliminate Y(s) to obtain X(s) first.
Multiplying Equation 8.19 by (s2 + 2) and then adding the two equa-
tions and after simplifying an expression for X(s) can be found. Hence,
solving the preceding two equations for X(s) gives

4 s 3 + 10 s
X (s ) =
(s 2 + 1)(s 2 + 3)

Step 4: Inverse Laplace transform using partial fractions and tables:

3s s
X (s ) = +
(s 2 + 1) (s 2 + 3)

Hence,

 3s  −1  s 
x (t ) = ℒ −1  2 +ℒ  2 
 ( s + 1)   + 3) 
( s
 s  −1  s 
x (t ) = 3ℒ −1  2 + ℒ 
 ( s + 1 )  ( ( ))
 s 2 + 3
2 


x (t ) = 3 cos(t ) + cos ( 3t ) (8.21)

Now y(t) can be found using the original Equation 8.17. Using
Equation 8.17 and rearranging gives

d 2x
y(t ) = + 2x (8.22)
dt 2

Differentiating x(t) (i.e., Equation 8.21) once gives

dx
= −3 sin t − 3 sin 3t
dt

Differentiating this again gives

d 2x
= −3 cos t − 3 cos 3t
dt 2
 8.8 Applications 219

Then using Equation 8.22 gives the solution for y(t) as

y(t ) = 3 cos(t ) − cos ( 3t )

In many applications, there is a relationship between two variables and
the interdependence of the variables produces coupled differential equa-
tions. In fire engineering, the variables are usually the temperature and
the fire radius, which then generate a coupled differential equation when
modeling the phenomenon. Modeling predator–prey relationships of sim-
ple ecosystems often produces coupled first-order differential equations.
However, in this case the equations are nonlinear in nature and generally
more difficult to solve.
Having seen how the Laplace transform method can be applied to solv-
ing general differential equations, in the next section a range of different
real-world applications are given where the Laplace transform method can
be used to solve the problem.

8.8 Applications
Example 8.16: Temperature Variation of the Hot Gas and Smoke Layer
A simple two-zone model of a room splits the room into an upper smoke
layer and a lower layer with the fire as shown in Figure 8.7.
A simple equation for a two-zone model of a fire in a room can be derived
from using the heat balance equation:

Heat in the room = Heat generated from the fire − Heat losses in the room

which gives the following equation:

dT 
mc p = Q − Ah∆T (8.23)
dt

assuming that the initial temperature in the room is T0 (i.e., T(0) = T0) and if
radiation is not a dominant phenomenon (i.e., nonflashover fires).

Layer of hot gas and smoke d

Fuel array

Figure 8.7 Two-zone model of a fire in an enclosed room.

220 Laplace Transforms

The differential Equation 8.23 can be simplified by dividing by mcp and

letting ΔT = T – T0 as follows:

dT Q Ah
= − (T − T0 ) (8.24)
dt mc p mc p

Simplification can be done further by making the following substitutions

for the constants as

Ah Q
a= and b = + aT0
mc p mc p

into Equation 8.24. This gives the following simple first-order differential
equation:

dT
= b − aT (8.25)
dt

with initial condition that T(0) = T0 to solve.

This differential Equation 8.25 can be solved to find how the tempera-
ture T of the hot smoke varies as a function of time t using the Laplace
transform method.

Solution: Starting with Equation 8.25 and first rearranging gives

dT
+ aT = b with T (0) = T0 (8.26)
dt

Step 1: Laplace transforming Equation 8.26:

 dT 
ℒ + aℒ[T ] = ℒ[b]
 dt 
b
sT (s) − T (0) + a T (s) =
s

Step 2: Substitute in the initial condition:

b
sT (s) − T0 + aT (s) =
s

Step 3: Rearranging algebraically to find T(s):

b
sT (s) + aT (s) = + T0
s

b + sT0
(s + a)T (s) =
s

sT0 + b
T (s ) =
s (s + a)
 8.8 Applications 221

Step 4: Using partial fraction decomposition and inverse Laplace trans-

forms gives

sT0 + b A B
= + (8.27)
s (s + a) s s + a

Multiplying throughout by s(s + a) gives

sT0 + b = A(s + a) + Bs

b b
Using s = 0 gives A = and using s = –a gives B = T0 − . This gives
Equation 8.27 as a a

sT0 + b b 1  b 1
= +  T0 − 
s (s + a) a s  a s + a

So,

b1  b 1
T (s ) = +  T0 −  (8.28)
as  a s + a

Applying the inverse Laplace transform to Equation 8.28 to give T(t)

using the standard tables as

T (t ) = ℒ −1 T (s) 
 b1  b  1  b −1  1   b  −1  1 
T (t ) = ℒ −1  +  T0 −   = ℒ   +  T0 −  ℒ 
 as  a s + a a s a  s + a 
b b
T (t ) =  T0 −  e − at
a  a

Substituting back for the constants a and b and collecting terms gives
T(t):

Ah
−
mc p
t  Q   Ah 
 1 − e
−
T (t ) = T0 e +  T0 + mc p
t
 (8.29)
 Ah  

In Equation 8.29, as t → ∞ the exponential terms → 0, giving the long-

Q
term temperature as T (∞) → T0 + .
Ah
The solution can be shown graphically as in Figure 8.8.
It can be seen that the temperature in the room increases to a steady-
state value as long as there is still fuel burning. The fire will grow toward
222 Laplace Transforms

T (t)

Q̇
T0 +
Ah

T0

0 t

Figure 8.8 Graph showing temperature variation against time.

a “quasi-steady” temperature. It will keep settling closer and closer to this

value until the fuel source eventually runs out. When Q = 0, then the tem-
perature will start to decay again. So, Equation 8.24 can now be solved
with Q = 0, and the approximate initial conditions produce an exponential
decay for the temperature distribution once the fuel has run out.

Example 8.17: Basic Fire Growth Model Using Unit Step Functions
As an example, in fire combustion a basic crude model of fire growth is that
it can be represented as a pulse wave similar to that given in Figure 8.4 with
the heat release rate Q (t) against time t shown in Figure 8.9.
Here in the first 2 minutes it is assumed that the fire is just getting started
and then there is a constant fire for around 20 minutes before it dies out. The
heat release above could be represented in terms of the unit step functions
with the units here being MW as follows:

Q (t ) = H (t − 2) − H (t − 22)

The next example shows how the unit step function can be used to represent
more general functions and their applications.

˙ (t)
Q

1 MW

0 2 22 t (mins)

Figure 8.9 Heat release rate against time.

 8.8 Applications 223

Example 8.18: Piecewise Function Representation

In some practical situations, a function f(t) may wary differently over time
as follows:

 f (t ) 0 ≤ t < t1
 1
f (t ) = 
 f (t ) t1 ≤ t ≤ t2
 2

Graphically, the function may be represented as shown in Figure 8.10.

However, again using the definition of the unit step function, this func-
tion f(t) can now be represented as a single function as follows:

f (t ) = 1 − H (t − t1 )  f1 (t ) +  H (t − t1 ) − H (t − t2 )  f2 (t ) (8.30)

Since when 0 < t < t1, then H (t − t1) = 0 and H (t − t2) = 0. Then f(t) = f1(t).
When t1 ≤ t < t2, then H (t − t1) = 1, H (t − t2) = 0, and f(t) = f 2(t).
And when t ≥ t2, then now H (t – t1) = 1 and H (t – t2) = 1, so f(t) = 0 again
as required.
This representation of f(t) is true for all times.
Figure 8.10 can now be used to model a more realistic development of
fire growth then that given in Figure 8.9 in which the first function f1(t) is
a t-squared function and the second function f 2(t) is a constant K. Now, the
value of t1 would vary for the different types of fires from 10 minutes for a
slow fire, 5 minutes for a medium, 2.5 minutes for a fast, and 1.25 minutes
for an ultrafast fire.
If f(t) is now the heat release rate Q (t), then this can be represented by a
single function as follows using Equation 8.30:

Q (t ) = 1 − H (t − t1 )  t 2 +  H (t − t1 ) − H (t − t2 )  K

f (t)

f2 (t)
K

f1 (t)

0 t1 t2 t

Figure 8.10 A piecewise function over time.

224 Laplace Transforms

In other engineering fields, the use of the unit step function can offer a
method to derive solutions for problems in which the boundary conditions
depend upon time. An example of this is found in oil and gas exploration
where there is a pressure buildup following constant pressure production.
The boundary conditions for the oil well are now different for the constant
pressure and for the pressure buildup regions. Representing the boundary
conditions as a single function using the unit step functions as in Equation
8.30 allows solutions to be found subsequently using the Laplace Transform
method. This scenario often arises in drill stem testing of gas wells.

Example 8.19: Transient Analysis to Determine the Current i(t) in a

RCL Circuit
In Figure 8.11 is an electrical circuit with a voltage source V(t), a resis-
tor with resistance R, an inductor with inductance L, and a capacitor with
capacitance C all connected in series.
Using Kirchhoff’s law for the voltage around the circuit, it is known that
the voltage source V(t) is equal to the voltage across the resistor, inductor,
and the capacitor. In terms of the current i(t) and charge q(t) this yields the
following first-order differential equation:

di q
L + Ri + = V (8.31)
dt C

Since the current in the circuit is given by

dq
i(t ) = (8.32)
dt

V(t)

AC
i(t) L

Figure 8.11 RCL electrical circuit.

 8.8 Applications 225

di d 2q
this implies that = . So Equation 8.31 can be written in terms of the
dt dt 2
di
charge q(t) by replacing for and for i(t) as follows:
dt
d 2q dq q
L +R + =V (8.33)
dt 2 dt C

This differential equation has initial conditions for the current and the
 0) and q(0), respectively.
charge in the circuit, which can be written as q(
This electrical circuit problem can now be solved using the Laplace
transform method to first find the charge q(t) and then subsequently the cur-
rent in the circuit i(t) using Equation 8.32.

Solution:

Step 1: Laplace transforming Equation (8.33):

 d 2q   dq  q
ℒ  L 2  + R ℒ   + ℒ   = ℒ[V ]
 dt  dt
  c

1
L  s 2Q(s) − s q(0) − q (0)  + R  s Q(s) − q(0)  + Q(s) = V (s)
C

 0) here and so can be

Step 2: The initial conditions are just q(0) and q(
left as such.
Step 3: Rearrange the equation algebraically to find Q(s).

 2 1
 L s + Rs +  Q(s) = V (s) + L s q(0) + L q (0) + R q(0)
C

V (s) + Ls q(0) + L q (0) + R q(0)

Q (s ) = (8.34)
1
L s 2 + Rs +
C

Step 4: Inverse Laplace transform gives

 V (s) + Ls q(0) + L q (0) + R q(0) 

q(t ) = ℒ −1 Q(s)  = ℒ −1   (8.35)
1
 L s 2 + Rs + 
 C 

 0) are known and so is the voltage source

The initial conditions q(0), q(
V(t). Hence V(s) is known. Also, L, R, and C are all constants, which are all
known.
So, from Equation 8.35, the charge q(t) in the circuit can be found. The
dq
current i(t) can then be found from Equation 8.32 using i(t ) = , that is, by
differentiating the expression for the charge q(t). dt
226 Laplace Transforms

Note: An important part of fire studies is fire investigation. After the fire has
been put out by firefighters, fire investigators have to determine what was the
original cause of the fire. Most fire investigation studies carried out are insur-
ance related, where the cause of the fire can often have an electrical cause as
the suspicion. For example, with a voltage source V = 230 V and a device that
has been running for say 2 hours, the temperatures reached can be calculated
approximately using the appropriate energy equations, that is, VIt = mcΔT. So,
understanding well the basic concepts of electrical flow are essential to the fire
investigators’ knowledge base.

Example 8.20: Application in Control Systems

In control systems, the output response of a system is designed to relate in
a particular way to the input of the system. It works by comparing the input
response of the system to the output response using appropriate feedback
control so as to reduce the difference between the input and the output to
zero.
For a general component of a control system, the transfer function H(s)
relates the Laplace transform Y(s) of the output y(t) to the Laplace transform
R(s) of the input r(t) as seen in Figure 8.12.

H(s)
R(s) Y(s)

Input Output
Control
system

Figure 8.12 Control system representation Y(s) = H(s) R(s).

A general symbolic representation of a control feedback system is shown

in Figure 8.13.
Using some basic block diagram representation, this gives E(s) = R(s) – Z(s).

Y ( s ) = G ( s ) E ( s ) = G ( s ) [ R( s ) − Z ( s ) ] (8.36)

So, Y(s) = G(s) E(s) = G(s) [R(s) – Z(s)].

R(s) + E(s) Y(s)

G(s)
–

Z(s)

H(s)

Figure 8.13 Control feedback system representation.

 Problems 227

But Z(s) = H(s) Y(s). Substituting this in above for Z(s) gives

Y (s) = G (s) [ R(s) − H (s)Y (s) ]

Rearranging to find Y(s) gives

Y (s) = G (s) R(s) − G (s) H (s)Y (s)

Y (s) + G (s) H (s)Y (s) = G (s) R(s)
Y (s) [1+ G (s) H (s) ] = G (s) R(s)

G (s )
Y (s ) = R( s ) (8.37)
[1 + G(s) H (s)]
This can be written as

Y ( s ) = T ( s ) R( s )

where

G (s )
T (s ) =
1 + G (s) H (s) 

Thus, the transfer function for the control system is T(s) in which the
feedback transfer function H(s) can be manipulated for design purposes as
required.
The solution for Equation 8.37 can be found by the Laplace transforms
and inverse Laplace transforms method once G(s), H(s), and R(s) are known.

 G (s ) 
y(t ) = ℒ −1 [Y (s) ] = ℒ −1  R( s ) 
 
 1 + G (s) H (s)  

This approach is of great importance in many real-world applications

in a variety of different engineering disciplines, such as automated air-
craft control. The main task here is ensuring that the feedback design
produces a stable solution, that is, the aircraft remains in the desired
position during flight.

Note: Although not studied extensively through this approach, flashover room
fires can also be understood as thermal feedback problems.

Problems
8.1 Using the formula definition, find the Laplace transform F(s) of f(t) = t2.
(Hint: Use integration by parts twice.)
228 Laplace Transforms

8.2 Solve the following differential equations by using the Laplace

transformation:

a. y ′(t ) − 7 y(t ) = e −2t; y( 0 ) = 1

b. T ′(t ) − 6 T (t ) = 3; T (0) = 1

c. y ′′(t ) − 9 y ′(t ) + 8 y(t ) = 8; y(0) = 2, y (0) = 1

d. x ′′(t ) − 18 x ′(t ) + 6 = 0; x (0) = 2, x (0) = 1

8.3 Consider the following electrical RL circuit as shown in Figure 8.14,

where V(t) is the voltage source and a resistor with resistance R com-
bined with an inductor with inductance L are connected in series.
Kirchhoff’s voltage law gives the differential equation for the current in
the circuit as follows:

di
L + Ri = V (t )
dt

Find an expression for the current i(t) in the circuit given that i(0) = 0 for the
following cases:

a. If V(t) = V0 (i.e., constant).

b. If V(t) = V0 sin wt.

V(t)
AC
i(t) L

Figure 8.14 RL circuit.

8.4 Solve the following coupled first-order differential equations:

x ′(t ) − k y(t ) = C

y′(t ) − k x (t ) = 0

with initial conditions x(0) = y(0) = 0, leaving your answers in terms of

the constants k and C.
 9 Fourier Series and
Fourier Transforms

The Fourier series is named after the French mathematician Jean-Baptiste Joseph
Fourier. A Fourier series is a mathematical way of representing a wavelike func-
tion (or signal) as a sum of simple sine and cosine waves. It decomposes a periodic
function or periodic signal into a sum of an infinite set of sinusoidal functions.
When these sines and cosines are expressed as complex exponentials this gives
the Fourier series in complex form. As seen in Section 9.4 the Fourier transform
is the generalization to nonperiodic functions. Since the Fourier series deals with
periodic phenomena, it is important to first understand what is meant by a func-
tion being periodic in nature.

9.1 Periodic Functions
A periodic function (or signal) f(x) is said to have a period T or be periodic with
period T if for all values of x

f (x + T ) = f (x) (9.1)

where T is a positive constant. The function then just repeats itself with period T
over the whole interval −∞ < x < ∞.

1
Note: The number of oscillation per second, that is, frequency (Hz) f = , and
2π T
angular frequency (radians per second) is w = = 2π f .
T

The function sin x is an example of a periodic function with period 2π. The
function repeats itself again after an interval of 2π (Figure 9.1).
Similar trigonometric functions that are periodic are the cosine function cos x
with period 2π, and the tangent function tan x with period π.

229
230 Fourier Series and Fourier Transforms

f (x)
Period

0 2π x

Figure 9.1 Sine wave function or signal.

f (x)

Period

Figure 9.2 Square wave function.

f (x)

Period

Figure 9.3 Sawtooth wave function.

Other examples of periodic functions are the square wave and the sawtooth
function shown in Figures 9.2 and 9.3, respectively.

9.2 Fourier Series
9.2.1 Periodic Functions of Period T
Repeating functions can have different values for the period. For example, the
sine function has period 2π A more general treatment of the Fourier series is
 9.2 Fourier Series 231

f (t)

Period T

Figure 9.4 General periodic waveform with period T.

given where the period can be of any value, say, T, and considering the interval
as being time t, a general periodic function f(t) can be represent as in Figure 9.4.
2π
Here, f(t + T) = f(t). Generally, w = , where w is the angular frequency (radi-
T
ans per second) and T is the period (seconds).
Using the idea that a periodic function can be represented as an infinite sum of
sinusoidal functions, it can be shown that the Fourier series to represent f(t) can
be written as follows:

f (t ) =
a0
2
+ ∑ a cos nwt + b sin nwt
n =1
n n (9.2)

Notes:

• When the period T = 2π, that is, w = 1, the terms in Equation 9.2 are
made up of cos nt and sin nt, which are periodic on the interval 2π for
any integer n.
• The coefficients an and bn measure the strength of the contribution from
each “harmonic” in the series.

The task is to see if f(t) can be written in the form given by Equation 9.2, then
what the coefficients a 0, an, and bn have to be.
Before these can be found, it is important to have some understanding of
orthogonality of functions as well as other general properties. These are covered
in the next section.

9.2.2 General Properties and Orthogonal Functions

It is the case that the functions cos nt and sin nt have the following property that if
they are integrated over a period then the result is zero, that is,

π π

∫ sin nt dt = ∫ cos nt dt = 0,
−π −π
for all integers n (9.3)
232 Fourier Series and Fourier Transforms

These last two results can easily be shown by integrating the preceding two
functions over the limits of integration but can also be seen from the graphs of
these functions. The cos nt and sin nt functions being periodic with period 2π,
then the integral over half the period cancels out the integral over the other half
of the period, as can be seen in Figure 9.1.
The functions cos nt and sin nt can now be used to help find the coefficients in
Equation 9.2 because they satisfy the following orthogonality properties:

∫ sin mt cos nt dt = 0,
−π
for all m and n (9.4)

∫ sin mt sin nt dt = 0,
−π
for m ≠ n
(9.5)
=π (half the period), for (m = n) > 0

∫ cos mt cos nt dt = 0 ,
−π
for m ≠ n

= 2π (period), for m = n = 0 (9.6)

=π (half the period), for (m = n) > 0

The orthogonality properties given in Equations 9.4, 9.5, and 9.6 can all be
proved by either considering the graphs of the product functions or by expressing
the product functions in terms of the sums of individual sine and cosine functions
and integrating out (see trigonometric functions and integration sections). These
results will all be very useful when calculating the Fourier coefficients in the next
section.
Another important result for the Fourier series in complex form is that in the
exponential notation the orthogonality conditions where m and n are integers
become

∫e
−π
jnt − jmt
e dt = 0; m≠n
(9.7)
= 2π (period); m=n

Again, the proof of this is evident from direct integration and putting in the limits.

9.2.3 Fourier Coefficients
Starting with Equation 9.2 for the Fourier series:

f (t ) =
a0
2
+ ∑ a cos nwt + b sin nwt
n =1
n n
 9.2 Fourier Series 233

The coefficients a0, an, and bn can be found by multiplying f(t) by 1, cos mwt, and
sin mwt, and integrating over a period, respectively, to give the following results:

2
a0 =
T ∫ f (t) dt
T
(9.8)

2
an =
T ∫ f (t) cos nwt dt
T
(9.9)

2
bn =
T ∫ f (t) sin nwt dt
T
(9.10)

Proofs:
For the coefficient a 0, starting with f(t)

f (t ) =
a0
2
+ ∑ a cos nwt + b sin nwt
n =1
n n

integrating over the period T gives

∫ f (t ) dt =
∫
a0
2
dt + ∑ a ∫ cos nwt dt + b ∫ sin nwt dt 
n n
T T n =1  T T 

Using Equation 9.3 gives this as

a0
∫ 2 dt = ∫ f (t) dt
T T
(9.11)

since

∫ cos nwt dt = ∫ sin nwt dt = 0

T T

Integrating the left-hand side of Equation 9.11 gives

a0 T
2
=
∫ f (t) dt
T

and so a 0 is

2
a0 =
T ∫ f (t) dt
T

as required.
234 Fourier Series and Fourier Transforms

Similarly, proofs for an and bn can be shown using the orthogonality properties.

