DIFFERENTIAL EQUATIONS FOR ENGINEERS
This book presents a systematic and comprehensive introduction to ordinary differential
equations for engineering students and practitioners. Mathematical concepts and various
techniques are presented in a clear, logical, and concise manner. Various visual features
are used to highlight focus areas. Complete illustrative diagrams are used to facilitate
mathematical modeling of application problems. Readers are motivated by a focus on
the relevance of differential equations through their applications in various engineering
disciplines. Studies of various types of differential equations are determined by engi
neering applications. Theory and techniques for solving differential equations are then
applied to solve practical engineering problems. Detailed stepbystep analysis is pre
sented to model the engineering problems using differential equations from physical
principles and to solve the differential equations using the easiest possible method. Such
a detailed, stepbystep approach, especially when applied to practical engineering prob
lems, helps the readers to develop problemsolving skills.
This book is suitable for use not only as a textbook on ordinary differential equa
tions for undergraduate students in an engineering program but also as a guide to self
study. It can also be used as a reference after students have completed learning the
subject.
WeiChau Xie is a Professor in the Department of Civil and Environment Engineering
and the Department of Applied Mathematics at the University of Waterloo. He is the
author of Dynamic Stability of Structures and has published numerous journal articles
on dynamic stability, structural dynamics and randomvibration, nonlinear dynamics and
stochastic mechanics, reliability and safety analysis of engineering systems, and seismic
analysis and design of engineering structures. He has been teaching differential equa
tions to engineering students for almost twenty years. He received the Teaching Excel
lence Award in 2001 in recognition of his exemplary record of outstanding teaching,
concern for students, and commitment to the development and enrichment of engineer
ing education at Waterloo. He is the recipient of the Distinguished Teacher Award in
2007, which is the highest formal recognition given by the University of Waterloo for a
superior record of continued excellence in teaching.
Differential Equations for Engineers
WeiChau Xie
University of Waterloo
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore,
São Paulo, Delhi, Dubai, Tokyo
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
First published in print format
ISBN13 9780521194242
ISBN13 9780511776229
© WeiChau Xie 2010
2010
Information on this title: www.cambridge.org/9780521194242
This publication is in copyright. Subject to statutory exception and to the
provision of relevant collective licensing agreements, no reproduction of any part
may take place without the written permission of Cambridge University Press.
Cambridge University Press has no responsibility for the persistence or accuracy
of urls for external or thirdparty internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
eBook (NetLibrary)
Hardback
TO
My Family
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Motivating Examples 1
1.2 General Concepts and Deﬁnitions 6
2
FirstOrder and Simple HigherOrder Differential Equations . 16
2.1 The Method of Separation of Variables 16
2.2 Method of Transformation of Variables 20
2.2.1 Homogeneous Equations 20
2.2.2 Special Transformations 25
2.3 Exact Differential Equations and Integrating Factors 31
2.3.1 Exact Differential Equations 32
2.3.2 Integrating Factors 39
2.3.3 Method of Inspection 45
2.3.4 Integrating Factors by Groups 48
2.4 Linear FirstOrder Equations 55
2.4.1 Linear FirstOrder Equations 55
2.4.2 Bernoulli Differential Equations 58
2.5 Equations Solvable for the Independent or Dependent Variable 61
2.6 Simple HigherOrder Differential Equations 68
2.6.1 Equations Immediately Integrable 68
2.6.2 The Dependent Variable Absent 70
2.6.3 The Independent Variable Absent 72
2.7 Summary 74
Problems 78
3
Applications of FirstOrder and Simple HigherOrder Equations 87
3.1 Heating and Cooling 87
3.2 Motion of a Particle in a Resisting Medium 91
3.3 Hanging Cables 97
vii
viii contents
3.3.1 The Suspension Bridge 97
3.3.2 Cable under SelfWeight 102
3.4 Electric Circuits 108
3.5 Natural Puriﬁcation in a Stream 114
3.6 Various Application Problems 120
Problems 130
4
Linear Differential Equations . . . . . . . . . . . . . . . . . . . 140
4.1 General Linear Ordinary Differential Equations 140
4.2 Complementary Solutions 143
4.2.1 Characteristic Equation Having Real Distinct Roots 143
4.2.2 Characteristic Equation Having Complex Roots 147
4.2.3 Characteristic Equation Having Repeated Roots 151
4.3 Particular Solutions 153
4.3.1 Method of Undetermined Coefﬁcients 153
4.3.2 Method of Operators 162
4.3.3 Method of Variation of Parameters 173
4.4 Euler Differential Equations 178
4.5 Summary 180
Problems 183
5
Applications of Linear Differential Equations . . . . . . . . . . 188
5.1 Vibration of a Single DegreeofFreedom System 188
5.1.1 Formulation—Equation of Motion 188
5.1.2 Response of a Single DegreeofFreedom System 193
5.1.2.1 Free Vibration—Complementary Solution 193
5.1.2.2 ForcedVibration—Particular Solution 200
5.2 Electric Circuits 209
5.3 Vibration of a Vehicle Passing a Speed Bump 213
5.4 BeamColumns 218
5.5 Various Application Problems 223
Problems 232
6
The Laplace Transform and Its Applications . . . . . . . . . . . 244
6.1 The Laplace Transform 244
contents ix
6.2 The Heaviside Step Function 249
6.3 Impulse Functions and the Dirac Delta Function 254
6.4 The Inverse Laplace Transform 257
6.5 Solving Differential Equations Using the Laplace Transform 263
6.6 Applications of the Laplace Transform 268
6.6.1 Response of a Single DegreeofFreedom System 268
6.6.2 Other Applications 275
6.6.3 Beams on Elastic Foundation 283
6.7 Summary 289
Problems 291
7
Systems of Linear Differential Equations . . . . . . . . . . . . 300
7.1 Introduction 300
7.2 The Method of Operator 304
7.2.1 Complementary Solutions 304
7.2.2 Particular Solutions 307
7.3 The Method of Laplace Transform 318
7.4 The Matrix Method 325
7.4.1 Complementary Solutions 326
7.4.2 Particular Solutions 334
7.4.3 Response of Multiple DegreesofFreedom Systems 344
7.5 Summary 347
7.5.1 The Method of Operator 347
7.5.2 The Method of Laplace Transform 348
7.5.3 The Matrix Method 349
Problems 351
8
Applications of Systems of Linear Differential Equations . . . 357
8.1 Mathematical Modeling of Mechanical Vibrations 357
8.2 Vibration Absorbers or Tuned Mass Dampers 366
8.3 An Electric Circuit 372
8.4 Vibration of a TwoStory Shear Building 377
8.4.1 Free Vibration—Complementary Solutions 378
8.4.2 ForcedVibration—General Solutions 380
Problems 384
x contents
9
Series Solutions of Differential Equations . . . . . . . . . . . . 390
9.1 Review of Power Series 391
9.2 Series Solution about an Ordinary Point 394
9.3 Series Solution about a Regular Singular Point 403
9.3.1 Bessel’s Equation and Its Applications 408
9.3.1.1 Solutions of Bessel’s Equation 408
9.3.2 Applications of Bessel’s Equation 418
9.4 Summary 424
Problems 426
10
Numerical Solutions of Differential Equations . . . . . . . . . 431
10.1 Numerical Solutions of FirstOrder Initial Value Problems 431
10.1.1 The Euler Method or Constant Slope Method 432
10.1.2 Error Analysis 434
10.1.3 The Backward Euler Method 436
10.1.4 Improved Euler Method—Average Slope Method 437
10.1.5 The RungeKutta Methods 440
10.2 Numerical Solutions of Systems of Differential Equations 445
10.3 Stiff Differential Equations 449
10.4 Summary 452
Problems 454
11
Partial Differential Equations . . . . . . . . . . . . . . . . . . . 457
11.1 Simple Partial Differential Equations 457
11.2 Method of Separation of Variables 458
11.3 Application—Flexural Motion of Beams 465
11.3.1 Formulation—Equation of Motion 465
11.3.2 Free Vibration 466
11.3.3 ForcedVibration 471
11.4 Application—Heat Conduction 473
11.4.1 Formulation—Heat Equation 473
11.4.2 TwoDimensional SteadyState Heat Conduction 476
11.4.3 OneDimensional Transient Heat Conduction 480
11.4.4 OneDimensional Transient Heat Conduction on a SemiInﬁnite
Interval 483
contents xi
11.4.5 ThreeDimensional SteadyState Heat Conduction 488
11.5 Summary 492
Problems 493
12
Solving Ordinary Differential Equations Using Maple . . . . . 498
12.1 ClosedForm Solutions of Differential Equations 499
12.1.1 Simple Ordinary Differential Equations 499
12.1.2 Linear Ordinary Differential Equations 506
12.1.3 The Laplace Transform 507
12.1.4 Systems of Ordinary Differential Equations 509
12.2 Series Solutions of Differential Equations 512
12.3 Numerical Solutions of Differential Equations 517
Problems 526
Appendix A Tables of Mathematical Formulas . . . . . . . . . . . . . 531
a.1 Table of Trigonometric Identities 531
a.2 Table of Derivatives 533
a.3 Table of Integrals 534
a.4 Table of Laplace Transforms 537
a.5 Table of Inverse Laplace Transforms 539
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542
Preface
Background
Differential equations have wide applications in various engineering and science
disciplines. In general, modeling of the variation of a physical quantity, such as
temperature, pressure, displacement, velocity, stress, strain, current, voltage, or
concentration of a pollutant, with the change of time or location, or both would
result in differential equations. Similarly, studying the variation of some physical
quantities on other physical quantities would also lead to differential equations.
In fact, many engineering subjects, such as mechanical vibration or structural
dynamics, heat transfer, or theory of electric circuits, are founded on the theory of
differential equations. It is practically important for engineers to be able to model
physical problems using mathematical equations, and then solve these equations so
that the behavior of the systems concerned can be studied.
I have been teaching differential equations to engineering students for the past
two decades. Most, if not all, of the textbooks are written by mathematicians
with little engineering background. Based on my experience and feedback from
students, the following lists some of the gaps frequently seen in current textbooks:
❧ A major focus is put on explaining mathematical concepts
For engineers, the purpose of learning the theory of differential equations is
to be able to solve practical problems where differential equations are used.
For engineering students, it is more important to know the applications and
techniques for solving application problems than to delve into the nuances of
mathematical concepts andtheorems. Knowing the appropriate applications can
motivate them to study the mathematical concepts and techniques. However,
it is much more challenging to model an application problem using physical
principles and then solve the resulting differential equations than it is to merely
carry out mathematical exercises.
❧ Insufﬁcient emphasis is placed on the stepbystep problem solving techniques
Engineering students do not usually have the same mathematical background
and interest as students who major in mathematics. Mathematicians are more
interested if: (1) there are solutions to a differential equation or a system of
differential equations; (2) the solutions are unique under a certain set of con
ditions; and (3) the differential equations can be solved. On the other hand,
xiii
xiv preface
engineers are more interested in mathematical modeling of a practical problem
and actually solving the equations to ﬁnd the solutions using the easiest possible
method. Hence, a detailed stepbystep approach, especially applied to practical
engineering problems, helps students to develop problem solving skills.
❧ Presentations are usually formuladriven with little variation in visual design
It is very difﬁcult to attract students to read boring formulas without variation
of presentation. Readers often miss the points of importance.
Objectives
This book addresses the needs of engineering students and aims to achieve the
following objectives:
❧ To motivate students on the relevance of differential equations in engineering
through their applications in various engineering disciplines. Studies of various
types of differential equations are motivated by engineering applications; the
ory and techniques for solving differential equations are then applied to solve
practical engineering problems.
❧ To have a balance between theory and applications. This book could be used as a
reference after students have completedlearning the subject. As a reference, it has
to be reasonably comprehensive and complete. Detailed stepbystep analysis is
presented to model the engineering problems using differential equations and
to solve the differential equations.
❧ To present the mathematical concepts and various techniques in a clear, logical
and concise manner. Various visual features, such as sidenotes (preceded by
the symbol), different fonts and shades, are used to highlight focus areas.
Complete illustrative diagrams are used to facilitate mathematical modeling of
application problems. This book is not only suitable as a textbook for classroom
use but also is easy for selfstudy. As a textbook, it has to be easy to understand.
For selfstudy, the presentation is detailed with all necessary steps and useful
formulas given as sidenotes.
Scope
This book is primarily for engineering students and practitioners as the main
audience. It is suitable as a textbook on ordinary differential equations for under
graduate students in an engineering program. Such a course is usually offered in
the second year after students have taken calculus and linear algebra in the ﬁrst
year. Although it is assumed that students have a working knowledge of calculus
and linear algebra, some important concepts and results are reviewed when they are
ﬁrst used so as to refresh their memory.
preface xv
Chapter 1 ﬁrst presents some motivating examples, which will be studied in
detail later in the book, to illustrate how differential equations arise in engineer
ing applications. Some basic general concepts of differential equations are then
introduced.
In Chapter 2, various techniques for solving ﬁrstorder and simple higherorder
ordinary differential equations are presented. These methods are then applied in
Chapter 3 to study various application problems involving ﬁrstorder and simple
higherorder differential equations.
Chapter 4 studies linear ordinary differential equations. Complementary solu
tions are obtained through the characteristic equations and characteristic numbers.
Particular solutions are obtained using the method of undetermined coefﬁcients,
the operator method, and the method of variation of parameters. Applications
involving linear ordinary differential equations are presented in Chapter 5.
Solutions of linear ordinary differential equations using the Laplace transform
are studied in Chapter 6, emphasizing functions involving Heaviside step function
and Dirac delta function.
Chapter 7 studies solutions of systems of linear ordinary differential equations.
The method of operator, the method of Laplace transform, and the matrix method
are introduced. Applications involving systems of linear ordinary differential equa
tions are considered in Chapter 8.
In Chapter 9, solutions of ordinary differential equations in series about an
ordinary point and a regular singular point are presented. Applications of Bessel’s
equation in engineering are considered.
Some classical methods, including forward and backward Euler method, im
proved Euler method, and RungeKutta methods, are presented in Chapter 10 for
numerical solutions of ordinary differential equations.
In Chapter 11, the method of separation of variables is applied to solve partial
differential equations. When the method is applicable, it converts a partial differ
ential equation into a set of ordinary differential equations. Flexural vibration of
beams and heat conduction are studied as examples of application.
Solutions of ordinary differential equations using Maple are presentedinChapter
12. Symbolic computation software, such as Maple, is very efﬁcient in solving
problems involving ordinary differential equations. However, it cannot replace
learning andthinking, especially mathematical modeling. It is important todevelop
analytical skills and proﬁciency through “hand” calculations, as has been done in
previous chapters. This will also help the development of insight into the problems
and appreciation of the solution process. For this reason, solutions of ordinary
differential equations using Maple is presented in the last chapter of the book
instead of a scattering throughout the book.
xvi preface
The book covers a wide range of materials on ordinary differential equations
and their engineering applications. There are more than enough materials for a
oneterm (semester) undergraduate course. Instructors can select the materials
according to the curriculum. Drafts of this book were used as the textbook in a
oneterm undergraduate course at the University of Waterloo.
Acknowledgments
First and foremost, my sincere appreciation goes to my students. It is the students
who give me a stage where I can cultivate my talent and passion for teaching. It is
for the students that this book is written, as my small contribution to their success
in academic and professional careers. My undergraduate students who have used
the draft of this book as a textbook have made many encouraging comments and
constructive suggestions.
I am very grateful to many people who have reviewed and commented on the
book, including Professor HongJian Lai of West Virginia University, Professors S.T.
Ariaratnam, XinZhi Liu, Stanislav Potapenko, and EdwardVrscay of the University
of Waterloo.
My graduate students Mohamad Alwan, Qinghua Huang, Jun Liu, Shunhao Ni,
and RichardWiebe have carefully read the book and made many helpful and critical
suggestions.
My sincere appreciation goes to Mr. Peter Gordon, Senior Editor, Engineering,
Cambridge University Press, for his encouragement, trust, and hard work to publish
this book.
Special thanks are due to Mr. John Bennett, my mentor, teacher, and friend, for
his advice and guidance. He has also painstakingly proofread and copyedited this
book.
Without the unfailing love and support of my mother, who has always believed in
me, this work wouldnot have beenpossible. Inaddition, the care, love, patience, and
understanding of my wife CongRong and lovely daughters Victoria and Tiffany
have been of inestimable encouragement and help. I love them very much and
appreciate all that they have contributed to my work.
I appreciate hearing your comments through email (xie@uwaterloo.ca) or regu
lar correspondence.
WeiChau Xie
Waterloo, Ontario, Canada
1
C H A P T E R
Introduction
1.1 Motivating Examples
Differential equations have wide applications in various engineering and science
disciplines. In general, modeling variations of a physical quantity, such as tempera
ture, pressure, displacement, velocity, stress, strain, or concentration of a pollutant,
with the change of time t or location, such as the coordinates (x, y, z), or both
would require differential equations. Similarly, studying the variation of a physi
cal quantity on other physical quantities would lead to differential equations. For
example, the change of strain on stress for some viscoelastic materials follows a
differential equation.
It is important for engineers to be able to model physical problems using mathe
matical equations, andthensolve these equations sothat the behavior of the systems
concerned can be studied.
In this section, a fewexamples are presented to illustrate howpractical problems
are modeled mathematically and how differential equations arise in them.
Motivating Example 1
First consider the projectile of a mass m launched with initial velocity v
0
at angle
θ
0
at time t =0, as shown.
O
A
y
θ
0
x
v
0
θ
θ
x
v(t)
v(t)
βv
mg
y
1
2 1 introduction
The atmosphere exerts a resistance force on the mass, which is proportional
to the instantaneous velocity of the mass, i.e., R=βv, where β is a constant,
and is opposite to the direction of the velocity of the mass. Set up the Cartesian
coordinate system as shown by placing the origin at the point from where the mass
m is launched.
At time t, the mass is at location
_
x(t), y(t)
_
. The instantaneous velocity of the
mass in the x and ydirections are ˙ x(t) and ˙ y(t), respectively. Hence the velocity
of the mass is v(t) =
_
˙ x
2
(t)+ ˙ y
2
(t) at the angle θ(t) = tan
−1
_
˙ y(t)/˙ x(t)
_
.
The mass is subjected to two forces: the vertical downward gravity mg and the
resistance force R(t) =βv(t).
The equations of motion of the mass can be established using Newton’s Second
Law: F =
ma. The xcomponent of the resistance force is −R(t) cos θ(t). In
the ydirection, the component of the resistance force is −R(t) sin θ(t). Hence,
applying Newton’s Second Law yields
xdirection: ma
x
=
F
x
=⇒ m¨ x(t) = −R(t) cos θ(t),
ydirection: ma
y
=
F
y
=⇒ m¨ y(t) = −mg − R(t) sin θ(t).
Since
θ(t) = tan
−1
˙ y(t)
˙ x(t)
=⇒ cos θ =
˙ x(t)
_
˙ x
2
(t)+ ˙ y
2
(t)
, sin θ =
˙ y(t)
_
˙ x
2
(t)+ ˙ y
2
(t)
,
the equations of motion become
m¨ x(t) = −βv(t) ·
˙ x(t)
_
˙ x
2
(t)+ ˙ y
2
(t)
=⇒ m¨ x(t) + β ˙ x(t) = 0,
m¨ y(t) = −mg − βv(t) ·
˙ y(t)
_
˙ x
2
(t)+ ˙ y
2
(t)
=⇒ m¨ y(t) + β ˙ y(t) = −mg,
in which the initial conditions are at time t =0: x(0) =0, y(0) =0, ˙ x(0) =v
0
cos θ
0
,
˙ y(0) =v
0
sin θ
0
. The equations of motion are two equations involving the ﬁrst and
secondorder derivatives ˙ x(t), ˙ y(t), ¨ x(t), and ¨ y(t). These equations are called, as
will be deﬁned later, a system of two secondorder ordinary differential equations.
Because of the complexity of the problems, in the following examples, the prob
lems are described and the governing equations are presented without detailed
derivation. These problems will be investigated in details in later chapters when
applications of various types of differential equations are studied.
Motivating Example 2
Atank contains a liquid of volume V(t), which is polluted with a pollutant concen
tration in percentage of c(t) at time t. To reduce the pollutant concentration, an
1.1 motivating examples 3
inﬂow of rate Q
in
is injected to the tank. Unfortunately, the inﬂow is also polluted
but to a lesser degree with a pollutant concentration c
in
. It is assumed that the
inﬂow is perfectly mixed with the liquid in the tank instantaneously. An outﬂow
of rate Q
out
is removed from the tank as shown. Suppose that, at time t =0, the
volume of the liquid is V
0
with a pollutant concentration of c
0
.
Inflow
Outflow
Volume V(t)
Concentration c(t)
Q
out
, c(t)
Q
in
, c
in
The equation governing the pollutant concentration c(t) is given by
_
V
0
+ (Q
in
−Q
out
)t
_
dc(t)
dt
+ Q
in
c(t) = Q
in
c
in
,
with initial condition c(0) =c
0
. This is a ﬁrstorder ordinary differential equation.
Motivating Example 3
Hanger
Deck
Cable
w(x)
O
y
x
Consider the suspension bridge as shown, which consists of the main cable, the
hangers, and the deck. The selfweight of the deck and the loads applied on the
deck are transferred to the cable through the hangers.
4 1 introduction
Set up the Cartesian coordinate systemby placing the origin Oat the lowest point
of the cable. The cable can be modeled as subjected to a distributed load w(x). The
equation governing the shape of the cable is given by
d
2
y
dx
2
=
w(x)
H
,
where H is the tension in the cable at the lowest point O. This is a secondorder
ordinary differential equation.
Motivating Example 4
k
Reference position
m
c
x(t)
x
0
(t) y(t)
Consider the vibration of a singlestory shear building under the excitation of
earthquake. The shear building consists of a rigid girder of mass m supported by
columns of combined stiffness k. The vibration of the girder can be described by
the horizontal displacement x(t). The earthquake is modeled by the displacement
of the ground x
0
(t) as shown. When the girder vibrates, there is a damping force
due to the internal friction between various components of the building, given by
c
_
˙ x(t)−˙ x
0
(t)
_
, where c is the damping coefﬁcient.
The relative displacement y(t) =x(t)−x
0
(t) between the girder and the ground
is governed by the equation
m¨ y(t) + c ˙ y(t) + ky(t) = −m¨ x
0
(t),
which is a secondorder linear ordinary differential equation.
Motivating Example 5
In many engineering applications, an equipment of mass m is usually mounted on
a supporting structure that can be modeled as a spring of stiffness k and a damper
of damping coefﬁcient c as shown in the following ﬁgure. Due to unbalanced mass
inrotating components or other excitationmechanisms, the equipment is subjected
to a harmonic force F
0
sin t. The vibration of the mass is described by the vertical
displacement x(t). When the excitation frequency is close to ω
0
=
_
k/m, which
is the natural circular frequency of the equipment and its support, vibration of large
amplitudes occurs.
1.1 motivating examples 5
In order to reduce the vibration of the equipment, a vibration absorber is
mounted on the equipment. The vibration absorber can be modeled as a mass
m
a
, a spring of stiffness k
a
, and a damper of damping coefﬁcient c
a
. The vibration
of the absorber is described by the vertical displacement x
a
(t).
x(t)
Vibration
Absorber
Supporting
Structure
Equipment
x
a
(t)
F
0
sint
c
m
k
c
a
m
a
k
a
The equations of motion governing the vibration of the equipment and the
absorber are given by
m¨ x + (c +c
a
) ˙ x + (k+k
a
)x − c
a
˙ x
a
− k
a
x
a
= F
0
sin t,
m
a
¨ x
a
+ c
a
˙ x
a
+ k
a
x
a
− c
a
˙ x − k
a
x = 0,
which comprises a system of two coupled secondorder linear ordinary differential
equations.