Note: For a general period T, it is better to first sketch the function f(t), then
T T
choose the appropriate periodic interval, that is, from 0 to T or − to .
2 2

As an example of finding the trigonometric Fourier series of a periodic func-

tion, the next example will show how to calculate the coefficients and what the
series looks like that represents the function.

Example 9.1
Determine the Fourier series for the periodic function defined by:

 2(1 + t ) −1 < t < 0

f (t ) =  (9.12)
 0 0 < t <1

where f(t) = f(t + 2), that is, the period T = 2.

Solution: To see what the function given by Equation 9.12 looks like, it is
first sketched as shown in Figure 9.5.
Starting with the general Fourier series for f(t),

∞
a
f (t ) = 0 +
2 ∑ a cos nwt + b sin nwt
n =1
n n (9.13)

2π
Since the period of f(t) is equal to T = 2, this implies that w = = π.
Calculating the coefficient for a 0 using Equation 9.8 gives T

1 1
2 2
a0 =
T ∫T
f (t ) dt =
2 ∫ f (t) dt = ∫ f (t) dt
−1 −1

f (t)

–3 –2 –1 0 1 2 3 t

Figure 9.5 Graph of the function f(t).

 9.2 Fourier Series 235

Now this integral has to be split into two regions since the function f(t)
is different in the two regions −1 < t < 0 and 0 < t < 1, as can be seen from
Figure 9.5.

1 0 1

a0 =
∫
−1
f (t ) dt =
∫
−1
f (t ) dt +
∫ f (t) dt
0
(9.14)

Putting in the function values for the function in the different regions
into Equation 9.14 gives

0 1

a0 =
∫
−1
2(1 + t ) dt +
∫ 0 dt
0
0

a0 =
∫ 2(1 + t) dt = 1
−1

a 0 = 1 in Equation 9.13 for f(t).

Calculating the coefficient for an using Equation 9.9 gives

2
an =
T ∫ f (t) cos nwt dt
T

Using w = π gives

1 0
2
an =
2 ∫ f (t) cos nπ t dt = ∫ f (t) cos nπ t dt
−1 −1

Since f(t) is zero otherwise

an =
∫ 2(1 + t) cos nπ t dt
−1
(9.15)

Now this integral is a product of two functions and has to be performed

dv
by using integration by parts, that is, let u = 2(1 + t) and then = cos nπ t
dt
gives

2
an = [1 − cos nπ ] (9.16)
n 2π 2
236 Fourier Series and Fourier Transforms

where n = 1, 2, 3, ….
This coefficient an has different values according to whether n is odd or
even as follows:

4 (9.17)
If n = odd (1, 3, 5,  ), an = because cos nπ = −1
n π22

If n = even (2, 4, 6,  ), an = 0 because cos nπ = 1 (9.18)

These are the values of an that will be used in Equation 9.13 for f(t).
Similarly, calculating for the coefficient for bn using Equation 9.10 gives

2
bn =
T ∫ f (t) sin nwt dt
T

with w = π, gives

1 0
2
bn =
2 ∫ f (t)sin nπ t dt = ∫ f (t)sin nπ t dt
−1 −1

Since f(t) is zero otherwise

bn =
∫ 2(1 + t)sin nπ t dt
−1
(9.19)

Now this integral is again a product of two functions and has to be done
dv
by using integration by parts, that is, let u = 2(1 + t) and then = sin nπ t
gives dt

2
bn = − (9.20)
nπ

where n = 1, 2, 3, ….
Putting together all the coefficients found gives

a0 = 1
4
an = , n = odd
n 2π 2
an = 0, n = even (9.21)

2
bn = −
nπ
 9.2 Fourier Series 237

Having all the coefficients, the Fourier series for f(t) in Equation 9.2 can
be written as

1 4  cos 3π t cos 5π t 
f (t ) = +  cos π t + + +…
2 π2  9 25 
(9.22)
2 sin 2π t sin 3π t 
−  sin π t + + +…
π 2 3 

and in a more compact form f(t) can be expressed as

∞ ∞
 4   
f (t ) =
1
+ ∑   cos nπ t +
2 n= odd  n 2π 2  ∑  − n2π  sin nπ t
n=1
(9.23)

The above sum continues to an infinite number of terms. It can be seen how
it converges to the original function by plotting a truncated sum of a finite num-
ber of terms. If the sum containing n-trigonometric terms is defined as fn(t), then

1
f0 (t ) = (which is just the average value of the function over the period)
2
1 4 2
f1 (t ) = + 2 cos π t − sin π t
2 π π
1 4 2 2
f2 (t ) = + cos π t − sin π t − sin 2π t , etc.
2 π2 π 2π
The graphs of these functions can now be plotted to see how the series
converges to the original function, as shown in Figure 9.6.
As more and more terms in the series are taken, the resulting function
approximates the original signal more closely. This can be seen in Figure
9.6 with n = 40. The series is starting to approximate the original function
given by Equation 9.12 and Figure 9.5 reasonably well.

Note: The oscillations seen in Figure 9.6b and c do become smaller and smaller
as n gets larger and larger but do not disappear altogether since a discontinuous
function is being represented by smooth sinusoidal functions. This is known as
Gibbs phenomenon after J.W. Gibbs.

1.2 1.8 2
1.0 1.6
1.4
0.8 1.5
1.2
1
0.6 1
0.8
0.4 0.6
0.4 0.5
0.2
0.2
–3 –2 1 2 3 –3 –2 –1 0 1 2 3
–0.2 t –3 –2 –1 0 1 2 3
t t
(a) (b) (c)

Figure 9.6 Graphs of f n(t) for values of n: (a) n = 1, (b) n = 6, and (c) n = 40.
238 Fourier Series and Fourier Transforms

9.3 Complex Form of the Fourier Series

Another way of representing the Fourier series is in terms of the complex expo-
nentials and this turns out to have many useful applications in engineering.
It can be shown that the Fourier series given by Equations 9.2

f (t ) =
a0
2
+ ∑ a cos nwt + b sin nwt
n =1
n n (9.24)

can be written in a more compact form known as the complex form, which simpli-
fies the calculations and is more useful especially when the Fourier transform is
considered later.
Starting with Euler’s formula:

e jθ = cos θ + j sin θ (9.25)

then

e − jθ = cos θ − j sin θ (9.26)

By adding and subtracting Equations 9.25 and 9.26 gives the following formu-
lae for the cosine and sine function terms of exponentials:

e jθ + e − jθ
cos θ = (9.27)
2

e jθ − e − jθ
sin θ = (9.28)
2j

Letting θ = nwt, gives the following expressions:

e jnwt + e − jnwt
cos nwt = (9.29)
2

e jnwt − e − jnwt
sin nwt = (9.30)
2j

In Equation 9.24 for the trigonometric Fourier series, the term inside the ∑,
which is an cos nwt + bn sin nwt, now becomes

1 ( jnwt
+ e − jnwt ) +
1 ( jnwt
an e bn e − e − jnwt )
2 2j
(9.31)
1
2
( an − jbn ) e jnwt + 12 ( an + jbn ) e− jnwt
 9.3 Complex Form of the Fourier Series 239

Letting

1
cn =
2
( an − jbn ) and kn = 12 ( an + jbn ) (9.32)

then

an cos nwt + bn sin nwt = cn e jnwt + kn e − jnwt (9.33)

Using Equation 9.33 into Equation 9.24 gives

f (t ) = c0 + ∑c e
n =1
n
jnwt
+ kn e − jnwt (9.34)

a0
where now c0 is given as c0 = .
2
Also, using the definitions of cn and kn given in Equation 9.32, it can be shown
further that kn = c−n, where cn is given by

1
cn =
T ∫ f (t) e
T
− jnwt
dt (9.35)

Equation 9.34 then becomes

f (t ) = c0 + ∑c e
n =1
n
jnwt
+ c− n e − jnwt (9.36)

If the summation for n goes from −∞ to ∞, this then gives the most compact
form for the Fourier series as

f (t ) = ∑c e
n =−∞
n
jnwt
(9.37)

where,

1
cn =
T ∫ f (t) e
T
− jnwt
dt (9.38)

Again, this result for cn follows directly from the orthogonality condition for
exponential notation, that is, from Equation 9.7.

Note: The real coefficients an and bn can be obtained from Equation 9.32 using
cn = 1
2
( an − jbn ) and c− n = 12 ( an + jbn ) .
240 Fourier Series and Fourier Transforms

Solving these for an and bn gives

an = cn + c− n and jbn = c− n − cn (9.39)

These allow for converting back to the real form of the Fourier series if required.
The next example shows how to find the Fourier series of a periodic function
using the complex form.

Example 9.2
Determine the complex Fourier series for the following periodic function:

f (t ) = t −π < t < π (9.40)

f (t ) = f (t + 2π )

Solution: Sketch the function f(t) as shown in Figure 9.7.

Using Equations 9.37 and 9.38 for the complex Fourier series,

f (t ) = ∑c e
n =−∞
n
jnwt
(9.41)

1
cn =
T ∫ f (t) e
T
− jnwt
dt (9.42)

2π
Since f(t) = f(t + 2π) gives T = 2π and so w = =1.
Calculating the coefficients cn using T

1
cn =
T ∫ f (t) e
T
− jnwt
dt (9.43)

f (t)

–3π –2π –π 0 π 2π 3π t

Figure 9.7 Graph of the function f(t).

 9.3 Complex Form of the Fourier Series 241

π
1
=
2π ∫t e
−π
− jnt
dt (9.44)

Using integration by parts to solve Equation 9.44

dv
u=t and = e − jnt
dt

gives

e − jnt
du = dt and v=
− jn

And using the formula for integration by parts gives

  − jnt π π
− jnt 
1   t e  − e dt 
cn =
2π   − jn  − π
 −π
− jn 

∫
  − jnt  
π

cn =
1
(
 −π π e − jnπ − (−π ) e jnπ −  e ) 
2
2π  jn  (− jn)  − π 

1  −π ( − jnπ 
+ e jnπ ) + 2 ( e − jnπ − e jnπ ) 
1
cn = e
2π  jn n 

Since, e−jnπ = ejnπ = cos nπ = (−1)n

1  −π 
cn =  2(−1)n 
2π  jn 

This gives

j(−1)n
cn =
n

Therefore, the complex Fourier series becomes

j ( −1) jnwt
∞ n

f (t ) = ∑
n =−∞
n
e (9.45)

Converting back to the real form of the Fourier series can be done using
Equation 9.39, that is, an = cn + c−n and jbn = c−n − cn if required.
242 Fourier Series and Fourier Transforms

9.4 Fourier Transforms
So far, the Fourier series has been representing periodic functions by a com-
bination of infinite sinusoidal functions. What happens when the function is
not periodic in nature? Can it still be made up of a combination of simpler
functions?
The idea is to transfer from periodic phenomena to nonperiodic phenomena.
This can be achieved by viewing a nonperiodic function as a limiting case of a
periodic function as the period tends to infinity.

9.4.1 Nonperiodic Functions
As stated earlier, Fourier series are applicable to periodic functions only, but non-
periodic functions can also be decomposed into Fourier components. This pro-
cess is called a Fourier transform of a function or signal.
First consider a function that is of finite extent but much less than its periodic-
ity, T, as shown in Figure 9.8. If the period T becomes very large, that is, it tends to
infinity, then the above function shown in Figure 9.8 becomes an isolated aperiodic
function as required. This limiting process is used to develop a heuristic approach
to finding the equations for the Fourier transform from the Fourier series.

9.4.2 Fourier Transform Pair

Starting with the Fourier series and the Fourier coefficient formulae 9.41 and 9.42

f (t ) = ∑c e
n =−∞
n
jnwt
(9.46)

1
cn =
T ∫ f (t) e
T
− jnwt
dt (9.47)

It would be nice to just let T →∞ for cn in Equation 9.47, but this will not work
since it can be shown that Cn → 0 in this case.

f (t)

t
–T –T/2 0 T/2 –T

Figure 9.8 A finite function with a large period.

 9.4 Fourier Transforms 243

What is considered first is to multiply the equation for cn by the period T and
look at the integral given by

∫ f (t) e
T
− jnwt
dt (9.48)

T T
The period can be any period, so consider going from − to and replacing
2π 2 2
for w = into Equation 9.48 giving
T
T
2  n
 n
∫
−2π j  t
T
F  = f (t ) e dt (9.49)
T
T
−
2

Then Equation 9.46 for the Fourier series can now be written as

∞  n
 n  2π j   t 1
f (t ) = ∑
n =−∞
F  e T
T T
(9.50)

n
Now as T →∞, the “discrete variable” gets closer together and is replaced
T
by the continuous variable s, where −∞ < s < ∞.
Therefore, Equation 9.49 can now be called the Fourier transform and can be
written as
∞

F (s ) =
∫ f (t ) e
−∞
−2π jst
dt (9.51)

And the Fourier series Equation 9.49 is given by the summation changing to
1
an integral, and the gets smaller and smaller ≈ ds in integration for the limiting
T
process giving the following:

f (t ) =
∫ F (s)e
−∞
2π jst
ds (9.52)

Therefore, the equations for the Fourier transform for a nonperiodic function are
∞

F (s ) =
∫ f (t) e
−∞
−2π jst
dt (9.53)

f (t ) =
∫ F (s ) e
−∞
2π jst
ds (9.54)

Equation 9.53 is called the Fourier transform of f(t) and Equation 9.54 is called
the inverse Fourier transform of F(s).
244 Fourier Series and Fourier Transforms

Note: The definitions and notations for the Fourier transform and its inverse
are not rigidly fixed; they can vary by factors of 2π or 2π in their equations.

It is useful to see how Equation 9.53 can be used to find the Fourier transform
of functions, as shown in the next example.

Example 9.3
Find the Fourier transform of the following rectangular function
f (t ) = ∏(t ) given by

1 1
f (t ) = 1 − <t<
2 2
(9.55)
1
=0 t ≥
2

Solution: A sketch of the rectangular function is shown in Figure 9.9.

Using Equation 9.53, the definition of the Fourier transform gives

F (s ) =
∫ f (t) e
−∞
−2π jst
dt

1
1
2
 e −2π jst  2 e − π js − eπ js
1
∫
= e −2π jst dt =   =
 −2π js  − 1 −2π js
− 2
2

e − π js − eπ js 1  eπ js − e − π js 
= =  
−2π js πs  2j 
sin π s
F (s) =
πs

f (t)

–1/2 0 1/2 t

Figure 9.9 The rectangular function f (t ) = ∏(t ) .

 9.4 Fourier Transforms 245

F(s)

–2π –π 0 π 2π s

Figure 9.10 The Sinc(s) function.

The function sin π s is given a special name: the Sinc(s) function.

πs
Graphically, it looks like Figure 9.10.
Fourier transforms of different functions can be found in a similar man-
ner to Example 9.3 but usually there are more involved integration tech-
niques like integration by parts is required (see “Problems” section).

9.4.3 What Does the Fourier Transform Represent?

The essence of the Fourier transform of a waveform is to decompose or separate
the waveform into a sum of sinusoids of different frequencies. If these sinusoids
sum together to form the original signal waveform, then the Fourier transform of
the waveform has been found.
A pictorial representation of the Fourier Transform is a diagram that displays
the amplitude and frequency of each of the determined sinusoids as shown in the
following example.

Example 9.4
See the general waveform f(t) in Figure 9.11. To find the Fourier transform
of f(t) is to ask what combination of sinusoids added together will give f(t).
Suppose that f(t) is made up of two functions f1(t) and f 2(t), then the Fourier

f (t)

–T/2 0 T/2 t

Figure 9.11 A general non-periodic waveform.

 246 Fourier Series and Fourier Transforms

Transform of f(t) has been found. If the two functions are shown as in
Figure 9.12, then a diagram can be constructed that displays the amplitude
and frequency of each of the sinusoids f1(t) and f 2(t).
From Figure 9.11, it can be seen that

1
f1 (t ): amplitude = 1; frequency =
T

1 3
f2 (t ): amplitude = ; frequency =
2 T
Putting this information onto a single diagram gives Figure 9.13.
In terms of the delta (or impulse) function this can be written as

1  3 1  3 1  1 1  1
F (s) = δ s−  + δ s+  − δ s−  − δ s+ 
4  T  4  T  2  T  2  T

A summary of this is to say that every signal has a spectrum and the spec-
trum then determines the signal.

f1(t) f2(t)

1 +
1/2

–T/2 T/2 t –T/3 T/3 t

Figure 9.12 Combination of two functions to give f(t).

F (s)

1/2

1/4

–4/T –3/T –2/T –1/T 1/T 2/T 3/T 4/T

Frequency
–1/2

–1

Figure 9.13 Amplitude and frequency components of functions f1(t) and f 2(t).
 9.4 Fourier Transforms 247

9.4.4 Properties of the Fourier Transform

Letting the symbol  denote the Fourier transform operator, then there are vari-
ous properties of the Fourier transform similar to the properties of the Laplace
transform seen in the last chapter. The first of these is that of the linearity property.
Given a function f(t), then the Fourier transform can be written as

 [ f (t )] = F (s)

9.4.4.1 Linearity Property
If f(t) and g(t) are functions of time and a and b are constants, then

  a f (t ) + b g(t )  = a F  f (t )  + b F  g(t ) 
= a F (s ) + b G (s )

There are other important general properties of the Fourier transform and
some of these are given in Table 9.1. The proof of the properties can be shown
using the definition formula of the Fourier transform given by Equation 9.53. The
next example shows the proof for property 3, that is, multiplying a function in the
time domain by an exponential produces a shift in the frequency domain.

Table 9.1 Basic Properties of Fourier Transforms

Operation Time Domain s-Domain
1. Time shifting f(t − a) e−2π jsa F(s)
1  s
2. Time scaling f(at) F 
a  a
3. Multiplying by an exponential e2π jat f(t) F(s − a)
in t-domain
1 d
4. Multiplying by (t) t f (t) − [ F (s)]
2π js ds
5. First derivative f ′(t) 2π js F(s)
6. Second-order derivative f ″(t) (2π js)2 F(s)
7. nth derivative f n(t) (2π js)n F(s)

Example 9.5
Show that the  [e2πjat f(t)] = F(s − a).

Solution: Starting with the formula for the Fourier transform

∞

F (s ) =
∫ f (t) e
−∞
−2π jst
dt (9.56)

Substituting for f(t) by e2πjat f(t) gives

 e 2π jat f (t )  =
∫e
−∞
2π jat
f (t )e −2π jst dt (9.57)
248 Fourier Series and Fourier Transforms

Now putting the exponentials together gives

=
∫ f (t) e
−∞
−2π j ( s − a ) t
dt = F (s − a) (9.58)

Here, s is being replaced by s − a in the definition of the Fourier transform

as required.
This is a very important property and is used in amplitude modulation of
signals as shown in the applications section at the end of the chapter.