Motivating Example 6
L
v
P P
EI, ρA
Ut
t =0
x
A bridge may be modeled as a simply supported beam of length L, mass density
per unit length ρA, and ﬂexural rigidity EI as shown. Avehicle of weight P crosses
the bridge at a constant speed U. Suppose at time t =0, the vehicle is at the left end
of the bridge and the bridge is at rest. The deﬂection of the bridge is v(x, t), which
is a function of both location x and time t. The equation governing v(x, t) is the
partial differential equation
ρA
∂
2
v(x, t)
∂t
2
+ EI
∂
4
v(x, t)
∂x
4
= Pδ(x−Ut),
6 1 introduction
where δ(x−a) is the Dirac delta function. The equation of motion satisﬁes the
initial conditions
v(x, 0) = 0,
∂v(x, t)
∂t
¸
¸
¸
t=0
= 0,
and the boundary conditions
v(0, t) = v(L, t) = 0,
∂
2
v(x, t)
∂x
2
¸
¸
¸
x=0
=
∂
2
v(x, t)
∂x
2
¸
¸
¸
x=L
= 0.
1.2 General Concepts and Deﬁnitions
In this section, some general concepts and deﬁnitions of ordinary and partial
differential equations are presented.
Let x be an independent variable and y be a dependent variable. An equation
that involves x, y and various derivatives of y is called a differential equation (DE).
For example,
dy
dx
= 2y + sin x,
_
dy
dx
_
3
+ e
x
+ 2 =
d
2
y
dx
2
are differential equations.
Deﬁnition — Ordinary Differential Equation
In general, an equation of the form
F
_
x, y,
dy
dx
, . . . ,
d
n
y
dx
n
_
= 0
is an Ordinary Differential Equation (ODE).
It is called an ordinary differential equation because there is only one independent
variable and only ordinary derivatives (not partial derivatives) are involved.
Deﬁnition — Order of a Differential Equation
The order of a differential equation is the order of the highest derivative appearing
in the differential equation.
Deﬁnition — Linear and Nonlinear Differential Equations
If y and its various derivatives y
, y
, . . . appear linearly in the equation, it is a
linear differential equation; otherwise, it is nonlinear.
For example,
d
2
y
dx
2
+ ω
2
y = sin x, ω =constant, Secondorder, linear
_
dy
dx
_
2
+ 4y = cos x, Firstorder, nonlinear because of the term
_
dy
dx
_
2
1.2 general concepts and deﬁnitions 7
x
3
d
3
y
dx
3
+ 5x
dy
dx
+ 6y = e
x
, Thirdorder, linear
d
2
y
dx
2
+ y
dy
dx
+ 2y = x. Secondorder, nonlinear because of the term y
dy
dx
Sometimes, the roles of independent and dependent variables can be exchanged
to render a differential equation linear. For example,
d
2
x
dy
2
− x
√
y = 5
is a secondorder linear equation with y being regarded as the independent variable
and x the dependent variable.
In some applications, the roles of independent and dependent variables are obvi
ous. For example, in a differential equation governing the variation of temperature
T with time t, the time variable t is the independent variable and the temperature T
is the dependent variable; time t cannot be the dependent variable. In other appli
cations, the roles of independent and dependent variables are interchangeable. For
example, in a differential equation governing the relationship between temperature
T and pressure p, the temperature T can be considered as the independent variable
and the pressure p the dependent variable, or vice versa.
Deﬁnition — Linear Ordinary Differential Equations
The general form of an nthorder linear ordinary differential equation is
a
n
(x)
d
n
y
dx
n
+ a
n−1
(x)
d
n−1
y
dx
n−1
+ · · · + a
1
(x)
dy
dx
+ a
0
(x) y = f (x).
If a
0
(x), a
1
(x), . . . , a
n
(x) are constants, the ordinary differential equation is said
to have constant coefﬁcients; otherwise it is said to have variable coefﬁcients.
For example,
d
2
y
dx
2
+ 0.1
dy
dx
+ 4y = 10 cos 2x, Secondorder linear, constant coefficients
x
2
d
2
y
dx
2
+ x
dy
dx
+ (x
2
−ν
2
) y = 0, x >0, ν 0 is a constant.
Secondorder linear, variable coefficients (Bessel's equation)
Deﬁnition — Homogeneous and Nonhomogeneous Differential Equations
A differential equation is said to be homogeneous if it has zero as a solution;
otherwise, it is nonhomogeneous.
8 1 introduction
For example,
d
2
y
dx
2
+ 0.1
dy
dx
+ 4y = 0, Homogeneous
d
2
y
dx
2
+ 0.1
dy
dx
+ 4y = 2 sin 2x + 5 cos 3x. Nonhomogeneous
Note that a homogeneous differential equation may have distinctively different
meanings in different situations (see Section 2.2).
Partial Differential Equations
Deﬁnition — Partial Differential Equations
If the dependent variable u is a function of more than one independent variable,
say x
1
, x
2
, . . . , x
m
, an equation involving the variables x
1
, x
2
, . . . , x
m
, u and
various partial derivatives of u with respect to x
1
, x
2
, . . . , x
m
is called a Partial
Differential Equation (PDE).
For example,
∂
2
u
∂x
2
=
1
α
∂u
∂t
, α =constant, Heat equation in onedimension
∂
2
u
∂x
2
+
∂
2
u
∂y
2
= f (x, y),
Poisson's equation in twodimensions
Laplace's equation if f (x, y) =0
∂
4
u
∂x
4
+ 2
∂
4
u
∂x
2
∂y
2
+
∂
4
u
∂y
4
= 0, Biharmonic equation in twodimensions
∂
2
u
∂x
2
+
∂
2
u
∂y
2
+
∂
2
u
∂z
2
=
1
α
∂u
∂t
, Heat equation in threedimensions
∂
2
u
∂x
2
+
∂
2
u
∂y
2
+
∂
2
u
∂z
2
= 0. Laplace's equation in threedimensions
General and Particular Solutions
Deﬁnition — Solution of a Differential Equation
For an nthorder ordinary differential equation F
_
x, y, y
, . . . , y
(n)
_
=0, a func
tion y = y(x), which is n times differentiable and satisﬁes the differential equation
in some interval a<x<b when substituted into the equation, is called a solution
of the differential equation over the interval a<x<b.
Consider the ﬁrstorder differential equation
dy
dx
= 3.
1.2 general concepts and deﬁnitions 9
Integrating with respect to x yields the general solution
y = 3x + C, C =constant.
The general solution of the differential equation, which includes all possible solu
tions, is a family of straight lines with slope equal to 3. On the other hand, y =3x
is a particular solution passing through the origin, with the constant C being 0.
Consider the differential equation
d
3
y
dx
3
= 48x.
Integrating both sides of the equation with respect to x gives
d
2
y
dx
2
= 24x
2
+ C
1
.
Integrating with respect to x again yields
dy
dx
= 8x
3
+ C
1
x + C
2
.
Integrating with respect to x once more results in the general solution
y = 2x
4
+
1
2
C
1
x
2
+ C
2
x + C
3
,
where C
1
, C
2
, C
3
are arbitrary constants. When the constants C
1
, C
2
, C
3
take speciﬁc
values, one obtains particular solutions. For example,
y = 2x
4
+ 3x
2
+ 1, C
1
= 6, C
2
= 0, C
3
= 1,
y = 2x
4
+ x
2
+ 3x + 5, C
1
= 2, C
2
= 3, C
3
= 5,
are two particular solutions.
Remarks: In general, an nthorder ordinary differential equation will contain
n arbitrary constants in its general solution. Hence, for an nthorder ordinary
differential equation, n conditions are required to determine the n constants to
yield a particular solution.
In applications, there are usually two types of conditions that can be used to deter
mine the constants.
Illustrative Example
Consider the motion of an object dropped vertically at time t =0 from x =0 as
shown in the following ﬁgure. Suppose that there is no resistance fromthe medium.
10 1 introduction
t =0, x=0
t , x, v
mg
x
The equation of motion is given by
d
2
x
dt
2
= g,
and the general solution is, by integrating both sides of the equation with respect to
t twice,
x(t) = C
0
+ C
1
t +
1
2
gt
2
.
The following are two possible ways of specifying the conditions.
Initial Value Problem
If the object is dropped with initial velocity v
0
, the conditions required are
at time t = 0: x(0) = 0, ˙ x(0) =
dx
dt
¸
¸
¸
t=0
= v
0
.
The constants C
0
and C
1
can be determined from these two conditions and the
solution of the differential equation is
x(t) = v
0
t +
1
2
gt
2
.
In this case, the differential equation is required to satisfy conditions speciﬁed at
one value of t, i.e., t =0.
Deﬁnition — Initial Value Problem
If a differential equation is required to satisfy conditions on the dependent vari
able and its derivatives speciﬁed at one value of the independent variable, these
conditions are called initial conditions and the problem is called an initial value
problem.
Boundary Value Problem
If the object is required to reach x =L at time t =T, L
1
2
gT
2
, the conditions can
be speciﬁed as
at time t =0: x(0) =0; at time t =T : x(T) =L.
1.2 general concepts and deﬁnitions 11
The solution of the differential equation is
x(t) =
_
L
T
−
1
2
gT
_
t +
1
2
gt
2
.
In this case, the differential equation is required to satisfy conditions speciﬁed at
two values of t, i.e., t =0 and t =T.
Deﬁnition — Boundary Value Problem
If a differential equationis requiredtosatisfy conditions onthe dependent variable
and possibly its derivatives speciﬁed at two or more values of the independent
variable, these conditions are called boundary conditions and the problemis called
a boundary value problem.
Existence and Uniqueness of Solutions
Note that y
is the slope of curve y = y(x) on the xy plane. Hence, solving
differential equation y
= f (x, y) means ﬁnding curves whose slope at any given
point (x, y) is equal to f (x, y). Solving the initial value problem y
= f (x, y) with
y(x
0
) = y
0
means ﬁnding curves passing through point (x
0
, y
0
) whose slope at any
given point (x, y) is equal to f (x, y).
This can be better visualized using direction ﬁelds. At a given point (x, y) in
region R, one can drawa short straight line whose slope is f (x, y). Adirection ﬁeld
as shown in Figure 1.1 is then obtained if this is done for a large number of points.
x
C
(x
0
, y
0
)
y
R
Figure 1.1 Direction ﬁeld.
Determining the general solution of y
= f (x, y) is then ﬁnding the curves that
are tangent to the short straight line at each point (x, y). Determining the solution
of the initial value problem y
= f (x, y) with y(x
0
) = y
0
means ﬁnding curves
passing through point (x
0
, y
0
) and are tangent to the short straight line at each
point (x, y).
12 1 introduction
Theorem — Existence and Uniqueness
Consider the initial value problem
y
= f (x, y), y(x
0
) = y
0
,
where f (x, y) is a continuous function in the rectangular region
R :
¸
¸
x−x
0
¸
¸
a,
¸
¸
y−y
0
¸
¸
b, a>0, b>0.
Suppose f (x, y) also satisﬁes the Lipschitz condition with respect to y in R, i.e.,
there exists a constant L>0 such that, for every (x, y
1
) and (x, y
2
) in R,
¸
¸
f (x, y
1
) − f (x, y
2
)
¸
¸
L
¸
¸
y
1
− y
2
¸
¸
.
Then there exists a unique solution y =ϕ(x), continuous on
¸
¸
x−x
0
¸
¸
h and
satisfying the initial condition ϕ(x
0
) = y
0
, where
h = min
_
a,
b
M
_
, M = max
¸
¸
f (x, y)
¸
¸
in R.
The graphical interpretation of the Existence and Uniqueness Theorem is that, in
region R in which the speciﬁed conditions hold, passing through any given point
(x
0
, y
0
) there exists one and only one curve C such that the slope of curve C at any
point (x, y) in R is equal to f (x, y).
Remarks:
❧ It can be shown that if ∂f (x, y)/∂y is continuous in R, then f (x, y) satisfies
the Lipschitz condition. Because it is generally difficult to checkthe Lipschitz
condition, the Lipschitz condition is often replaced by the stronger condition
of continuous partial derivative ∂f (x, y)/∂y in R.
❧ The Existence and Uniqueness Theorem is a sufficient condition, meaning
that the existence and uniqueness of the solution is guaranteed when the
specified conditions hold. It is not a necessary condition, implying that,
evenwhen the specified conditions are not all satisfied, there may still exist a
unique solution.
Example 1.1 1.1
Knowing that y =Cx
2
satisﬁes x y
=2y, discuss the existence and uniqueness of
solutions of the initial value problem
x y
= 2y, y(x
0
) =y
0
,
for the following three cases
(1) x
0
=0; (2) x
0
=0, y
0
=0; (3) x
0
=0, y
0
=0.
1.2 general concepts and deﬁnitions 13
Since f (x, y) =
2y
x
,
∂f (x, y)
∂y
=
2
x
,
the conditions of the Existence and Uniqueness Theorem are not satisﬁed in a
region including points with x =0.
With the help of the direction ﬁeld as shown in the following ﬁgure, the solution
of the initial value problem can be easily obtained.
x
y
(x
0
, y
0
)
(1) x
0
=0
(a) If x
0
>0, then, in the region R with x >0, there exists a unique solution to
the initial value problem
y =
y
0
x
2
0
x
2
, x >0.
(b) If x
0
<0, then, in the region R with x<0, there exists a unique solution to
the initial value problem
y =
y
0
x
2
0
x
2
, x<0.
(c) If x
0
>0, then, in the region R including x =0, the solution to the initial value
problem is not unique
y =
⎧
⎨
⎩
y
0
x
2
0
x
2
, x 0,
ax
2
, x<0, a is a constant.
(d) If x
0
<0, then, in the region R including x =0, the solution to the initial
value problem is not unique
y =
⎧
⎨
⎩
ax
2
, x >0, a is a constant,
y
0
x
2
0
x
2
, x 0.
14 1 introduction
(2) x
0
=0, y
0
=0
Passing through (0, 0), there are inﬁnitely many solutions
y =
_
ax
2
, x 0,
bx
2
, x<0,
a, b are constants.
(3) x
0
=0, y
0
= 0
There are no solutions passing through point (x
0
, y
0
) with y
0
=0.
Remarks: Whether or not a givenfunctionis a solutionof a differential equation
can be checked by substituting the function into the differential equation along
with the initial or boundary conditions if there are any.
Example 1.2 1.2
Show that
y = C
1
e
−4x
+ C
2
e
x
−
3
125
(13 sin 3x + 9 cos 3x)
is a solution of the differential equation
y
+ 3y
− 4y = 6 sin 3x.
Differentiating y successively twice yields
y
= −4C
1
e
−4x
+ C
2
e
x
−
3
125
(39 cos 3x − 27 sin 3x),
y
= 16C
1
e
−4x
+ C
2
e
x
−
3
125
(−117 sin 3x − 81 cos 3x).
Substituting into the differential equation gives
y
+ 3y
− 4y = 16C
1
e
−4x
+ C
2
e
x
−
3
125
(−117 sin 3x − 81 cos 3x)
− 12C
1
e
−4x
+ 3C
2
e
x
−
3
125
( −81 sin 3x + 117 cos 3x)
− 4C
1
e
−4x
− 4C
2
e
x
−
3
125
( −52 sin 3x − 36 cos 3x)
= 6 sin 3x.
Hence
y = C
1
e
−4x
+ C
2
e
x
−
3
125
(13 sin 3x + 9 cos 3x)
is a solution of the differential equation.
1.2 general concepts and deﬁnitions 15
Example 1.3 1.3
Show that
u(x, t) = 2 sin
3πx
L
exp
_
−
π
2
L
2
t
_
is a solution of the partial differential equation
1
9
∂
2
u
∂x
2
=
∂u
∂t
,
with the initial condition
u(x, 0) = 2 sin
3πx
L
, for 0x L;
and the boundary conditions
u(0, t) = 0, u(L, t) = 0, for t >0.
Evaluate the partial derivatives
∂u
∂t
= 2 sin
3πx
L
·
_
−
π
2
L
2
_
exp
_
−
π
2
L
2
t
_
,
∂u
∂x
= 2·
_
3π
L
_
cos
3πx
L
· exp
_
−
π
2
L
2
t
_
,
∂
2
u
∂x
2
= 2· (−)
_
3π
L
_
2
sin
3πx
L
· exp
_
−
π
2
L
2
t
_
.
Substitute into the differential equation
L.H.S. =
1
9
∂
2
u
∂x
2
= −
2π
2
L
2
sin
3πx
L
exp
_
−
π
2
L
2
t
_
R.H.S. =
∂u
∂t
= −
2π
2
L
2
sin
3πx
L
exp
_
−
π
2
L
2
t
_
⎫
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎭
=⇒ L.H.S. = R.H.S.
Check the initial and boundary conditions
u(x, 0) = 2 sin
3πx
L
exp
_
−
π
2
L
2
· 0
_
= 2 sin
3πx
L
, satisﬁed,
u(0, t) = 2 sin
3π· 0
L
exp
_
−
π
2
L
2
t
_
= 0, satisﬁed,
u(L, t) = 2 sin
3π· L
L
exp
_
−
π
2
L
2
t
_
= 0, satisﬁed.
Hence
u(x, t) = 2 sin
3πx
L
exp
_
−
π
2
L
2
t
_
is a solution of the partial differential equation with the given initial and boundary
conditions.
2
C H A P T E R
FirstOrder
and Simple HigherOrder
Differential Equations
There are various techniques for solving ﬁrstorder and simple higherorder ordi
nary differential equations. The key in the application of the speciﬁc technique
hinges on the identiﬁcation of the type of a given equation. The objectives of this
chapter are to introduce various types of ﬁrstorder and simple higherorder dif
ferential equations and the corresponding techniques for solving these differential
equations.
In this chapter, it is assumed that x is the independent variable and y is the
dependent variable. Solutions in the explicit form y =η(x) or in the implicit form
u(x, y) =0 are sought.
2.1 The Method of Separation of Variables
Consider a ﬁrstorder ordinary differential equation of the form
dy
dx
= F(x, y).
Suppose that the righthand side F(x, y), which is a function of x and y, can be
written as a product of a function of x and a function of y, i.e.,
F(x, y) = f (x) · φ( y).
For example, the functions
e
x +y
2
= e
x
· e
y
2
, x y + x + 2y + 2 = (x+2) · ( y+1)
16
2.1 the method of separation of variables 17
can be separated into a product of a function of x and a function of y, but the
following functions cannot be separated
ln(x+2y), sin(x
2
+y), x y
2
+ x
2
.
This type of differential equationis calledvariable separable or separable differential
equations. The equations can be solved by the method of separation of variables.
Rewrite the equation as
dy
dx
= f (x) · φ( y).
Case 1. If φ( y) =0, moving terms involving variable y to the lefthand side and
terms of variable x to the righthand side yields
g( y)dy = f (x)dx, g( y) =
1
φ( y)
.
. ,, . . ,, .
function of y only function of x only
Integrating both sides of the equation results in the general solution
_
g( y)dy =
_
f (x)dx + C,
where C is an arbitrary constant.
Remarks: When dividing a differential equation by a function, it is important
to ensure that the function is not zero. Otherwise, solutions may be lost in the
process. Hence, the casewhenthe function is zero should be considered separately
to determine if it yields extra solutions.
Case 2. If φ( y) =0, solve for the roots of this equation. Let y = y
0
be one of
the solutions of equation φ( y) =0. Then y = y
0
is a solution of the differential
equation. Note that sometimes the solution y = y
0
may already be included in the
general solution obtained from Case 1.
Remarks: It should be emphasized that, only when one side of the equation
contains only variable x and the other side of the equation contains only variable
y, the equation can be integrated to obtain the general solution.
Example 2.1 2.1
Solve
dy
dx
+
1
y
e
y
2
+3x
= 0, y =0.
Separating the variables yields
−ye
−y
2
dy = e
3x
dx.
18 2 ﬁrstorder and simple higherorder differential equations
Integrating both sides to obtain the general solution leads to
−
_
ye
−y
2
dy =
_
e
3x
dx + C,
1
2
_
e
−y
2
d( −y
2
) =
1
3
_
e
3x
d( 3x ) + C, d(−y
2
) = −2ydy
∴
1
2
e
−y
2
=
1
3
e
3x
+ C.
_
e
x
dx =e
x
General solution
Example 2.2 2.2
Solve tan ydx − cot xdy = 0, cos y =0, sin x =0.
The equation can be written as
sin y
cos y
dx =
cos x
sin x
dy.
To separate the variables, multiply the differential equation by
sin x cos y
cos x sin y
; it is
required that sin y not be zero.
Case 1. If sin y =0, separating the variables yields
sin x
cos x
dx =
cos y
sin y
dy.
Integrating both sides results in the general solution
_
sin x
cos x
dx =
_
cos y
sin y
dy + C,
−
_
1
cos x
d( cos x ) =
_
1
sin y
d( sin y ) + C,
d(sin x) = cos xdx
d(cos x) = −sin xdx
∴ −ln
¸
¸
cos x
¸
¸
= ln
¸
¸
sin y
¸
¸
+ C.
_
1
x
dx = ln
¸
¸
x
¸
¸
The result can be simpliﬁed as follows
ln
¸
¸
cos x · sin y
¸
¸
= −C, ln a+ln b= ln(a· b)
¸
¸
cos x · sin y
¸
¸
= e
−C
=⇒ cos x · sin y = A. ±e
−C
=⇒ A
Since C is an arbitrary constant, A is an arbitrary constant, which can in turn be
renamed as C. The general solution becomes
cos x · sin y = C. General solution
Case 2. If sin y =0, one has y =kπ, k =0, ±1, ±2, . . . . It is obvious that y =kπ
or sin y =0 is a solution of the differential equation. However, sin y =0 is already
included in the general solution cos x · sin y =C, with C=0, obtained in Case 1.
2.1 the method of separation of variables 19
Example 2.3 2.3
Solve x y
3
dx + ( y+1)e
−x
dy = 0.
Case 1. If y =0, separating the variables leads to
xe
x
dx = −
y+1
y
3
dy.
Integrating both sides results in the general solution
_
xe
x
dx = −
_
y+1
y
3
dy + C.
The integrals are evaluated as
_
xe
x
dx =
_
xd(e
x
) d(e
x
) =e
x
dx, Integration by parts
= xe
x
−
_
e
x
dx = xe
x
− e
x
,
−
_
y+1
y
3
dy = −
_
( y
−2
+ y
−3
)dy
= −
_
y
−1
−1
+
y
−2
−2
_
=
1
y
+
1
2y
2
.
_
x
n
dx =
x
n+1
n+1
The general solution becomes
(x−1)e
x
=
1
y
+
1
2y
2
+ C.
Case 2. It is easy to verify that y =0 is a solution of the differential. This solution
cannot be obtained from the general solution for any value of the constant C.
Deﬁnition — Singular Solution
Any solutions of a differential equation that cannot be obtained from the general
solution for any values of the arbitrary constants are called singular solutions.
Hence, combining Cases 1 and 2, the solutions of the differential equation are
(x−1)e
x
= −
1
y
−
1
2y
2
+ C, General solution
y = 0. Singular solution
Remarks: Avariable separable equation is very easy to identify, and it is easy to
express the general solution in terms of integrals. However, the actual evaluation
of the integrals may sometimes be quite challenging.
20 2 ﬁrstorder and simple higherorder differential equations
2.2 Method of Transformation of Variables
2.2.1 Homogeneous Equations
Equations of the type
dy
dx
= f
_
y
x
_
(1)
are called homogeneous differential equations. For example,
g(x, y) =
x
2
+3y
2
x
2
−x y+y
2
=
1+3
_
y
x
_
2
1−
_
y
x
_
+
_
y
x
_
2
= f
_
y
x
_
,
g(x, y) = ln x − ln y = ln
_
x
y
_
= −ln
_
y
x
_
= f
_
y
x
_
.
Remarks: A homogeneous differential equation has several distinct meanings:
❧ A firstorder ordinary differential equation of the form
dy
dx
= f
_
y
x
_
is of
the type of homogeneous equation.
❧ A homogeneous differential equation, defined in Chapter 1, means that the
differential equation has zero as a solution.
Ahomogeneous equation can be converted to a variable separable equation using a
transformation of variables. Let v =
y
x
be the new dependent variable, while x is
still the independent variable. Hence
y = xv =⇒
dy
dx
= v + x
dv
dx
.
Substituting into differential equation (1) leads to
v + x
dv
dx
= f (v) =⇒ x
dv
dx
= f (v) − v.
Case 1. f (v)−v =0. One has y =v
0
x, where v
0
is the solution of f (v
0
)−v
0
=0.
Case 2. f (v)−v =0. Separating the variables leads to
dv
f (v) − v
=
dx
x
.
The transformed differential equation is variable separable. Integrating both sides
gives the general solution
_
dv
f (v) − v
=
_
dx
x
+ C.
2.2 method of transformation of variables 21
Example 2.4 2.4
Solve
dy
dx
+
x
y
+ 2 = 0, y =0, y(0) =1.
The differential equation is homogeneous. Letting v =
y
x
,
y = xv =⇒
dy
dx
= v + x
dv
dx
,
the differential equation becomes
v+x
dv
dx
+
1
v
+2 = 0 =⇒ x
dv
dx
= −
_
v+
1
v
+2
_
=⇒ x
dv
dx
= −
(v+1)
2
v
.
Case 1. v = −1 =⇒ y = −x. But it does not satisfy the condition y(0) =1.
Case 2. v = −1, separating the variables yields
v
(v+1)
2
dv = −
1
x
dx.
Integrating both sides gives
_
v
(v+1)
2
dv = −
_
1
x
dx + C.
Since
_
v
(v+1)
2
dv =
_
(v+ 1 ) − 1
(v+1)
2
dv =
_ _
1
v+1
−
1
(v+1)
2
_
dv
=
_
1
v+1
d( v+1 ) −
_
1
( v+1 )
2
d( v+1 )
= ln
¸
¸
v+1
¸
¸
+
1
v+1
,
_
1
x
dx = ln
¸
¸
x
¸
¸
,
_
1
x
2
dx = −
1
x
one obtains
ln
¸
¸
v+1
¸
¸
+
1
v+1
= −ln
¸
¸
x
¸
¸
+ C.
Converting back to the original variables x and y results in the general solution
ln
¸
¸
¸
y
x
+1
¸
¸
¸ + ln
¸
¸
x
¸
¸
+
1
y
x
+1
= C, ln a+ln b= ln(a· b)
∴ ln
¸
¸
y+x
¸
¸
+
x
y+x
= C. General solution
The constant C is determined using the initial condition y(0) =1
ln
¸
¸
1+0
¸
¸
+
0
1+0
= C =⇒ C = 0.
22 2 ﬁrstorder and simple higherorder differential equations
The particular solution satisfying y(0) =1 is
ln
¸
¸
y+x
¸
¸
+
x
y+x
= 0. Particular solution
Example 2.5 2.5
Solve x(ln x − ln y)dy − ydx = 0, x >0, y >0.
Dividing both sides of the equation by x gives
_
ln
x
y
_
dy −
y
x
dx = 0 =⇒
dy
dx
=
y
x
−ln
y
x
,
which is homogeneous. Putting v =
y
x
, v >0,
y = xv =⇒
dy
dx
= v + x
dv
dx
,
the equation becomes
v + x
dv
dx
=
v
−ln v
=⇒ x
dv
dx
= −
v
ln v
− v = −v
1+ln v
ln v
.
Case 1. For ln v = −1, the equation can be separated as
ln v
v(1+ln v)
dv = −
1
x
dx.
Integrating both sides yields
_
ln v
v(1+ln v)
dv = −
_
1
x
dx + C.
Since d(ln v) =
1
v
dv, one has
_
ln v
v(1+ln v)
dv =
_
( 1 + ln v) − 1
1+ln v
d(ln v)
=
_
_
1 −
1
1+ln v
_
d(ln v) = ln v − ln
¸
¸
1+ln v
¸
¸
.
Hence,
ln
¸
¸
¸
v
1+ln v
¸
¸
¸ = −ln x + C.
Replacing v by the original variables results in the general solution
ln
¸
¸
¸
¸
y/x
1+ln( y/x)
¸
¸
¸
¸
+ ln x = C,
2.2 method of transformation of variables 23
ln
¸
¸
¸
¸
y
1+ln y−ln x
¸
¸
¸
¸
= ln C,
Since C is an arbitrary constant,
it is rewritten as ln C.
∴
y
1+ln y−ln x
= C.
Case 2. If ln v = −1, one has v =e
−1
. Hence, v = y/x =e
−1
or ey =x is a
solution. This solution cannot be obtained from the general solution for any value
of the constant C and is therefore a singular solution.
Combining Cases 1 and 2, the solutions of the differential equations are
y
1+ln y−ln x
= C, General solution
e y = x. Singular solution
Example 2.6 2.6
Solve ( y+x)dy + (x−y)dx = 0.
Since y = −x is not a solution, the equation can be written as
dy
dx
= −
x − y
y + x
= −
1 −
y
x
y
x
+ 1
,
Dividing both the numerator
and denominator by x, x =0
which is a homogeneous equation. Letting v =
y
x
,
y = xv =⇒
dy
dx
= v + x
dv
dx
,
the equation becomes
v + x
dv
dx
= −
1−v
v+1
,
x
dv
dx
= −
1−v
v+1
− v = −
1−v+v(v+1)
v+1
= −
v
2
+1
v+1
.
Since v
2
+1=0, separating the variables gives
v+1
v
2
+1
dv = −
1
x
dx.
Integrating both sides yields
_
v
v
2
+1
dv +
_
1
v
2
+1
dv = −
_
1
x
dx + C,
1
2
_
1
v
2
+1
d( v
2
+1 ) + tan
−1
v = −ln
¸
¸
x
¸
¸
+ C,
_
1
x
2
+1
dx = tan
−1
x
24 2 ﬁrstorder and simple higherorder differential equations
1
2
ln
¸
¸
v
2
+1
¸
¸
+ tan
−1
v = −ln
¸
¸
x
¸
¸
+ C.
_
1
x
dx = ln
¸
¸
x
¸
¸
Replacing v by the original variables x and y results in the general solution
ln
¸
¸
¸
_
y
x
_
2
+1
¸
¸
¸ + 2 ln
¸
¸
x
¸
¸
+ 2 tan
−1
y
x
= 2C,
a ln b = ln b
a
ln a+ln b = ln(a· b)
∴ ln( y
2
+x
2
) + 2 tan
−1
y
x
= C. 2C is renamed as C.
Example 2.7 2.7
Solve
dy
dx
=
2x−y+1
x−2y+1
, x−2y+1=0.
The equation is not homogeneous because of the constant terms in both the nu
merator and denominator. 2x−y+1=0 and x−2y+1=0 are the equations of
straight lines. The coordinates of the point of intersection are the solution of the
equations
2x − y + 1=0
x − 2y + 1=0
_
=⇒ x = −
1
3
, y =
1
3
.
0 1
1
−1
−1 −2
2x−y +1=0
x−2y +1=0
y
x
X
Y
P
Hence, the point of intersection is P(h, k), (h, k) =
_
−
1
3
,
1
3
_
. Shift to new axes
(X, Y) through point P(h, k). Then
x = X+h = X−
1
3
, y = Y +k = Y +
1
3
, and
dy
dx
=
dY
dX
.
In the new variables X and Y, the constant terms are removed
dY
dX
=
2
_
X−
1
3
_
−
_
Y +
1
3
_
+1
_
X−
1
3
_
− 2
_
Y +
1
3
_
+1
=
2X−Y
X−2Y
=
2−
Y
X
1−2
Y
X
,
Dividing both the numerator and denominator by X, X =0
2.2 method of transformation of variables 25
which is now a homogeneous equation. Putting v =
Y
X
,
Y = Xv =⇒
dY
dX
= v + X
dv
dX
,
and the equation becomes
v + X
dv
dX
=
2−v
1−2v
,
X
dv
dX
=
2−v
1−2v
− v =
2−v−(v−2v
2
)
1−2v
= 2
v
2
−v+1
1−2v
.
Since v
2
−v+1=0, separating the variables yields
1−2v
v
2
−v+1
dv =
2
X
dX.
Integrating both sides leads to
_
1−2v
v
2
−v+1
dv = 2
_
1
X
dX + C.
Since d(v
2
−v+1) =(2v−1)dv, one obtains
−
_
1
v
2
−v+1
d( v
2
−v+1 ) = 2
_
1
X
dX + C,
−ln
¸
¸
v
2
−v+1
¸
¸
= 2 ln
¸
¸
X
¸
¸
+ C.
_
1
x
dx = ln
¸
¸
x
¸
¸
Replacing v by X and Y gives
ln
¸
¸
X
2
¸
¸
+ ln
¸
¸
¸
¸
_
Y
X
_
2
−
_
Y
X
_
+1
¸
¸
¸
¸
= ln
¸
¸
C
¸
¸
, −C is written as ln
¸
¸
C
¸
¸
.
ln
¸
¸
Y
2
−XY +X
2
¸
¸
= ln
¸
¸
C
¸
¸
=⇒ Y
2
−XY +X
2
= C.
In terms of the original variables x and y, the general solution becomes
_
y−
1
3
_
2
−
_
x+
1
3
__
y−
1
3
_
+
_
x+
1
3
_
2
= C.
2.2.2 Special Transformations
Example 2.8 2.8
Solve
dy
dx
=
x−y+5
2x−2y−2
, x−y−1=0.
Unlike the previous example, the two lines x−y+5=0 and 2x−2y−2=0 are
parallel so that there is no ﬁnite point of intersection.
26 2 ﬁrstorder and simple higherorder differential equations
Because both the numerator and denominator have the term x−y, take a new
dependent variable v =x−y,
v = x − y =⇒
dv
dx
= 1 −
dy
dx
=⇒
dy
dx
= 1 −
dv
dx
.
The differential equation becomes
1 −
dv
dx
=
v+5
2v−2
,
dv
dx
= 1 −
v+5
2v−2
=
2v−2−v−5
2v−2
=
v−7
2(v−1)
.
Case 1. v =7, separating the variables gives
v−1
v−7
dv =
1
2
dx. Variable separable
Integrating both sides yields
_
(v−7)+6
v−7
dv =
_
1
2
dx + C =⇒
_
_
1 +
6
v−7
_
dv =
1
2
x + C,
v + 6 ln
¸
¸
v−7
¸
¸
=
1
2
x + C.
Replacing v by the original variables gives the general solution
x − y + 6 ln
¸
¸
x−y−7
¸
¸
=
1
2
x + C,
∴
1
2
x − y + 6 ln
¸
¸
x−y−7
¸
¸
= C.
Case 2. v =7 =⇒ x−y =7. It can be easily veriﬁed that x−y =7 is a solution
of the differential equation. This solution cannot be obtained from the general
solution for any value of the constant C and is therefore a singular solution.
Combining Cases 1 and 2, the solutions of the differential equation are
1
2
x − y + 6 ln
¸
¸
x−y−7
¸
¸
= C, General solution
x − y = 7. Singular solution
Summary
Consider a differential equation of the form
dy
dx
=
ax+by+c
αx+β y+γ
, ab=0, αβ =0.
Case 1.
a
α
=
b
β
2.2 method of transformation of variables 27
The two straight lines ax+by+c =0 and αx+β y+γ =0 intersect at point
P(h, k), where (h, k) is the solution of
ah + bk + c = 0, αh+βk+γ = 0.
Letting
x = X + h, y = Y + k,
the differential equation becomes
dY
dX
=
aX+bY
αX+βY
. Homogeneous
Case 2.
a
α
=
b
β
=
1
r
=⇒ α = r a, β = r b
Letting
v = ax + by =⇒
dv
dx
= a + b
dy
dx
=⇒
dy
dx
=
1
b
_
dv
dx
− a
_
,
the differential equation becomes
1
b
_
dv
dx
− a
_
=
v+c
r v+γ
=⇒
dv
dx
=
b(v+c)
r v+γ
+ a. Variable separable
Example 2.9 2.9
Solve
dy
dx
= ( x+y )
2
.
This equation is neither variable separable nor homogeneous. The “special” term
in the equation is x+y. Hence, letting v =x+y be the new dependent variable
dv
dx
= 1 +
dy
dx
=⇒
dy
dx
=
dv
dx
− 1,
the equation becomes
dv
dx
− 1 = v
2
=⇒
dv
dx
= v
2
+1.
Since v
2
+1=0, separating the variables gives
1
v
2
+1
dv = dx. Variable separable
Integrating both sides leads to
_
1
v
2
+1
dv =
_
dx + C =⇒ tan
−1
v = x + C.
28 2 ﬁrstorder and simple higherorder differential equations
Replacing v by the original variable results in the general solution
tan
−1
(x+y) = x + C, or x + y = tan(x+C).
Remarks: There are no systematic procedures to followin applying the method
of special transformations. It is important to carefully inspect the differential
equation to uncover the ''special'' term and then determine the transformation
accordingly.
Example 2.10 2.10
Solve
dy
dx
=
y
6
−2x
2
2x y
5
+x
2
y
2
, x =0, y =0, 2y
3
+x =0.
The differential equation can be rewritten as
dy
dx
=
y
6
−2x
2
x y
2
(2y
3
+x)
=
_
y
3
x
_
2
−2
y
2
_
2
_
y
3
x
_
+1
_
.
Dividing both the numerator
and denominator by x
2
, x =0
The “special” term in the equation is
y
3
x
. Hence, letting
v =
y
3
x
=⇒ y
3
= xv =⇒ 3y
2
dy
dx
= v + x
dv
dx
,
the differential equation becomes
1
3
_
v + x
dv
dx
_
=
v
2
−2
2v+1
,
x
dv
dx
=
3(v
2
−2)
2v+1
− v =
3v
2
−6−2v
2
−v
2v+1
=
v
2
−v−6
2v+1
.
Case 1. v
2
−v−6=0, i.e., (v−3)(v+2) =0 =⇒ v = −2 or v =3, which gives
y
3
= −2x or y
3
= 3x.
Case 2. v
2
−v−6=0, separating the variables gives
2v+1
v
2
−v−6
dv =
1
x
dx. Variable separable
Integrating both sides yields
_
2v+1
v
2
−v−6
dv =
_
1
x
dx + C.
2.2 method of transformation of variables 29
The ﬁrst integral can be evaluated using partial fractions (see pages 259–261 for a
brief review on partial fractions)
2v+1
v
2
−v−6
=
2v+1
(v−3)(v+2)
=
A
v−3
+
B
v+2
.
Using the coverup method, the coefﬁcients A and B can be easily determined
A =
2v+1
v+2
¸
¸
¸
¸
v=3
=
7
5
, B =
2v+1
v−3
¸
¸
¸
¸
v=−2
=
3
5
,
_
2v+1
v
2
−v−6
dv =
1
5
_
_
7
v−3
+
3
v+2
_
dv =
7
5
ln
¸
¸
v−3
¸
¸
+
3
5
ln
¸
¸
v+2
¸
¸
.
Hence
7
5
ln
¸
¸
v−3
¸
¸
+
3
5
ln
¸
¸
v+2
¸
¸
= ln
¸
¸
x
¸
¸
+ C,
7 ln
¸
¸
v−3
¸
¸
+ 3 ln
¸
¸
v+2
¸
¸
= 5 ln
¸
¸
x
¸
¸
+ ln
¸
¸
C
¸
¸
, 5C⇒ ln
¸
¸
C
¸
¸
ln
¸
¸
(v−3)
7
(v+2)
3
¸
¸
= ln
¸
¸
Cx
5
¸
¸
, ln a + ln b = ln(a· b)
∴ (v−3)
7
(v+2)
3
= Cx
5
.
Replacing v by the original variables results in the general solution
_
y
3
x
−3
_
7
_
y
3
x
+2
_
3
= Cx
5
=⇒ ( y
3
−3x)
7
( y
3
+2x)
3
= Cx
15
.
Note that the solutions y
3
= −2x and y
3
=3x obtained in Case 1 are contained in
the general solution ( y
3
−3x)
7
( y
3
+2x)
3
=Cx
15
obtained in Case 2, with C=0.
Hence, the solution of the differential equation is
( y
3
− 3x)
7
( y
3
+ 2x)
3
= Cx
15
. General solution
Example 2.11 2.11
1. Show that equations of the form
x
y
dy
dx
= f (x y), y =0,
can be converted to variable separable by the transformation x y =v.
2. Using the result obtained above, solve
x
y
dy
dx
=
2+x
2
y
2
2−x
2
y
2
.
1. Letting x y =v,
y + x
dy
dx
=
dv
dx
=⇒
x
y
dy
dx
=
1
y
dv
dx
− 1 =
x
v
dv
dx
− 1,
30 2 ﬁrstorder and simple higherorder differential equations
the differential equation becomes
x
v
dv
dx
− 1 = f (v) =⇒
x
v
dv
dx
= f (v) + 1.
Case 1. f (v)+1=0. If v
0
is a root of f (v
0
)+1=0, then a solution is x y =v
0
.
Case 2. f (v)+1=0. Separating the variables gives
1
v
_
f (v)+1
_ dv =
1
x
dx. Variable separable
Hence, the transformation x y =v converts the original differential equation to
variable separable.
2. In this case,
f (v) =
2+v
2
2−v
2
=⇒
x
v
dv
dx
=
4
2 − v
2
,
and separating the variables gives
2 − v
2
2v
dv =
2
x
dx.
Integrating both sides leads to
_
_
1
v
−
v
2
_
dv = 2
_
1
x
dx + C =⇒ ln
¸
¸
v
¸
¸
−
1
4
v
2
= 2 ln
¸
¸
x
¸
¸
+ C.
Replacing v by the original variables gives
ln
¸
¸
¸
y
x
¸
¸
¸ −
1
4
(x y)
2
= C.
Example 2.12 2.12
Solve
dy
dx
=
_
x+y −
_
x−y
_
x+y +
_
x−y
, x >0, x
¸
¸
y
¸
¸
.
The differential equation is a homogeneous equation. However, it can be solved
more easily using a special transformation. The equation can be written as
dy
dx
=
(
_
x+y −
_
x−y)
2
(
_
x+y +
_
x−y)(
_
x+y −
_
x−y)
=
(x+y) − 2
_
x
2
−y
2
+ (x−y)
(x+y) − (x−y)
=
x −
_
x
2
−y
2
y
.
The “special” term is x
2
−y
2
. In order to remove the square root, let x
2
−y
2
=v
2
x
2
− y
2
= v
2
=⇒ 2x − 2y
dy
dx
= 2v
dv
dx
=⇒ y
dy
dx
= x − v
dv
dx
.
2.3 exact differential equations and integrating factors 31
The differential equation becomes
x − v
dv
dx
= x − v =⇒ v
_
dv
dx
− 1
_
= 0.
Case 1. v =0 =⇒
dv
dx
−1=0 =⇒ dv =dx. Integrating both sides yields
v = x + C =⇒ v
2
= (x+C)
2
.
Replacing v by the original variables results in the general solution
x
2
− y
2
= (x+C)
2
=⇒ y
2
+ 2Cx + C
2
= 0.
Case 2. v =0 =⇒ x
2
−y
2
=0 =⇒ y = ±x. This solution cannot be obtained
from the general solution for any value of the constant C and is therefore a singular
solution.
Combining Cases 1 and 2, the solutions of the differential equation are
y
2
+ 2Cx + C
2
= 0, General solution
y = ±x. Singular solution
2.3 Exact Differential Equations and Integrating
Factors
Consider differential equations of the form
M(x, y)dx + N(x, y)dy = 0, or
dy
dx
= −
M(x, y)
N(x, y)
, N(x, y) =0, (1)
where
∂M
∂y
and
∂N
∂x
are continuous. Suppose the solution of equation (1) is
u(x, y) =C, C=constant. Taking the differential yields
du =
∂u
∂x
dx +
∂u
∂y
dy = dC = 0 =⇒
∂u
∂x
dx +
∂u
∂y
dy = 0. (2)
Equation (2) should be the same as equation (1) if u(x, y) =C is the solution of (1),
except for a common factor μ(x, y), i.e., the coefﬁcients of dx and dy in equations
(1) and (2) are proportional
∂u
∂x
M(x, y)
=
∂u
∂y
N(x, y)
= μ(x, y) =⇒
∂u
∂x
= μM,
∂u
∂y
= μN.
32 2 ﬁrstorder and simple higherorder differential equations
Substituting into equation (2) gives
μMdx + μNdy = 0. (2
)
Since the lefthand side is an exact differential of some function u(x, y),
∴ du(x, y) = 0 =⇒ u(x, y) = C.
Hence, if one could ﬁnd a function μ(x, y), called an integrating factor (IF) mul
tiplying it to equation (1) yields an exact differential equation (2
), which means
that the lefthand side is the exact differential of some function. The resulting
differential equation can then be easily solved.
Motivating Example
Solve ( y + 2x y
2
)dx + (2x + 3x
2
y)dy = 0.
It happens that an integrating factor is y. Multiplying the differential equation by y
results in
( y
2
+ 2x y
3
)dx + (2x y + 3x
2
y
2
)dy = 0.
The lefthand side is the exact differential of u(x, y) =x y
2
+x
2
y
3
. Hence,
d(x y
2
+ x
2
y
3
) = 0 =⇒ x y
2
+ x
2
y
3
= C. General solution
2.3.1 Exact Differential Equations
If the differential equation
M(x, y)dx + N(x, y)dy = 0 (1)
is exact, then there exists a function u(x, y) such that
du = M(x, y)dx + N(x, y)dy. (2)
But, by deﬁnition of differential,
du =
∂u
∂x
dx +
∂u
∂y
dy. (3)
Comparing equations (2) and (3) leads to
M =
∂u
∂x
, N =
∂u
∂y
=⇒
∂M
∂y
=
∂
2
u
∂y∂x
,
∂N
∂x
=
∂
2
u
∂x∂y
.
If
∂M
∂y
and
∂N
∂x
are continuous, one has
∂
2
u
∂y∂x
=
∂
2
u
∂x∂y
. (4)
2.3 exact differential equations and integrating factors 33
Hence, a necessary condition for exactness is, from equations (4),
∂M
∂y
=
∂N
∂x
.
It can be shown that this condition is also sufﬁcient.
Exact Differential Equations
Consider the differential equation
M(x, y)dx + N(x, y)dy = 0.
If
∂M
∂y
=
∂N
∂x
, Exactness condition
then the differential equation is exact, meaning that the lefthand side is the exact
differential of some function.
Example 2.13 2.13
Solve (6x y
2
+ 4x
3
y)dx + (6x
2
y + x
4
+ e
y
)dy = 0.
The differential equation is of the form
M(x, y)dx + N(x, y)dy = 0,
where
M(x, y) = 6x y
2
+ 4x
3
y, N(x, y) = 6x
2
y + x
4
+ e
y
.
Test for exactness:
∂M
∂y
= 12x y + 4x
3
,
∂N
∂x
= 12x y + 4x
3
,
∴
∂M
∂y
=
∂N
∂x
=⇒ The differential equation is exact.
Two methods are introduced in the following to ﬁnd the general solution.
Method 1: Since the differential equation is exact, there exists a function u(x, y)
such that
du =
∂u
∂x
dx +
∂u
∂y
dy = (6x y
2
+ 4x
3
y)dx + (6x
2
y + x
4
+ e
y
)dy,
i.e.,
∂u
∂x
= 6x y
2
+ 4x
3
y, (1)
∂u
∂y
= 6x
2
y + x
4
+ e
y
. (2)
34 2 ﬁrstorder and simple higherorder differential equations
To determine u(x, y), integrate equation (1) with respect to x
u(x, y) =
_
(6x y
2
+ 4x
3
y)dx + f ( y)
When integrating w.r.t. x,
y is treated as constant or fixed.