9.4.5 Convolution of Two Functions

One of the most important operations in signal processing is the idea of the con-
volution of functions in the time domain. Signal processing uses one signal to
modify another. Most often looking to modify the spectrum of a signal this can
be achieved at a basic level using the linearity property:

  f (t ) + g(t )  =   f (t )  +   g(t ) 

This is just modifying [f(t)] by adding the spectrum of [g(t)] to it. Therefore,
what about multiplying the Fourier transform of two functions does this have a
similar correspondence in the time domain of just multiplying the functions? The
answer to this is not as simple as just simply multiplication of functions in the
time domain but of a convolution of two functions in the time domain instead.
To show this process, start with the multiplication of the Fourier transform of
two functions:

  g(t )  ×   f (t )  (9.59)

Replacing for the definitions of the Fourier transforms for both g(t) and f(t)
gives

∞  ∞ 
∫
=  g(t ) e
−∞
−2π jst
 
−∞
∫
dt   f ( x ) e −2π jsx dx 

(9.60)

This is a separated integral and can be expressed as a mixed double integral

as follows:

∞ ∞

=
∫ ∫e
−∞ −∞
−2π jst −2π jsx
e g(t ) f ( x ) dt dx (9.61)

Collecting the exponential terms together gives

∞ ∞

=
∫ ∫e
−∞ −∞
−2π js ( t + x )
g(t ) f ( x ) dt dx (9.62)
 9.4 Fourier Transforms 249

Regrouping the integral as

∞ ∞ 
=
∫ ∫
 e −2π js (t + x ) g(t ) dt  f ( x ) dx

−∞  −∞

(9.63)

Now, making a change of variables by letting u = t + x, t = u − x, and du = dt,

and substituting into Equation 9.63 gives

∞ ∞ 
 ∫ ∫
=  e −2π jsu g(u − x ) du  f ( x ) dx
−∞  −∞

(9.64)

Interchanging the terms gives

∞ ∞ 
=
∫ ∫
 g(u − x ) f ( x ) dx  e −2π jsu du

−∞  −∞

(9.65)

Now defining h(u) as

h(u) =
∫ g(u − x) f (x) dx
−∞
(9.66)

Then defining the Fourier transform h(u) by

 [h(u)] =
∫ h(u) e
−∞
−2π jsu
du (9.67)

Therefore, it has been shown that

  g(t )  ×   f (t )  =  [ h(u) ] (9.68)

Defining the convolution of functions g and f as

(g * f ) x =
∫ g(x − y) f ( y) dy
−∞
(9.69)

Then finally, it can be stated that the Fourier transform of the convolution of
two functions is equal to the product of the Fourier transform of the two functions
as shown next:

  g * f  =   g(t )  ×   f (t )  (9.70)

This is again an important property for signal processing purposes as seen in the
applications section.
250 Fourier Series and Fourier Transforms

9.5 Applications
Example 9.6: Designing of Fourier Transform Infrared (FTIR)
Smoke Detectors
Fire detector systems should have the ability to discriminate between real
fire sources and nonfire sources. Smoke detectors that can respond quickly
may suffer from the inability to discriminate between a real fire smoke
and smoke from other sources. In high-value installations where there is
expensive and sensitive equipment, it is clear that reliable fire detection
systems are needed. The detection systems are generally used to activate
fixed fire suppression systems like sprinklers, so any false alarms are an
undesirable outcome causing valuable time loss and having potential cost
implications.
Fourier transform infrared (FTIR) smoke detectors make use of Fourier
Transform techniques to analyze combustion products and so aid new more
reliable detection systems to be built. Using FTIR measurements of con-
centrations of gases (e.g., CO2, CO, H2O, and CH4) given off during differ-
ent modes of combustion, the detection system can classify input data as
a flaming fire, smoldering fire, nuisance, or other environmental sources.
Therefore, FTIR spectroscopy can give multiple gas concentrations that
enable an advanced fire detection system to be built.

Example 9.7: Antenna Design Using Frequency Modulation

Consider the design of an antenna for the transmission of an audio signal
x(t) with maximum frequency of approximately 10 KHz.
The relationship between the frequency and wavelength is given by the
equation

c
λ=
f

where λ is the wavelength, f is the frequency, and c is the speed of electro-

magnetic waves.
Using the values of f = 10000 Hz and c = 3 × 108 ms–1 gives a value for
the wavelength of λ = 30,000 m.
1
In the design of antennae, they usually have dimensions of ≈ λ . This
4
implies an antenna size of approximately 7500 m, which of course is not
practical for design purposes.
One way around this problem is to use frequency modulation and use a
carrier signal, say, cos 2πωct, where ωc is a very high frequency combined
with the original signal. The new modulated function can be written as
φ(t) = x(t) cos 2πωct.
Using the definition of the exponential form for the cosine function gives

φ( t ) = x ( t ) 1 e 2π jω t + e −2π jω t
2
 9.5 Applications 251

If the Fourier transform of x(t) is X(s) and the Fourier Transform of φ(t)
is Φ(s) then making use of property 3 in Table 9.1 gives

1
Φ (s ) =  X (s − ω c ) + X (s + ω c ) 
2

Graphically, this is shown in Figure 9.14.

Clearly, the frequency for the modulated signal is now higher than the
original signal, which then makes for a smaller λ and so making for a more
realistic smaller antenna size.

X(s)

Φ(s)

1/2

–wc wc s

Figure 9.14 Frequency for the modulated signal Φ(s).

Example 9.8: Smoothing Process Using Filters

In Section 9.4.5, it was found that the Fourier transform of the convolution
of two functions g and f is given by the product of the Fourier transforms of
the individual functions. This property can be used in signal processing as
a method of filtering the required signal.
Taking as an example a set of results from an experiment. The results
have been plotted on a graph as a function (or signal) φ(t) that is periodic
in nature but has a jagged appearance around the edges (see Figure 9.15).
Taking the Fourier transform of this signal will show the spectrum of the
frequencies associated with the signal Φ(s). This is shown in Figure 9.16.
252 Fourier Series and Fourier Transforms

φ (t)

0 t (months)

Figure 9.15 Signal showing jagged edges.

Φ(s)

0 10 20 30 40 50
Frequency

Figure 9.16 Graph showing the frequency spectrum of the signal Φ(s).

Now the high-end frequencies tend to cause the jaggedness or rapid

oscillations with the signal. If the high frequencies could be eliminated
somehow, then the signal may become smoother in appearance.
This can be achieved by multiplying the signal Φ(s) by a scaled rect-
angular function ∏ 2v (s) in the frequency domain as shown in Figure 9.17.

–ν 0 ν
Frequency

Figure 9.17 Multiplying Φ(s) by the rectangular function ∏ 2v (s).

 Problems 253

φ' (t)

0 t (months)

Figure 9.18 The convoluted function φ′(t).

This is called a low-pass filter as it allows the lower frequencies through

and cuts out the higher frequencies. In the frequency domain this is given
by multiplying the two functions, that is, ∏ 2v (s) × Φ (s) and in the time
domain, this will correspond to a convolution of the Sinc function with φ(t)
to produce φ′(t), that is,

' t) = 2 ν Sinc ( 2 ν t)* φ(t)

φ(

The result φ′(t) is a much smoother function with the jaggedness taken
out, as shown in Figure 9.18.
Examples can also be found where the use of other types of filtering,
such as, high-pass filters and band-pass filters that allow signals to be
manipulated to produce the required outputs.

Problems
9.1 Determine the trigonometric Fourier series for the following functions:

a. f (t ) = 0 −2 < t < 0
=t 0<t<2
f (t ) = f (t + 4)

b. g(t ) = 1 0≤t <π

=0 −π ≤ t < 0
g(t ) = g(t + 2π )

9.2 Given the following periodic function h(t):

h(t ) = t − 3 −2 < t < 0

=t+3 0<t<2
h(t ) = h(t + 4)
254 Fourier Series and Fourier Transforms

a. Determine the complex Fourier series for the function h(t).

b. Hence, obtain its trigonometric Fourier series.

9.3 Prove the orthogonality property for the following exponential func-
tions, where m and n are integers.

∫e
−π
jnt − jmt
e dt = 0 m≠n

= 2π m=n

9.4 Given the triangular function Λ(t) is shown in Figure 9.19 and defined
below as

Λ(t ) = 1 − t t ≤1
=0 t >1

Λ(t)

–1 0 1 t

Figure 9.19 The triangular function Λ(t).

show that the Fourier transform is given by [Λ (t)] = Sinc2 s.

2
9.5 Given the Gaussian function f (t ) = e − π t , show that its Fourier trans-
2
form is given by F (s) = e − π s . (Harder problem)

(Hint: Start with definition, consider F′(s) and then integration by parts.)
 10 Multivariable
Calculus

Multivariable calculus is essentially an extension of the calculus with one variable

in which the system now depends on many variables. In engineering, multivari-
able calculus can be used to model higher dimensional system behavior, that is,
stress may depend on the x, y, and z positions. Most of the concepts introduced in
the one-variable calculus discussions can be extended to multivariable calculus
starting with the ideas of partial derivatives. The higher-order partial derivatives
lead onto the multivariable chain rule and ideas of a general directional derivative
with applications to tangent planes. Higher-order integration of double and triple
integrals using different coordinate systems is considered as well as how these
concepts are used in real-world applications.

10.1 Partial Derivatives
10.1.1 Introduction and Definition
A function of two or more variables such as

f ( x , y) = xy − ye x

has two inputs x and y and it produces one output. A many-input function could
have many outputs and is usually termed as a vector function.
Recall the definition from 1-dimensional differentiation (or derivative) as seen
in Chapter 6,

df ( x )  f ( x + h) − f ( x ) 
= f ′( x ) = lim   (10.1)
dx h→0  h

What is the interpretation of derivatives in 1-dimension calculus? The first is to

consider it as the slope of the tangent line as shown in Figure 10.1.

255
256 Multivariable Calculus

y
y = f (x)

Tangent line

Slope f ʹ(a)

a x

Figure 10.1 The derivative as the slope of the tangent line at some point.

Second, the derivative can be represented as the instantaneous rate of change

of a function, that is, using the first two terms of a Taylor series gives

f (a + h) ≈ f (a) + hf ′(a) (10.2)

If there is a change in the input by an amount h, the output changes by the rate
of change output f′(a) multiplied by h. So f′(a) represents the instantaneous rate
of change of f(a).
Now what is the situation when there is two or more variables? Here there is
still a definition of derivatives, but since there is more than one variable these are
called partial derivatives. Considering a function with two variables, the partial
derivatives can be defined in a similar manner to Equation 10.1.

10.1.1.1 Partial Derivatives Defined

A function f(x, y) has two partial derivatives:

∂f  f (a + h, b) − f (a, b) 
Partial with respect to x: (a, b) = f x  lim  
∂x h→0  h

∂f  f (a, b + h) − f (a, b) 
Partial with respect to y: (a, b) = f y  lim  
∂y h→ 0  h

So, with partial derivatives the derivatives are taken with respect to one vari-
able while keeping the other variables fixed.

∂f
Note: The shorthand notation for partial derivatives is (a, b) = f x .
∂x

In the same way as with 1-D derivatives, the interpretation of the partial deriv-
atives can be seen as an instantaneous rate of change as follows.
 10.1 Partial Derivatives 257

10.1.1.2 Instantaneous Rate of Change

∂f
f (a + h, b) ≈ f (a, b) + h (a, b)
∂x

∂f
f (a, b + h) ≈ f (a, b) + h (a, b)
∂y

These can be put together as a single total change given by

∂f ∂f
f (a + h, b + k ) ≈ f (a, b) + h (a, b) + k (a, b) (10.3)
∂x ∂y

Here a small change in the input variables a by h and b by k produces a total

output change given by Equation 10.3. Also, this can be thought of as the slope of
a tangent line to a surface as shown in Figure 10.2.
Another important way in considering derivatives in higher dimensions is to
fix all the inputs except one and think of it as a function of the one variable only.
For example,

g( x ) = f ( x , b) with y = b

then
∂f
g′(a) = (a, b)
∂x

at the point x = a.

z = f (x, y)

Slope of tangent line is fy(a, b)

(a, b)
Fix x = a

Figure 10.2 Shows the slope of the tangent line keeping one variable fixed.
258 Multivariable Calculus

This is generally the best way to understand partial derivatives and is the way
that is used to compute them. The next examples illustrate this process.

Example 10.1
Given f(x, y) = x + xy2, compute the following.
∂f
1. (1, 3)
∂x
∂f
2. (2, 4)
∂y

Solution:

∂f
1. For (1, 3), here x is varying and so consider y as fixed or constant.
∂x
f(x, y) = x + xy2 differentiating with respect to x keeping y fixed
∂f
gives = 1 + y 2.
∂x
Now substituting in the values of the point (1,3) gives

∂f
(1, 3) = 1 + 32 = 10
∂x

∂f
2. For (2, 4), here y is varying and consider x as fixed or constant.
∂y
∂f ∂f
= 2 xy , so (2, 4) = 2(2)(4) = 16 .
∂y ∂y

Example 10.2
Given f(x, y) = x2y + xey, compute (1) fx(1, 0) and (2) f y(1, 1).

Solution:

∂f
1. f x = = 2 xy + e y
∂x

f x (1, 0) = 2(1)(0) + e 0 = 1

∂f
2. f y = = x 2 + xe y
∂y

f y (1,1) = (1)2 + (1)e1 = 1 + e

 10.1 Partial Derivatives 259

10.1.2 Higher Derivatives
As with the one variable case higher derivatives can be calculated. The only dif-
ference now is that with the multiple variable case, there are many higher deriva-
tives to consider. With a single variable case, there is only one second derivative
but with a two-variable problem. It turns out that there are four second derivatives
that can be calculated. Some of the notation used is given next.
The following notation is used:

∂2 f ∂  ∂f 
=   = f xx
∂x 2 ∂x  ∂x 

∂2 f ∂  ∂f 
=   = f yy
∂y 2 ∂y  ∂y 

∂2 f ∂  ∂f 
= = f xy
∂y∂x ∂y  ∂x 

∂2 f ∂  ∂f 
= = f yx
∂x ∂y ∂x  ∂y 

Example 10.3
Given f(x,y) = xey + xy3, find the following second-order derivatives:

(1) fxx (2) fxy (3) f yx (4) f yy.

Solution:

1. fx = ey + y3, fxx = 0
2. fx = ey + y3, fxy = ey + 3y2
3. fy = xey + 3xy2, f yx = ey + 3y2
4. fy = xey + 3xy2, f yy = xey + 6xy

It can be seen that fxy = f yx. Generally, this property is observed and the
following theorem known as Clairaut’s theorem expresses this.

10.1.2.1 Clairaut’s Theorem
Most of the time fxy(a, b) = f yx (a, b). The order of the differentiation does not mat-
ter. With more variables, that is, x, y, z, the order does not matter, for example,
fxyzz = fzxzy. However, there are conditions on f for this to be true.

10.1.2.2 Antiderivatives When There Are Multiple Variables

Example 10.4
Find a function f(x,y) such that

fx = 2 x + 3 y

f y = 3x + e y
260 Multivariable Calculus

Solution: You need to find the antiderivative of fx by keeping y constant. So,

starting with fx = 2x + 3y and integrating with respect to x keeping y as a
constant implies

f = x 2 + 3 xy + g( y) (any function of y) (10.4)

Finding the antiderivative of f y keeping x constant implies

f = 3 xy + e y + h( x ) (any function of x ) (10.5)

Now the function f(x, y) has to fit both Equation 10.4 and Equation 10.5.
Therefore,

f ( x , y) = x 2 + 3 xy + e y

will satisfy both the required conditions. This will be an important idea for
working with conservative vector fields as seen later in Chapter 11.

Example 10.5
Find f(x, y) such that

fx = 4 x − 3 y

fy = x + y

Solution:

f x = 4 x − 3 y implies f = 2 x 2 − 3 xy + g( y)

y2
fy = x + y implies f = xy + + h( x )
2

Now in this case there cannot be found any function f(x, y) that fits both
profiles. Therefore, it is not possible to find any f(x, y) that has these required
fx and f y functions simultaneously.

10.1.3 Chain Rule
10.1.3.1 Chain Rule with One Variable
The chain rule in one variable is a process of calculating derivatives in situations
where there is a function of a function involved. This can be stated in different
ways, including

d 
f  g( x )   = f ′  g( x )  g′( x ) (10.6)
dx 
 10.1 Partial Derivatives 261

But this can also be thought of as

f = f ( y) and g = g( x )

df df dy
= × (10.7)
dx dy dx

Sometimes it is easier to see how the variables are related to each other by
drawing a dependency chart as shown in Figure 10.3. Equations 10.6 and 10.7 are
the same process.
Using Equation 10.7 as the way of thinking about how variables depend on
each other is more useful as it leads to a natural extension in higher dimensions
as shown in the next sections.

g=y

Figure 10.3 Dependency chart showing relationship between variables.

10.1.3.2 Chain Rule with Multivariables

Suppose there is a function f(x, y) and

x = x (u, v)

y = y(u, v)

Then what is the change in the function due to a change in the variable u, that
∂f
is, ?
∂u
First, drawing a dependency chart helps to show the situation of how the vari-
ables are related to each other, as in Figure 10.4.
So,

∂f ∂f ∂x ∂f ∂y
= × + ×
∂u ∂x ∂u ∂y ∂u

This gives the net chain in f due to the change u through the variables x
and y.
262 Multivariable Calculus

x y

u v

Figure 10.4 Dependency chart showing relationship between variables.

Example 10.6

Given that f = f(z,w), z = z(x,y), w = w(y), x = x(u), and y = y(u,v),

∂f ∂f
find and .
∂u ∂v

Solution: First draw a dependency chart showing the relationship between

the variables as in Figure 10.5.
So, the change in f due to a change in u is given as

∂f ∂f ∂z ∂x ∂f ∂w ∂y ∂f ∂z ∂y
= + +
∂u ∂z ∂x ∂u ∂w ∂y ∂u ∂z ∂y ∂u

Also, the change in f due to a change in v is given by

∂f ∂f ∂z ∂y ∂f ∂w ∂y
= +
∂v ∂z ∂y ∂v ∂w ∂y ∂v

Consider the chain rule by using the diagram to find the relationships
between the different variables.

z w

x y

u v

Figure 10.5 Dependency chart showing relationships between variables.

 10.1 Partial Derivatives 263

Example 10.7
∂f
Given that f = f (u,v,t) where u = u(t) and v = v(t), find.
∂t
Solution: First draw the dependency chart as in Figure 10.6.

df ∂f ∂u ∂f ∂v ∂f
= + +
dt ∂u ∂t ∂v ∂t ∂t

Total derivative Partial derivative, how f changes with

(with the variable t) the third variable input t

u v

Figure 10.6 Dependency chart showing relationships between variables.

The next section considers derivatives in a more general direction called

the directional derivative.

10.1.4 Directional Derivatives and Gradients

To find the change of a function f(x,y) in any general direction not just the x and y
∂f ∂f
directions (i.e., the and ), it is important to consider the following situation
∂x ∂y
shown in Figure 10.7. What is the directional derivative of f(x,y) at (x0 ,y0) in the
direction of 〈a, b〉?
Parameterization of the line in the direction 〈a, b〉 is given by

a
x = x0 + t (10.8)
a + b2
2

b
y = y0 + t (10.9)
a + b2
2

These makes us travel along 〈a, b〉 with unit speed so unit time is now equal to
unit distance in the direction 〈a, b〉. A dependency chart for this situation is shown
in Figure 10.8.
Now the change in f due to a change in t is given by

∂f ∂f dx ∂f dy
= +
∂t ∂x dt ∂y dt
264 Multivariable Calculus

<a, b>

(x0, y0)

Figure 10.7 General directional vector.

x y

Figure 10.8 Dependency chart showing relationships between variables.

Using Equations 10.8 and 10.9 and differentiating with respect to t gives

∂f ∂f a ∂f b
= + (10.10)
∂t ∂x a +b
2 2 ∂y a + b2
2

To represent this change of f with a change in t in the direction of 〈a,b〉 as a

special notation it is written as D‹a,b›  f, the directional derivative of f in the direc-
tion of 〈a,b〉. Equation 10.10 can now be written in a more compact form using
the dot product form as

∂f ∂f 〈 a, b 〉
D‹ a ,b › f = 〈 , 〉⋅ (10.11)
∂x ∂y a2 + b2

∂f ∂f
Using some further notation as 〈 , 〉 = ∇f (called the gradient of f). Also,
∂x ∂y
∇ is sometimes called “del” or “nabla.” So, Equation 10.11 can be written in sim-
pler notation as follows.
 10.1 Partial Derivatives 265

Given f(x, y) with its gradient ∇f = 〈 f x , f y 〉, the directional derivative of f in the

direction u is given by

u
Du f = ∇f ⋅ = ∇f . uˆ (10.12)
u

Now it can be checked to see what the directional derivatives would be in the x
and y directions from the formula given in Equation 10.12 as follows.
In the x-direction, the unit vector is uˆ = 〈1, 0 〉. This then gives Du f as

D〈1, 0 〉 f = ∇f .〈1, 0 〉 = 〈 f x , f y 〉.〈1, 0 〉 = f x

and in the y-direction Du f is

D〈 0 , 1〉 f = ∇f .〈0, 1〉 = 〈 f x , f y 〉.〈0, 1〉 = f y

as expected for both.

Example 10.8
Given f(x,y) = xy2 – 10x:

1. Compute ∇f.
2. What is the directional derivative of f in the direction 〈2,5〉 at the
point (1, 1).

Solution:

1. ∇f = 〈 fx, f y〉 = 〈y2 − 10, 2xy〉

〈2, 5〉
2. D〈 2,5〉 f (1,1) = ∇f (1,1).
29
〈2, 5〉 8
= 〈−9, 2〉 ⋅ =−
29 29

Example 10.9
Given f(x,y) = x – xy2, find:

1. D〈2,−1〉 f(1,0)
2. In what direction û is Du f the biggest?

Solution:

1. Now ∇f = 〈 fx, f y〉 = 〈1 − y2, −2xy〉.

So, ∇f(1,0) = 〈1,0〉.
266 Multivariable Calculus

Therefore,

〈2, −1〉 2
D〈 2,−1〉 f (1, 0) = 〈1, 0 〉. =
5 5

2.