= 3x
2
y
2
+ x
4
y + f ( y). (3)
Differentiating equation (3) with respect to y and comparing with equation (2)
yield
∂u
∂y
= 6x
2
y + x
4
+
df ( y)
dy
When differentiating w.r.t. y,
x is treated as constant or fixed.
= 6x
2
y + x
4
+ e
y
; Equation (2)
hence,
df ( y)
dy
= e
y
=⇒ f ( y) = e
y
.
Substituting into equation (3) leads to
u(x, y) = 3x
2
y
2
+ x
4
y + e
y
.
The general solution is then given by
u(x, y) = C =⇒ 3x
2
y
2
+ x
4
y + e
y
= C.
Method 2: The Method of Grouping Terms
The essence of Method 1 is to determine function u(x, y) by
❧ integrating the coefﬁcient of dx with respect to x,
❧ differentiating the result with respect to y and comparing with the coefﬁcient
of dy.
This procedure can be recast to result in the method of grouping terms, which is
noticeably more succinct and is the preferred method. The method is illustrated
stepbystep as follows:
1. Pick up a term, for example 6x y
2
dx.
❧ Since the term has dx, integrate the coefﬁcient 6x y
2
with respect to x
to yield 3x
2
y
2
.
❧ Differentiate the result with respect to y to yield the coefﬁcient of dy
term, i.e., 6x
2
y.
❧ The two terms 6x y
2
dx+6x
2
ydy are grouped together.
6x y
2
dx
_
dx
+ 6x
2
y dy
3x
2
y
2
∂
∂y
¸
_
dx stands for integrating w.r.t. x,
∂
∂y
denotes differentiating w.r.t. y.
2.3 exact differential equations and integrating factors 35
2. Pick up one of the remaining terms, for example 4x
3
ydx.
❧ Similarly, since the term has dx, integrate the coefﬁcient 4x
3
y with
respect to x to yield x
4
y.
❧ Differentiate the result with respect to y to yield the coefﬁcient of dy
term, i.e., x
4
.
❧ The two terms 4x
3
ydx+x
4
dy are grouped together.
4x
3
y dx
_
dx
+
x
4
dy
x
4
y
∂
∂y
¸¸
3. Pick up one of the remaining terms. Since there is only one termleft, e
y
dy is
picked.
❧ Since the term has dy, integrate the coefﬁcient e
y
with respect to y to
yield e
y
.
❧ Differentiate the result with respect to x to yield the coefﬁcient of dx
term, i.e., 0.
❧ The term e
y
dy is in a group by itself.
e
y
dy
_
dy
+
0 · dx
e
y
∂
∂x
¸¸
4. All the terms on the lefthand side of the equation have now been grouped.
5. Steps 1 to 3 can be combined to give a single expression as follows
_
6x y
2
dx
_
dx
+ 6x
2
y dy
_
3x
2
y
2
∂
∂y
¸
+
_
4x
3
y dx
_
dx
+
x
4
dy
_
x
4
y
∂
∂y
¸¸
+
e
y
dy
_
dy
e
y
=0.
6. Hence
d
_
3x
2
y
2
+ x
4
y + e
y
_
= 0,
which gives the general solution
3x
2
y
2
+ x
4
y + e
y
= C.
Remarks:
❧ The method of grouping terms is easier to apply. The sum of the functions in
the second row is the required function u(x, y), and the general solution can
36 2 ﬁrstorder and simple higherorder differential equations
be readily obtained as u(x, y) =C. Hence, the method of grouping terms is the
preferred method.
❧ If a differential equation is exact, then all the terms on the lefthand side of the
equation will be grouped. If there are terms left that cannot be grouped, there
must be mistakes made in the calculation.
❧ Terms of the form
f (x)dx or g( y)dy
are in groups by themselves, because
_
f (x) dx
_
dx
+ 0· dy
_
_
f (x)dx
∂
∂y
¸¸
or
_
g( y) dy
_
dy
+ 0· dx
_
.
_
g( y)dy
∂
∂x
¸
Example 2.14 2.14
Solve
dy
dx
=
y sin x − e
x
sin 2y
cos x + 2e
x
cos 2y
.
The differential equation can be written in the standard form Mdx+Ndy =0:
(−y sin x + e
x
sin 2y)dx + (cos x + 2e
x
cos 2y)dy = 0.
. ,, . . ,, .
M(x, y) N(x, y)
Test for exactness:
∂M
∂y
= −sin x + 2e
x
cos 2y,
∂N
∂x
= −sin x + 2e
x
cos 2y,
∴
∂M
∂y
=
∂N
∂x
=⇒ The differential equation is exact.
The general solution is obtained using the method of grouping terms:
_
−y sin x dx
_
dx
+ cos x dy
_
y cos x
∂
∂y
¸
+
_
e
x
sin 2y dx
_
dx
+ 2e
x
cos 2y dy
_
e
x
sin 2y
∂
∂y
¸
=0.
Hence, by summing up the terms in the second row one obtains the function
u(x, y), and the general solution is given by
y cos x + e
x
sin 2y = C.
2.3 exact differential equations and integrating factors 37
Example 2.15 2.15
Solve 2x(3x + y − ye
−x
2
)dx + (x
2
+ 3y
2
+ e
−x
2
)dy = 0.
The differential equation is of the standard form Mdx+Ndy =0, where
M(x, y) = 6x
2
+ 2x y − 2x ye
−x
2
, N(x, y) = x
2
+ 3y
2
+ e
−x
2
.
Test for exactness:
∂M
∂y
= 2x − 2xe
−x
2
,
∂N
∂x
= 2x − 2xe
−x
2
,
∴
∂M
∂y
=
∂N
∂x
=⇒ The differential equation is exact.
The general solution is determined using the method of grouping terms:
_
2x y dx
_
dx
+
x
2
dy
_
x
2
y
∂
∂y
¸¸
+
_
e
−x
2
dy
_
dy
+
−2x ye
−x
2
dx
_
ye
−x
2
∂
∂x
¸
+
6x
2
dx
_
dx
2x
3
+ 3y
2
dy
_
dy
y
3
= 0,
which gives
x
2
y + ye
−x
2
+ 2x
3
+ y
3
= C. General solution
Remarks: In the second group of terms above, it is easier to pick up the term
e
−x
2
dy first, integrate its coefficient with respect to y, and then differentiate the
result with respect to x to find the matching term.
Example 2.16 2.16
Solve
_
1
y
sin
x
y
−
y
x
2
cos
y
x
+1
_
dx +
_
1
x
cos
y
x
−
x
y
2
sin
x
y
+
1
y
2
_
dy = 0.
The differential equation implies that x =0 and y =0. The equation is of the
standard form Mdx+Ndy =0, where
M(x, y) =
1
y
sin
x
y
−
y
x
2
cos
y
x
+ 1, N(x, y) =
1
x
cos
y
x
−
x
y
2
sin
x
y
+
1
y
2
.
38 2 ﬁrstorder and simple higherorder differential equations
Test for exactness:
∂M
∂y
= −
1
y
2
sin
x
y
+
1
y
cos
x
y
·
_
−
x
y
2
_
−
1
x
2
cos
y
x
−
y
x
2
_
−sin
y
x
·
1
x
_
= −
1
y
2
sin
x
y
−
x
y
3
cos
x
y
−
1
x
2
cos
y
x
+
y
x
3
sin
y
x
,
∂N
∂x
= −
1
x
2
cos
y
x
+
1
x
_
−sin
y
x
·
_
−
y
x
2
__
−
1
y
2
sin
x
y
−
x
y
2
_
cos
x
y
·
1
y
_
= −
1
x
2
cos
y
x
+
y
x
3
sin
y
x
−
1
y
2
sin
x
y
−
x
y
3
cos
x
y
,
∴
∂M
∂y
=
∂N
∂x
=⇒ The differential equation is exact.
The general solution is determined using the method of grouping terms
_
1
y
sin
x
y
dx
_
dx
+
−
x
y
2
sin
x
y
dy
_
−cos
x
y
∂
∂y
¸
+
1 dx
_
dx
x
+
1
y
2
dy
_
dy
−
1
y
+
_
1
x
cos
y
x
dy
_
dy
+ −
y
x
2
cos
y
x
dx
_
sin
y
x
∂
∂x
¸
= 0,
which gives
−cos
x
y
+ x −
1
y
+ sin
y
x
= C. General solution
Note that in the fourth group of terms above, the term
_
1
x
cos
y
x
dy
_
is picked
ﬁrst, because it is easier to integrate the coefﬁcient
_
1
x
cos
y
x
_
with respect to y.
Remarks: When applying the method of grouping terms, whether to pick a
term f (x, y)dx or g(x, y)dy first depends onwhether it is easier to integrate
_
f (x, y)dx or
_
g(x, y)dy.
2.3 exact differential equations and integrating factors 39
2.3.2 Integrating Factors
Consider the differential equation
M(x, y)dx + N(x, y)dy = 0. (1)
❧ If
∂M
∂y
=
∂N
∂x
, the differential equation is exact.
❧ If
∂M
∂y
=
∂N
∂x
, the differential equation can be rendered exact by multiplying by
a function μ(x, y), known as an integrating factor (IF), i.e.,
μ(x, y)M(x, y)dx + μ(x, y)N(x, y)dy = 0, (2)
is exact.
To ﬁnd an integrating factor μ(x, y), apply the exactness condition on equation (2)
∂(μM)
∂y
=
∂(μN)
∂x
,
i.e.,
M
∂μ
∂y
+ μ
∂M
∂y
= N
∂μ
∂x
+ μ
∂N
∂x
=⇒ μ
_
∂M
∂y
−
∂N
∂x
_
= N
∂μ
∂x
− M
∂μ
∂y
. (3)
This is a partial differential equation for the unknown function μ(x, y), which
is usually more difﬁcult to solve than the original ordinary differential equation
(1). However, for some special cases, equation (3) can be solved for an integrating
factor.
Special Cases:
If μ is a function of x only, i.e., μ=μ(x), then
∂μ
∂x
=
dμ
dx
,
∂μ
∂y
= 0,
and equation (3) becomes
N
dμ
dx
= μ
_
∂M
∂y
−
∂N
∂x
_
=⇒
1
μ
dμ
dx
=
1
N
_
∂M
∂y
−
∂N
∂x
_
. (4)
Since μ(x) is a function of x only, the lefthand side is a function of x only. Hence,
if an integrating factor of the form μ=μ(x) is to exist, the righthand side must
also be a function of x only. Equation (4) is variable separable, which can be solved
easily by integration
ln μ =
_
1
N
_
∂M
∂y
−
∂N
∂x
_
dx =⇒ μ(x) = exp
__
1
N
_
∂M
∂y
−
∂N
∂x
_
dx
_
. (5)
40 2 ﬁrstorder and simple higherorder differential equations
Note that, since only one integrating factor is sought, there is no need to include a
constant of integration C.
Interchanging M and N, and x and y inequation(5), one obtains anintegrating
factor for another special case
μ( y) = exp
__
1
M
_
∂N
∂x
−
∂M
∂y
_
dy
_
. (6)
. ,, .
function of y only
Integrating Factors
Consider the differential equation
M(x, y)dx + N(x, y)dy = 0.
❧ If
1
N
_
∂M
∂y
−
∂N
∂x
_
is a function of x only,
μ(x) = exp
__
1
N
_
∂M
∂y
−
∂N
∂x
_
dx
_
.
❧ If
1
M
_
∂N
∂x
−
∂M
∂y
_
is a function of y only,
μ( y) = exp
__
1
M
_
∂N
∂x
−
∂M
∂y
_
dy
_
.
Example 2.17 2.17
Solve 3(x
2
+ y
2
)dx + x(x
2
+ 3y
2
+ 6y)dy = 0.
The differential equation is of the standard form Mdx+Ndy =0, where
M(x, y) = 3(x
2
+ y
2
), N(x, y) = x
3
+ 3x y
2
+ 6x y.
Test for exactness:
∂M
∂y
= 6y,
∂N
∂x
= 3x
2
+ 3y
2
+ 6y,
∴
∂M
∂y
=
∂N
∂x
=⇒ The differential equation is not exact.
Since
1
M
_
∂N
∂x
−
∂M
∂y
_
=
1
3(x
2
+ y
2
)
_
(3x
2
+ 3y
2
+ 6y) − 6y
_
= 1, A function of y only
2.3 exact differential equations and integrating factors 41
∴ μ( y) = exp
__
1
M
_
∂N
∂x
−
∂M
∂y
_
dy
_
= exp
_
_
1· dy
_
= e
y
.
Multiplying the differential equation by the integrating factor μ( y) =e
y
yields
(3x
2
e
y
+ 3y
2
e
y
)dx + (x
3
e
y
+ 3x y
2
e
y
+ 6x ye
y
)dy = 0.
The general solution is determined using the method of grouping terms
_
3x
2
e
y
dx
_
dx
+
x
3
e
y
dy
_
x
3
e
y
∂
∂y
¸
+
_
3y
2
e
y
dx
_
dx
+ (6x ye
y
+3x y
2
e
y
) dy
_
=0,
3x y
2
e
y
∂
∂y
¸
which gives
x
3
e
y
+ 3x y
2
e
y
= C =⇒ xe
y
(x
2
+ 3y
2
) = C. General solution
Example 2.18 2.18
Solve y(2x − y + 2)dx + 2(x − y)dy = 0.
The differential equation is of the standard form Mdx+Ndy =0, where
M(x, y) = 2x y − y
2
+ 2y, N(x, y) = 2(x − y).
Test for exactness:
∂M
∂y
= 2x − 2y + 2,
∂N
∂x
= 2 =⇒
∂M
∂y
=
∂N
∂x
=⇒ The DE is not exact.
Since
1
N
_
∂M
∂y
−
∂N
∂x
_
=
1
2(x − y)
_
(2x − 2y + 2) − 2
_
= 1, A function of x only
∴ μ(x) = exp
__
1
N
_
∂M
∂y
−
∂N
∂x
_
dx
_
= exp
_
_
1· dx
_
= e
x
.
Multiplying the differential equation by the integrating factor μ(x) =e
x
yields
(2x ye
x
− y
2
e
x
+ 2ye
x
)dx + (2xe
x
− 2ye
x
)dy = 0.
The general solution is determined using the method of grouping terms
_
2xe
x
dy
_
dy
+ (2ye
x
+ 2x ye
x
) dx
_
2x ye
x ∂
∂x
¸
+
_
−y
2
e
x
dx
_
dx
+ −2ye
x
dx
_
=0,
−y
2
e
x
∂
∂y
¸
which gives
2x ye
x
− y
2
e
x
= C =⇒ ye
x
(2x − y) = C. General solution
42 2 ﬁrstorder and simple higherorder differential equations
Example 2.19 2.19
Solve y(cos
3
x + y sin x)dx + cos x(sin x cos x + 2y)dy = 0.
The differential equation is of the standard form Mdx+Ndy =0, where
M(x, y) = y cos
3
x + y
2
sin x, N(x, y) = sin x cos
2
x + 2y cos x.
Test for exactness:
∂M
∂y
= cos
3
x + 2y sin x,
∂N
∂x
= cos
3
x − 2 sin
2
x cos x − 2y sin x,
∴
∂M
∂y
=
∂N
∂x
=⇒ The differential equation is not exact.
Since
1
N
_
∂M
∂y
−
∂N
∂x
_
=
(cos
3
x + 2y sin x) − (cos
3
x − 2 sin
2
x cos x − 2y sin x)
sin x cos
2
x + 2y cos x
=
2 sin x(2y + sin x cos x)
cos x(2y + sin x cos x)
=
2 sin x
cos x
, A function of x only
∴ μ(x) = exp
__
1
N
_
∂M
∂y
−
∂N
∂x
_
dx
_
= exp
_
_
2 sin x
cos x
dx
_
= exp
_
−2
_
1
cos x
d(cos x)
_
= exp
_
−2 ln
¸
¸
cos x
¸
¸
_
=
1
cos
2
x
.
Multiplying the differential equation by the integrating factor μ(x) =
1
cos
2
x
yields
_
y cos x +
y
2
sin x
cos
2
x
_
dx +
_
sin x +
2y
cos x
_
dy = 0.
The general solution is determined using the method of grouping terms
_
y cos x dx
_
dx
+
sin x dy
_
y sin x
∂
∂y
¸
+
_
2y
cos x
dy
_
dy
+
y
2
sin x
cos
2
x
dx
_
y
2
cos x
∂
∂x
¸
= 0,
which gives
y sin x +
y
2
cos x
= C. General solution
2.3 exact differential equations and integrating factors 43
Example 2.20 2.20
1. Show that if the equation M(x, y)dx + N(x, y)dy = 0 is such that
x
2
xM + yN
_
∂N
∂x
−
∂M
∂y
_
= F
_
y
x
_
then an integrating factor is given by
μ(x, y) = exp
_
_
F(u) du
_
, u =
y
x
.
2. Using the result of Part 1, solve the differential equation
_
2x − y + 2x y − y
2
_
dx +
_
x + x
2
+ x y
_
dy = 0.
1. If
μ = exp
_
_
F(u)du
_
, u =
y
x
,
is an integrating factor, then μMdx +μNdy = 0 is an exact differential equation,
and the exactness condition must be satisﬁed
∂(μM)
∂y
−
∂(μN)
∂x
=
_
∂μ
∂y
M + μ
∂M
∂y
_
−
_
∂μ
∂x
N + μ
∂N
∂x
_
= exp
_
_
F(u)du
_
F(u)
1
x
· M − exp
_
_
F(u)du
_
F(u)
_
−
y
x
2
_
· N
+ μ
_
∂M
∂y
−
∂N
∂x
_
= μF(u)
_
M
x
+
yN
x
2
_
− μ
_
∂N
∂x
−
∂M
∂y
_
= μ
_
F(u)
x
2
(xM + yN) −
_
∂N
∂x
−
∂M
∂y
_
_
= 0.
Since μ=0, one has
F(u)
x
2
(xM + yN) −
_
∂N
∂x
−
∂M
∂y
_
= 0,
i.e.,
x
2
xM + yN
_
∂N
∂x
−
∂M
∂y
_
= F(u), u =
y
x
.
2. The differential equation is of the standard form Mdx+Ndy =0, where
M(x, y) = 2x + 2x y − y − y
2
, N(x, y) = x
2
+ x + x y.
44 2 ﬁrstorder and simple higherorder differential equations
It can be easily evaluated
∂N
∂x
−
∂M
∂y
= (2x + 1 + y) − (2x − 1 − 2y) = 2 + 3y,
xM + yN = (2x
2
+ 2x
2
y − x y − x y
2
) + (x
2
y + x y + x y
2
)
= 2x
2
+ 3x
2
y = x
2
(2 + 3y),
x
2
xM + yN
_
∂N
∂x
−
∂M
∂y
_
=
x
2
x
2
(2 + 3y)
(2 + 3y) = 1 = F(u),
μ(x, y) = exp
_
_
F(u)du
_
= exp
_
_
1· du
_
= exp(u) = exp
_
y
x
_
.
Multiplying the differential equation by the integrating factor yields
_
2x − y + 2x y − y
2
_
exp
_
y
x
_
dx +
_
x + x
2
+ x y
_
exp
_
y
x
_
dy = 0.
Note the following integrals
_
x exp
_
y
x
_
dy = x
2
exp
_
y
x
_
,
_
(x
2
+ x y) exp
_
y
x
_
dy = x
3
exp
_
y
x
_
+ x
2
_
yd
_
exp
_
y
x
__
= x
3
exp
_
y
x
_
+ x
2
_
y exp
_
y
x
_
−
_
exp
_
y
x
_
dy
_
Integration by parts
= x
3
exp
_
y
x
_
+ x
2
_
y exp
_
y
x
_
− x exp
_
y
x
__
= x
2
y exp
_
y
x
_
.
The general solution is determined using the method of grouping terms
_
x exp
_
y
x
_
dy
_
dy
+ (2x − y) exp
_
y
x
_
dx
_
x
2
exp
_
y
x
_
∂
∂x
¸
+
_
(x
2
+x y) exp
_
y
x
_
dy
_
dy
+ (2x y−y
2
) exp
_
y
x
_
dx
_
x
2
y exp
_
y
x
_
∂
∂x
¸
= 0,
which gives
x
2
(1 + y) exp
_
y
x
_
= C. General solution
2.3 exact differential equations and integrating factors 45
2.3.3 Method of Inspection
Useful Formulas
1. d(x y) = ydx + xdy,
2. d
_
y
x
_
=
−ydx + xdy
x
2
, d
_
x
y
_
=
ydx − xdy
y
2
,
3. d
_
tan
−1
y
x
_
=
−ydx + xdy
x
2
+ y
2
, d
_
tan
−1
x
y
_
=
ydx − xdy
x
2
+y
2
,
4. d
_
1
2
ln
_
x
2
+y
2
_
_
=
xdx + ydy
x
2
+y
2
,
5. d
__
x
2
± y
2
_
=
xdx ± ydy
_
x
2
± y
2
.
By rearrangement of terms, multiplication or division of suitable functions, an
integrating factor may be determined using these formulas.
Example 2.21 2.21
Solve (3x
4
+ y)dx + (2x
2
y − x)dy = 0.
The differential equation can be rearranged as
x
2
(3x
2
dx + 2ydy) + ( ydx − xdy) = 0.
Dividing the equation by x
2
yields
3x
2
dx + 2ydy −
−ydx + xdy
x
2
= 0.
Hence
d(x
3
) + d( y
2
) − d
_
y
x
_
= 0,
and the general solution is given by
x
3
+ y
2
−
y
x
= C.
Remarks: Using the normal procedure as introduced in Section 2.3.2, it is easy
to determine that μ(x) = 1/x
2
is an integrating factor.
46 2 ﬁrstorder and simple higherorder differential equations
Example 2.22 2.22
Solve (x + x
2
y + y
3
)dx + ( y + x
3
+ x y
2
− x
2
y
2
− y
4
)dy = 0.
The differential equation can be rearranged as
_
x + y(x
2
+y
2
)
_
dx +
_
y + x(x
2
+y
2
) − y
2
(x
2
+y
2
)
_
dy = 0.
Dividing the equation by (x
2
+y
2
) yields
_
x
x
2
+y
2
+ y
_
dx +
_
y
x
2
+y
2
+ x − y
2
_
dy = 0.
Rearranging the equation leads to
xdx + ydy
x
2
+y
2
+ ( ydx + xdy) − y
2
dy = 0.
Hence
d
_
1
2
ln(x
2
+y
2
)
_
+ d(x y) −
1
3
d( y
3
) = 0,
and the general solution is given by
1
2
ln(x
2
+y
2
) + x y −
1
3
y
3
= C.
Example 2.23 2.23
Solve
_
2x
√
x + x
2
+ y
2
_
dx + 2y
√
xdy = 0.
Rearrange the differential equation as
2
√
x(xdx + ydy) + (x
2
+y
2
)dx = 0.
Dividing the equation by
√
x(x
2
+y
2
) leads to
2
xdx + ydy
x
2
+y
2
+
1
√
x
dx = 0,
which gives
2d
_
1
2
ln(x
2
+y
2
)
_
+ d(2
√
x) = 0.
Hence, the general solution is
ln(x
2
+y
2
) + 2
√
x = C.
Example 2.24 2.24
Solve y
2
dx + (x y + y
2
− 1)dy = 0.
The differential equation can be rearranged as
y( ydx + xdy) + ( y
2
− 1)dy = 0.
2.3 exact differential equations and integrating factors 47
It is easy to see that y =0 is a solution of the differential equation. For y =0,
dividing the equation by y leads to
ydx + xdy +
_
y −
1
y
_
dy = 0,
which gives
d(x y) + d
_
1
2
y
2
_
− d
_
ln
¸
¸
y
¸
¸
_
= 0.
Hence, the solutions are given by
x y +
1
2
y
2
− ln
¸
¸
y
¸
¸
= C, y =0.
Remarks: Using the normal procedure as introduced in Section 2.3.2, it is easy
to determine that μ( y) = 1/y is an integrating factor.
Example 2.25 2.25
Solve x ydx − (x
2
+ x
2
y + y
3
)dy = 0.
Rearrange the differential equation as
x( ydx − xdy) − y(x
2
+ y
2
)dy = 0.
Divide the equation by x
2
+y
2
x
ydx − xdy
x
2
+y
2
− ydy = 0 =⇒ x d
_
tan
−1
x
y
_
− ydy = 0.
It is easy to see that y =0 is a solution of the differential equation. For y =0,
dividing the equation by y yields
x
y
d
_
tan
−1
x
y
_
− dy = 0.
Since
_
ud
_
tan
−1
u
_
=
_
u·
1
1+u
2
du =
1
2
_
1
1+u
2
d(1+u
2
) =
1
2
ln(1+u
2
),
one has
ud
_
tan
−1
u
_
=
1
2
d
_
ln(1+u
2
)
_
=⇒
x
y
d
_
tan
−1
x
y
_
=
1
2
d
_
ln
_
1+
x
2
y
2
__
.
Hence
1
2
d
_
ln
_
1+
x
2
y
2
__
−dy =0 =⇒ ln
_
1+
x
2
y
2
_
−2y =C. General solution
48 2 ﬁrstorder and simple higherorder differential equations
2.3.4 Integrating Factors by Groups
Theorem
If μ(x, y) is an integrating factor of the differential equation
M(x, y)dx + N(x, y)dy = 0, (1)
which implies that
μ(x, y)M(x, y)dx + μ(x, y)N(x, y)dy = 0
is an exact differential equation, i.e.,
μ(x, y)M(x, y)dx + μ(x, y)N(x, y)dy = dv(x, y),
then μ(x, y) · g
_
v(x, y)
_
is also an integrating factor of equation (1), where g(·)
is any differentiable nonzero function.
Using this theorem, the following method of integrating factors by groups can be
derived.
Method of Integrating Factors by Groups
Suppose equation (1) can be separated into two groups
_
M
1
(x, y)dx + N
1
(x, y)dy
_
+
_
M
2
(x, y)dx + N
2
(x, y)dy
_
= 0. (2)
. ,, . . ,, .
First group Second group
If the ﬁrst and second groups have integrating factors μ
1
(x, y) and μ
2
(x, y),
respectively, such that
μ
1
M
1
dx + μ
1
N
1
dy = dv
1
,
μ
2
M
2
dx + μ
2
N
2
dy = dv
2
,
then from the theorem, for any differentiable functions g
1
and g
2
,
❧ μ
1
(x, y) · g
1
_
v
1
(x, y)
_
is the integrating factor of the ﬁrst group, and
❧ μ
2
(x, y) · g
2
_
v
2
(x, y)
_
is the integrating factor of the second group.
If one can choose g
1
and g
2
suitably such that
μ
1
(x, y) · g
1
_
v
1
(x, y)
_
= μ
2
(x, y) · g
2
_
v
2
(x, y)
_
=⇒ μ(x, y),
then μ(x, y) is an integrating factor of equation (1).
2.3 exact differential equations and integrating factors 49
Example 2.26 2.26
Solve (4x y + 3y
4
)dx + (2x
2
+ 5x y
3
)dy = 0.
The differential equation is of the standard form Mdx+Ndy =0, where
M(x, y) = 4x y + 3y
4
, N(x, y) = 2x
2
+ 5x y
3
.
Test for exactness:
∂M
∂y
= 4x + 12y
3
,
∂N
∂x
= 4x + 15y
3
.
It can be seen that there does not exist an integrating factor that is a function of x
only or y only.
Separate the differential equation into two groups as
(4x ydx + 2x
2
dy) + (3y
4
dx + 5x y
3
dy) = 0.
. ,, . . ,, .
First group Second group
❧ For the ﬁrst group: M
1
= 4x y, N
1
= 2x
2
,
∂M
1
∂y
= 4x,
∂N
1
∂x
= 4x;
hence, the ﬁrst group is exact or an integrating factor is μ
1
=1.
It is easy to ﬁnd that
4x y dx
_
dx
+
2x
2
dy
2x
2
y
∂
∂y
¸¸
=⇒ 2d(x
2
y) =⇒ v
1
(x, y) = x
2
y.
❧ For the second group: M
2
= 3y
4
, N
2
= 5x y
3
,
∂M
2
∂y
= 12y
3
,
∂N
2
∂x
= 5y
3
,
1
N
2
_
∂M
2
∂y
−
∂N
2
∂x
_
=
1
5x y
3
_
12y
3
− 5y
3
_
=
7
5x
, A function of x only
μ
2
(x) = exp
_
_
1
N
2
_
∂M
2
∂y
−
∂N
2
∂x
_
dx
_
= exp
_
7
5
_
1
x
dx
_
= exp
_
7
5
ln
¸
¸
x
¸
¸
_
= x
7
5
.
50 2 ﬁrstorder and simple higherorder differential equations
Multiplying μ
2
(x) to the second group yields
3x
7
5
y
4
dx
_
dx
+
5x
12
5
y
3
dy
5
4
x
12
5
y
4
∂
∂y
¸
=⇒
5
4
d
_
x
12
5
y
4
_
=⇒ v
2
(x, y) = x
12
5
y
4
.
To ﬁnd an integrating factor for the original differential equation, one needs to ﬁnd
functions g
1
and g
2
such that
μ
1
· g
1
(v
1
) = μ
2
· g
2
(v
2
) =⇒ 1· g
1
(x
2
y) = x
7
5
· g
2
(x
12
5
y
4
).
Letting g
1
(v
1
) = v
α
1
and g
2
(v
2
) = v
β
2
leads to
(x
2
y)
α
= x
7
5
(x
12
5
y
4
)
β
=⇒ x
2α
y
α
= x
12 β+7
5
y
4β
,
∴
2α =
12β + 7
5
α = 4β
⎫
⎬
⎭
=⇒
⎧
⎨
⎩
α = 1,
β =
1
4
.
Hence, an integrating factor is given by
μ(x, y) = μ
1
· g
1
(v
1
) = 1· (x
2
y)
1
= x
2
y.
Multiplying μ=x
2
y to the original differential equation leads to
_
4x
3
y
2
dx
_
dx
+ 2x
4
y dy
_
x
4
y
2
∂
∂y
¸
+
_
3x
2
y
5
dx
_
dx
+ 5x
3
y
4
dy
_
x
3
y
5
∂
∂y
¸
= 0.
The general solution is then obtained as
x
4
y
2
+ x
3
y
5
= C.
Example 2.27 2.27
Solve (5x y − 3y
3
)dx + (3x
2
− 7x y
2
)dy = 0.
The differential equation is of the standard form Mdx+Ndy =0, where
M(x, y) = 5x y − 3y
3
, N(x, y) = 3x
2
− 7x y
2
.
Test for exactness:
∂M
∂y
= 5x − 9y
2
,
∂N
∂x
= 6x − 7y
2
,
2.3 exact differential equations and integrating factors 51
∂M
∂y
−
∂N
∂x
= (5x − 9y
2
) − (6x − 7y
2
) = −x − 2y
2
.
It can be seen that there does not exist an integrating factor that is a function of x
only or y only.
Separate the differential equation into two groups as
(5x ydx + 3x
2
dy) + (−3y
3
dx − 7x y
2
dy) = 0.
. ,, . . ,, .
First group Second group
❧ For the ﬁrst group: M
1
= 5x y, N
1
= 3x
2
,
∂M
1
∂y
= 5x,
∂N
1
∂x
= 6x,
1
N
1
_
∂M
1
∂y
−
∂N
1
∂x
_
=
1
3x
2
(5x − 6x) = −
1
3x
, A function of x only
μ
1
(x) = exp
_
_
1
N
1
_
∂M
1
∂y
−
∂N
1
∂x
_
dx
_
= exp
_
−
1
3
_
1
x
dx
_
= exp
_
−
1
3
ln
¸
¸
x
¸
¸
_
= x
−
1
3
.
Multiplying μ
1
(x) to the ﬁrst group yields
5x
2
3
y dx
_
dx
+
3x
5
3
dy
3x
5
3
y
∂
∂y
¸¸
=⇒ 3d
_
x
5
3
y
_
=⇒ v
1
(x, y) = x
5
3
y.
❧ For the second group: M
2
= −3y
3
, N
2
= −7x y
2
,
∂M
2
∂y
= −9y
2
,
∂N
2
∂x
= −7y
2
,
1
N
2
_
∂M
2
∂y
−
∂N
2
∂x
_
=
1
−7x y
2
(−9y
2
+ 7y
2
) =
2
7x
, A function of x only
μ
2
(x) = exp
_
_
1
N
2
_
∂M
2
∂y
−
∂N
2
∂x
_
dx
_
= exp
_
2
7
_
1
x
dx
_
= exp
_
2
7
ln
¸
¸
x
¸
¸
_
= x
2
7
.
52 2 ﬁrstorder and simple higherorder differential equations
Multiplying μ
2
(x) to the second group yields
−3x
2
7
y
3
dx
_
dx
+
−7x
9
7
y
2
dy
−
7
3
x
9
7
y
3
∂
∂y
¸
=⇒ −
7
3
d
_
x
9
7
y
3
_
=⇒ v
2
(x, y) = x
9
7
y
3
.
To ﬁnd an integrating factor for the original differential equation, one needs to ﬁnd
functions g
1
and g
2
such that
μ
1
· g
1
(v
1
) = μ
2
· g
2
(v
2
) =⇒ x
−
1
3
· g
1
(x
5
3
y) = x
2
7
· g
2
(x
9
7
y
3
).
Letting g
1
(v
1
) =v
α
1
and g
2
(v
2
) =v
β
2
leads to
x
−
1
3
(x
5
3
y)
α
= x
2
7
(x
9
7
y
3
)
β
=⇒ x
5α−1
3
y
α
= x
9β+2
7
y
3β
,
∴
5α − 1
3
=
9β + 2
7
α = 3β
⎫
⎬
⎭
=⇒
⎧
⎨
⎩
α =
1
2
,
β =
1
6
.
Hence, an integrating factor is given by
μ(x, y) = μ
1
· g
1
(v
1
) = x
−
1
3
· (x
5
3
y)
1
2
= x
1
2
y
1
2
.
Multiplying μ=x
1
2
y
1
2
to the original differential equation leads to
_
5x
3
2
y
3
2
dx
_
dx
+
3x
5
2
y
1
2
dy
_
2x
5
2
y
3
2
∂
∂y
¸
+
_
−3x
1
2
y
7
2
dx
_
dx
+
−7x
3
2
y
5
2
dy
_
−2x
3
2
y
7
2
∂
∂y
¸
= 0.
The general solution is then obtained as
2x
5
2
y
3
2
− 2x
3
2
y
7
2
= C =⇒ x
3
2
y
3
2
(x − y
2
) = C.
Remarks: In the following two examples, the techniques used in the method of
inspection can be combined with the method of integrating factors by groups to
result in an efficient way of finding an integrating factor.
Example 2.28 2.28
Solve ( ydx − xdy) +
_
x ydy = 0, x >0, y >0.
2.3 exact differential equations and integrating factors 53
Dividing the equation by x
2
leads to
ydx − xdy
x
2
+ x
−
3
2
y
1
2
dy = 0, (1)
∴ d
_
y
x
_
+ x
−
3
2
y
1
2
dy = 0. (1
)
. ,, . . ,, .
First group Second group
Obviously, the ﬁrst group has an integrating factor μ
1
=1 and the correspond
ing v
1
= y/x. The second group has an integrating factor μ
2
=x
3
2
y
−
1
2
and the
corresponding v
2
=y.
To ﬁnd an integrating factor for the original differential equation, one needs to
ﬁnd functions g
1
and g
2
such that
μ
1
· g
1
(v
1
) = μ
2
· g
2
(v
2
) =⇒ 1· g
1
_
y
x
_
= x
3
2
y
−
1
2
· g
2
( y).
Letting g
1
(v
1
) =v
α
1
and g
2
(v
2
) =v
β
2
leads to
1·
_
y
x
_
α
= x
3
2
y
−
1
2
· ( y)
β
=⇒ x
−α
y
α
= x
3
2
y
β−
1
2
,
∴
−α =
3
2
α = β −
1
2
⎫
⎬
⎭
=⇒
_
α = −
3
2
,
β = −1.
Hence, an integrating factor is given by
μ(x, y) = μ
1
· g
1
(v
1
) = x
3
2
y
−
3
2
.
Multiplying μ=x
3
2
y
−
3
2
to equation (1) leads to
_
x
−
1
2
y
−
1
2
dx
_
dx
+
−x
1
2
y
−
3
2
dy
_
2x
1
2
y
−
1
2
∂
∂y
¸
+ y
−1
dy
_
dy
ln y
= 0.
The general solution is then obtained as
2
_
x
y
+ ln y = C.
Example 2.29 2.29
Solve
_
x ye
x
y
+ y
4
_
dx − x
2
e
x
y
dy = 0, y =0.
54 2 ﬁrstorder and simple higherorder differential equations
Rearrange the equation as
xe
x
y
_
ydx − xdy) + y
4
dx = 0.
Dividing the equation by x y
2
leads to
e
x
y
ydx − xdy
y
2
+
y
2
x
dx = 0 =⇒ e
x
y
d
_
x
y
_
+
y
2
x
dx = 0,
∴ d
_
e
x
y
_
+
y
2
x
dx = 0. (1)
. ,, . . ,, .
First group Second group
Obviously, the ﬁrst group has an integrating factor μ
1
=1 and the corresponding
v
1
=e
x
y
. The second group has an integrating factor μ
2
=
x
y
2
and the correspond
ing v
2
=x.
To ﬁnd an integrating factor for the original differential equation, one needs to
ﬁnd functions g
1
and g
2
such that
μ
1
· g
1
(v
1
) = μ
2
· g
2
(v
2
) =⇒ 1· g
1
_
e
x
y
_
=
x
y
2
· g
2
(x).
To remove the exponential functiononthe lefthandside, one must take g
1
as a log
arithmic function. Because of
1
y
2
onthe righthandside, letting g
1
(v
1
) =
_
ln
¸
¸
v
1
¸
¸
_
2
and g
2
(v
2
) =v
2
leads to
1·
_
x
y
_
2
=
x
y
2
· x =⇒ μ =
x
2
y
2
.
Multiplying μ=
x
2
y
2
to equation (1) leads to
_
x
y
_
2
d
_
e
x
y
_
+ xdx = 0.
Since
_
_
x
y
_
2
d
_
e
x
y
_
=
_
z
2
d(e
z
), z =
x
y
= z
2
e
z
−
_
e
z
· 2z dz Integrating by parts
= z
2
e
z
− 2
_
z d(e
z
) = z
2
e
z
− 2
_
z e
z
−
_
e
z
dz
_
= e
z
(z
2
− 2z + 2) = e
x
y
__
x
y
_
2
− 2
_
x
y
_
+ 2
_
,
2.4 linear ﬁrstorder equations 55
one has
d
_
e
x
y
__
x
y
_
2
− 2
_
x
y
_
+ 2
_
_
+ d
_
x
2
2
_
= 0.
The general solution is then obtained as
e
x
y
__
x
y
_
2
− 2
_
x
y
_
+ 2
_
+
x
2
2
= C.
2.4 Linear FirstOrder Equations
2.4.1 Linear FirstOrder Equations
Linear ﬁrstorder equations occur in many engineering applications and are of the
form
dy
dx
+ P(x) · y = Q(x). (1)
The equation is ﬁrstorder because the highest order of derivative present is ﬁrst
dy
dx
, and it is linear because y and
dy
dx
appear linearly.
Sometimes, the roles of x and y may be exchangedto result ina linear ﬁrstorder
equation of the form
dx
dy
+ P( y) · x = Q( y),
in which x is treated as a function of y, and x and
dx
dy
appear linearly.
Equation (1) can be written in the form Mdx + Ndy = 0:
_
P(x) y − Q(x)
_
dx + dy = 0, (2)
in which M(x, y) =P(x) y−Q(x), N(x, y) =1. Test for exactness:
∂M
∂y
= P(x),
∂N
∂x
= 0 =⇒ Differential equation (2) is not exact.
However, since
1
N
_
∂M
∂y
−
∂N
∂x
_
= P(x)
is a function of x only, there exists an integrating factor that is a function of x only
given by
μ(x) = exp
_
_
1
N
_
∂M
∂y
−
∂N
∂x
_
dx
_
= e
_
P(x)dx
.
Multiplying equation (2) by the integrating factor μ(x) yields
_
P(x) y − Q(x)
_
e
_
P(x)dx
dx + e
_
P(x)dx
dy = 0.
56 2 ﬁrstorder and simple higherorder differential equations
The general solution can be determined using the method of grouping terms
_
e
_
P(x)dx
dy
_
dy
+
P(x) ye
_
P(x)dx
dx
_
ye
_
P(x)dx
∂
∂x
¸
+
−Q(x)e
_
P(x)dx
dx
_
dx
−
_
Q(x)e
_
P(x)dx
dx
= 0,
which results in
ye
_
P(x)dx
−
_
Q(x)e
_
P(x)dx
dx = C. General solution
The above results can be summarized as follows.
Linear FirstOrder Equations
1.
dy
dx
+ P(x) · y = Q(x) =⇒ y = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
_
.
2.
dx
dy
+ P( y) · x = Q( y) =⇒ x = e
−
_
P( y)dy
_
_
Q( y)e
_
P( y)dy
dy + C
_
.
Example 2.30 2.30
Solve y
= 1 + 3y tan x.
The differential equation can be written as
y
− 3 tan x · y = 1, Linear firstorder
which is of the form y
+P(x) · y =Q(x), P(x) =−3 tan x, Q(x) =1. The follow
ing quantities can be evaluated
_
P(x)dx = −3
_
tan x dx = 3 ln
¸
¸
cos x
¸
¸
,
e
_
P(x)dx
= e
3 lncos x
= cos
3
x, e
−
_
P(x)dx
=
1
cos
3
x
,
_
Q(x)e
_
P(x)dx
dx =
_
1· cos
3
xdx =
_
cos
2
x · cos xdx
=
_
( 1−sin
2
x )d(sin x) = sin x −
1
3
sin
3
x.
The general solution of the differential equation is
y = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
_
=
1
cos
3
x
_
sin x −
1
3
sin
3
x + C
_
.
2.4 linear ﬁrstorder equations 57
Example 2.31 2.31
Solve ydx − (e
y
+ 2x y − 2x)dy = 0.
It is easy to see that y =0 is a solution of the differential equation. For y =0, the
differential equation can be written as
dx
dy
+
2(1−y)
y
· x =
e
y
y
, Linear firstorder
which is of the form
dx
dy
+ P( y) · x = Q( y), P( y) =
2(1−y)
y
, Q( y) =
e
y
y
.
The following quantities can be evaluated
_
P( y)dy = 2
_
_
1
y
− 1
_
dy = 2
_
ln
¸
¸
y
¸
¸
− y
_
= ln
¸
¸
y
2
e
−2 y
¸
¸
,
e
_
P( y)dy
= e
ln y
2
e
−2 y

= y
2
e
−2 y
, e
−
_
P( y)dy
=
e
2 y
y
2
,
_
Q( y)e
_
P( y)dy
dy =
_
e
y
y
· y
2
e
−2 y
dy =
_
ye
−y
dy = −
_
y d(e
−y
)
= −
_
ye
−y
−
_
e
−y
dy
_
Integration by parts
= −( ye
−y
+ e
−y
) = −e
−y
( y + 1).
The general solution of the differential equation is
x = e
−
_
P( y)dy
_
_
Q( y)e
_
P( y)dy
dy + C
_
=
e
2 y
y
2
_
−e
−y
( y + 1) + C
_
,
∴ x y
2
= −e
y
( y + 1) + Ce
2 y
. General solution
Note that the solution y =0 is included in the general solution with C=1.
Example 2.32 2.32
Solve
dy
dx
=
1
x cos y + sin 2y
, cos y =0.
The differential equation can be written as
dx
dy
− cos y · x = sin 2y, Linear firstorder
which is of the form
dx
dy
+ P( y) · x = Q( y), P( y) = −cos y, Q( y) = sin 2y.
58 2 ﬁrstorder and simple higherorder differential equations
The following quantities can be evaluated
_
P( y)dy = −
_
cos ydy = −sin y, e
_
P( y)dy
= e
−sin y
, e
−
_
P( y)dy
= e
sin y
,
_
Q( y)e
_
P( y)dy
dy =
_
sin 2ye
−sin y
dy = 2
_
sin y cos y · e
−sin y
dy
= −2
_
sin y d(e
−sin y
) Integration by parts
= −2
_
sin y · e
−sin y
−
_
e
−sin y
d(sin y)
_
= −2e
−sin y
(sin y+1).
The general solution of the differential equation is
x = e
−
_
P( y)dy
_
_
Q( y)e
_
P( y)dy
dy + C
_
= e
sin y
_
−2e
−sin y
(sin y+1) + C
_
,
i.e.,
x = −2(sin y+1) + Ce
sin y
. General solution
2.4.2 Bernoulli Differential Equations
A differential equation of the form
dy
dx
+ P(x) · y = Q(x) · y
n
, n = Constant, (1)
is called a Bernoulli differential equation. Except when n=0 or 1, the equation is
nonlinear.
❧ When n=0, the equation reduces to a linear ﬁrstorder equation
dy
dx
+ P(x) · y = Q(x).
❧ When n=1, the equation can be written as
dy
dx
=
_
Q(x) − P(x)
_
y. Variable separable
❧ In the following, the case when n=0, 1 is considered.
For n<0, y =0. When n>0, y =0 is a solution of the differential equation. For
y =0, dividing both sides of equation (1) by y
n
yields
y
−n
dy
dx
+ P(x) · y
1−n
= Q(x).
Letting u= y
1−n
, when it is deﬁned, one has
du
dx
= (1−n) y
−n
dy
dx
=⇒ y
−n
dy
dx
=
1
1−n
du
dx
,
2.4 linear ﬁrstorder equations 59
and the equation becomes
1
1−n
du
dx
+ P(x)u = Q(x),
or
du
dx
+ (1−n)P(x) · u = (1−n)Q(x), n = 1. (2)
. ,, . . ,, .
¯
P(x)
¯
Q(x)
Hence, a Bernoulli differential equation is transformed to a linear ﬁrstorder equa
tion (2) in the new variable u.
Remarks: It is important to consider exchanging the roles of x and y so that
a differential equation can be cast into a linear firstorder equation or a Bernoulli
differential equation.
Bernoulli Differential Equations
1.
dy
dx
+ P(x) · y = Q(x) · y
n
u = y
1−n
=⇒
du
dx
+ (1−n)P(x) · u = (1−n)Q(x), Linear firstorder
. ,, . . ,, .
¯
P(x)
¯
Q(x)
2.
dx
dy
+ P( y) · x = Q( y) · x
n
u =x
1−n
=⇒
du
dy
+ (1−n)P( y) · u = (1−n)Q( y), Linear firstorder
. ,, . . ,, .
¯
P( y)
¯
Q( y)
Example 2.33 2.33
Solve 2x y y
= y
2
− 2x
3
, y(1) = 2.
The differential equation can be written as
y
−
1
2x
· y = −x
2
·
1
y
, Bernoulli DEwith n= −1
Multiplying both sides of the equation by y yields
y y
−
1
2x
· y
2
= −x
2
.
60 2 ﬁrstorder and simple higherorder differential equations
Letting u= y
2
=⇒
du
dx
=2y
dy
dx
, the equation becomes
1
2
du
dx
−
1
2x
· u = −x
2
=⇒
du
dx
−
1
x
· u = −2x
2
.
The equation is linear ﬁrstorder of the form
du
dx
+
¯
P(x) · u =
¯
Q(x),
¯
P(x) = −
1
x
,
¯
Q(x) = −2x
2
.