Du f (1, 0) = ∇f (1, 0).uˆ

= ∇f uˆ cos θ

To make this the biggest, cos θ needs to be the largest, which implies
θ = 0. Therefore, angle between ∇f and û should be in the same direction. So,
û should be in the same direction as ∇f but the unit vector version of it, that is,

∇f
û =
∇f

For the above problem ∇f = 〈1,0〉, therefore uˆ = 〈1, 0 〉. So, ∇f = 〈 fx, f y〉

always points in the direction of steepest increase.

10.1.5 Stationary Points (Maxima, Minima, and Saddle Points)

To try to find the maximum and minimum of the function of two variables, let’s
first review the theory in the one-variable problem again as shown in Figure 10.9.
The tangent line gives a first-order approximation:

f ( x ) ≈ f (a) + ( x − a) f ′(a)

And if y ≈ f (a) + ( x − a) f ′(a) this is just the tangent line, similar to the argu-
ments for the tangent plane in higher-order problems.
What do you consider if it is a maximum? At a maximum, f ′(a) = 0 . So, the
second-order approximation is shown in Figure 10.10.

1
f ( x ) ≈ f (a) + ( x − a) f ′(a) + ( x − a)2 f ′′(a)
2

y = f (x)

a x

Figure 10.9 Tangent line at a point.

 10.1 Partial Derivatives 267

y = f (x)

a x

Figure 10.10 Second-order approximation at given point.

This is the Taylor series expansion of a function at a point. This is the parabola
of best fit at x = a. Now, if f′(a) = 0 at a maximum, then f(x) becomes

1
f ( x ) ≈ f (a) + ( x − a)2 f ′′(a)
2

So, this parabola is shaped if f ″(a) > 0 (i.e., a minimum) and this parab-
ola is shaped if f″(a) < 0 (i.e., a maximum).

This can be extended to a function of higher variables. Using a Taylor series

expansion of a function of two variables gives
1
f ( x , y) = f (a, b) + ∇f (a, b).〈 x − a, y − b 〉 + 〈 x − a, y − b 〉.∇ 2 f (a, b).〈 x − a, y − b 〉
2

There needs to be ∇f (a, b) = 0 for maximum and minimum. And

 f f 
xx xy
∇2 f =  
 f yx f yy 

is a matrix called the “Hessian.”

Note: The determinant of the above matrix is also sometimes referred to as the
Hessian.

To determine if you have a maximum or minimum you need to consider the

Hessian matrix. If D is called the Hessian, then it is defined as

 f f 
xx xy
D = det  
 f yx f yy 

and it is this that determines whether there is a maximum or minimum.

Conditions for determining maximum or minimum are stated next.
268 Multivariable Calculus

10.1.5.1 Summary to Find Maximum or Minimum Points

1. Find the critical points given by ∇f (a, b) = 0.

 f f 
xx xy
2. Let D = det   = f xx (a, b) f yy (a, b) − f xy2 = Discriminant
 f yx f yy 

If D > 0 and fxx(a,b) > 0, then (a,b) is a local minimum.

If D > 0 and fxx(a,b) < 0, then (a,b) is a local maximum.

If D < 0, then (a,b) is a saddle point.

Example 10.10
Find and identify the critical points of the following function f(x, y):

f ( x , y) = x 3 − 12 xy + 8 y3

Solution:

0 = f x = 3 x 2 − 12 y gives x 2 = 4 y

0 = f y = −12 x + 24 y 2 gives x = 2 y 2

Using x = 2y2 into x2 = 4y gives (2y2)2 = 4y. Solving for y gives 4y(y3 – 1) = 0.
This gives y = 0 or y = 1 as the solutions to this equation which, then gives
the critical points as (0,0) and (2,1).
To see if the critical points are a maximum or minimum, find the Hessian D:

 f f 
xx xy
D = det   = f xx (a, b) f yy (a, b) − f xy2
 f yx f yy 

Calculating all the higher-order derivatives gives

f xx = 6 x f xx (0, 0) = 0 f xx (2,1) = 12

f yy = 48 y f yy (0, 0) = 0 f yy (2,1) = 48

f xy = −12 f xy (0, 0) = −12 f xy (2,1) = −12

So, D(0,0) = (0)(0) – 144 gives D = – 144 < 0, which implies a saddle
point.
D(2,1) = (12)(48) – 144 gives D = 432 > 0, but fxx(2,1) > 0, so it is a
minimum.
 10.2 Higher-Order Integration 269

Example 10.11
Find three numbers that sum to 100 and have the largest product.

Solution: Let x, y, and 100 – x – y be the three numbers. Therefore, maxi-

mize f(x,y) = xy(100 – x – y).
Look for critical points:

0 = f x = y(100 − x − y) + x (− y) = y(100 − 2 x − y)

0 = f y = x (100 − x − y) + y(− x ) = x (100 − x − 2 y)

Since the product of two things equals zero implies y = 0, the product = 0,
so ignore this.
Therefore, 100 – 2x – y = 0
and x = 0, then again the product = 0, so ignore this.
Therefore, 100 – x –2y = 0.
Solving these two equations simultaneously gives

2 x + y = 100

x + 2 y = 100

100
−3 y = −100 y=
3

This gives x = 100 and the third number 100 − x − y = 100 .

3 3
 100 100 
So the numbers are  100 , , .
 3 3 3 

10.2 Higher-Order Integration
10.2.1 Double Integrals and Fubini’s Theorem
Instead of integrating over an interval, integration is carried out as shown in
Figure 10.11 over a 2-D region.
First, chop the region into small pieces and pick a point (x*, y*). Then,

∫∫ f (x, y) dA  lim ∑
D
∆A→0
all pieces
f ( x*, y*)∆A

To compute this double integral, start with a special case of a rectangle as

shown in Figure 10.12.

Note: The integration sign is used twice.

270 Multivariable Calculus

(x*, y*)

∆A

Figure 10.11 A general 2-D region.

x
a b

Figure 10.12 A general rectangular area.

∫∫ f (x, y) dA  lim ∑
D
∆A→0
all pieces
f ( x*, y*)∆A

Split the rectangular region into horizontal and vertical strips as shown in
Figure 10.13.

lim lim
∆x →0 ∆y→0 ∑ ∑ f (x*, y*)∆y∆x
i j

= lim
∆x →0 ∑ ∑ lim f (x*, y*)∆y  ∆x
∆y→0
i  j 
d

= lim
∆x →0 ∑ ∫ f (x*, y) dy∆x
i c
d

=
∫ lim ∑ f (x*, y)dy∆x
c
∆x →0
i
b d

∫ ∫ f (x, y) dy dx
a c
 10.2 Higher-Order Integration 271

d
(x*, y*)

∆y
c
∆x

x
a b

Figure 10.13 Rectangular region into smaller pieces.

This is just two iterated integrals to be calculated separately. The following

examples show how to calculate these types of double integrals.

Example 10.12
3 1

Evaluate ∫ ∫ (2xy − 4 y) dy dx
0 0

Solution: Do the inner integral with respect to y by fixing x, then do the x

integral.
3 3 3
 x2  9 3
∫ ∫
1
 xy 2 − 2 y 2  dx = ( x − 2) dx =  − 2 x  = − 6 = −
0
2 0 2 2
0 0

Example 10.13
π
3 2

Evaluate
∫ ∫ x cos y dy dx.
0 0
2

Solution:

3 π 3 3
 x3 
∫
0
 x 2 sin y  2 dx =
0 ∫
0
x 2 dx =   = 9
 3 0

Example 10.14
1 1

Evaluate
∫ ∫ xy
0 0
x 2 + y 2 dy dx (more tricky problem).
272 Multivariable Calculus

Solution: Keeping x constant and integrating with respect to y gives

1 1

( )
1  3 
1  3   2 5  5
 x (x + y ) 2   x ( x + 1) − x  dx =  ( x + 1) − x  = 1 2 2 − 2
2 2 2 4 5

∫ ∫
2 2

 3  dx = 
3 3
  15 15  0 15
0
0 0

Fubini’s theorem is an important theorem that essentially allows the

integration order to be interchanged. It is written as

b d d b

∫ ∫ f (x, y) dy dx ≡ ∫ ∫ f (x, y) dx dy
a c c a
(10.13)

This is just saying that it does not matter if you sum x first or y first.

10.2.1.1 An Application of Double Integration

Find the volume under a surface z = f(x,y) as shown in Figure 10.14. To compute
this volume chop the region D into pieces. The volume of that “tower” piece is
approximately:

Vol ≈ lim
∆A→0 ∑
all pieces
f ( x*, y*)∆A

This approaches the volume below the surface:

Therefore,

Volume =
∫∫ f (x, y) dA
D
(10.14)

z
z = f (x, y)

D y

x ∆A

Figure 10.14 Volume under a surface.

 10.2 Higher-Order Integration 273

Example 10.15
Find the volume of the solid in the first octant (see Figure 10.15) bounded
by z = 9 – x2 and y = 4.

Solution: What surface does z = 9 – x2 look like? Since y is not present it can
take on any value. This region is shown in Figure 10.15.

Volume =
∫∫ f (x, y) dA = ∫∫ ( 9 − x ) dA
D D
2

3 4

=
∫ ∫ (9 − x ) dy dx
0 0
2

Integrating with respect to y and keeping x constant gives

3 3 3
 4x3 
∫ ∫
4
9 y − yx 2  dx = (36 − 4 x ) dx = 36 x −
2
 = 72
0
 3 0
0 0

z = 9 – x2
9

4
D
3
y

Figure 10.15 Volume bounded by the surfaces.

10.2.2 Double Integration Using Polar Coordinates

First. let’s review concepts of polar coordinates. Figure 10.16 shows a point on the
x-y plane. The point P can be identified using the usual (x, y) coordinates but also
can be given by the angle θ with the x-axis and a distance r from the origin. Using
this (r, θ) gives the polar form of a point in 2-D space.
Using basic trigonometry gives

x = r cos θ

y = r sin θ
274 Multivariable Calculus

P
r
y

θ
x
x

Figure 10.16 General point P in 2-D space.

Using the Pythagorean theorem gives

r = x 2 + y2

 y
Note: Sometimes θ = tan −1   , but in general must be careful if the point is in a
 x
different quadrant, hence the need to use trigonometry to find the angle θ accord-
ing to the problem being solved.

10.2.2.1 Using Polar Coordinates to Calculate Double Integrals

Suppose the region lies as shown in Figure 10.17. Calculate the area of the region
D such that α ≤ θ ≤ β and r1(θ) ≤ r ≤ r 2(θ). Again using the formula for general
area gives

r = r2 (θ)

β r = r1 (θ)

α
x

Figure 10.17 Area bounded by polar coordinates.

 10.2 Higher-Order Integration 275

∫∫ f (x, y) dA  lim ∑
D
∆A→0
all pieces
f ( x*, y*)∆A

= lim lim
∆r →0 ∆θ →0 ∑ ∑ f (r* cosθ*, r* sinθ* ) r∆r∆θ
i j
i j i j

The r in r∆r∆θ is needed here for the change of variable (see later section defin-
ing Jacobian).

β r2

Area =
∫ ∫ f (r cosθ , r sinθ )r dr dθ
α r1
(10.15)

Example 10.16
Consider the disc x2 + y2 ≤ 4 has charge density σ(x,y) = 3x + x2 + y2.
Compute the total charge on the disc. The diagram of this region is shown
in Figure 10.18.

Solution: The area is given by formula

A=
∫∫ (3x + x
D
2
)
+ y 2 dA

You could use traditional Cartesian coordinates x and y, but it is easier

here to use polar coordinates r and θ.
Changing to polar coordinates, x = r cos θ, y = r sin θ, and r 2 = x2 + y2
with dA = r dr dθ.

0 x
–2 2

–2

Figure 10.18 Disc with density σ(x,y) and radius 2.

276 Multivariable Calculus

Also, the limits of integration now change to 0 ≤ θ ≤ 2π and 0 ≤ r ≤ 2.

2π 2

∫ ∫ (3r cosθ + r )r dr dθ
0 0
2

Multiplying by r and integrating while keeping θ constant gives

2π 2
 3 1 4
=
∫
0
r cos θ + 4 r  dθ
0

2π

=
∫ [8 cosθ + 4] dθ = 8π
0

Since the integral of both the cos θ and sin θ over a period are both equal
to zero.

Example 10.17

Find the volume of the “snow cone” bounded by z = x 2 + y 2 and x2 + y2 + z2 = 4.

Figure 10.19 shows the diagram of the region bounded.

Solution:

Volume =
∫∫ Upper surface − ∫∫ Lower surface
z

D
2
2

Figure 10.19 The snow cone bounded region.

 10.2 Higher-Order Integration 277

Upper surface is the sphere z = 4 − x 2 − y 2.

Lower surface is the cone z = x 2 + y 2.

Therefore,

Volume =
∫∫ (D
4 − x 2 − y2 − )
x 2 + y 2 dA (10.16)

The region D is a circle, which is the intersection of the sphere with the
cone as shown in Figure 10.20.

x 2 + y 2 + z 2 = 4 with z = x 2 + y2

This gives

( 2)
2
x 2 + y 2 = 2 or x 2 + y2 =

In polar coordinates Equation 10.16 becomes

2π 2

∫ ∫(
0 0
)
4 − r 2 − r r dr dθ

2π 2

∫ ∫ (r
0 0
)
4 − r 2 − r 2 dr dθ

Which is just a standard integral with respect to r first then with respect
to θ giving the volume as

2π 2 2π  5 
 1 3
1 3 8 − 2
∫ ∫
2
 − (4 − r ) 2 − r  dθ =
2   dθ
 
 3 3 0 3 
0 0

0 x
√2

Figure 10.20 The circle that is the intersection of the sphere and cone.
278 Multivariable Calculus

Volume =
2π
3
(
8 − 22
5
)
10.2.3 General Regions
Previously, calculations of double integrals over a rectangular region were consid-
ered as shown in Figure 10.21.
The area of the region D is given by

b d d b

∫∫
D
f ( x , y) dA =
∫∫
a c
f ( x , y) dy dx =
∫ ∫ f (x, y) dx dy
c a

Suppose the region D has the general form as shown in Figure 10.22.
The area of the region D can be found as

b g2 ( x )

∫∫ f (x, y) dA = ∫ ∫
D a g1 ( x )
f ( x , y) dy dx

x
a b

Figure 10.21 Rectangular region as area.

y = g2(x)

y = g1(x)
D

x
a b

Figure 10.22 General region with area D.

 10.2 Higher-Order Integration 279

y
x = g1(y) x = g2(y)

D
c

Figure 10.23 General region with area D.

Again here integrate out y first then x.

Also, if D has the following form as in Figure 10.23.

Again this is given by
d g2 ( y )

∫∫ f (x, y) dA = ∫ ∫
D c g1 ( y )
f ( x , y) dy dx

Here integrate out x first then y to find the required area.

Example 10.18
Integrate f(x,y) = 4xy – x2 over the region D given in Figure 10.24.

Solution: Using the definition gives

1 1− x

∫∫ f (x, y) dA = ∫ ∫ (4 xy − x ) dy dx
D 0 0
2

Note: Usually, the limits are the most difficult part of the integral to sort
out correctly.

1 y = –x + 1

0 x
1

Figure 10.24 Triangular region D.

280 Multivariable Calculus

Here if x limits are from 0 to 1 then the y limits are from 0 to the line y = 1 − x.
1 1

∫ ∫ 2x(1 − x) − x (1 − x) dx
1− x
=  2 xy 2 − x 2 y  dx = 2 2
0
0 0

1 1
 5x 3 3x 4  1
=
∫
0
(2 x − 5 x + 3 x ) dx =  x 2 −
2

3
3
+  =
4  0 12

Example 10.19
Find the volume enclosed by the paraboloid z = x2 + 3y2 and the planes
x = 0, y = 1, y = x, and z = 0 shown in Figure 10.25.

Solution: The volume below the surface within the region D is required. Again
if the limits for x are from 0 to 1, then the limits for y are from 0 to y = x.
1 1

∫∫
D
( x 2 + 3 y 2 ) dA =
∫ ∫ (x
0 x
2
+ 3 y 2 ) dy dx

1 1

∫ ∫ (x
1
=  x 2 y + y3  dx = 2
+ 1) − 2 x 3  dx
x
0 0
1
 x3 x4  5
= +x−  =
3 2 0 6

The next section looks at triple integrals and some of their applications.

y=1

1
D
1
y

x y=x

Figure 10.25 The volume enclosed under the paraboloid.

 10.2 Higher-Order Integration 281

10.2.4 Triple Integrals
Consider a 3-D region E in space as shown in Figure 10.26.
What is meant by triple integration? This is exactly the same concept as double
integration, that is, splitting the region into smaller volumes and summing.

∫∫∫ f (x, y, z) dV  lim ∑

E
∆V →0
all pieces
f ( x*, y*, z*)∆V (10.17)

Again, there are similar applications for triple integrals as double integrals.

Vol (E ) =
∫∫∫ 1. dV
E
with f ( x*, y*, z*) = 1

Mass(E ) =
∫∫∫ d (x, y, z) dV
E

In 2-D space the double integrals were calculated using either Cartesian or
polar coordinates. These concepts can be extended to 3-D space.
To calculate triple integrals in 3-D space, different coordinate systems can
be used, including Cartesian, cylindrical, or spherical coordinates depending on
which will make the integration easier to compute.
The next section starts with evaluating triple integrals in Cartesian coordinates.

10.2.4.1 Cartesian Coordinates
Express the triple integrals as

∫∫∫ f (x, y, z) dV = ∫∫∫ f (x, y, z) dz dy dx = ∫∫∫ f (x, y, z) dy dx dz

E Limits Limits

Split into small volumes

Figure 10.26 3-D volume enclosed by a region.

282 Multivariable Calculus

This integral can be done six different ways since the order of the integration
can be taken with respect to the different variables x, y, or z as required.

Example 10.20
Let E be the region shown in Figure 10.27 which is bounded by the planes
x + y + z = 1, x = 0, y = 0 and z = 0. Suppose E has mass density d(x,y,z) = x2y.
Compute the mass of E of the region bounded by the above planes.

Solution: See Figure 10.27.

Using the formula for the mass and replacing density d(x,y,z) = x2y gives

Line in x-y plane Plane z = 1 – x – y

1 1–x1–x–y
Mass (E) = ∫∫∫ x. dV = ∫∫ ∫ x2y dz dy dx
E 0 0 0

Integrate out with respect to z first gives

1 1− x 1 1− x

=
∫ ∫ x y[z]
0 0
2 1− x − y
0 dy dx =
∫ ∫ x y(1 − x − y) dy dx
0 0
2

Integrate out with respect to y gives

1 1− x
 x 2 y 2 x 3 y 2 x 2 y3 
=
∫
0

 2
−
2
−
3 0
 dx

y
z
y=1–x
1

1 x
(0, 0, 1)
(b)
E

(0, 1, 0)

(1, 0, 0) y

x
(a)

Figure 10.27 (a) The bounded volume E. (b) The triangular region in the x-y plane.
 10.2 Higher-Order Integration 283

1
1 1 1 
=
∫  2 x (1 − x) − 2 x (1 − x) − 3 x (1 − x)  dx
0
2 2 3 2 2 3

1
1 1 3 1 4 1 5
=
∫  6 x
0
2
−
2
x + x − x  dx
2 6 

This is worked out using single-variable integration with respect to x and

1
the answer turns out to be 360
.

10.2.5 3-D Coordinate Systems

In 3-D space the following coordinate systems can be used: Cartesian, cylindri-
cal, and spherical coordinates. First considering cylindrical coordinates (θ, r, z)
as shown in Figure 10.28. This is basically using polar coordinates plus the z
coordinate. Relationships are given by

x = r cos θ

y = r sin θ

r = x 2 + y2

z=z

θ
r
y

Figure 10.28 General point P in space in terms of cylindrical coordinates.

Now consider spherical coordinates (θ, φ, ρ). Here there are two angles and
a distance to travel from the origin as shown in Figure 10.29. Relationships are
given by the following
284 Multivariable Calculus

z P

φ
ρ
z

θ
r
y

Figure 10.29 General point P in space in terms of spherical coordinates.

z = ρ cos ϕ

r = ρ sin ϕ

x = ρ sin ϕ cos θ

y = ρ sin ϕ sin θ

ρ= x 2 + y2 + z 2

10.2.5.1 Integrals in the New Coordinate Systems

As with the 2-D case when Cartesian coordinates were changed to polar coordi-
nates, there was an introduction of the extra term known as the Jacobian and a
similar term appears again in the 3-D change of coordinates as shown next.
Cylindrical:
Jacobian

∫∫∫ f (x, y, z) dV =
∫∫∫
Limits!
f (r cos θ, r sin θ, z) r dz dr dθ

E
Jacobian
Spherical:

∫∫∫ f (x, y, z) dV =
∫∫∫
Limits!
f (ρ sin φ cos θ, ρ sin φ sin θ, ρ cos φ) ρ2 sin φ dρ dφ dθ
E
These formulae can be used to find and prove the volumes of standard shapes.