The following quantities can be evaluated
_
¯
P(x)dx =
_
−
1
x
dx =−ln
¸
¸
x
¸
¸
, e
_
¯
P(x)dx
=e
−lnx
=
1
x
, e
−
_
¯
P(x)dx
=x,
_
¯
Q(x)e
_
¯
P(x)dx
dx =
_
−2x
2
·
1
x
dx = −x
2
.
Hence
u = e
−
_
¯
P(x)dx
_
_
¯
Q(x)e
_
¯
P(x)dx
dx + C
_
= x(−x
2
+ C),
i.e.,
y
2
= x(−x
2
+ C). General solution
The constant C can be determined from the initial condition y(1) =2
2
2
= 1· (−1
2
+ C) =⇒ C = 5.
The particular solution is
y
2
= x(5−x
2
). Particular solution
Example 2.34 2.34
Solve 3ydx − x
_
3x
3
y ln
¸
¸
y
¸
¸
+ 1
_
dy = 0, y =0.
The differential equation can be written as
dx
dy
−
1
3y
· x = ln
¸
¸
y
¸
¸
· x
4
. Bernoulli DEwith n=4
Dividing both sides of the equation by x
4
yields
1
x
4
dx
dy
−
1
3y
·
1
x
3
= ln
¸
¸
y
¸
¸
.
Letting u=
1
x
3
=⇒
du
dy
= −
3
x
4
dx
dy
, the equation becomes
−
1
3
du
dy
−
1
3y
· u = ln
¸
¸
y
¸
¸
=⇒
du
dy
+
1
y
· u = −3 ln
¸
¸
y
¸
¸
.
2.5 equations solvable for the independent or dependent variable 61
The equation is linear ﬁrstorder of the form
du
dy
+
¯
P( y) · u =
¯
Q( y),
¯
P( y) =
1
y
,
¯
Q( y) = −3 ln
¸
¸
y
¸
¸
.
The following quantities can be evaluated
_
¯
P( y)dy =
_
1
y
dy = ln
¸
¸
y
¸
¸
, e
_
¯
P( y)dy
= e
ln y
= y, e
−
_
¯
P( y)dy
=
1
y
,
_
¯
Q( y)e
_
¯
P( y)dy
dy =
_
−3 ln
¸
¸
y
¸
¸
· y dy = −
3
2
_
ln
¸
¸
y
¸
¸
d( y
2
)
Integration
by parts
= −
3
2
_
y
2
ln
¸
¸
y
¸
¸
−
_
y
2
·
1
y
dy
_
= −
3
2
_
y
2
ln
¸
¸
y
¸
¸
−
1
2
y
2
_
.
Hence
u = e
−
_
¯
P( y)dy
_
_
¯
Q( y)e
_
¯
P( y)dy
dy + C
_
=
1
y
_
−
3
4
y
2
_
2 ln
¸
¸
y
¸
¸
− 1
_
+ C
_
.
Replacing u by the original variable results in the general solution
4
x
3
= −3y
_
2 ln
¸
¸
y
¸
¸
− 1
_
+
C
y
.
2.5 Equations Solvable for the Independent or
Dependent Variable
A general ﬁrstorder differential equation is of the form
F(x, y, y
) = 0, (1)
or
F(x, y, p) =0, y
= p. (1
)
In the following, cases when variable x or y can be solved are considered.
Case 1. Equation Solvable for Variable y
Suppose, fromequation (1
), variable y can be expressed explicitly as a function of
x and p to yield
y = f (x, p). (2)
Differentiating equation (2) with respect to x gives
dy
dx
=
∂f
∂x
+
∂f
∂p
dp
dx
=⇒ p = f
x
+ f
p
dp
dx
, (3)
which is a differential equation between x and p. If equation (3) can be solved to
obtain the general solution
φ(x, p, C) = 0, (4)
62 2 ﬁrstorder and simple higherorder differential equations
then the general solution of equation (1) can be obtained as follows.
❧ Eliminate variable p between equations (2) and (4) to obtain the solution in
terms of x and y.
❧ If it is difﬁcult to eliminate p between equations (2) and (4), then equations (2)
and (4) can be treated as parametric equations with p being the parameter.
For example, consider the parametric equations
x = a + r cos θ, y = b + r sin θ,
where a, b, and r are constants, and θ is the parameter. Rewrite the equations as
x − a
r
= cos θ,
y − b
r
= sin θ.
Using the trigonometric identity cos
2
θ +sin
2
θ =1, the parameter θ can be elimi
nated to yield
cos
2
θ + sin
2
θ =
_
x − a
r
_
2
+
_
y − b
r
_
2
= 1 =⇒ (x−a)
2
+ ( y−b)
2
= r
2
,
which is the equation of a circle with center at (a, b) and radius r.
Case 2. Equation Solvable for Variable x
Suppose, fromequation (1
), variable x can be expressed explicitly as a function of
y and p to yield
x = g( y, p). (5)
Differentiating equation (5) with respect to y gives
dx
dy
=
∂g
∂y
+
∂g
∂p
dp
dy
=⇒
1
p
= g
y
+ g
p
dp
dy
, (6)
which is a differential equation between y and p. If equation (6) can be solved to
obtain the general solution
ψ( y, p, C) = 0, (7)
then the general solution of equation (1) can be obtained as follows.
❧ Eliminate variable p between equations (5) and (7) to obtain the solution in
terms of x and y.
❧ If it is difﬁcult to eliminate p between equations (5) and (7), then equations (5)
and (7) can be treated as parametric equations with p being the parameter.
2.5 equations solvable for the independent or dependent variable 63
Example 2.35 2.35
Solve x =
dy
dx
+
_
dy
dx
_
4
.
Letting p=
dy
dx
, the equation can be written as
x = p + p
4
= f ( p),
which is the case of equation solvable for x. Differentiating with respect to y yields
1
p
=
df
dp
dp
dy
= (1 + 4p
3
)
dp
dy
,
dx
dy
=
1
p
,
which can be written as
dy = ( p + 4p
4
)dp. Variable separable
Integrating both sides leads to
y =
1
2
p
2
+
4
5
p
5
+ C.
Hence, the general solution is given by the parametric equations
_
x = p + p
4
,
y =
1
2
p
2
+
4
5
p
5
+ C,
where p is a parameter.
Example 2.36 2.36
Solve x
_
dy
dx
_
2
− 2y
dy
dx
− x = 0.
Solution 1. Letting p=
dy
dx
, one has
x p
2
− 2y p − x = 0. (1)
Since p=0 is not a solution, one has p=0. Variable y can be expressed explicitly
in terms of x and p to yield
y =
1
2
x
_
p −
1
p
_
= f (x, p). (2)
Differentiating equation (2) with respect to x gives
p =
∂f
∂x
+
∂f
∂p
dp
dx
=
1
2
_
p −
1
p
_
+
1
2
x
_
1 +
1
p
2
_
dp
dx
,
64 2 ﬁrstorder and simple higherorder differential equations
which can be simpliﬁed as
p
2
+1
p
_
x
p
dp
dx
− 1
_
= 0.
Since ( p
2
+1)/p=0, one has
x
p
dp
dx
− 1 = 0 =⇒
1
p
dp =
1
x
dx, Variable separable
which can be solved to give
_
1
p
dp =
_
1
x
dx + C =⇒ ln
¸
¸
p
¸
¸
= ln
¸
¸
x
¸
¸
+ ln
¸
¸
C
¸
¸
,
∴ ln
¸
¸
p
¸
¸
= ln
¸
¸
Cx
¸
¸
=⇒ p = Cx.
Substituting into equation (2) results in the general solution
y =
1
2
x
_
Cx −
1
Cx
_
=⇒ 2y = Cx
2
−
1
C
.
Solution 2. Since p= ±1 is not a solution, one has p= ±1. From equation (1),
variable x can also be expressed explicitly in terms of y and p to yield
x =
2y p
p
2
−1
= g( y, p). (2
)
Differentiating equation (2
) with respect to y yields
1
p
=
∂g
∂y
+
∂g
∂p
dp
dy
=
2p
p
2
−1
− 2y
p
2
+1
( p
2
−1)
2
dp
dy
,
which can be simpliﬁed as
−
p
2
+1
p( p
2
−1)
= −2y
p
2
+1
( p
2
−1)
2
dp
dy
=⇒
2p
p
2
−1
dp =
1
y
dy.
The equation is variable separable and can be solved by integrating both sides
_
1
p
2
−1
d( p
2
−1) =
_
1
y
dy + D,
ln
¸
¸
p
2
−1
¸
¸
= ln
¸
¸
2Cy
¸
¸
,
Set D= ln
¸
¸
2C
¸
¸
for the purpose of
comparing the results of two methods.
∴ p
2
− 1 = 2Cy. (3)
Parameter p can be eliminated between equations (2
) and (3). Substituting equa
tion (3) into (2
) yields
x =
2y p
2Cy
=⇒ p = Cx.
Substituting into equation (3) results in the general solution
C
2
x
2
− 1 = 2Cy.
2.5 equations solvable for the independent or dependent variable 65
Example 2.37 2.37
Solve y = x
_
dy
dx
+
_
1+
_
dy
dx
_
2
_
.
Letting p=
dy
dx
, the differential equation becomes
y = x
_
p +
_
1+p
2
_
= f (x, p). (1)
Differentiating equation (1) with respect to x yields
p =
∂f
∂x
+
∂f
∂p
dp
dx
= p +
_
1+p
2
+ x
_
1 +
p
_
1+p
2
_
dp
dx
,
which can be simpliﬁed as
−
_
1+p
2
= x
_
1 +
p
_
1+p
2
_
dp
dx
,
∴ −
1
x
dx =
_
1
_
1+p
2
+
p
1+p
2
_
dp. Variable separable
Integrating both sides leads to
−
_
1
x
dx =
_
_
1
_
1+p
2
+
p
1+p
2
_
dp + C,
−ln
¸
¸
x
¸
¸
= ln
¸
¸
p+
_
1+p
2
¸
¸
+
1
2
ln
¸
¸
1+p
2
¸
¸
+ ln
¸
¸
D
¸
¸
, C⇒ ln
¸
¸
D
¸
¸
∴
1
x
= D
_
p +
_
1+p
2
_ _
1+p
2
. (2)
The parameter p can be eliminated between equations (1) and (2). From equation
(1), one has
p +
_
1+p
2
=
y
x
. (3)
Substituting into equation (2) yields
1
x
= D
_
1+p
2
·
y
x
=⇒
_
1+p
2
=
C
y
or p
2
=
_
C
y
_
2
− 1.
1
D
⇒C
Substituting into equation (3) leads to
p =
y
x
−
_
1+p
2
=⇒ p
2
=
_
y
x
−
_
1+p
2
_
2
∴
_
C
y
_
2
− 1 =
_
y
x
−
C
y
_
2
,
66 2 ﬁrstorder and simple higherorder differential equations
which can be further simpliﬁed as
_
C
y
_
2
−
_
y
x
−
C
y
_
2
= 1,
_
C
y
+
_
y
x
−
C
y
___
C
y
−
_
y
x
−
C
y
__
= 1, a
2
−b
2
= (a+b)(a−b)
∴
y
x
_
2C
y
−
y
x
_
= 1 =⇒
x
y
+
y
x
=
C
y
. General solution, 2C⇒C
Example 2.38 2.38
Solve y
2
_
dy
dx
_
2
+ 3x
_
dy
dx
_
− y = 0.
Letting p=
dy
dx
, the differential equation can be written as
y
2
p
2
+ 3x p − y = 0.
If p=0, one must have y =0, which is a solution of the differential equation.
For p=0, solving for x gives
x = −
1
3
_
y
2
p −
y
p
_
= g( y, p). (1)
Differentiating equation (1) with respect to y yields
1
p
=
∂g
∂y
+
∂g
∂p
dp
dy
= −
1
3
_
2y p −
1
p
_
−
1
3
_
y
2
+
y
p
2
_
dp
dy
.
Multiplying both sides by −3p and arranging yield
( y p
2
+ 1)
_
2 +
y
p
dp
dy
_
= 0.
Case 1. 2 +
y
p
dp
dy
= 0
The equation can be written as
1
p
dp = −
2
y
dy. Variable separable
Integrating both sides yields
_
1
p
dp = −2
_
1
y
dy + C =⇒ ln
¸
¸
p
¸
¸
= −2 ln
¸
¸
y
¸
¸
+ ln
¸
¸
C
¸
¸
=⇒ p =
C
y
2
.
2.5 equations solvable for the independent or dependent variable 67
Substituting into equation (1) results in the general solution
x = −
1
3
_
C
y
2
_
_
y
2
_
C
y
2
_
2
−y
_
=⇒ 3Cx−y
3
+C
2
= 0. General solution
Case 2. y p
2
+ 1 = 0 =⇒ p
2
= −
1
y
Substituting into equation (1) results in
x
2
=
1
9p
2
( y
2
p
2
− y)
2
=⇒ x
2
=
1
9
_
−
1
y
_
_
y
2
_
−
1
y
_
− y
_
2
,
∴ 9x
2
+ 4y
3
= 0. Singular solution
This solution is not obtainable from the general solution for any value of C and is
therefore a singular solution.
The Clairaut Equation
A ﬁrstorder differential equation of the form
y = x y
+ f ( y
),
or
y = x p + f ( p), p =
dy
dx
.
is called the Clairaut equation. The equation is of the type of equation solvable for
variable y and can be solved using the approach introduced in this section.
Example 2.39 2.39
Solve y = x y
+
a
2
y
.
Letting y
= p, the differential equation becomes
y = x p +
a
2
p
= f (x, p). (1)
Differentiating with respect to x yields
p =
∂f
∂x
+
∂f
∂p
dp
dx
= p +
_
x −
a
2
p
2
_
dp
dx
=⇒
_
x −
a
2
p
2
_
dp
dx
= 0.
Case 1.
dp
dx
= 0 =⇒ p = C.
68 2 ﬁrstorder and simple higherorder differential equations
Substituting into equation (1) results in the general solution
y = Cx +
a
2
C
, General solution (2)
which is a family of straight lines with slope C and yintercept a
2
/C.
Case 2. x −
a
2
p
2
= 0 =⇒ p
2
=
a
2
x
.
Substituting into equation (1) gives
py = x p
2
+ a
2
=⇒ p
2
y
2
= (x p
2
+ a
2
)
2
,
a
2
x
· y
2
=
_
x ·
a
2
x
+ a
2
_
2
=⇒ y
2
= 4a
2
x. Singular solution
This solution is a parabola, which cannot be obtained fromthe general solution (2)
for any value of C and is therefore a singular solution.
2.6 Simple HigherOrder Differential Equations
2.6.1 Equations Immediately Integrable
An nthorder differential equation of the form
d
n
y
dx
n
= f (x)
can be solved easily by integrating n times the function f (x) with respect to x.
The general solution of an nthorder differential equation contains n constants of
integration.
Example 2.40 2.40
Solve x
2
y
= 120x
3
+ 8x
2
e
2x
+ 1, x =0.
Dividing both sides of the equation by x
2
yields
y
= 120x + 8e
2x
+
1
x
2
.
Integrating the equation with respect to x once gives
y
= 60x
2
+ 4e
2x
−
1
x
+ C
1
.
Integrating the equation with respect to x again leads to
y
= 20x
3
+ 2e
2x
− ln
¸
¸
x
¸
¸
+ C
1
x + C
2
. (1)
2.6 simple higherorder differential equations 69
Since
_
ln
¸
¸
x
¸
¸
dx = x ln
¸
¸
x
¸
¸
−
_
x ·
1
x
dx = x
_
ln
¸
¸
x
¸
¸
− 1
_
, Integrating by parts
integrating equation (1) with respect to x results in the general solution
y = 5x
4
+ e
2x
− x
_
ln
¸
¸
x
¸
¸
− 1
_
+
1
2
C
1
x
2
+ C
2
x + C
3
,
∴ y = C
0
+ C
1
x + C
2
x
2
+ 5x
4
+ e
2x
− x ln
¸
¸
x
¸
¸
. General solution
Example 2.41 2.41
Solve e
−x
y
− sin x = 2, y(0) = y
(0) =1.
The equation can be written as
y
= e
x
sin x + 2e
x
. (1)
It can be easily determined
_
e
x
sin x dx =
_
sin x d(e
x
) = e
x
sin x −
_
e
x
cos x dx Integrating by parts
= e
x
sin x −
_
cos x d(e
x
) = e
x
sin x −
_
e
x
cos x +
_
e
x
sin x dx
_
,
=⇒
_
e
x
sin x dx =
1
2
e
x
(sin x − cos x),
_
e
x
cos x dx =
_
cos x d(e
x
) = e
x
cos x +
_
e
x
sin x dx
= e
x
cos x +
1
2
e
x
(sin x − cos x) =
1
2
e
x
(sin x + cos x).
Integrating equation (1) with respect to x yields
y
=
1
2
e
x
(sin x − cos x) + 2e
x
+ C
1
.
The constant C
1
can be determined from the initial condition y
(0) =1:
1 =
1
2
e
0
(sin 0 − cos 0) + 2e
0
+ C
1
=⇒ C
1
= −
1
2
.
Hence
y
=
1
2
e
x
(sin x − cos x) + 2e
x
−
1
2
.
Integrating with respect to x again gives
y =
1
2
_
1
2
e
x
(sin x − cos x) −
1
2
e
x
(sin x + cos x)
_
+ 2e
x
−
1
2
x + C
2
= −
1
2
e
x
cos x + 2e
x
−
1
2
x + C
2
.
70 2 ﬁrstorder and simple higherorder differential equations
The constant C
2
is determined from the initial condition y(0) =1:
1 = −
1
2
e
0
cos 0 + 2e
0
−
1
2
· 0 + C
2
=⇒ C
2
= −
1
2
.
∴ y = −
1
2
e
x
cos x + 2e
x
−
1
2
x −
1
2
. Particular solution
2.6.2 The Dependent Variable Absent
In general, an nthorder differential equation is of the form
f
_
x; y, y
, y
, . . . , y
(n)
_
= 0.
If the dependent variable y is absent, or more generally, y, y
, y
, . . . , y
(k−1)
are
absent, the differential equation is of the form
f
_
x; y
(k)
, y
(k+1)
, . . . , y
(n)
_
= 0.
Let u= y
(k)
be the new dependent variable, the differential equation becomes
f
_
x; u, u
, u
, . . . , u
(n−k)
_
= 0,
which is an (n−k)thorder equation. Hence the order of the differential equation
is reduced from n to (n−k).
Remarks: It should be emphasized that for the differential equation to be con
sidered as y
(k−1)
absent, all derivatives with order lower than (k−1) must also be
absent.
For example, equation
y
+ y
= g(x)
is of the type y absent, whereas equation
y
+ y
= g(x)
is of the type y
absent.
Example 2.42 2.42
Solve 2y
= ( y
)
3
sin 2x, y(0) = y
(0) = 1.
The equation is of the type y absent. Let y
=u, y
=u
. Since y
=u=0 does
not satisfy the initial condition y
(0) =1, one has u=0. The equation becomes
2
du
dx
= u
3
sin 2x =⇒
2
u
3
du = sin 2x dx. Variable separable
Integrating both sides yields
_
2
u
3
du =
_
sin 2x dx + C
1
=⇒ −
1
u
2
= −
cos 2x
2
+ C
1
.
2.6 simple higherorder differential equations 71
The constant C
1
can be determined from the initial condition u(0) = y
(0) =1:
−1 = −
1
2
+ C
1
=⇒ C
1
= −
1
2
.
Hence
−
1
u
2
= −
1 + cos 2x
2
= −cos
2
x =⇒ u = ±
1
cos x
.
Since u(0) =1, only the positive sign is taken, which leads to
u =
1
cos x
=⇒
dy
dx
= sec x. Immediately integrable
Integrating both sides results in
y =
_
sec xdx + C
2
= ln
¸
¸
sec x + tan x
¸
¸
+ C
2
.
The constant C
2
can be determined using the initial condition y(0) =1:
1 = ln
¸
¸
1 + 0
¸
¸
+ C
2
=⇒ C
2
= 1,
∴ y = ln
¸
¸
sec x + tan x
¸
¸
+ 1. Particular solution
Example 2.43 2.43
Solve x y
− ( y
)
3
− y
= 0.
The equation is of the type y absent. Denoting y
=u, y
=u
, one has
xu
− u
3
− u = 0 =⇒ u
−
1
x
· u =
1
x
· u
3
. Bernoulli DE
Case 1. u=0 =⇒ y =C is a solution of the differential equation.
Case 2. u=0. Dividing both sides by u
3
leads to
1
u
3
u
−
1
x
·
1
u
2
=
1
x
.
Letting v =
1
u
2
,
dv
dx
=−
2
u
3
du
dx
=⇒
1
u
3
du
dx
=−
1
2
dv
dx
, the equation becomes
−
1
2
dv
dx
−
1
x
· v =
1
x
=⇒
dv
dx
+
2
x
· v = −
2
x
,
which is linear ﬁrstorder with
P(x) =
2
x
, Q(x) = −
2
x
.
The following quantities can be easily determined
_
P(x)dx =
_
2
x
dx = 2 ln
¸
¸
x
¸
¸
, e
_
P(x)dx
= e
2 lnx
= x
2
, e
−
_
P(x)dx
=
1
x
2
,
72 2 ﬁrstorder and simple higherorder differential equations
_
Q(x)e
_
P(x)dx
dx =
_
−
2
x
· x
2
dx = −x
2
.
Hence, the solution is
v = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
1
_
=
1
x
2
(−x
2
+ C
1
),
1
u
2
=
C
1
− x
2
x
2
=⇒ u =
dy
dx
= ±
x
_
C
1
−x
2
. Immediately integrable
Integrating both sides results in the general solution
y = ±
_
x
_
C
1
−x
2
dx + C
2
= ±
_
C
1
−x
2
+ C
2
=⇒ y − C
2
= ±
_
C
1
−x
2
.
Squaring both sides gives
( y − C
2
)
2
= C
1
− x
2
=⇒ x
2
+ ( y − C
2
)
2
= C
1
. General solution
2.6.3 The Independent Variable Absent
When the independent variable x does not appear explicitly, an nthorder differ
ential equation is of the form
f
_
y, y
, y
, . . . , y
(n)
_
= 0. (1)
Let y be the new independent variable and u= y
the new dependent variable.
Using the chain rule, it is easy to show that
dy
dx
= u,
d
2
y
dx
2
=
du
dx
=
du
dy
dy
dx
= u
du
dy
, Chain rule
d
3
y
dx
3
=
d
dx
_
d
2
y
dx
2
_
=
d
dy
_
u
du
dy
_
dy
dx
= u
_
du
dy
_
2
+ u
2
d
2
u
dy
2
,
· · · · · ·
It may be shown that
d
k
y
dx
k
may be expressed in terms of u,
du
dy
, . . . ,
d
k−1
u
dy
k−1
, for
k n. Hence, differential equation (1) becomes
f
_
y; u , u
du
dy
, u
_
du
dy
_
2
+u
2
d
2
u
dy
2
, . . .
_
= 0,
or
g
_
y; u,
du
dy
,
d
2
u
dy
2
, . . . ,
d
n−1
u
dy
n−1
_
= 0,
in which the order of the differential equation is reduced by 1.
2.6 simple higherorder differential equations 73
Example 2.44 2.44
Solve 3y y
y
− ( y
)
3
+ 1 = 0.
The equation is of the type x absent. Let y be the new independent variable and
u= y
the new dependent variable, y
=u du/dy.
Case 1. u
3
=1 =⇒ u=1 =⇒ y =x, which is a solution of the equation.
Case 2. For u=1, the equation becomes
3y · u· u
du
dy
− u
3
+ 1 = 0 =⇒
3u
2
u
3
− 1
du =
1
y
dy. Variable separable
Integrating both sides yields
_
1
u
3
−1
d(u
3
−1) =
_
1
y
dy + C
1
=⇒ ln
¸
¸
u
3
−1
¸
¸
= ln
¸
¸
y
¸
¸
+ ln
¸
¸
C
1
¸
¸
,
u
3
−1 = C
1
y =⇒
dy
dx
= u = (C
1
y+1)
1
3
.