Example 10.21
Find the volume of a sphere (E) of radius r and center the origin as shown
in Figure 10.30.
 10.2 Higher-Order Integration 285

r
r

Figure 10.30 Sphere of radius r and center origin.

Solution: Volume is given by the formula

Vol ( E ) =
∫ ∫ ∫ 1.dv
E

Using spherical coordinates gives

2π π r

=
∫ ∫ ∫ 1ρ sin ϕ d ρ dϕ dθ
0 0 0
2

Integrate out w.r.t dρ gives

2π π r 2π π
 ρ3  r3
=
∫∫
0 0
 sin ϕ  d ϕ dθ =
 3 0
∫∫
0 0
3
sin ϕ d ϕ dθ

Integrate out w.r.t dφ gives

2π π 2π
 r3  2r 3
=
∫
0
 − cos ϕ  dθ =
 3 0
∫ 0
3
dθ

Integrate out w.r.t dθ gives

2r 3 4
= × 2π = π r 3
3 3
So,
4 3
Vol (E ) = πr
3
This is just the usual expression for the volume of a sphere of radius r.
286 Multivariable Calculus

10.2.6 General Change of Coordinate Systems

It can be important in some applications in engineering to be able to change to a
different coordinate system to enable the integration to be easily carried out. The
next section shows how this is done in general, starting with the one-variable case.
For the one-variable case, let x = g(u)

b g −1 ( b ) upper limit

∫ f (x) dx = ∫
a −1
g (a)
f  g(u)  g′(u) du =
∫
lower limit
f  g(u)  g′(u) du

For the two-variable case, making a change of variables to u and v gives the
transformation shown in Figure 10.31:

∫∫ f(x, y) dy dx = ∫∫ f [x (u, v), y(u, v)]

∂ (x, y)
∂ (u, v)
dvdu
D T (D)
Jacobian

where

 ∂x ∂y 
∂( x , y)  
∂u ∂u
= det  
∂(u, v)  ∂x ∂y 
 ∂v ∂v 

The idea behind this coordinate transformation is shown in Figure 10.32.

y v

T
D
T(D)

x u

Figure 10.31 How a region D in the x-y space gets transformed into the u-v space.

v y
<xv dv, yv dv>

T
(u,v + dv) (u + du, v + dv)
<xu du, yu du>
(x(u,v), y(u,v))
(u, v) (u + du, v)
u x

Old Area = dv du New area

Figure 10.32 Area gets mapped to a new area under a general transformation.
 10.2 Higher-Order Integration 287

Using the Taylor series expansion gives

 x (u + du, v) ≈ x (u, v) + du xu (u, v)


 y(u + du, v) ≈ y(u, v) + du yu (u, v)

 x (u, v + dv) ≈ x (u, v) + dv x v (u, v)
 y(u, v + dv) ≈ y(u, v) + dv yv (u, v)


The new area is

 x du yu du 
= det  u 
 x v dv yv dv 

x yu 
= det  u  dvdu
 x v yv 

∂( x , y)
= dvdu
∂(u, v)

Example 10.22

Evaluate
∫∫ (64xy) dA, where D is the parallelogram with vertices, (–1,3),
D
(1,–3), (3,–1), and (1,5) as shown in Figure 10.33.

Solution: The region is complicated using just the x and y coordinates sys-
tem. So changing variables can make things easier. Let

u = y− x

v = y + 3x

y v
y =x+4 y = –3x + 8 T 8

T(D)

x u
–4 0 4

y = –3x
y =x– 4

Figure 10.33 Area transformation with variable change.

288 Multivariable Calculus

These can be solved to find x and y and give

1 1
x= ( v − u) and y = (3u + v)
4 4

So in this change of coordinate system of u and v gives a nice rectangular

region in u – v space.
The transformation mapping becomes

y=x–4 maps to u = –4
y=x+4 maps to u=4
y = –3x + 8 maps to v=8
y = –3x maps to v=0

Now the change of variables gives

4 8
1 1  ∂( x , y)
∫∫ (64 xy) dA = ∫ ∫ 64  4 (v − u)  4 (3u + v) ∂(u, v) dv du
D −4 0

So

 ∂x ∂x   1 1 
∂( x , y)    −  1
∂u ∂v  = det  4 4  = −
= det 
∂(u, v)  ∂y ∂y   3 1  4
 ∂u ∂v   4 4 

Now the integral becomes

4 8
1
∫ ∫ 4( v
−4 0
2
+ 2uv − 3u 2 ) −
4
dv du

4 8 4
 v3   512  1024
=
∫
−4
 + uv − 3u v  du =
3
2 2

0
∫ 
−4
3
+ 64u − 24u 2  du =
 3

Note: Sometimes it is easier to use the following to calculate

∂( x , y) 1
=
∂(u, v) ∂(u, v)
∂( x , y)

The above analysis can be further generalized to a three-variable case

as follows:
 10.3 Applications 289

∂( x , y, z )
∫∫∫ f (x, y, z) dz dy dx = ∫∫∫ f ( x(u, v, w)……) ∂(u, v, w) dw dv du
E

with x = x(u,v,w), y = y(u,v,w), z = z(u,v,w), where

 x x x 
u v w
∂( x , y, z )  
= det  yu yv yw  .
∂(u, v, w)  z z z 
 u v w 

So cylindrical and spherical coordinates are examples of three-vari-

able problems and their Jacobians can be found using the aforementioned
formula.

10.3 Applications
There are many physical applications of double and triple integration, most of
which depend on the idea of splitting a region into smaller pieces and summing
over all the pieces. Start by looking at applications of double integrals, which
shows the principle involved.

10.3.1 Application of Double Integration

Remember what is meant by the double integration (see Figure 10.34). The defini-
tion of the double integral is

∫∫ f (x, y) dA  lim ∑
D
∆A →0
all pieces
f ( x*, y*)∆A

The region is chopped into pieces. Then pick a point (x*, y*) and sum over all
the pieces with f(x*, y*) multiplied by the width of all the pieces ΔA.
This forms the basis of all applications as shown in the next few examples.

(x*, y*)

∆A

Figure 10.34 General region in the x-y plane.

290 Multivariable Calculus

Example 10.23: Volume under a Surface

Find the volume under a surface z = f(x,y) shown in Figure 10.35.

Solution: The approximate volume under the single column is f(x*,y*)∆A.

Then the total volume is given by

Volume ≈ lim
∆A→0 ∑
all pieces
f ( x*, y*)∆A

Volume =
∫∫ f (x, y) dA
D
(10.18)

z
z = f (x, y)

D y

x ∆A

Figure 10.35 Volume under a surface.

Example 10.24: Area of a Bounded Region

Find the area of a region D shown in Figure 10.36.

D
(x*, y*)

∆A

Figure 10.36 Area of a region.

 10.3 Applications 291

Solution: Here, if f(x*,y*) = 1, then one has just the area of the small piece
and then summing gives

Area ≈ lim
∆A→0 ∑ 1.∆A
all pieces

Area =
∫∫ 1dA
D
(10.19)

Example 10.25: Mass of a Region

Find the mass of a region D shown in Figure 10.37, given that the mass
density is d(x,y) (mass/area).

Solution:

Mass ≈ lim
∆A→0 ∑
all pieces
d ( x*, y*)∆A

Mass =
∫∫ d (x, y) dA
D
(10.20)

Flat metal sheet with

different densities along it,
d (x, y).

D
(x*, y*)

∆A

Figure 10.37 Mass of a region.

Example 10.26: Radiative Heat Transfer between Surfaces Using

View Factors
Calculating the radiative heat transfer between different surfaces is an
important area in many fields of engineering. In fire engineering, it can
be an important consideration when calculating the radiative heat transfer
from one building surface to nearby buildings to see if nearby buildings are
in danger of catching fire.
Consider two finite surface areas Ai and Aj with small infinitesimal
areas dAi and dAj with normal vectors ni and nj, respectively, as shown in
Figure 10.38. The view factor between two finite areas Ai and Aj is denoted
292 Multivariable Calculus

nj

θj
dAj

ni Aj

θi

dAi

Ai

Figure 10.38 View factors between two finite surfaces.

by Fij and defined as the fraction of the radiation leaving the surface i that is
intercepted then by the surface j. This can be represented as

[ Intensity leaving i and hitting j]

Fij =
Total inttensity leaving i 

The mathematical formula for this view factor needs some work to
derive it and is given as

1 cos θ i cos θ j
Fij =
Ai ∫∫
Ai A j
π R2
dA j dAi (10.21)

Essentially, this double integral arises by taking a particular dAi and

calculating its contributions to each of the dAj and then summing all these
contributions, which is the inner integral. Then the process is repeated for
all the dAi and summing, which is then the outer integral. This then gives
the total contribution from surface i to surface j.
From a practical basis, this can involve lots of computational work and
some simplifications can be made using view factor algebra to avoid using
Equation 10.21. Simplifications using properties of reciprocity and summa-
tion can help in calculating view factors for multisurfaces.

10.3.2 Application of Triple Integration (Center of Mass)

The region shown in Figure 10.39 has a mass density of d(x,y,z). Then the mass
of the region is given by

Mass(E ) = M =
∫∫∫ d (x, y, z) dV
E
(10.22)
 10.3 Applications 293

Figure 10.39 Center of mass of a region.

Therefore, the center of mass has position ( x , y , z ), where

1
x=
M ∫∫∫ x d (x, y, z) dV
E
(10.23)

1
y=
M ∫∫∫ y d (x, y, z) dV
E
(10.24)

1
z=
M ∫∫∫ z d (x, y, z) dV
E
(10.25)

Example 10.27: Center of mass of a body

Compute the center of mass of the region E shown in Figure 10.40 enclosed
by z = 1 – x2 – y2 and z = 0. Assume a constant mass density d = 1.

Solution: See Figure 10.40. By radial symmetry, the center of mass must be
on the z axis with more mass lower down so it should be closer to the z = 0
plane. By symmetry, x = 0 and y = 0 are already known. Now to work out
z use the formula given by Equation 10.25 with d = 1 gives

1 1
z=
M ∫∫∫ z d (x, y, z) dV = M ∫∫∫ zd dV
E E
294 Multivariable Calculus

z = 1− x2 − y2
1

1
D
1
y

Figure 10.40 Region enclosed by the surfaces.

First, work out the mass M of the region using the mass formula given
by Equation 10.22:

M=
∫∫∫ 1dV
E
1− x 2 − y 2

=
∫∫ ∫
D 0
1 dz dy dx

=
∫∫ (1 − x
D
2
− y 2 ) dy dx

Now D is the circle of radius 1 in the x-y plane. So changing to polar

coordinates gives

2π 1
π
∫ ∫ (1 − r ) r dr dθ = 2
0 0
2

This is the expression for mass M.

 Problems 295

For the center of mass calculation, use Equation 10.25 to give

1
z=
M ∫∫∫ zd dV
E
1− x 2 − y 2
2
=
π ∫∫ ∫
D 0
z dz dy dx

2 1
=
π ∫∫ 2 (1 − x
D
2
− y 2 )2 dy dx

Changing to polar coordinates with r 2 = x2 + y2 gives

2π 1 2π 1
1 1  1 2 3
=
π ∫∫
0 0
(1 − r ) r dr dθ =
2 2
π ∫0
 − 6 (1 − r )  dθ
0

2π
1 1 1
=
π ∫ 6 dθ = 3
0

 1
∴ center of mass is at ( x , y , z ) =  0, 0,  .
 3

Problems
10.1 Given that f(x,y) = x3y + y3, evaluate the following.

∂f ∂f
a. b.
∂x (1,1) ∂y ( 2,1)

10.2 Given that f(x,y) = x sin(xy), evaluate the following.

a. fx b. f y c. fxy d. f yx

10.3 Given that f(x,y) = xy2 –5xy, calculate the directional derivative of f in
the direction of 〈3,1〉 at the point (1,1).

10.4 Evaluate the following double integral:

3 1

∫ ∫ (x y − 5x) dy dx
0 0
2
296 Multivariable Calculus

10.5 Find the volume of the solid tetrahedron enclosed by the plane
2x + y + z = 4 and the coordinate planes.

10.6 In Figure 10.41, a cylinder of height H and radius R is shown. Show that
the volume of the cylinder is given by the formula V = πR2 H.

R
R

Figure 10.41 A general cylinder of height H and radius R.

 11 Vector Calculus

11.1 Differentiation and Integration of Vectors

Vector calculus deals with the differentiation and integration of vector fields in
two- and three-dimensional space. In this chapter important concepts such as
the different types of line integrals are first considered and then how closed line
integrals can be equivalent to double integrals over a region known as Green’s
theorem. Further important ideas on the gradient and curl of vector fields are
developed leading to surface integrals and their applications to describe fluid flow
and many different force fields that occur naturally in the world.

11.1.1 Derivatives of Vector Functions

A curve is defined in 3-D space in parametric form as r (t ), where

r (t ) = 〈 f (t ), g(t ), h(t )〉

Note: ‹u,v,w› represents a general three-dimensional vector.

The derivative of r(t) with respect to time t is defined as r ′(t ) and is given by

r (t + ∆t ) − r (t )
r ′(t ) = lim (11.1)
∆t →0 ∆t
This is a tangent vector to a space curve. Also, this can be thought of as the instan-
taneous velocity, as shown in Figure 11.1.
As, Δt → 0 the vector PQ tends to the tangent vector at the point P.
  f (t + ∆t ) i + g(t + ∆t ) j + h(t + ∆t ) k  −  f (t ) i + g(t ) j + h(t ) k  
= lim     
∆t → 0  ∆t 

297
298 Vector Calculus

r(t + ∆t) − r(t)

r(t)
r(t + ∆t)

Figure 11.1 The tangent vector to a curve.

   f (t + ∆t ) − f (t )  i +  g(t + ∆t ) − g(t )  j + [ h(t + ∆t ) − h(t ) ] k  

= lim   
∆t →0 
 ∆t 

  f (t + ∆t ) − f (t )  i  g(t + ∆t ) − g(t )  j [ h(t + ∆t ) − h(t ) ] k 

= lim   + + 
∆t →0  ∆t ∆t ∆t 

So,

r ′(t ) = f ′(t ) i + g′(t ) j + h′(t ) k (11.2)

Therefore, to find the derivative of r(t) with respect to time t, just differentiate
each of the components of r(t), that is, given that

r (t ) = 〈 f (t ), g(t ), h(t )〉

then

r ′(t ) = 〈 f ′(t ), g′(t ), h′(t )〉.

Example 11.1

If r (t ) = 〈t , t 2 , t 3 〉, then

r ′(t ) = 〈 f ′(t ), g′(t ), h′(t )〉 = 〈1, 2t , 3t 2 〉

This is the directional vector of the tangent line space curve at any point
t for the vector function r (t ).
 11.1 Differentiation and Integration of Vectors 299

Also differentiating a second time gives

r ′′(t ) = 〈 f ′′(t ), g′′(t ), h′′(t )〉 = 〈0, 2, 6t 〉

11.1.2 Integrating Vector Functions

Integrals or antiderivatives work like normal integrals do as shown in the next
few examples.

Example 11.2

∫ (t i + t
0
2
)
j + t 3 k dt

1
1 1 1 
=  t2 i + t3 j + t4k 
2 3 4 0
1 1 1 1 1 1
= i + j + k or 〈 , , 〉
2 3 4 2 3 4

Example 11.3

1 3t  t3 e 3t
∫  i + t j + e k  dt = ln t i + j +
t
2
3 3
k+a

(
Where a is a constant vector, that is, a = c1 i + c2 j + c3 k . )
Example 11.4

Given that r (t ) = 2 i + 4tj − 6t 2 k and r (0) = j + k , find r (t ) .

Solution:

r (t ) =
∫ ( 2 i + 4tj − 6t k ) dt 2

r (t ) = 2t i + 2t 2 j − 2t 3 k + c

where c is a constant vector.

r (0) = 0 + 0 + 0 + c = j + k

Therefore,

r (t ) = 2t i + 2t 2 j − 2t 3 k + j + k or r (t ) = 2t i + (2t 2 + 1) j + (1 − 2t 3 )k
300 Vector Calculus

11.2 Vector Fields
Previously, the functions had different input variables but only one output variable.
For example,

f(x) 1 input, 1 output

f(x, y) 2 inputs, 1 output

f(x, y, z) 3 inputs, 1 output

Now, looking at vector fields, these can be represented as follows:

F ( x , y) = 〈 M ( x , y), N ( x , y)〉 2 inputs, 2 outputs

F ( x , y, z ) = 〈 M ( x , y, z ), N ( x , y, z ), P( x , y, z )〉 3 inputs, 3 outputs

a vector field is a more general function. So to every point in the x-y plane or 3-D
space it assigns a vector.
Some simple vector fields can be seen by graphing them as follows.

Example 11.5
What does the vector field F ( x , y) = 〈 x , y〉 look like? See Figure 11.2.

What is the magnitude of the vector field, that is, F ( x , y) ? The magni-
tude of a vector is given by the square root of all the components squared
and added together. Therefore,

(x, y)

1 2 x

Look at points and see what vector is given. Take a general point (x, y)
in space and this vector from the origin is the vector you put at the
location (x, y).

Figure 11.2 The vector field for F ( x , y) = 〈 x , y 〉.

 11.2 Vector Fields 301

F ( x , y) = x 2 + y 2 = r

So the magnitude of the vector field at any position is just the dis-
tance from the origin to that point, pointing radially outward as shown
in Figure 11.2.

Example 11.6
Sketch the vector field given by

〈 x , y〉
F ( x , y) =
x 2 + y2

which is not defined at (0,0).

Solution: Again this is the same vector field as in Example 11.5 but now the
magnitude is always equal to unity as shown in Figure 11.3 (the arrows are
all of length 1 unit).
The magnitude of the vector field is given by

〈 x , y〉 x 2 + y2
F ( x , y) = = =1
x 2 + y2 x 2 + y2

1 2 x

〈 x , y〉
Figure 11.3 The vector field for F ( x , y) = .
x 2 + y2

Example 11.7
Sketch the vector field given by, F ( x , y) = 〈 y, 0 〉.
302 Vector Calculus

Solution: This vector field shown in Figure 11.4 is known as shear flow in
fluid dynamics and models flow of water near boundaries.
y

Figure 11.4 The vector field for F ( x , y) = 〈 y, 0 〉.

Example 11.8
Consider the vector force field for gravity. For gravitational attraction the
magnitude of the force varies as reciprocal of the distance squared, that is,

1
Force of gravity = F ( x , y, z ) ~ (distance squared)
r2

It turns out that the vector field for the force due to gravity can be arrived
at by starting with the basic vector field F ( x , y, z ) = 〈 x , y, z 〉, multiplying by
a constant and a negative sign, and making the modulus proportional to the
reciprocal of the distance squared.
The sketch is shown in Figure 11.5.
z

Figure 11.5 The vector field for gravitational attraction force.

 11.3 Line Integrals 303

The vector field representation is given as

− k 〈 x , y, z 〉
F ( x , y) = 3
( x 2 + y2 + z 2 ) 2

11.3 Line Integrals
The basic idea of a line integral is to integrate along a curve C in 2-D or 3-D, as
shown in Figure 11.6.

11.3.1 The ds-Type Integral

∫ f (x, y) ds = lim ∑
C
∆s → 0
all pieces
f ( x*, y*)∆s

where the Δs is the length of a small piece.

Applications of the ds-type integral include the following:

1. Length of the curve is given by

Length = 1.ds
∫
C
(11.3)

2. If d(x, y) is the mass density (mass/length) of the wire, then the total mass
of the wire is given by

Mass of wire =
∫ d (x, y) ds
C
(11.4)

Note: These formulae can be used to calculate how long or heavy are the steel
cables needed in the design of suspension bridges.

∆s

Figure 11.6 The line integral in 2-D space.

304 Vector Calculus

To compute the ds-type of integral, first, the curve C, as shown in Figure 11.7,
needs to be parameterized.
Here, x = x(t), y = y(t), and a ≤ t ≤ b. Then ds can be found using the Pythagorean
theorem as

ds = dx 2 + dy 2

Then

∫ f (x, y) ds = ∫ f (x, y)
C C
dx 2 + dy 2

From which the integral in parametric form becomes

b 2 2
 dx   dy 
∫
C
f ( x , y) ds =
∫
a
f ( x (t ), y(t ))   +   dt
 dt   dt 
(11.5)

∆s ds
dy
dx
C

Figure 11.7 Parameterization of the curve.

Example 11.9
If the density of a wire is d(x,y) = x, compute its mass when the wire is the
quarter circle of radius 2 as shown in Figure 11.8.

Solution: The mass is given by the formula

Mass =
∫ d (x, y) ds = ∫ x ds
C C
(11.6)

To parameterize the curve, since the wire is in the form of a circle, then
the natural parameter is in terms of the angle made, that is, t.

x (t ) = 2 cos t x (t ) = −2 sin t

y(t ) = 2 sin t y (t ) = 2 cos t

 11.3 Line Integrals 305

2
C

x
2

Figure 11.8 A wire in the form of a quarter circle.

π
0≤t ≤
2

The mass is given by Equation 11.6 as

π
2

Mass =
∫ 2 cos t
0
(−2 sin t )2 + (2 cos t )2 dt

π
2 π
=
∫ 0
4 cos t dt = 4[sin t ]02 = 4

11.3.2 The dr − Type Integral

Here a vector field acts in the region given by F ( x , y). Again, the line is split into
small segments as shown in Figure 11.9.
The component of the vector field in the direction of the line is given by
F ( x*, y*).∆r , so summing along the whole line gives

∫ F (x, y). dr = lim ∑ F (x*, y*).∆r

C
∆r → 0
all pieces
(11.7)

A useful application of this type of dr integral is the calculation of the work

done along a curve through a vector field.

So, if F ( x , y) is a force field, then

Work =
∫ F. dr
C
(11.8)
306 Vector Calculus

is the work done in moving along the curve.

To compute this d r -type integral, one has to parameterize the curve C. Again
Figure 11.10 shows a path through a vector field.
Let x = x (t), y = y (t), a ≤ t ≤ b, and d r = 〈 dx , dy〉. Then to calculate the work
done with F ( x , y) = 〈M ( x , y), N ( x , y)〉 gives

=
∫ F (x, y).d r
C

=
∫ M (x, y) dx + N (x, y) dy
C

b
 dx dy 
=
∫  M (x(t), y(t)) dt + N (x(t), y(t)) dt  dt
a (11.9)

Vector field F(x, y) = M(x, y), N(x, y)

∆r

Figure 11.9 A path traveled through a vector field.

y
Vector field F(x, y) = M(x, y), N(x, y)

dr
dy

C dx

Figure 11.10 Parameterization of the path and vector field.

 11.3 Line Integrals 307

Example 11.10

If there is a force field F ( x , y) = 〈 x 2, y 〉, a particle travels along a curve C

(i.e., y = –x2 + 1) as shown in Figure 11.11.
So, the particle moves along the curve C to find the work done using
Equation 11.8 gives

Work =
∫ F. d r
C

=
∫ x dx + y dy
C
2

Now to parameterize the curve C, one way is to let

dx
x=t =1
dt
dy
y = −t 2 + 1 = −2t
dt
y

C y = –x2 + 1

x
–1 1

Fgure 11.11 Path traveled by the particle.

And then let t be between –1 ≤ t ≤ 1.

Using Equation 11.9 gives
b
 dx dy 
∫ F (x, y).d r = ∫  M (x(t), y(t)) dt + N (x(t), y(t)) dt  dt
C a

=
∫ (t) (1) + (−t
−1
2 2
+ 1)(−2t )  dt

∫ [ 2t + t 2 − 2t ] dt =
2
= 3
3
−1
308 Vector Calculus

11.3.3 Summary of Results
A general path in 3-D space is shown in Figure 11.12.
The two types of line integrals can be summarized by the formulae as follows:

ds–type:

b 2 2 2
 dx   dy   dz 
∫
C
f ( x , y, z ) ds =
∫a
f ( x (t ), y(t ), z (t ))   +   +   dt
 dt   dt   dt 

dr − type:

b
 dx dy dz 
∫
C
F ( x , y, z ).d r =
∫  M (x(t), y(t), z(t)) dt + N (x(t), y(t), z(t)) dt + P(x(t), y(t), z(t)) dt  dt
a

x = x(t)
y = y(t)
z = z(t)
a≤t≤b
C

Figure 11.12 A general path in space.

11.4 Gradient Fieds
A special type of vector field is known as a gradient field.
Given f(x,y), defining a vector field, the F ( x , y) = ∇f
∇f is called its “gradient field.”

Example 11.11
1. Given the function f(x,y) = xy + y3, the gradient field is

F ( x , y) = ∇f = 〈 f x , f y 〉 = 〈 y, x + 3 y 2 〉
 11.4 Gradient Fieds 309

1
2. Given the function f ( x , y) = ( y − x 2 ), then the gradient field is
2
given by

1
F ( x , y) = ∇f = 〈 f x , f y 〉 = 〈− x , 〉
2

1
Now f ( x , y) = ( y − x 2 ). The level curves for this function, that is,
2

1
f =0= (y − x2) y = x2
2

1
f =1= (y − x2) y = x2 + 2
2

1
f = −1 = (y − x2) y = x2 − 2
2

and so on as shown in Figure 11.13. These show the level curves of f(x,y).
Also the gradient vector ∇f is always perpendicular to the level curves
pointing to the greatest increase.

1
∇f = 〈 f x , f y 〉 = 〈− x , 〉
2

is shown with arrow heads.

f=1

f=0

f = –1
2

–2

Figure 11.13 The level curves for the function.

310 Vector Calculus

11.4.1 Conservative Vector Fields

Why are gradient fields of importance? If F ( x , y) = ∇f for some vector field f,

then F ( x , y) is a conservative vector field.
In Figure 11.14 there is a path C within a vector field F ( x , y) . If F ( x , y) = ∇f ,
then the fundamental theorem of line integral can be expressed by

∫ F.d r = f (end) − f (start)

C
(11.10)

This is analogous to the fundamental theorem of calculus.

y
Vector field F(x, y)

End
Start
C

Figure 11.14 A path given within a vector field.

Example 11.12

Consider the vector field F ( x , y) = 〈2 x + 3 y, 3 x 〉. Compute

∫ F (x, y). d r
C

where C is the curve given in Figure 11.15.

This could be tackled in the usual way. Work out

∫ F. d r
C

using parameterization of the curve as before.

But since it is known that F ( x , y) = ∇f , for f(x, y) = x2 + 3xy, this implies
that F is a conservative vector field. It is easier to use the fundamental
theorem of line integrals, which gives

∫ F.d r = f (end) − f (start) = f (3, 0) − f (0, 9)

C

= (32 + 0) − (0 2 + 0) = 9
 11.4 Gradient Fieds 311

y = 9 – x2

x
3

Figure 11.15 The path is the curve y = 9 – x2.

Following are some important properties of the fundamental theorem of

line integrals.
Property 1: If F is conservative, and C and C have the same start and
end point as shown in Figure 11.16, then it can be shown that

∫ F. d r = ∫ F. d r
C C
(11.11)

This is called path independence for conservative vector fields.

C End

Start

Figure 11.16 Two paths with the same start and end points.

Note: Systems that are reversible are conservative and those that are non-
conservative in nature mean that energy is lost during the process as waste
or to entropy.
312 Vector Calculus

Figure 11.17 A closed curve with the same start and end points.

Property 2: If C is a closed curve (a curve that comes back on its self)

as shown in Figure 11.17, then it can be shown that the integral around the
closed path is

∫ F. d r = 0
C
(conservativeness) (11.12)

Can it be said that all vector fields are conservative? No, not all vector fields
are conservative. Consider the following vector field (also see Figure 11.18)
〈− y, x 〉
F ( x , y) =
x 2 + y2

〈− y, x 〉
Figure 11.18 The vector field given by F ( x , y) = .
x 2 + y2
 11.4 Gradient Fieds 313

This is not a conservative vector field since clearly

∫ F. d r > 0
C

11.4.2 Testing for Conservativeness

So, if

F ( x , y) = 〈 M ( x , y), N ( x , y)〉 = ∇f

then we need to find some function f such that we have the following as shown in
Figure 11.19.

∂ ∂
∂x ∂y

M N

∂ ∂
∂y ∂x

Same

Figure 11.19 Showing that the order of differentiating does not matter.

Example 11.13

Let F ( x , y) = 〈 M ( x , y), N ( x , y)〉 = 〈 x 2 + y, y〉 be the vector field. So,

∂N
= Nx = 0
∂x

and

∂M
= My = 1
∂y

These are not equal so the vector field F ( x , y) is not conservative.

Example 11.14

Is F ( x , y) = 〈3 x 2 + y, x 〉 a conservative vector field?

Check the first partial derivative tests: Nx = 1 and My = 1. These are nec-
essary but not sufficient conditions. This is a good sign that maybe

F ( x , y) = ∇f
314 Vector Calculus

To see if this is the case integrating the functions as follows: fx = 3x2 + y

with respect to x gives, f = x3 + xy + g(y) and f y = x with respect to y gives f =
xy + h(x). Now, the function f has to fit these two equations simultaneously.
Therefore, it is clear that if f(x, y) = x3 + xy, then this satisfies the necessary
conditions and so F ( x , y) is a conservative vector field.

Example 11.15
Compute the following for line integral

∫ F. d r
C

for F ( x , y) = 〈3 x 2 + y, x 〉 along the curve C, where C is given in Figure 11.20.

Maybe F is a conservative vector field. Yes it is, as was shown in
Example 11.14.
F ( x , y) = ∇f , where f(x, y) = x3 + xy, so the working out is easy using the
fundamental theorem of line integrals:

∫ F. d r = f (end) − f (start)
C

= f (4, 0) − f (0, 2) = 4 3 − 0 = 64

(0, 2)
C

x
(4, 0)

Figure 11.20 The straight-line curve C.

Notation: Given curves C1, C2, C3, and C4 as shown in Figure 11.21.
The whole curve is partitioned as C = C1 + C2 + C3 + C4.
Also, shown in Figure 11.22 is the negative of a curve.
The same curve as C but in the opposite direction is called –C.
Property 3: If F ( x , y) = 〈 M ( x , y), N ( x , y)〉 is conservative, then My = Nx.
 11.4 Gradient Fieds 315

Does My = Nx imply that F is conservative, that is, F ( x , y) = ∇f ? No,

this is not always the case.
A region or domain (D) is “simply connected” if any closed curve C in
D can be contracted to a point in D.

Note: Connected means that you can get from a point to another without
leaving the region.

C4
C3

C1

C2

Figure 11.21 Combining different curves.

–C

Figure 11.22 A curve and is negative.

Example 11.16
Some examples of regions D that are connected and simply connected are
shown in Figure 11.23.
Theorem: If F ( x , y) = 〈 M ( x , y), N ( x , y)〉 is defined on a simply connected
domain and My = Nx, then it can be said that F is conservative.

D
D D

Connected: Y Y N

Simply connected: Y N Y

Figure 11.23 Different regions in space.

316 Vector Calculus

Example 11.17
An example of a famous conservative vector field is the gravitational vector
force field as shown in Figure 11.24. Here the vector field is given by

− k 〈 x , y, z 〉
F ( x , y) = 3
( x 2 + y2 + z 2 ) 2

This is the gradient of f, where

k
f ( x , y, z ) = 1
( x 2 + y2 + z 2 ) 2

and so F ( x , y) = ∇f .

Figure 11.24 The vector field for gravity.

11.5 Green’s Theorem
Having calculated line integrals for different paths as shown in Figure 11.25,
using the formula

∫ M (x, y) dx + N (x, y) dy
C

and having also done double integrals as follows

∫∫ D
f ( x , y) dA
 11.5 Green’s Theorem 317

Figure 11.25 A path showing the direction of the line integral.

then Green’s theorem relates these two concepts.

If the curve C is a simple, closed and positively oriented curve that surrounds
a two-dimensional region D.
The following definitions are used about the curve.

Closed – The end point is the same as the start point.

Simple – Has no complexity such as no intersection with itself.

Positively oriented – As you travel along the curve C the region D is toward
the left.

Pictorially, this is shown in Figure 11.26.

Then Green’s theorem states

∫ M (x, y) dx + N (x, y) dy = ∫∫ (N
C D
x − M y ) dA (11.13)

Figure 11.26 A region bounded by a closed curve.

318 Vector Calculus

This states that the line integral along the curve C is identical to a double inte-
gral over the region D.

Note: Just as the fundamental theorems of calculus and line integrals turn line
integrals into a calculation of the function at end points, Green’s theorem turns
an area integral into an integral around a boundary line of the area. It is a
dimension reducing method.

Example 11.18

Given F ( x , y) = 〈 M ( x , y), N ( x , y)〉 = 〈 y 2, xy〉 is a vector field, suppose the

curve C is the curve shown in Figure 11.27. Find the work done moving
along C, that is, calculate the following:

∫ F.d r
C

Now, since F is not a conservative vector field, then it is the case that

∫ F.d r ≠ 0
C

So, this problem can be done either directly by parameterizing the three
curves and solving the F .dr integral along each one.
Alternatively, you can use Green’s theorem and change this problem into
a double integral using Equation 11.13:

∫ F.d r = ∫∫ (N
C D
x − M y ) dA

C = C1 + C2 + C3

C2
C3

x
C1 5

Figure 11.27 The closed curve is a combination of curves.

 11.5 Green’s Theorem 319

Here, M (x,y) = y2 implies that My = 2y.

N(x, y) = xy implies that Nx = y.

Therefore, the double integral now becomes,

∫∫ (N
D
x − M y ) dA =
∫∫ y dA
D

Since the region is a quarter circle, changing to polar coordinates is bet-

ter and gives

π
2 5

∫∫ (N
D
x − M y ) dA =
∫∫ y dA = ∫ ∫ (r sinθ )r dr dθ
D 0 0

π π
2 2
1 125 125
=
∫ 3 [r sinθ ] dθ = ∫
0
3 5
0
0
3
cos θ dθ =
3

With a closed curve, Green’s theorem can be used to save some effort in the
calculations.

11.5.1 Properties of Green’s Theorem

Property 1: If F ( x , y) = 〈 M ( x , y), N ( x , y)〉 is a conservative vector field, then My =

Nx. But Green’s theorem states

∫ F.d r = ∫ ∫ (N
C D
x − M y ) dA = 0

This ties into the previous result for conservative closed vector fields (Equation 11.12).
Property 2: The above only relies on My= Nx.
〈− y, x 〉
If F ( x , y) = 2 as shown in Figure 11.28, then it can be shown that My = Nx,
x + y2
but the vector field is not conservative.
So,

∫ F. d r = ∫ ∫ (N
C1 D1
x − M y ) dA = 0

Also, what about along C2, since Nx and My are not defined at (0, 0).

∫ F. d r = ∫ ∫ (N
C2 D2
x − M y ) dA ≠ 0

Property 3: Where My = Nx, there is a restricted form of path independence,

paths can be perturbed so long as to stay in the domain as shown in Figure 11.29.
320 Vector Calculus

C1

D1

C2
D2 x

Figure 11.28 Vector field not defined at the origin.

C2

D1
C1

Figure 11.29 Perturbed paths in the domain.

〈− y, x 〉
F ( x , y) = with My = Nx
x 2 + y2

So,

∫ F.d r − ∫ F. d r = ∫
C2 C1 C2 −C1
F. d r =
∫ ∫ (N
D2
x − M y ) dA = 0
 11.6 Divergence and Curl of Vector Fields 321

This implies that

∫ F.d r = ∫ F.d r
C1 C2

Property 4: Green’s theorem relates a 2-D region D to its 1-D boundary C, as

shown in Figure 11.30.
A similar concept is the fundamental theorem of line integrals, which relates a
1-D curve to its 0-D boundary points, as shown in Figure 11.31.
y

Figure 11.30 A region with a closed boundary.

Figure 11.31 1-D curve with 0-D boundary points.

11.6 Divergence and Curl of Vector Fields

11.6.1 2-Dimensional Definitions

In 2-D space if F ( x , y) = 〈 M ( x , y), N ( x , y)〉 is a vector field, then the following

definitions are used.

div F = M x + N y is called the divergence of F . (11.14)

322 Vector Calculus

curl F = N x − M y is called the curl of F . (11.15)

How do we interpret these quantities? It first helps to think of F ( x , y) as a

velocity field of a flow of water. Then,

Divergence – Measures the net flow out a point.

Curl – Measures the counterclockwise (ccw) rotation at a point. This implies

that curl is positive if there is a ccw rotation.”

Now let’s look at some examples that illustrate these concepts with vector
fields.

Example 11.19

A vector field is given by F ( x , y) = 〈 x , y〉 and shown in Figure 11.32.

Consider the divergence at any point. Place a tiny box at a point and asking
if there is more flow into the box or more flow coming out of the box. Here
it is the case that there should be more flow out of a box at a point then flows
in, that is, div F > 0.
For the curl of this field, consider placing a paddle wheel into the field
and see if it rotates. The paddle wheel will flow outward but does not rotate,
that is, curl F = 0.
So, computing the divergence and curl of F using the formulae given by
Equations 11.14 and 11.15 gives

div F = M x + N y = 1 + 1 = 2 > 0

curl F = N x − M y = 0 + 0 = 0

x
1 2

Figure 11.32 Vector field given by F ( x , y) = 〈 x , y 〉.

 11.6 Divergence and Curl of Vector Fields 323

Example 11.20

Given the vector field, F ( x , y) = 〈 y, 0 〉, as shown in Figure 11.33, calculating

the divergence and curl gives

div F = M x + N y = 0 + 0 = 0

curl F = N x − M y = 0 − 1 = −1 (i.e., an clockwise rotation)

Figure 11.33 Vector field given by F ( x , y) = 〈 y, 0 〉.

Example 11.21

In a vector field, F ( x , y) = 〈 y, xy〉. This is a complicated vector field to see

easily, but it is still possible to compute the divergence and curl of F using
the definitions.

div F = M x + N y = 0 + x = x

curl F = N x − M y = y − 1 = y − 1

These depend on the location in the x-y plane.

11.6.2 Alternative Forms of Green’s Theorem

Now having the definitions of the divergence and curl of a vector field, it is pos-
sible to write Green’s theorem using these forms. The region is D and is bounded
by a curve C, as shown in Figure 11.34.

11.6.2.1 
Curl Form
Using the definition of Green’s theorem given by Equation 11.13,
324 Vector Calculus

Figure 11.34 Region bounded by a curve.

∫ M (x, y) dx + N (x, y) dy = ∫ ∫ (N
C D
x − M y ) dA

This can now also be written as using the definition of the curl:

∫ F. d r = ∫ ∫ curlF dA
C D
(11.16)

So, the curl F represents how much counterclockwise rotation there is at each
point and if all of these are added up, this then gives the total rotation around the
boundary.
Also note that now if F is a conservative (My = Nx), then this implies that
curl F = 0.

11.6.2.2 
Divergence Form
Figure 11.35 shows a bounded region within a vector field.

∫ ( F.nˆ ) ds = ∫ M dy − N dx
C C

y
n
F(x, y) = M(x, y), N(x, y)
dr
dy
D C dx

n = nds = dy, –dx

n
x

Figure 11.35 A bounded region within a vector field.

 11.6 Divergence and Curl of Vector Fields 325

But using Green’s theorem gives

∫ M dy − N dx = ∫ ∫ (M
C D
x − − N y ) dA =
∫ ∫ (M
D
x + N y ) dA =
∫ ∫ div F dA
D

So, this gives the following result:

∫ ( F.nˆ ) ds = ∫ ∫ div F dA
C D
(11.17)

This shows that the net flow out at each point all added up together gives the
total flux out of the curve shown in Figure 11.35.

11.6.3 3-Dimensional Definitions

Given a vector field in 3-D space, F ( x , y, z ) = 〈 M ( x , y, z ), N ( x , y, z ), P( x , y, z )〉,

then the following are the definitions for the divergence and curl of the vector
field:

div F = M x + N y + Pz is called the divergence of F in 3-D. (11.18)

curl F = 〈 Py − N z , M z − Px , N x − M y 〉 is called the curl of F in 3--D and is a vector.

(11.19)

It is easier to remember the definitions for grad F , div F , and the curl F in
terms of the “del or nabla notation.”
Let

∂ ∂ ∂
∇ 〈 , , 〉
∂x ∂y ∂z

then the following can be defined:

grad F = ∇F (11.20)

div F = ∇.F (11.21)

curl F = ∇ × F (11.22)

Note: The div F still measures the net flow out at a point. However, the curl
F is more difficult to see. Generally, the curl F points in the direction of the
“broom handle” with the largest counterclockwise rotation of the paddle wheel.
The magnitude of curl F is the amount of counterclockwise rotation.
326 Vector Calculus

Example 11.22
Given the vector field,

F ( x , y, z ) = 〈 M ( x , y, z ), N ( x , y, z ), P( x , y, z )〉 = 〈 x 2 , xy, z 〉,

find the divergence and curl of the vector field. Using the definitions gives

div F = M x + N y + Pz = 2 x + x + 1 = 3 x + 1

i j k
∂ ∂ ∂
curl F = ∇ × F =
∂x ∂y ∂z
x2 xy z

= 〈 0 − 0, 0 − 0, y − 0 〉

= 〈 0, 0, y 〉

Some facts can now be stated as follows:

1. (2-D) curl (∇f) = 0 alternative notation gives ∇ × ∇f = 0.