Since y =constant is not a solution, one has (C
1
y+1) =0; hence
(C
1
y+1)
−
1
3
dy = dx. Variable separable
Integrating both sides results in the general solution
1
C
1
_
(C
1
y+1)
−
1
3
d(C
1
y+1) = x + C
2
=⇒
1
C
1
·
3
2
(C
1
y+1)
2
3
= x + C
2
,
∴ 3(C
1
y+1)
2
3
− 2C
1
x = C
2
. General solution
Example 2.45 2.45
Solve y y
= ( y
)
2
(1 − y
sin y − y y
cos y).
The equation is of the type x absent. Let y be the new independent variable and
u= y
the new dependent variable, y
=u du/dy. The equation becomes
y · u
du
dy
= u
2
(1 − u sin y − yu cos y).
Case 1. u = 0 =⇒ y
= 0 =⇒ y = C.
Case 2. y
du
dy
= u(1 − u sin y − yu cos y).
The differential equation can be written as
du
dy
−
1
y
· u = −
sin y + y cos y
y
· u
2
. Bernoulli DE
74 2 ﬁrstorder and simple higherorder differential equations
Dividing both sides by u
2
yields
1
u
2
du
dy
−
1
y
·
1
u
= −
sin y + y cos y
y
.
Letting v =
1
u
,
dv
dy
=−
1
u
2
du
dy
, one obtains
dv
dy
+
1
y
· v =
sin y + y cos y
y
. Linear firstorder
.,,. . ,, .
P( y) Q( y)
The following quantities can be easily determined
_
P( y)dy =
_
1
y
dy = ln
¸
¸
y
¸
¸
, e
_
P( y)dy
= e
ln y
= y, e
−
_
P( y)dy
=
1
y
,
_
Q( y)e
_
P( y)dy
dy =
_
sin y + y cos y
y
· ydy =
_
(sin y + y cos y)dy
= −cos y +
_
y d(sin y) = −cos y + y sin y −
_
sin y dy = y sin y.
Hence, the solution is
v = e
−
_
P( y)dy
_
_
Q( y)e
_
P( y)dy
dy + C
1
_
=
1
y
( y sin y + C
1
) = sin y +
C
1
y
,
∴ v =
1
u
=
dx
dy
= sin y +
C
1
y
. Immediately integrable
Integrating both sides yields
x =
_
_
sin y +
C
1
y
_
dy + C
2
=⇒ x =−cos y + C
1
ln
¸
¸
y
¸
¸
+ C
2
.
Therefore, the solutions are y = C, x =−cos y + C
1
ln
¸
¸
y
¸
¸
+ C
2
.
2.7 Summary
In this chapter, various types of ﬁrstorder and simple higherorder ordinary dif
ferential equations and the associated methods are introduced. The key in solving
such equations depends on identifying the type of the equations. Sometimes, an
equation can be classiﬁed as more than one type and solved using more than one
method.
When one is given a differential equation, the following steps may be followed to
identify and solve the equation.
☞
When dividing a differential equation by a function, care must be taken to
ensure that the function is not zero. The case when the function is zero should be
considered separately to determine if it will give extra solutions.
2.7 summary 75
FirstOrder Ordinary Differential Equation
1. Method of Separation of Variables
If the differential equation can be written as
dy
dx
= f (x)g( y) or f
1
(x)g
1
( y)dx + f
2
(x)g
2
( y)dy = 0,
then the equation is variable separable. Moving all terms involving variable x to
one side and all terms involving variable y to the other side of the equation gives
1
g( y)
dy = f (x)dx or
g
2
( y)
g
1
( y)
dy = −
f
1
(x)
f
2
(x)
dx.
Integrating both sides of the equation yields the general solution
_
1
g( y)
dy =
_
f (x)dx + C or
_
g
2
( y)
g
1
( y)
dy = −
_
f
1
(x)
f
2
(x)
dx + C.
This solution is valid for g( y) =0 or g
1
( y) =0. Check if g( y) =0 or g
1
( y) =0 is
a solution of the differential equation. If it is a solution and is not included in the
general solution, then it is a singular solution.
2. Homogeneous Equations
If the differential equation can be written in the form
dy
dx
= f
_
y
x
_
, then it is a
homogeneous equation. Applying the transformation
y
x
= v =⇒ y = xv =⇒
dy
dx
= v + x
dv
dx
,
the differential equation becomes variable separable =⇒ x
dv
dx
= f (v) − v.
3. Linear FirstOrder Equations and Bernoulli Equations
Rewrite the differential equation in the form
dy
dx
+ P(x) · y = f (x, y) or
dx
dy
+ P( y) · x = f (x, y).
Depending on the righthand side, the equation may be linear ﬁrstorder
dy
dx
+ P(x) · y = Q(x) =⇒ y = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
_
,
dx
dy
+ P( y) · x = Q( y) =⇒ x = e
−
_
P( y)dy
_
_
Q( y)e
_
P( y)dy
dy + C
_
,
76 2 ﬁrstorder and simple higherorder differential equations
or Bernoulli equations, for n=0, 1, which can be converted to linear ﬁrstorder
equations using a change of variable,
dy
dx
+ P(x) · y = Q(x) · y
n
u = y
1−n
==⇒
du
dx
+ (1−n)P(x) · u = (1−n)Q(x),
. ,, . . ,, .
¯
P(x)
¯
Q(x)
dx
dy
+ P( y) · x = Q( y) · x
n
u = x
1−n
==⇒
du
dy
+ (1−n)P( y) · u = (1−n)Q( y).
. ,, . . ,, .
¯
P( y)
¯
Q( y)
4. Exact Differential Equations and Integrating Factors
Consider the differential equation M(x, y)dx + N(x, y)dy = 0.
Exact Differential Equations: If
∂M
∂y
=
∂N
∂x
, then the equation is exact.
The method of grouping terms can be applied to ﬁnd the general solution
Mdx + Ndy = 0
Grouping terms
=====⇒
du(x, y) = 0 =⇒ u(x, y) = C.
Integrating Factors: If
∂M
∂y
=
∂N
∂x
, an integrating factor μ(x, y) may be deter
mined so that μMdx + μNdy = 0 is exact.
❧ If
1
N
_
∂M
∂y
−
∂N
∂x
_
= g(x) =⇒ μ(x) = exp
_
_
g(x)dx
_
.
❧ If
1
M
_
∂N
∂x
−
∂M
∂y
_
= g( y) =⇒ μ( y) = exp
_
_
g( y)dy
_
.
❧ Method of Inspection—By rearranging terms, multiplying or dividing suit
able functions, an integrating factor may be found
1. d(x y) = ydx + xdy,
2. d
_
y
x
_
=
−ydx + xdy
x
2
, d
_
x
y
_
=
ydx − xdy
y
2
,
3. d
_
tan
−1
y
x
_
=
−ydx + xdy
x
2
+ y
2
, d
_
tan
−1
x
y
_
=
ydx − xdy
x
2
+ y
2
,
4. d
_
1
2
ln
_
x
2
+ y
2
_
_
=
xdx + ydy
x
2
+ y
2
,
5. d
__
x
2
± y
2
_
=
xdx ± ydy
_
x
2
± y
2
.
2.7 summary 77
❧ Integrating Factors by Groups—Separate the equation into two groups
_
M
1
(x, y)dx + N
1
(x, y)dy
_
+
_
M
2
(x, y)dx + N
2
(x, y)dy
_
= 0.
Suppose μ
1
(x, y) and μ
2
(x, y) are the integrating factors for the ﬁrst and
second groups, respectively, and
μ
1
M
1
dx + μ
1
N
1
dy = dv
1
, μ
2
M
2
dx + μ
2
N
2
dy = dv
2
.
If functions g
1
and g
2
can be found such that
μ
1
(x, y) · g
1
_
v
1
(x, y)
_
= μ
2
(x, y) · g
2
_
v
2
(x, y)
_
= μ(x, y),
then μ(x, y) is an integrating factor.
5. Equation Solvable for the Independent or Dependent Variable
❧ Equation Solvable for Variable y: y = f (x, p), p = y
Differentiate with respect to x
p =
∂f
∂x
+
∂f
∂p
dp
dx
=⇒ φ(x, p, C) = 0.
General solution (parametric): y = f (x, p), φ(x, p, C) =0, p=parameter.
❧ Equation Solvable for Variable x: x = g( y, p), p = y
Differentiate with respect to y
1
p
=
∂g
∂y
+
∂g
∂p
dp
dy
=⇒ ψ( y, p, C) = 0.
General solution (parametric): x =g( y, p), ψ( y, p, C) =0, p=parameter.
6. Method of Special Transformations
There are no systematic procedures and rules to follow in applying the method
of special transformations. The key is to uncover the “special” term in a given
equation and determine a transformation accordingly.
Simple HigherOrder Differential Equation
1. Equations Immediately Integrable
d
n
y
dx
n
= f (x)
Integrate n times w.r.t. x
=======⇒
General solution.
2. The Dependent Variable Absent
f
_
x; y
(k)
, y
(k+1)
, . . . , y
(n)
_
= 0 =⇒ y, y
, y
, . . . , y
(k−1)
absent.
78 2 ﬁrstorder and simple higherorder differential equations
Let u= y
(k)
=⇒ f
_
x; u, u
, u
, . . . , u
(n−k)
_
= 0. The order of the equation is
reduced by k.
3. The Independent Variable Absent
f
_
y, y
, . . . , y
(n)
_
= 0 =⇒ x absent.
Let y be the new independent variable and u= y
the new dependent variable,
dy
dx
= u,
d
2
y
dx
2
= u
du
dy
,
d
3
y
dx
3
= u
_
du
dy
_
2
+ u
2
d
2
u
dy
2
, · · · .
The order of the differential equation is reduced by 1.
Problems
Solve the following differential equations, if not stated otherwise.
Variable Separable
2.1 cos
2
ydx+(1+e
−x
) sin ydy = 0
A
NS ln(e
x
+1) = −
1
cos y
+C; cos y =0
2.2
dy
dx
=
x
3
e
x
2
y ln y
A
NS y
2
_
ln y−
1
2
_
= e
x
2
(x
2
−1) + C
2.3 x cos
2
y dx + e
x
tan y dy = 0
A
NS e
−x
(x+1) =
1
2 cos
2
y
+ C; cos y =0
2.4 x ( y
2
+ 1) dx + (2y + 1) e
−x
dy = 0
A
NS (x−1)e
x
+ ln( y
2
+1) + tan
−1
y = C
2.5 x y
3
dx + e
x
2
dy = 0
A
NS e
−x
2
+
1
y
2
= C; y =0
2.6 x cos
2
y dx + tan y dy = 0
A
NS x
2
+ tan
2
y = C
2.7 x y
3
dx + ( y+1)e
−x
dy = 0
A
NS e
x
(x−1) −
1
y
−
1
2y
2
= C; y =0
Homogeneous and Special Transformations
2.8
dy
dx
+
x
y
+ 2 = 0
A
NS ln
¸
¸
x+y
¸
¸
+
x
x+y
= C; y = −x
2.9 x dy − y dx = x cot
_
y
x
_
dx
A
NS cos
_
y
x
_
=
C
x
2.10
_
x cos
2
_
y
x
_
− y
_
dx + x dy = 0
A
NS ln
¸
¸
x
¸
¸
+ tan
y
x
= C; cos
y
x
=0
2.11 x dy = y(1 + ln y − ln x) dx
A
NS y = xe
Cx
Problems 79
2.12 x y dx + (x
2
+ y
2
) dy = 0
A
NS y
2
(2x
2
+ y
2
) = C
2.13
_
1 + exp
_
−
y
x
__
dy +
_
1 −
y
x
_
dx = 0
A
NS x exp
_
y
x
_
+ y = C
2.14 (x
2
− x y + y
2
) dx − x y dy = 0
A
NS ( y−x)e
y/x
= C
2.15 (3 + 2x + 4y) y
= 1 + x + 2y
A
NS 8y−4x+ln
¸
¸
4x+8y+5
¸
¸
=C; 4x+8y+5=0
2.16 y
=
2x + y − 1
x − y − 2
A
NS
√
2 tan
−1
y + 1
√
2(x−1)
=ln
_
( y+1)
2
+2(x−1)
2
_
+C
2.17 ( y + 2)dx =(2x + y − 4)dy
A
NS ( y+2)
2
=C(x+y−1); y =1−x
2.18 y
= sin
2
(x−y)
A
NS x = tan(x−y)+C; x−y =
π
2
±kπ, k =0, 1, 2, . . .
2.19
dy
dx
= (x+1)
2
+ (4y+1)
2
+ 8x y + 1
A
NS
2
3
(x+4y+1) = tan(6x+C)
Exact Differential Equations
2.20 (3x
2
+ 6x y
2
) dx + (6x
2
y + 4y
3
) dy = 0
A
NS x
3
+ 3x
2
y
2
+ y
4
= C
2.21 (2x
3
− x y
2
− 2y + 3) dx − (x
2
y + 2x) dy = 0
A
NS x
4
− x
2
y
2
− 4x y + 6x = C
2.22 (x y
2
+ x − 2y + 3) dx + x
2
y dy = 2(x + y) dy
A
NS x
2
y
2
+ x
2
+ 6x − 4x y − 2y
2
= C
2.23 3y(x
2
− 1) dx + (x
3
+ 8y − 3x) dy = 0, when x = 0, y = 1
A
NS x
3
y − 3x y + 4y
2
= 4
2.24 (x
2
+ ln y) dx +
x
y
dy = 0
A
NS
1
3
x
3
+ x ln y = C
2.25 2x(3x + y − ye
−x
2
) dx + (x
2
+ 3y
2
+ e
−x
2
) dy = 0
A
NS 2x
3
+ x
2
y + ye
−x
2
+ y
3
= C
2.26 (3 + y + 2y
2
sin
2
x) dx + (x + 2x y − y sin 2x) dy = 0
A
NS 3x + x y + x y
2
−
1
2
y
2
sin 2x = C
2.27 (2x y + y
2
) dx + (x
2
+ 2x y + y
2
) dy = 0
A
NS x
2
y+x y
2
+
1
3
y
3
=C
Integrating Factors
2.28 (x
2
− sin
2
y) dx + x sin 2y dy = 0
A
NS x +
sin
2
y
x
= C
80 2 ﬁrstorder and simple higherorder differential equations
2.29 y(2x − y + 2) dx + 2(x − y) dy = 0
A
NS ye
x
(2x − y) = C
2.30 (4x y + 3y
2
− x) dx + x(x + 2y) dy = 0
A
NS 4x
4
y + 4x
3
y
2
− x
4
= C
2.31 y dx + x( y
2
+ ln x) dy = 0
A
NS 3y ln x + y
3
= C
2.32 (x
2
+ 2x + y) dx + (3x
2
y − x) dy = 0
A
NS x+2 ln
¸
¸
x
¸
¸
−
y
x
+
3
2
y
2
=C
2.33 y
2
dx + (x y + y
2
− 1) dy = 0
A
NS x y +
1
2
y
2
− ln
¸
¸
y
¸
¸
= C
2.34 3(x
2
+ y
2
) dx + x(x
2
+ 3y
2
+ 6y) dy = 0
A
NS xe
y
(x
2
+ 3y
2
) = C
2.35 2y(x + y + 2) dx + ( y
2
− x
2
− 4x − 1) dy = 0
A
NS x
2
+ 4x + 2x y + y
2
+ 1 = Cy
2.36 (2 + y
2
+ 2x) dx + 2y dy = 0
A
NS e
x
(2x + y
2
) = C
2.37 (2x y
2
− y) dx + ( y
2
+ x + y) dy = 0
A
NS x
2
−
x
y
+y+ln
¸
¸
y
¸
¸
=C
2.38 y(x + y) dx + (x + 2y − 1) dy = 0
A
NS e
x
(x y + y
2
− y) = C
2.39 2x(x
2
−sin y+1)dx + (x
2
+1) cos ydy =0
A
NS ln(x
2
+1)+
sin y
x
2
+1
=C
2.40 Consider a homogeneous differential equation of the form
M(u) dx + N(u) dy = 0, u =
y
x
.
If Mx+N y =0, i.e., M(u)+N(u)u=0, show that
1
xM
is an integrating factor.
Method of Inspection
2.41 (x
2
+ y + y
2
) dx − x dy = 0
A
NS x − tan
−1
y
x
= C
2.42
_
x −
_
x
2
+y
2
_
dx +
_
y −
_
x
2
+y
2
_
dy = 0
A
NS
_
x
2
+y
2
− x − y = C
2.43 y
_
1+y
2
dx +
_
x
_
1+y
2
− y
_
dy = 0
A
NS x y −
_
1+y
2
= C
2.44 y
2
dx − (x y + x
3
) dy = 0
A
NS
1
2
_
y
x
_
2
+ y = C
2.45 y dx − x dy − 2x
3
tan
y
x
dx = 0
A
NS sin
y
x
= Ce
−x
2
2.46 (2x
2
y
2
+ y) dx + (x
3
y − x) dy = 0
A
NS x
2
y + ln
¸
¸
¸
x
y
¸
¸
¸ = C
2.47 y
2
dx +
_
x y + tan(x y)
_
dy = 0
A
NS y sin(x y) = C
2.48 (2x
2
y
4
− y) dx + (4x
3
y
3
− x) dy = 0
A
NS 2x y
2
+
1
x y
= C
Problems 81
Integrating Factors by Groups
2.49 (x
2
y
3
+ y) dx + (x
3
y
2
− x) dy = 0
A
NS
1
2
x
2
y
2
+ ln
¸
¸
¸
x
y
¸
¸
¸ = C
2.50 y( y
2
+ 1) dx + x( y
2
− x + 1) dy = 0
A
NS
1
x y
−
1
y
− tan
−1
y = C
2.51 y
2
dx + (e
x
− y) dy = 0
A
NS −ye
−x
+ ln
¸
¸
y
¸
¸
= C
2.52 (x
2
y
2
− 2y) dx + (x
3
y − x) dy = 0
A
NS ln
¸
¸
x y
¸
¸
+
1
x
2
y
= C
2.53 (2x
3
y + y
3
) dx − (x
4
+ 2x y
2
) dy = 0
A
NS 4x
−
1
3
y
2
3
− x
8
3
y
−
4
3
= C
Linear FirstOrder Equations
2.54 (1 + y cos x) dx − sin x dy = 0
A
NS y = −cos x + C sin x
2.55 (sin
2
y + x cot y) y
= 1
A
NS x = sin y(C−cos y)
2.56 dx − ( y − 2x y) dy = 0
A
NS 2x = 1 + C exp(−y
2
)
2.57 dx − (1 + 2x tan y) dy = 0
A
NS 2x cos
2
y = y + sin y cos y + C
2.58
dy
dx
_
y
3
+
x
y
_
= 1
A
NS x =
1
3
y
4
+ Cy
2.59 dx + (x − y
2
) dy = 0
A
NS x = y
2
− 2y + 2 + Ce
−y
2.60 y
2
dx + (x y + y
2
− 1) dy = 0
A
NS y
2
+ 2x y − 2 ln
¸
¸
y
¸
¸
= C
2.61 y dx = (e
y
+ 2x y − 2x) dy
A
NS y
2
x = Ce
2 y
− ( y+1)e
y
2.62 (2x + 3) y
= y + (2x + 3)
1/2
, y(−1) =0
A
NS 2y =
√
2x+3 ln
¸
¸
2x+3
¸
¸
2.63 y dx + ( y
2
e
y
− x) dy = 0
A
NS x = Cy − ye
y
2.64 y
= 1 + 3y tan x
A
NS y =
1
cos
3
x
_
sin x −
1
3
sin
3
x + C
_
2.65 (1 + cos x) y
= sin x(sin x + sin x cos x − y)
A
NS y = (1 + cos x)(x − sin x + C)
2.66 y
= (sin
2
x − y) cos x
A
NS y = sin
2
x − 2 sin x + 2 + Ce
−sin x
2.67 x y
− ny − x
n+2
e
x
= 0, n =constant
A
NS y = x
n
_
e
x
(x−1)+C
_
2.68 (1+x)
dy
dx
− y = x(1+x)
2
A
NS y = (1+x)
_
1
2
x
2
+C
_
2.69 (1+y) dx +
_
x−y(1+y)
2
_
dy =0
A
NS x =
1
1+y
_
y
4
4
+
2y
3
3
+
y
2
2
+C
_
82 2 ﬁrstorder and simple higherorder differential equations
2.70 Consider the ﬁrstorder differential equation
dy
dx
= α(x) F( y) + β(x) G( y)
If
G
( y)F( y) − G( y)F
( y)
F( y)
= a = constant, then the transformation u =
G( y)
F( y)
reduces the differential equation to a ﬁrstorder linear differential equation. Show
that the general solution of the differential equation is given by
G( y)
F( y)
= exp
_
a
_
β(x)dx
_
_
a
_
α(x) exp
_
−a
_
β(x)dx
_
dx + C
_
.
2.71 The Riccati equation is given by y
= α(x) y
2
+ β(x) y + γ (x).
1. If one solutionof this equation, say y
1
(x), is known, thenthe general solution
can be found by using the transformation y = y
1
+
1
u
, where u is a new
dependent variable. Show that u is given by
u = e
−
_
P(x)dx
dx
_
_
Q(x) e
_
P(x)dx
dx
dx + C
_
,
where P(x) =2α(x) y
1
(x)+β(x) and Q(x) = −α(x).
2. For the differential equation y
+ y
2
= 1 + x
2
, ﬁrst guess a solution y
1
(x)
and then use the result of Part 1 to ﬁnd the general solution y(x).