2. (3-D) curl (∇f ) = 0 alternative notation gives ∇ × ∇f = 0.
3. div(curl F ) = 0 alternative notation gives ∇.∇ × F = 0.

Proofs:

1. ∇f = 〈fx, f y〉, then curl (∇f) = f yx – fxy = 0.

2. ∇f = 〈fx, f y, fz〉, then

i j k
∂ ∂ ∂
curl (∇f ) =
∂x ∂y ∂z
fx fy fz

= 〈 fzy − f yz , f xz − fzx , f yx − f xy 〉 = 〈0, 0, 0 〉 = 0

 11.7 Surface Integration 327

i j k

3. curl F = ∇ × F = ∂ ∂ ∂ = 〈 Py − N z , M z − Px , N x − M y 〉
∂x ∂y ∂z
M N P

Therefore, div(curl F ) = Pyx − N zx + M zy − Pxy + N xz − M yz = 0 .

Note: ∇.∇ × F = 0 . The ∇ × F is a vector perpendicular to both ∇ and F,

so its dot product with ∇ should be zero.

11.7 Surface Integration
11.7.1 Parametric Surfaces
For a line there was one parameter. Generally for surfaces there are more param-
eters starting with two, say, u and v.

x = x (u, v), y = y(u, v), z = z (u, v), where (u, v) ∈ D

Or these can be written as

r (u, v) = 〈 x (u, v), y(u, v), z (u, v)〉, where (u, v) ∈ D

Pictorially, this can be shown as in Figure 11.36.

z
v
(x(u, v), y(u, v), z(u, v))
S
(u, v)
D

u y

Figure 11.36 Region maps unto a surface.

328 Vector Calculus

Example 11.23
A graph of a function in terms of an equation given the surface z = 4 – x2
– y2 and z ≥ 0 is shown in Figure 11.37.
To parameterize the surface, let x = u, y = v, then z = 4 – u2 – v2, where
(u, v) ∈ D (Figure 11.38).

z = 4 – x2 – y2

2
y

Figure 11.37 Region bounded by a surface.

2 u2 + v2 ≤ 4

D
u
2

Figure 11.38 The region on the x-y plane.

Example 11.24
A spherical shell of radius 5 is shown in Figure 11.39. In spherical coordi-
nates, this a sphere with radius ρ = 5.
 11.7 Surface Integration 329

So letting u = θ and v = φ, then a logical parameterization is

x = 5sin v cos u

y = 5sin v sin u

z = 5cos v

(u, v) ∈ D and the limits on u and v are shown in Figure 11.40.

Note: The sphere in 3D (x, y, z) is mapped to a rectangle in 2D (u, v).

5
5

Figure 11.39 A spherical shell of radius 5.

π
D
u
0
2π

Figure 11.40 The limits of the region in the u-v plane.

330 Vector Calculus

Summary of the Main Types of Surfaces

11.7.1.1 

1. The graph of the function z = f(x, y) is shown in Figure 11.41.

z = f (x, y)

S “u = x”, “v = y”
x=u
y=v
z = f (x, y)
(u, v) D

D y
x

Figure 11.41 Graph of a function.

2. In some coordinate systems, one of the coordinates is constant (e.g.,

in cylindrical coordinates), the radius r = constant = R as shown in
Figure 11.42.

10

4
4

Figure 11.42 Fixing a coordinate in a system.

 11.7 Surface Integration 331

To parameterize this surface S, in this case R is fixed at 4. Then using cylindri-

cal coordinates gives x = 4 cos u, y = 4sin u, and z = v. Here u = θ and v = z, and

π
0≤u≤ and 0 ≤ v ≤ 10
2

The need to parameterize surfaces leads to surface integrals.

11.7.2 Surface Integrals
The idea of a surface is similar to a region as shown in Figure 11.43.
Splitting the surface into small pieces and summing gives the definition of a
surface integral as

∫ ∫ f (x, y, z) dS  lim ∑
S
∆S →0
all pieces
f ( x*, y*, z*)∆S (11.23)

Usual applications of surface integrals are

Area ( S ) =
∫ ∫ 1dS
S
(11.24)

Mass ( S ) =
∫ ∫ d (x, y, z) dS;
S
(11.25)

y
x

Figure 11.43 A general surface in 3-D space.

where d is the density of a piece.

There is more work to do here when computing surface integrals. This can be
thought of as a “change of variables,” as shown in Figure 11.44.
Using the definition of transforming variables gives

∂( x , y, z )
∫ ∫ g(x, y, z) dS = ∫ ∫ g[x(u, v), y(u, v), z(u, v)]
S D
∂(u, v)
dv du (11.26)
332 Vector Calculus

 x xv 
u
∂( x , y , z )  
Now, the Jacobian is = det  yu yv 
∂(u, v)  z
 u z v 
The generalized determinant of a rectangular matrix is given by

x xv  2 2 2

u
 x xv  x xv  y yv 
u u u
det  yu yv   det   + det   + det   (11.27)
 yu yv   zu z v   zu z v 
z z v 
 u

z
v

S
T
D

u y

Figure 11.44 A transformation of variables.

Example 11.25
Find the area of x + y + z = 1 in the first octant as shown in Figure 11.45.
Parameterizing the surface gives x = u, y = v, and z = 1 − u − v. So, the
area is given by the formula

∂( x , y, z )
Area( S ) =
∫ ∫ 1dS = ∫ ∫ 1
S D
∂(u, v)
dv du

x x v 2 2 2
u
∂( x , y, z )   1 0 1 0 + det  0 1
= det  yu yv = det   + det    
∂(u, v) z 0 1 −1 −1  −1 −1
 u z v

= (1)2 + (−1)2 + (1)2 = 3

Area( S ) = 3
∫ ∫ dv du
D

3
= 3 Area( D) =
2
 11.7 Surface Integration 333

v
z

D
1 u
(0, 0, 1)

(0, 1, 0)
D
(1, 0, 0) y

Figure 11.45 Region in the first quadrant.

11.7.3 Tangent Planes and Normal Vectors

Graphically, Figure 11.46 shows a surface and a tangent plane to it. (x, y) is a point
near (a, b) but using the Taylor series expansion gives

f ( x , y) ≈ f ( a , b ) + ( x − a ) f x ( a , b ) + ( y − b ) f y ( a , b )

z = f (x, y)
z

Tangent plane to the

surface

y
(a, b)

Figure 11.46 A surface given by z = f(x, y) and its tangent plane.

Call the right-hand side equal to z, then this can be written as

z = f (a, b) + ( x − a) f x (a, b) + ( y − b) f y (a, b) (11.28)

This is an approximation to f(x, y).

334 Vector Calculus

Now, this is a plane accurate at (a, b) but approximate near (a, b) and is called
the tangent plane (or best linear approximation).

Example 11.26
Find the tangent plane to the surface z = x + 5xy2 at point (1, 2, 3).

Solution: Let the function f(x, y) = x + 5xy2. Therefore, using Equation 11.28
gives

f ( x , y) ≈ f (1, 2) + ( x − 1) f x (1, 2) + ( y − 2) f y (1, 2)

Let
z = f (1, 2) + ( x − 1) f x (1, 2) + ( y − 2) f y (1, 2)

f (1, 2) = 21

fx = 1 + 5 y2 → f x (1, 2) = 21

f y = 10 xy → f y (1, 2) = 20

z = 21 + 21( x − 1) + 20( y − 2)

21x + 20y – z = 40 is the equation of the required tangent plane.

11.7.3.1 Normal Vector to the Tangent Plane

Example 11.27
Find the normal vector to the surface z = x2 + y2 at the point (−1, 2, 2), as
shown in Figure 11.47.

Solution: n is a normal vector perpendicular to the surface at (−1, 2, 2). First,

finding the tangent plane using f(x, y) = x2 + y2 with Equation 11.28 gives

z = f (−1, 2) + ( x + 1) f x (−1, 2) + ( y − 2) f y (−1, 2)

f ( x , y) = x 2 + y 2 f (−1, 2) = 5

fx = 2 x → f x (−1, 2) = −2

fy = 2 y → f y (−1, 2) = 4

z = 5 − 2( x + 1) + 4( y − 2)

2 x − 4 y + z = −5

This is the tangent plane at (−1, 2, 2).

But the normal to the surface is the normal to the tangent plane. So, the
normal vector to the tangent plane 2x – 4y + z = −5 is just n = 〈2, −4, 1〉
 11.7 Surface Integration 335

(–1, 2, 2)

Figure 11.47 Surface and its normal vector at the given point.

11.7.3.2 General Formula for the Normal Vector to a Surface

In Figure 11.48 is a general surface z = f(x, y) and its tangent plane. The tangent
plane is given by Equation 11.28 as

z = f (a, b) + ( x − a) f x (a, b) + ( y − b) f y ( a, b)

z
z = f (x, y)
n

Tangent plane to the

surface

(a, b) y

Figure 11.48 General surface and its normal vector.

This can be rewritten as x, y, and z terms on one side with all the constants on
the other side as

− f x (a, b) x − f y (a, b) y + z = some constant

336 Vector Calculus

Therefore, the normal vector also normal to the surface at the given point is

n = 〈− f x (a, b). − f y (a, b),1〉 (11.29)

This is pointing up as z = 1 and this formula will be useful in later sections on

surface integrals.

11.7.4 Normal Vectors to Surfaces

Normal vectors to surfaces can be pointing in different directions as shown in
Figure 11.49. Normal vectors can be pointing upward or outward as shown in
Figures 11.50 and 11.51, respectively.

n
Normal vectors can be pointing
up or down

Figure 11.49 Normal vectors to a surface.

“Oriented upwards”

Figure 11.50 Normal vector pointing upward.

 11.7 Surface Integration 337

n
r

“Oriented outwards”

E n

r
r

Figure 11.51 Normal vector pointing outward.

There are different ways to find the normal vectors to different surfaces. For a
general graph as shown in Figure 11.52 the normal vector is simply given by using
the formula for z = f(x,y) as
〈− f x , − f y , 1〉
nˆ = (11.30)
f x2 + f y2 + fz2

z = f (x, y)
S

Figure 11.52 Normal vector for a surface given by a graph.

The case of a sphere is shown in Figure 11.53. The normal vector is given by
using the formula given by Equation 11.29 as
〈 x , y, z 〉 (11.31)
nˆ =
x + y2 + z 2
2
338 Vector Calculus

n r n

r
r

Figure 11.53 Normal vectors to a sphere.

In the case of a cylindrical surface is shown in Figure 11.54. The normal vector
is given by using the formula given in Equation 11.31 as
〈 x , y, 0 〉 (11.32)
nˆ =
x 2 + y2

Figure 11.54 Normal vectors to a cylinder.

Generally, for a surface of the form r = (u, v), the normal vector is

ru × rv
n̂ = (11.33)
ru × rv

Note: This formula can always be used if required.

 11.7 Surface Integration 339

11.7.5 Applications of Surface Integrals

The flux of a vector F across a surface S is shown in Figure 11.55. The flux
of a vector F across the surface element ΔS is calculated by considering the
component of F perpendicular to the surface, that is, in the direction of n̂
as ( F .nˆ ). The flux coming out of the surface element ΔS is then given by
( F .nˆ ) ∆S .
The total flux coming out of the surface S can be found by a summing of
( F .nˆ ) ∆S as ΔS→0 which then produces a double integral:

Flux across the surface =

∫∫ (F.nˆ) dS
S
(11.34)

z
n

n F
S

∆S

Figure 11.55 Flux across a surface.

Example 11.28
Compute the flux of the vector field F = 〈 x , y, z 〉 across the surface S, where S is
given by z = 4 – x2 – y2 and z ≥ 0 and oriented upward as shown in Figure 11.56.
〈− f x , − f y , 1〉
Solution: The normal vector n̂ is given by the formula nˆ = ,
which gives f x2 + f y2 + fz2

〈2 x , 2 y, 1〉
nˆ =
4 x 2 + 4 y2 + 1

〈2 x , 2 y, 1〉
Flux =
∫∫ (F.nˆ) dS = ∫∫ 〈x, y, z 〉.
S S
4 x 2 + 4 y2 + 1
dS

2 x 2 + 2 y2 + z
=
∫∫ S
4 x 2 + 4 y2 + 1
dS
340 Vector Calculus

4
n

S z = 4 – x2 – y2

2
y

Figure 11.56 Flux out of a surface.

Now to parameterize the surface S, let x = u, y = v, and z = 4 – u2 – v2,

where (u, v) ∈ D (see Figure 11.57):

2 x 2 + 2 y2 + z (2u 2 + 2v 2 ) + (4 − u 2 − v 2 ) ∂( x , y, z )
=
∫∫
S
4 x 2 + 4 y2 + 1
dS =
∫∫
D
4u 2 + 4 v 2 + 1 ∂(u, v)
dv du

2 u2 + v2 ≤ 4

D
u
2

Figure 11.57 Region in the u-v plane.

 x xv 
u
∂( x , y, z )  
= det  yu yv 
∂(u, v)  z
 u z v 

 1 0 
= det  0 1
 = (1)2 + (2 v)2 + (2u)2 = 1 + 4u 2 + 4 v 2
 
 −2u −2 v 
 11.7 Surface Integration 341

2 x 2 + 2 y2 + z (4 + u 2 + v 2 )
=
∫∫
S
4 x 2 + 4 y2 + 1
dS =
∫∫
D
4u 2 + 4 v 2 + 1
1 + 4u 2 + 4 v 2 dv du

=
∫∫ (4 + u
D
2
+ v 2 ) dv du

This double integral can be performed using polar coordinates:

2π 2

∫ ∫ (4 + r )r dr dθ = 24π
0 0
2

Example 11.29
Find the surface area of the cylinder shown in Figure 11.58.
Area is given by the formula

2π H
∂( x , y, z )
Area( S ) =
∫∫S
1 dS =
∫ ∫1
0 0
∂(θ , z )
dz dθ

x = R cos θ
y = R sin θ
z=z
S 0 ≤ θ ≤ 2π
0≤z≤H

R
R

Figure 11.58 Cylinder of height H and radius R.

Now,

 x xz   − R sin θ 

θ
 0
∂( x , y, z )
= det  yθ yz  = det  R cos θ 0

∂(θ , z )  
 z z z   0 1 
 θ
342 Vector Calculus

= (0)2 + (− R sin θ )2 + ( R cos θ )2

∂( x , y, z )
=R
∂(θ , z )

2π H

=
∫ ∫ R dz dθ
0 0

2π

=
∫ RH dθ = 2π RH
0

Example 11.30
Compute the flux out of a sphere given by x2 + y2 + z2 = 4 for the vector field
given by F = 〈 x , y, z 〉 and shown in Figure 11.59.

2
n

2
2

Figure 11.59 Flux out of a sphere.

〈 x , y, z 〉
Flux =
∫∫ (F.nˆ) dS = ∫∫ 〈x, y, z 〉.
S S
x + y2 + z 2
2
dS

x 2 + y2 + z 2
=
∫∫
S
x +y +z
2 2 2
dS =
∫∫
S
x 2 + y 2 + z 2 dS
 11.8 Stokes’ Theorem 343

Parameterizing the surface with (p = 2) gives

x = 2 sin ϕ cos θ
y = 2 sin ϕ cos θ
z = 2 cos ϕ

0 ≤ θ ≤ 2 π and 0 ≤ φ ≤ π

2π π
∂( x , y, z )
=
∫ ∫ρ
0 0
∂(ϕ ,θ )
d ϕ dθ

 x xθ   2 cos ϕ cos θ
ϕ −2 sin ϕ sin θ 
∂( x , y, z )    
= det  yϕ yθ  = det  2 cos ϕ sin θ 2 sin ϕ cos θ  = 4 sin ϕ
∂(ϕ , θ )  
 zϕ zθ   2 sin ϕ 0 
 

2π π 2π

=
∫ ∫ 2(4 sin ϕ ) dϕ dθ = ∫ 16 dθ = 32π
0 0

11.8 Stokes’ Theorem
There are many theorems that relate the integral over a region to the integral over the
boundary of the region. Some that have already been considered are the following.
In 1-D, Figure 11.60 shows a line interval from [a, b].

x
a b

Figure 11.60 Line interval from [a,b].

∫ f ′(x) dx = f (b) − f (a)

a

This is the fundamental theorem of calculus.

In 2-D or 3-D, see Figure 11.61.

For conservative vector fields, the work done along a curve as shown in
Figure 11.61 is

∫ ∇f .d r = f (end) − f (start)
C

This is the fundamental theorem of line integrals.

344 Vector Calculus

C End

Start

Figure 11.61 A line path in space.

In 2-D, Figure 11.62 shows a bounded region in 2-D space.

∫∫ curlF dA = ∫ F.d r
D C

Figure 11.62 Region bounded by a curve.

This is Green’s theorem. This can be extended to a 3-D surface and its bound-
ary, as shown in Figure 11.63.
Suppose C is a simple, closed and positively oriented curve with respect to the
surface S as earlier. Stokes’ theorem states

∫∫
S
^ dS =
(curl F . n)
∫
C
F .dr (11.35)

Flux of the curl of F

 11.8 Stokes’ Theorem 345

An idea of what is being represented is shown in Figure 11.64.

n
S

Figure 11.63 A surface in 3-D surface bounded by a curve.

S Summing all the curl’s on the surface is

equal to the bulk rotation around the
boundary curve C.

Figure 11.64 Pictorial representation of Stokes’ theorem.

Example 11.31

Verify Stokes’ theorem for the vector field F = 〈 x , z , − y〉 and the surface S,
where S is defined by x + y + z = 1 and x2 + y2 ≤ 1.
The surface S is given by the plane being effectively chopped by cylinder
of radius 1 as shown in Figure 11.65.
To verify Stokes’ theorem the following has to be shown:

∫∫ (curlF.nˆ) dS = ∫ F.d r
S C
(11.36)
346 Vector Calculus

S
1 n

D
1
1

Figure 11.65 Surface in 3-D space.

Starting with the left-hand side of Equation 11.36 gives

∫∫ (curlF.nˆ) dS
S

i j k
∂ ∂ ∂
curl F = ∇ × F = = 〈−2, 0, 0 〉
∂x ∂y ∂z
x z −y

〈1, 1, 1〉
A unit normal vector to the surface is given by nˆ =
3

〈1, 1, 1〉 −2
∫∫ (curlF.nˆ) dS = ∫∫ 〈−2, 0, 0〉.
S S
3
ds =
∫∫
S
3
dS

Parameterize the surface S,

x=x

y= y

z = 1− x − y

where (x, y) ∈ D

−2 ∂( x , y, z )
=
∫∫
D
3 ∂( x , y)
dy dx
 11.9 Divergence Theorem 347

 x xy 
x  1 0 
∂( x , y, z )  
= det  yx yy  = det  0 1
 = (1)2 + (−1)2 + (−1)2 = 3
∂( x , y)    
 z x z y   −1 −1 

=
∫∫ −2 dy dx = −2Area(D) = −2π
D

The right-hand side of the Equation 11.37 gives

∫ F.d r = ∫ x dx + z dy + (− y) dz = ∫ x dx + z dy − y dz
C C C

where C is the boundary curve, which is the intersection of the plane

x + y + z = 1 and the circle x2 + y2 = 1.

x = cos t
y = sin t
z = 1 − cos t − sin t

0 ≤ t ≤ 2π

=
∫ x dx + z dy − y dz
C

2π

=
∫ [(cos t)(− sin t) + (1 − cos t − sin t)(cos t) − (sin t)(sin t − cos t)] dt
0

2π

=
∫ (−1 + cos t − cos t sin t) dt
0

Using trigonometric identities and properties of the cosine and sine func-
tions gives

2π
 1 
=
∫  −1 + cos t − 2 sin 2t  dt = −2π
0

So the left-hand side equals the right-hand side and Stokes’ theorem is
verified.

11.9 Divergence Theorem
In Figure 11.66 there is a 3-D region (E) bounded by a 2-D surface-oriented out-
ward. If all the flux at each point in the region (E) are added together, as shown
348 Vector Calculus

in Figure 11.67, then this will give the net flux out of the surface. Mathematically,
this is stated using the divergence theorem as

∫∫∫ div F dV = ∫∫ (F . n)^ dS (11.37)

E S

Adding all the flux at Net flux out of the

each point in the region surface.

Figure 11.66 3-D region bounded by a 2-D surface.

Figure 11.67 Adding all flux at each point in the region.

 11.9 Divergence Theorem 349

Example 11.32
Verify the divergence theorem for the vector field F = 〈 x , y, z 〉 for the follow-
ing sphere of radius 1 and with a boundary surface S as shown in Figure 11.68.

Solution: Starting with the definition, show the following:

∫∫∫ divF dV = ∫∫ (F.nˆ) dS

E S
(11.38)

Starting with the left-hand side of Equation 11.38 gives

div F = M x + N y + Pz = 1 + 1 + 1 = 3

4
∫∫∫ divF dv = ∫∫∫ 3 dV = 3Vol(E) = 3 3 π (1)
E E
3
= 4π

1
n

1
1

Figure 11.68 Flux out of a sphere.

The right-hand side of Equation 11.38 gives

〈 x , y, z 〉
∫∫ (F.nˆ) dS = ∫∫ 〈x, y, z 〉.
S S
x + y2 + z 2
2
dS

x 2 + y2 + z 2
∫∫
S
x 2 + y2 + z 2
dS

Now to parameterize the surface of the sphere with ρ = 1, let

x = sin ϕ cos θ
y = sin ϕ sin θ
z = cos ϕ
350 Vector Calculus

0 ≤ θ ≤ 2π and 0 ≤ φ ≤ π
x2 + y2 + z2 = ρ 2 and ρ = 1
This gives the following:

∂( x , y, z )
=
∫ ∫1S
∂(ϕ ,θ )
dS

2π π

=
∫ ∫ (sin ϕ ) dϕ dθ
0 0

2π

∫ − cosϕ 
π
= 0
dθ
0

2π

=
∫ 2 dθ = 2 × 2π = 4π
0

So the left-hand side equals the right-hand side as required.

11.10 Applications
Example 11.33: Fluid Dynamics of Smoke Flow
In fluid dynamics one of the main tasks is to find the velocity field describ-
ing the flow in a given region. One can make use of the laws of mechanics,
that is, the conservation of mass to derive the continuity equation.
Any “small” fluid element can be assigned a velocity v(x,t) and average
density ρ(x,t).
Considering a volume V bounded by a surface S, the mass inside the
volume is given by

∫ ρ dV
V

The rate of decrease of mass in the volume is

d ∂ρ
V =−
dt ∫ ρ dV = − ∫ ∂t dV
V V
(11.39)

This must be equal to the total rate of mass flux out of V if mass is con-
served. The rate of mass flux out of any small element dS of S is given as ρv.dS.
Integrating over the whole surface gives the rate of mass flux out of V as

=
∫ ρv. dS
S

Now using the divergence theorem this can be written as a volume integral as
 11.10 Applications 351

∫ ∇.(ρv) dV
V
(11.40)

This then implies that for mass to be conserved, Equations 11.39 and
11.40 must be equal for any volume V giving the continuity equation as

∂ρ
+ ∇.( ρ v) = 0
∂t

For fluids that are incompressible, ρ is a constant and so this reduces to

the following:
∇.( ρ v) = 0

In any flow region, the flow equations are solved to a set of conditions
that act at the boundary.
Applications in smoke control situations occur when the smoke temperature
of the smoke reservoir does not change. In order to maintain the smoke layer
height, the amount of smoke extracted from the smoke reservoir equals the
amount of smoke flow into the reservoir from the fire plume, that is, ∇.(ρv) = 0.

Example 11.34: Application in Thermodynamics

Heat engines such as automobile engines operate in a cyclic manner, adding
energy in the form of heat in one part of the cycle and using that energy to
do useful work in another part of the cycle. Since work is done only when
the volume of the gas changes, the PV diagram gives a visual interpretation
of work done. For a cyclic heat engine process, the PV diagram will be a
closed loop. The area inside the loop is a representation of the amount of
work done during a cycle.
Find the work done by the engine cycle C, where P is the pressure and V
is the volume as shown in Figure 11.69.
Parameterization of the cyclic path gives

V = 5 cos(−t ) + 50
P = 5 sin(−t ) + 30
0 ≤ t ≤ 2π

The formula for the work done is given by using Equation (11.8),

Work =
∫ P. dV
C

2π

=
∫ (−5 sin(t) + 30)(−5 sin(t)) dt
0

2π

=
∫ (25 sin (t) − 150 sin(t)) dt = 25π J
0
2
352 Vector Calculus

30 5

50 V

Figure 11.69 The closed path for the engine cycle.

Example 11.35: Velocity of the Earth’s Wind

The vorticity plays an important role in fluid dynamics. For the earth’s
wind, it is defined as the curl of the wind velocity (i.e., curl v):

i j k

curl v = ∇ × v = ∂ ∂ ∂ = i  ∂w − ∂v  +  ∂u ∂w 
j −
 ∂v ∂u 
+ k − 
∂x ∂y ∂z
 ∂y ∂z   ∂z ∂x   ∂x ∂y 
u v w

Since, in general, the horizontal velocities are much larger than the ver-
tical velocities and vertical scales are much smaller than the horizontal
scales, in the x and y components of the above expression we can neglect
the terms in the vertical velocity, giving

 ∂v   ∂u   ∂v ∂u 
curl v ~ i  −  + j  + k −
 ∂z   ∂z   ∂x ∂y 

∂v ∂u (10 ms −1 )
The first two terms and have typical magnitudes ~ 10 −3 s −1.
∂z ∂z 4
(10 m)
∂v ∂u (10 ms −1 )
The third term − has a typical magnitude ~ 10 −5 s −1.
∂x ∂y (10 6 m)
But for irrotational flows, the vorticity is 0 (i.e., ∇ × v = 0) and then the
velocity can be represented as the gradient of a scalar function. This is
known as the velocity potential f such that v = ∇f .

Example 11.36: Use of Green’s Theorem to Calculate Areas

Green’s theorem can be used to compute areas of regions as shown in
Figure 11.70.
Green’s theorem (Equation 11.13) is given by
 Problems 353

∫ M (x, y) dx + N (x, y) dy = ∫∫ (N
C D
x − M y ) dA

The closed integral is given as

∫ x dy = ∫∫ (N
C D
x − M y ) dA =
∫∫ (1 − 0) dA = ∫∫ dA = Area (D)
D D

or it can be given as

− ∫ y dx = ∫∫ (N
C D
x − M y ) dA =
∫∫ (0 − −1) dA = ∫∫ dA = Area (D)
D D

Figure 11.70 Region D bounded by a curve C.

Also, it can be shown that a combination of the above two, that is,

1 1 1
2 ∫ x dy − y dx = ∫∫  2 − − 2  dA = ∫∫ dA = Area (D)
C D D

This principle has been used to find areas of shapes as one draws a curve
around the region. It uses
∫ x dy and keeps a track of this quantity and adds
C
it up to calculate the area.
Some software applications have been developed that allows the calcula-
tions of areas of regions on maps using this type of principle.

Problems
11.1 If r (t ) = 3t i + t 2 j − 2t 3 k , calculate r ′(t ) and r ′′(t ).

11.2 Given that r ′(t ) = i + 3t j − 6t 2 k and r (0) = i + j + k , find r (t ).

11.3 Suppose the density of a wire is given by d(x,y) = x2. Compute its mass
when the wire is the semicircle of radius 1 as shown in Figure 11.71.
354 Vector Calculus

1
C

x
1 1

Figure 11.71 A wire in the form of a semicircle.

11.4 A force field is given by F ( x , y) = 〈 x , y 2 〉, a particle travels along a curve

C given by y = x2 as shown in Figure 11.72. Calculate the work done by
the particle.

4 y = x2

0 x
2

Figure 11.72 Path traveled by the particle.

11.5 Verify Stokes’ theorem for the vector field F = 〈4 z , −2 x , 2 x 〉, where C is

the intersection of x2 + y2 = 1 and z = y + 1.

11.6 Suppose F = 〈 x , y, 0 〉 with E is the cylinder of radius 4 cm and height

5 cm as shown in Figure 11.73. For the vector field verify that

∫∫∫ divF dV = ∫∫ (F.nˆ) dS = 160π

E S
 Problems 355

n
5

S3

n
E

S2

4
S1

n y

Figure 11.73 Flux coming out of a cylinder.

http://taylorandfrancis.com
Answers

Chapter 1

3.6V F −G h − v2
1.1 a. A = b. H= c. k = αρc d. t=
L 7 g
1
 I 4 sτ H 2E
e. d = 2 α t f. T = g. t= h. v =
 εσ A  H2 g m
3
3 + 5g  U 
i. h = j. Q =  H
g −1  0.96 

3
 H (Tj − T0 )  2
1.2 a. Q = r   b. 26649.5 kW
 5.38 
1.3 a. a = 3 b = –1 b. x = 1.5 y = 0.5 c. x = 0.54 or – 5.54
d. x = 3.16 or – 1.16

1.4 a. t = 10.6 s, θ = 81.5° b. V = 9.65ms–1

1.5 x = 11.04 σ = 3.61

Chapter 2
1 2 7
2.1 a. b. c.
4 13 13
5
2.2
14
2.3 a. 0.63 b. 0.97

1 3 3 1
2.4 P(0) = , P(1) = , P(2) = , P(3) =
8 8 8 8
2.5 P(B\D) = 0.62
357
358 Answers

13
2.6 E ( X ) = , σ = 0.68
6
2.7 E(t) = 2.7 years, σ = 0.84 years

1
2.8
407

Chapter 3
1
3.1 uˆ = 〈2, − 1, 4 〉
21
3.2 θ = 129.05°

3.3 α = –2

3.5 −3 i − 5 j − 7 k

3.6 4 i + 9 j + 5k

3.7 –3x + 5y + 2z = –3

3.8 a. 1

3.9 VR = 0.57ms–1 θ = 48.1°

Chapter 4
4.1 a. –13 b. –154

6
4.2 k =
5

 6 3 6  −2 −1 4 
4.3 a.   b.  −2 4 −9 
 4 2 5  

 9 −8   17 14 20 
 −16 −75   
4.4 a.  8 −1  b.
 
c.
 14 13 15 
   25 34 
 10 −15   20 15 25 

4.5 a. x = 2 y = 3 z = –1 b. x=3y=1z=2

4.6 i1 = 2 i 2 = 2 i3 = 1

 −1  −3   1
   
4.7 λ1 = 1 e1 = 1 , λ2 = 2 e2 = −3 , λ3 = 21 e1 =  1 
     
 0  1  0
 Answers 359

 
1  1 1  1 0   3 −1 b. A10 =  4882813 1627604 
4.8 a.
6  −3 3   0 5   3 1  14648436 4882813 

4.9 S = 11 minutes, F = 19.9 minutes, B = 21 minutes

Chapter 5

1 55 33
5.1 a. –2 ± j b. ± j c. −1 ± j
4 4 3
13 1
5.2 a. 4 + 7j b. –j c. –7 + 11j d. − j
5 5
1 1
5.3 a = 9, b = ; a = , b = 12
3 4
5.4 a. 1.81 + 0.42j, –1.27 + 1.36j, –0.55 – 1.78j

b. 1.21 – 0.17j, 0.54 + 1.09j, –0.87 + 0.85j, –1.08 – 0.57j, 0.21 – 1.20j

c. 5.80 – 1.65j, –5.80 + 1.55j d. 2.60 – 1.5j, 3j, –2.60 – 1.5j

5.5 a. z = 32 + 46.91j b. i = 0.119 + 0.175j, |i| = 212mA

5.6 a12 < 4 a2

Chapter 6

6.1 a. 12x2 – 10x + 7 b. 12 x + 40 c. xe3x (3x + 2)

x cos x − sin x
d. e. 15x2(x3 + 1)4
x2

6.2 (2,0) and (–2,0) both are maximum points

6.3 b. r = 11.8 cm, minimum

5 3
6.4 a. x4 + 3x3 – 5x2 + 4x + C b. 6 2 10 2
x + x +C
5 3
3
x4 1
c. − +C d. 2( x 2 + 5) 2 + C e. 1.106
4 x

xe 4 x e 4 x 1 x
f. − +C g. e (sin x − cos x )
4 16 2

x n+1 x n+1
h. ln x − +C
n +1 (n + 1)2
360 Answers

6.5 100 J

6.6 3.125 years

Chapter 7
7.1 y2 = x2 + 1
x2
7.2 ln( y − 3) = + 5x + C
2
y2
7.3 + y = ln x + C
2

x3
7.4 y = + Cx
2

x5 1
7.5 y = +
5(1 − x ) 1 − x 2
2

x 3
7.6 y = α e − x + β e 2 x − −
2 4

10
7.7 y = α e5 x + β xe5 x +
25

7.8 y = e–2x (–0.5 cos x – sin x) + 0.5e3x

1
7.9 T = (6e5t − 1)
5

Chapter 8
2
8.1
t3
1 10 1
8.2 a. y = − e −2t + e 7t b. T = (1 + e −6t )
9 9 2

c. y = 1 + et d. 1
x= (109 + 18t − e18t )
54

( )
R
− t
V0
R
− t V wLe L V
8.3 a. i = 1− e L b. i = 02 + 2 0 2 2 ( R sin wt − wL cos wt )
R R +w L R +w L
2 2

C kt − kt C kt − kt C
8.4 x = (e − e ) y = (e + e ) −
2k 2k k
 Answers 361

Chapter 9
 2
 0 n = even − n = even
 
9.1 a. a0 = 1, an =  4 , bn =  nπ
 − n 2π 2 n = odd  2 n = odd
  nπ

 0 n = even

b. a0 = 1, an = 0 all n, bn =  2
 nπ n = odd


 4
− n = even

b. a0 = 0, an = 0, bn =  nπ
1 
9.2 a. cn = 3 − 5(−1) 
n
jnπ  16 n = odd
 nπ

Chapter 10
∂f ∂f
10.1 = 3, = 13
∂x ∂y
10.2 a. xy cos xy + sin xy

b. x2cos xy c. fxy = –x2y sin xy + 2x cos xy

d. f yx = –x2y sin xy + 2x cos xy

10.3 − 15
10
10.4 –18

10.5 16
3

Chapter 11

11.1 r ′(t ) = 〈3, 2t , − 6t 2 〉, r ″ (t ) = 〈0, 2, − 12t 〉

11.2 r (t ) = (1 + t ) i + (1 + 1.5t 2 ) j + (1 − 2t 3 ) k
π
11.3
2
11.4 10
http://taylorandfrancis.com
Index

A D
Algebra, 9–17 Decimal places, 3–4
Alpert’s equation, 16, 41 Delta function, 212–213
Applications, 41–44, 60–63, 78–79, Determinants, 81–87
110–114, 125–129, 164–167, 2 x 2, 83
191–197, 219–227, 250–253, 3 x 3, 85–87
289–295, 339–343, 350–353 Cramer’s rule, 82
definition, 82
B properties, 83–84
Bayes’ theorem, 49–51, 62–63 Differential equations, 171, 172
complementary function, 183
C continuous time Markov process,
Centre of mass, 292–294 193–195
Change of scale, 39 degree, 173–174
Complex numbers, 117 first order, 174–182
addition and subtraction, 118 integrating factor, 179–182
Argand diagram, 120 order, 173
conjugate, 119 particular integral, 188–191
De Moivre’s theorem, 122 second order, 182–191
division, 119 separating variables, 175–179
electrical circuits, 127 smoke layer, 191–193
exponential form, 122 types of, 172–174
forces, 128 Differentiation, 131
imaginary j, 117–118 chain rule, 143–144
multiplication, 119 definition of a limit, 131–136
polar form, 120–122 general formula, 133–134
roots of equations, 123–125 gradient, 131
Conditional probability, 48–49 practical test, 137–138
Conservative vector fields, 310–313 products, 139–140
Continuous random variables, quotients, 140–141
58–60 second derivative test, 138–139
Cosine rule, 25, 27, 28 standard functions, 141–143
Cumulative distribution function, stationary points, 136–137,
56–57 266–267
Curl, 321–327 Dimensional analysis, 113

363
364 Index

Discrete random variables, 52–55 standard integrals, 149–150

Discriminant, 21, 185 substitution, 155–157
Divergence, 321–327 Interquartile range, 37
Divergence theorem, 347–350
L
E Laplace transform, 199
Eigenvalues and Eigenvectors, 96–101 basic relations, 205
Elimination method, 18–19 control systems, 226
Enclosure fire, 40–41 derivation, 200–201
Event tree analysis, 60–61 first shift theorem, 213
Expectation values, 54 inverse transforms, 207
Exponential notation, 6, 232, 239 linearity, 205
schematic representation, 202
F second shift theorem, 213
Fault tree analysis, 61–62 solving linear differential equations,
Filters, 251–253 214–219
Fluid dynamics, 350–351 standard transforms, 201–205
Fourier series, 229 two-zone model of a fire, 219
complex form, 238–241 Line Integrals, 303–308
Fourier coefficients, 232–237 Linear equations, 10–17, 103
orthogonal functions, 231 Linear simultaneous equations,
periodic function, 229–230 17–20
Fourier transforms, 242–249
convolution of functions, 248–249 M
filters, 251 Markov chain, 110, 112
Fourier transform pair, 242–245 Matrices, 87
frequency modulation, 250–251 addition and subtraction, 88
properties of, 247–248 diagonal factorization, 100–103
Fourier transforms Inferred (FTIR) smoke eigenvalues and eigenvectors,
detectors, 250 96–100
Frequency modulation, 250–251 inverse of, 94–96
Frequency Table, 34–35 multiplication, 89
order, 88
G powers, 94
Gaussian function, 254 special matrices, 92–93
Gradient fields, 308–309 Mean, 3, 33–36
Green’s theorem, 316–321 Measures of averages, 32–36
properties of, 319–321 Measures of spread, 36–39
Grouped data, 35–36 Median, 33, 35, 36, 37
Mode, 32, 34
H Multivariable calculus, 255
Heat release rate, 40, 41, 43, 165, 166, 168, applications, 289–295
222, 223 Cartesian coordinates, 281–283
chain rule, 261
I Clairaut’s theorem, 259–260
Independent events, 47–48 cylindrical coordinates, 292
Inferred (FTIR) smoke detectors, 250 directional derivatives, 263–266
Integration, 66, 144–155, 155, 161–164, double integral, 269
175, 195, 269–289 297, 327–343 Fubini’s theorem, 269
areas under curves, 149–152 general change of coordinate systems,
by parts, 162–164 286
improper integrals, 152–155 gradients, 263–266
mean value theorem, 147 higher derivatives, 259–263
partial fractions, 157–162 partial derivatives, 255–258
Riemann sum, 144–146 spherical coordinates, 283
 Index 365

triple integrals, 281 normal vectors, 336–338

view factors, 291 parametric surfaces, 327–331
Mutually exclusive events, 46–47 Stokes’ theorem, 343–347
surface integrals, 331–332, 339
N Systems of linear equations, 103–107
Normal vectors, 333, 336
T
O Thermodynamics, 351
Optimization, 164 Transposing equations, 13
Tree diagrams, 51
P Trigonometric identities, 29
Partial fractions, 157–162 Trigonometry, 29–30
Pool fires, 40
Probability density function, 56 U
Probability distribution, 52–54, 64 Unit step function, 210
Probability theory, 45–64 fire growth model, 222

Q V
Quadratic equations, 20–22, 117, 118, 184, Variance, 39, 54–55
185 Vector calculus, 297–355
conservative vector field, 310–313
R derivatives of vector functions, 297–299
Radian measure, 30 divergence theorem, 347–350
Range, 36–37 gradient field, 308–316
Relative frequency, 46 integrating vector functions, 299
Reliability theory, 166–167 path independence, 311
Resultant forces, 28 Stokes’ theorem, 343–348
Right angled triangles, 23–25 Vectors, 65
Round-to-even method, 2–3 addition and subtraction, 68
Rounding numbers, 1 definition, 65
flux of vector, 339
S magnetic induction, 79
Scalene triangles, 25–28 magnitude of, 66
Scientific notation, 6–9 normal vectors, 333–338
Significant places, 4–6 projection of, 72
Sine rule, 25–26 scalar product, 69–70
Smoke flow, 41–42, 78, 350–351 tangent planes, 333–336
Solving linear equations vector equation of a line, 74
Gaussian elimination, 104–107 vector equation of planes, 76
inverse matrix method, 108 vector fields, 68, 260, 297, 300,
Standard deviation, 37–39, 42, 54–55, 310–313, 319, 321–327
58–60 vector product, 70
Standard form, 6–9, 180–181, 195 View factors, 291
Statistics, 31–39 Vorticity, 352
Stokes’ theorem, 343–347
Substitution method, 20 Z
Surface integration, 327–336 Zone model, 192, 219