A
NS y = x +
e
−x
2
_
e
−x
2
dx + C
Bernoulli Differential Equations
2.72 3x y
− 3x y
4
ln x − y = 0
A
NS
1
y
3
= −
3
4
x(2 ln x−1) +
C
x
; y = 0
2.73
dy
dx
=
4x
3
y
2
x
4
y + 2
A
NS x
4
= −
1
y
+ Cy; y = 0
2.74 y(6y
2
− x − 1) dx + 2x dy = 0
A
NS
1
y
2
=
1
x
(6 + Ce
−x
); y = 0
2.75 (1 + x) ( y
+ y
2
) − y = 0
A
NS
1
y
=
1
1+x
_
x
2
2
+ x + C
_
; y = 0
2.76 x y y
+ y
2
− sin x = 0
A
NS x
2
y
2
= −2x cos x + 2 sin x + C
2.77 (2x
3
−y
4
)dx + x y
3
dy = 0
A
NS y
4
= 8x
3
+ Cx
4
2.78 y
− y tan x + y
2
cos x = 0
A
NS
1
y
= cos x(x+C); y = 0
2.79 6y
2
dx − x(2x
3
+ y) dy = 0
A
NS ( y−2x
3
)
2
= Cyx
6
; y = 0
Problems 83
Equation Solvable for the Independent or Dependent Variable
2.80 x y
3
− y y
2
+ 1 = 0
A
NS y = Cx +
1
C
2
; 4y
3
= 27x
2
2.81 y = x y
+ y
3
A
NS y = Cx + C
3
; 4x
3
+ 27y
2
= 0
2.82 x( y
2
− 1) = 2y
A
NS x =
2p
p
2
−1
, y =
2
p
2
−1
−ln
¸
¸
p
2
−1
¸
¸
+C
2.83 x y
( y
+ 2) = y
A
NS y = −x; y = ±2C
√
x + C
2
2.84 x = y
_
y
2
+ 1
A
NS x = p
_
p
2
+1, 3y =
_
p
2
+1(2p
2
−1)+C
2.85 2y
2
( y − x y
) = 1
A
NS y = Cx +
1
2C
2
; 8y
3
= 27x
2
2.86 y = 2x y
+ y
2
y
3
A
NS y
2
= 2Cx + C
3
; 32x
3
+ 27y
4
= 0; y = 0
2.87 y
3
+ y
2
= x y y
A
NS p
3
+ y
2
= x y p,
p
4
2y
2
− p = C; y = 0
2.88 2x y
− y = y
ln( y y
)
A
NS 2x = 1+2 ln
¸
¸
y
¸
¸
; y
2
= 2Cx−C lnC
2.89 y = x y
− x
2
y
3
A
NS y = 0; x p
2
= C
_
¸
¸
p
¸
¸
− 1, y = x p − x
2
p
3
2.90 y( y − 2xy
)
3
= y
2
A
NS 27x
2
y
2
=1; y
2
=2C
3
x + C
2
2.91 y + x y
= 4
_
y
A
NS x =
ln p+C
√
p
, y =
_
p (4−ln p−C); y =0
2.92 2x y
− y = ln y
A
NS x =
1
p
+
C
p
2
, y = 2
_
1 +
C
p
_
− ln p
Simple HigherOrder Differential Equations
2.93 y
= 2y y
3
A
NS y = C; 3x + y
3
+ C
1
y = C
2
2.94 y y
= y
2
− y
3
A
NS y = C; C
1
ln
¸
¸
y
¸
¸
+ y = x + C
2
2.95 x y
= (1−x) y
A
NS y = C
1
(x+2)e
−x
+ C
2
x + C
3
2.96 y
= e
x
y
2
A
NS y =
1
C
1
ln
¸
¸
1 + C
1
e
−x
¸
¸
+ C
2
; y =C
2.97 y y
+ y
2
= 0
A
NS y = C;
1
2
y
2
= C
1
x + C
2
2.98 1 + y
2
= 2y y
A
NS 4(C
1
y − 1) = C
2
1
(x + C
2
)
2
2.99 x y
= y
_
ln y
− ln x
_
A
NS y =
1
C
1
e
C
1
x+1
_
x −
1
C
1
_
+ C
2
; y =
1
2
ex
2
+C
84 2 ﬁrstorder and simple higherorder differential equations
2.100 3y y
y
− y
3
+ 1 = 0
A
NS 3(C
1
y + 1)
2/3
− 2C
1
x = C
2
; y =x
2.101 y
− y
2
− 1 = 0
A
NS y = −ln
¸
¸
cos(x+C
1
)
¸
¸
+ C
2
2.102 x
3
y
− x
2
y
= 3 − x
2
A
NS y =
1
x
+ x + C
1
x
2
+ C
2
2.103 2y
= y
3
sin 2x, y(0) = 1, y
(0) = 1
A
NS y =1+ln
¸
¸
sec x+tan x
¸
¸
2.104 x
d
2
y
dx
2
= 2 −
dy
dx
A
NS y = 2x + C
1
ln
¸
¸
x
¸
¸
+ C
2
2.105 y
= 3
√
y, y(0) = 1, y
(0) = 2
A
NS y =
_
±
1
2
x + 1
_
4
2.106 x
d
2
y
dx
2
=
dy
dx
+ x sin
_
1
x
·
dy
dx
_
A
NS y =
_
x
2
+
1
C
2
1
_
tan
−1
C
1
x−
x
C
1
+C
2
; y =
kπ
2
x
2
+C, k =0, ±1, ±2, . . .
2.107 y y
= y
2
(1 − y
sin y − y y
cos y)
A
NS y = C; x = −cos y + C
1
ln
¸
¸
y
¸
¸
+ C
2
2.108 y
+ x y
= x
A
NS y = x + C
1
_
e
−
1
2
x
2
dx + C
2
2.109 x y
− y
3
− y
= 0
A
NS x
2
+ ( y − C
1
)
2
= C
2
; y = C
2.110 y(1 − ln y) y
+ (1 + ln y) y
2
= 0
A
NS y = C; (C
1
x + C
2
)(ln y − 1) + 1 = 0
Review Problems
2.111 x y
2
(x y
+ y) = 1
A
NS 2x
3
y
3
− 3x
2
= C
2.112 5 y + y
2
= x(x + y
)
A
NS 4y = x
2
; 5y = −5x
2
+ 5Cx − C
2
2.113 y
=
y+2
x+1
+ tan
y−2x
x+1
A
NS sin
y−2x
x+1
= C(x+1);
y+2
x+1
= nπ +2, n=0, ±1, ±2, . . .
2.114 y
(e
x
+ 1) + y
= 0
A
NS y = C
1
(x − e
−x
) + C
2
2.115 x y
= y − xe
y/x
A
NS y = −x ln
¸
¸
ln
¸
¸
Cx
¸
¸
¸
¸
2.116 (1 + y
2
sin 2x)dx − 2 y cos
2
xdy = 0
A
NS x − y
2
cos
2
x = C
2.117 (2
√
x y −y)dx − xdy = 0, x >0, y >0
A
NS
√
x y − x = C
2.118 y
+ y
2
= 2e
−y
A
NS e
y
+ C
1
= (x + C
2
)
2
Problems 85
2.119 y
= e
x y
/ y
A
NS y = ex; Cx = ln
¸
¸
Cy
¸
¸
2.120 (2x
3
y
2
− y)dx + (2x
2
y
3
− x)dy = 0
A
NS x
2
+ y
2
+
1
x y
= C
2.121 ( y−1−x y) dx + x dy = 0
A
NS x y + 1 = Ce
x
2.122 x y
− y = x tan
y
x
A
NS sin
y
x
= Cx
2.123 y
+
y
x
= e
xy
A
NS −e
−x y
=
1
2
x
2
+ C
2.124 y y
− y y
= ( y
)
2
A
NS ln
¸
¸
¸ ln
¸
¸
y
¸
¸
+C
2
¸
¸
¸ = x + C
1
; y = C
2.125 2y dx − x
_
ln(x
2
y) − 1
_
dy = 0
A
NS ln(x
2
y) − Cy = 0
2.126 y
=
1
x y + x
3
y
3
A
NS
1
x
2
= 1 − y
2
+ Ce
−y
2
2.127 y
= 2
_
y + 2
x + y − 1
_
2
A
NS y + 2 = C exp
_
−2 tan
−1
y + 2
x − 3
_
2.128 (e
x
+ 3y
2
)dx + 2x ydy = 0
A
NS e
x
(x
2
−2x+2) + x
3
y
2
= C
2.129 (x y + 2x
3
y)dx + x
2
dy = 0
A
NS x ye
x
2
= C
2.130 x( y
)
2
− 2y y
+ 4x = 0
A
NS y =
C
2
x
2
+
2
C
; y = ±2x
2.131 y
= 2( y
−1) cot x
A
NS y = (C
2
+1)
x
2
2
+
C
2
4
cos 2x+C
1
x+C
0
2.132 ( y+3x
4
y
2
)dx + (x+2x
2
y
3
)dy = 0
A
NS −
1
x y
+x
3
+y
2
=C; y =0
2.133 x y
= y+
_
x
2
−y
2
, x >0,
¸
¸
y
¸
¸
¸
¸
x
¸
¸
A
NS sin
−1
y
x
= ln
¸
¸
x
¸
¸
+C; y = ±x
2.134 2y(xe
x
2
+y sin x cos x)dx + (2e
x
2
+3y sin
2
x)dy = 0
A
NS y
2
e
x
2
+ y
3
sin
2
x = C
2.135 cos y dx + sin y(x−sin y cos y) dy = 0
A
NS x = cos y
_
ln
¸
¸
sec y +tan y
¸
¸
− sin y + C
_
2.136 y
3
dx + (3x
2
−2x y
2
)dy = 0
A
NS y
3
= Ce
y
2
/x
2.137 ( y
+ 1) ln
y + x
x + 3
=
y + x
x + 3
A
NS ln
y + x
x + 3
= 1 +
C
x + y
2.138 2x
3
y y
+ 3x
2
y
2
+ 7 = 0
A
NS x
3
y
2
+ 7x = C
2.139
_
x−y cos
y
x
_
dx + x cos
y
x
dy = 0
A
NS sin
y
x
= C−ln
¸
¸
x
¸
¸
86 2 ﬁrstorder and simple higherorder differential equations
2.140 x
2
(xdy −ydx) = (x +y) ydx
A
NS ln
¸
¸
¸
x
y
+ 1
¸
¸
¸ =
1
x
+ C
2.141 ( y
4
+x y)dx + (x y
3
−x
2
)dy = 0
A
NS 2x y+
_
x
y
_
2
=C; y =0
2.142 (x
2
+ 3 ln y) dx −
x
y
dy = 0
A
NS x
2
+ ln y = Cx
3
2.143 x y
= y
+ x
A
NS 4 y = x
2
_
2 ln
¸
¸
x
¸
¸
+C
1
_
+ C
2
2.144 ydx + (x y−x−y
3
)dy = 0
A
NS x = y( y−1) + Cye
−y
; y = 0
2.145 y + 2 y
3
y
= (x +4 y ln y) y
A
NS
x
y
+ y
2
− 2(ln y)
2
= C
2.146 y ln x ln y dx + dy = 0
A
NS x ln x − x + ln
¸
¸
ln y
¸
¸
= C
2.147 (2x
√
x + x
2
+ y
2
) dx + 2y
√
x dy = 0
A
NS ln(x
2
+y
2
) + 2
√
x = C
2.148
_
2x + y cos(x y)
_
dx + x cos(x y) dy = 0
A
NS x
2
+ sin(x y) = C
2.149 y y
− y
2
y
− y
2
= 0
A
NS y = C;
1
C
1
ln
¸
¸
¸
¸
y
y+C
1
¸
¸
¸
¸
= x+C
2
2.150 2y
+ x = 4
√
y
A
NS (2
√
y−x) ln
¸
¸
C(2
√
y−x)
¸
¸
= x; 2
√
y = x
2.151 2 y
3
− 3 y
2
+ x = y
A
NS y = x−1; 4(x+C)
3
= 27( y+C)
2
2.152 y
− 6 x e
x−y
− 1 = 0
A
NS e
y−x
= 3x
2
+ C
2.153 (1 + y
2
) y
+ y
3
+ y
= 0
A
NS −C
1
y − (1 + C
2
1
) ln
¸
¸
y − C
1
¸
¸
= x + C
2
; y = C
2.154 ( y sin x + cos
2
x)dx − cos xdy = 0
A
NS −y cos x +
x
2
+
sin 2x
4
= C
2.155 y(6y
2
− x − 1) dx + 2x dy = 0
A
NS
x
y
2
= 6 + Ce
−x
; y = 0
2.156 y
(x − ln y
) = 1
A
NS x = ln p + p
−1
, y = p − ln p + C
2.157 (1 + cos x) y
+ sin x (sin x + sin x cos x − y) = 0
A
NS
1
2
x −
1
4
sin 2x +
1
3
sin
3
x + y cos x + y = C
2.158 x dx + sin
2
_
y
x
_
( y dx − x dy) = 0
A
NS
y
2x
−
1
4
sin
2y
x
= ln
¸
¸
x
¸
¸
+ C
2.159 (2x y
4
e
y
+ 2x y
3
+ y) dx + (x
2
y
4
e
y
− x
2
y
2
− 3x)dy = 0
A
NS x
2
e
y
+
x
2
y
+
x
y
3
= C
2.160 (x y
3
−1)dx + x
2
y
2
dy = 0
A
NS 2x
3
y
3
−3x
2
= C
3
C H A P T E R
Applications of FirstOrder and
Simple HigherOrder Equations
In this chapter, a number of examples are studied to illustrate the application of
ﬁrstorder and simple higherorder differential equations in various science and
engineering disciplines.
3.1 Heating and Cooling
Problems involving heating and cooling follow Newton’s Law of Cooling.
Newton’s Law of Cooling
The rate of change in the temperature T(t), dT/dt, of a body in a medium of
temperature T
m
is proportional to the temperature difference between the body
and the medium, i.e.,
dT
dt
= −k(T−T
m
),
where k >0 is a constant of proportionality.
Example 3.1 — Body Cooling in Air 3.1
A body cools in air of constant temperature T
m
=20
◦
C. If the temperature of the
body changes from 100
◦
C to 60
◦
C in 20 minutes, determine howmuch more time
it will need for the temperature to fall to 30
◦
C.
Newton’s Law of Cooling requires that
dT
dt
= −k(T−T
m
). Variable separable
87
88 3 applications of ﬁrstorder and simple higherorder equations
The general solution is
_
dT
T−T
m
= −k
_
dt + C =⇒ ln
¸
¸
T−T
m
¸
¸
= −kt + ln C
∴ T = T
m
+ Ce
−kt
.
At t =0, T =100
◦
C:
100 = 20 + Ce
−k·0
= 20 + C =⇒ C = 80.
At t =20 min, T =60
◦
C:
60 = 20 + 80e
−k·20
=⇒ k = −
1
20
ln
60−20
80
= 0.03466.
Hence
T = 20 + 80e
−0.03466t
or t = −
1
0.03466
ln
T−20
80
min.
When T =30
◦
C:
t = −
1
0.03466
ln
30−20
80
= 60 min.
Hence, it will needanother 60−20=40 minutes for the temperature to fall to 30
◦
C.
Example 3.2 — Heating in a Building 3.2
The rate of heat loss from a building is equal to K
1
_
T
B
(t)−T
A
(t)
_
, where T
B
(t)
and T
A
(t) are the temperatures of the building and the atmosphere at time t,
respectively, and K
1
is a constant. The rate of heat supplied to the building by the
heating system is given by Q+K
2
_
T
S
−T
B
(t)
_
, where T
S
is the “set” temperature
of the building, and Q and K
2
are constants. The value of Q is such that the
building is maintained at the “set” temperature when the atmosphere is at constant
temperature T
0
. The thermal capacity of the building is c.
1. Set up the differential equation governing the temperature of the building
T
B
(t).
2. If the atmospheric temperature ﬂuctuates sinusoidally about the mean value
T
0
with an amplitude of T
1
(
◦
C) and a period of 2π/ω (hour), i.e.,
T
A
= T
0
+ T
1
sin ωt,
determine the amplitude of temperature variation of the building due to
atmospheric temperature ﬂuctuation.
3. Suppose T
1
=12
◦
C, c/K
1
=4 hour, the atmospheric temperature ﬂuctuates
with a period of 24 hours, i.e., ω=π/12. The temperature of the building
is required to remain within 3
◦
C of the set value, i.e., the amplitude of
3.1 heating and cooling 89
temperature ﬂuctuation is less than or equal to 3
◦
C. Show that the value of
the ratio K
2
/K
1
must satisfy
K
2
K
1
4
_
1 −
π
2
144
− 1.
1. Consider a time period from t to t +t, the Principle of Conservation of
Energy requires
(Heat supply in time t) − (Heat loss in time t) = c T
B
,
where
Heat supply in time t =
_
Q + K
2
_
T
S
− T
B
(t)
__
t,
Heat loss in time t = K
1
_
T
B
(t) − T
A
(t)
_
t.
Hence,
_
Q + K
2
_
T
S
− T
B
(t)
__
t − K
1
_
T
B
(t) − T
A
(t)
_
t = c T
B
.
Dividing the equation by t and taking the limit as t →0 lead to
Q + K
2
_
T
S
− T
B
(t)
_
− K
1
_
T
B
(t) − T
A
(t)
_
= c
dT
B
dt
,
or
c
K
1
dT
B
(t)
dt
+
_
1 +
K
2
K
1
_
T
B
(t) =
Q
K
1
+
K
2
K
1
T
S
+ T
A
(t).
2. Since T
A
=T
0
+T
1
sin ωt, the differential equation becomes
c
K
1
dT
B
(t)
dt
+
_
1 +
K
2
K
1
_
T
B
(t) =
_
Q
K
1
+
K
2
K
1
T
S
+ T
0
_
+ T
1
sin ωt,
or
dT
B
(t)
dt
+
K
1
c
_
1 +
K
2
K
1
_
T
B
(t) =
K
1
c
_
Q
K
1
+
K
2
K
1
T
S
+ T
0
_
+
K
1
T
1
c
sin ωt.
. ,, . . ,, . . ,, .
k α
0
α
1
The differential equation is linear ﬁrstorder of the form
dT
B
(t)
dt
+ P(t) · T
B
(t) = Q(t), P(t) = k, Q(t) = α
0
+ α
1
sin ωt,
where
k =
K
1
c
_
1 +
K
2
K
1
_
, α
0
=
K
1
c
_
Q
K
1
+
K
2
K
1
T
S
+ T
0
_
, α
1
=
K
1
T
1
c
.
90 3 applications of ﬁrstorder and simple higherorder equations
It is easy to evaluate
_
P(t)dt = kt, e
_
P(t)dt
= e
kt
, e
−
_
P(t)dt
= e
−kt
,
_
Q(t)e
_
P(t)dt
dt =
_
(α
0
+ α
1
sin ωt)e
kt
dt
=
α
0
k
e
kt
+
α
1
k
2
+ω
2
e
kt
(k sin ωt − ωcos ωt)
= e
kt
_
α
0
k
+
α
1
√
k
2
+ω
2
sin(ωt −ϕ)
_
, ϕ = tan
−1
ω
k
.
See the Remarks on page 195 on finding the amplitude of Acos ωt +Bsin ωt.
Hence, the general solution is
T
B
(t) = e
−
_
P(t)dt
_
_
Q(t)e
_
P(t)dt
dt + C
_
= e
−kt
_
e
kt
_
α
0
k
+
α
1
√
k
2
+ω
2
sin(ωt −ϕ)
_
+ C
_
=
α
0
k
+
α
1
√
k
2
+ω
2
sin(ωt −ϕ) + Ce
−kt
.
. ,, .
Variation due to atmospheric temperature fluctuation
The amplitude of temperature variation due to atmospheric temperature ﬂuctua
tion is
a =
α
1
√
k
2
+ω
2
=
K
1
T
1
c
_
_
K
1
c
_
1+
K
2
K
1
_
_
2
+ ω
2
=
T
1
_
_
1+
K
2
K
1
_
2
+
_
cω
K
1
_
2
.
3. That the amplitude of temperature ﬂuctuation is less than or equal to 3
◦
C means
a3, i.e.,
T
1
_
_
1+
K
2
K
1
_
2
+
_
cω
K
1
_
2
3,
which gives
_
T
1
3
_
2
_
1 +
K
2
K
1
_
2
+
_
cω
K
1
_
2
=⇒
K
2
K
1
_
_
T
1
3
_
2
−
_
cω
K
1
_
2
− 1,
∴
K
2
K
1
_
_
12
3
_
2
−
_
4·
π
12
_
2
− 1 = 4
_
1 −
π
2
144
− 1.
3.2 motion of a particle in a resisting medium 91
3.2 Motion of a Particle in a Resisting Medium
Newton’s Second Law and D’Alembert’s Principle
Newton’s Second Law: The product of the mass of an object and its accelera
tion is equal to the sumof forces applied on the object, i.e., ma=
F.
D’Alembert’s Principle: Rewrite Newton’s Second Law as
F−ma=0. Treat
−ma as a force, known as the inertia force. An object is in (dynamic) equi
librium under the action of all the forces applied, including the inertia force.
This is known as D’Alembert’s Principle, which transforms a problemin dynamics
into a problem of static equilibrium.
ImpulseMomentum Principle
For a system of particles, the change in momentum of the system is equal to the
total impulse on the system, i.e.,
(Momentum at time t
2
) − (Momentum at time t
1
) = (Impulse during t
2
−t
1
).
The momentum of a mass m moving at velocity v is equal to mv. The impulse of
a force F during time interval t is equal to Ft.
Consider the motion of a particle moving in a resisting medium, such as air or
water. The medium exerts a resisting force R on the particle. In many applications,
the resisting force R is proportional to v
n
, where v is the velocity of the particle
and n>0, and is opposite to the direction of the velocity. Hence, the resisting force
can be expressed as R=βv
n
, where β is a constant. For particles moving in an
unbounded viscous medium at low speed, the resisting force is R=βv, i.e., n=1.
In the following, the case with R=βv will be studied for motion in the vertical
direction and speciﬁc initial conditions.
Case I: Upward Motion
Consider an object being launched vertically at time t =0 from x =0 with initial
velocity v
0
as shown in Figure 3.1.
The displacement x, the velocity v = ˙ x, and the acceleration a= ˙ v = ¨ x are taken
as positive in the upward direction. The particle is subjected to two forces: the
downward gravity mg and the resisting force from the medium R=βv, which is
opposite to the direction of the velocity and hence is downward.
From Newton’s Second Law, the equation of motion is
↑ ma =
F : m
dv
dt
= −R − mg, R=βv, m=
w
g
,
92 3 applications of ﬁrstorder and simple higherorder equations
m
t =0, v
0
t,
mg
R
x
v=x, a=v=x
Figure 3.1 Upward motion of a particle in a resisting medium.
∴
dv
dt
= −g (αv + 1), α =
β
w
>0, Variable separable
_
dv
αv + 1
= −
_
g dt + C =⇒
1
α
ln(αv+1) = −gt + C. (1)
Constant C is determined from the initial condition t =0, v =v
0
:
1
α
ln(αv
0
+1) = 0 + C.
Substituting into equation (1) yields
1
α
ln(αv+1) = −gt +
1
α
ln(αv
0
+1),
1
α
ln
_
αv + 1
αv
0
+ 1
_
= −gt =⇒
αv + 1
αv
0
+ 1
= e
−αgt
.
Solving for v leads to
αv = e
−αgt
(αv
0
+ 1) − 1. (2)
When the object reaches the maximum height at time t =t
max
, v =0, and
α· 0 = e
−αgt
max
(αv
0
+ 1) − 1 =⇒ t
max
=
1
αg
ln(αv
0
+1).
To determine the displacement x(t), note that v =
dx
dt
and use equation (2)
dx
dt
=
1
α
(αv
0
+ 1) e
−αgt
−
1
α
. Immediately integrable
Integrating with respect to x gives
x =
1
α
(αv
0
+ 1)
_
e
−αgt
dt −
t
α
+ D = −
1
α
2
g
(αv
0
+ 1) e
−αgt
−
t
α
+ D.
Constant D is determined from the initial condition t =0, x =0:
0 = −
1
α
2
g
(αv
0
+ 1) · e
0
− 0 + D =⇒ D =
1
α
2
g
(αv
0
+ 1).
3.2 motion of a particle in a resisting medium 93
Hence
x = −
1
α
2
g
(αv
0
+ 1)(e
−αgt
− 1) −
t
α
.
At time t =t
max
, the object reaches the maximum height given by
x = x
max
= x(t)
¸
¸
t=t
max
= −
1
α
2
g
(αv
0
+1)
_
e
−ln(αv
0
+1)
−1
_
−
1
α
·
1
αg
ln(αv
0
+1)
=
1
α
2
g
_
αv
0
− ln(αv
0
+1)
_
.
Case II: Downward Motion
t =0, v=0
mg
R
x
t, v=x, a=v=x
Figure 3.2 Downward motion of a particle in a resisting medium.
Consider an object being released and dropped at time t =0 from x =0 with v =0
as shown in Figure 3.2. In this case, it is more convenient to take x, v, and a as
positive in the downward direction. Newton’s Second Law requires
↓ ma =
F : m
dv
dt
= mg − R, R=βv, m=
w
g
,
∴
dv
dt
= g − αgv, α =
β
w
>0. Variable separable
The equation can be solved easily as
_
dv
1 − αv
=
_
g dt + C =⇒ −
1
α
ln
¸
¸
1 − αv
¸
¸
= gt + C,
where the constant C is determined from the initial condition t =0, v =0:
−
1
α
ln 1 = 0 + C =⇒ C = 0,
∴ −
1
α
ln
¸
¸
1 − αv
¸
¸
= gt =⇒ v =
1
α
(1 − e
−αgt
). (3)
When time t approaches inﬁnity, the velocity approaches a constant, the socalled
terminal velocity,
v = v
terminal
= lim
t→∞
v =
1
α
.
The change of velocity with time is shown in Figure 3.3.
94 3 applications of ﬁrstorder and simple higherorder equations
v
w
β
t
1
α
=
Figure 3.3 Velocity of a particle moving downward in a resisting medium.
To ﬁnd the displacement, rewrite equation (3) as
dx
dt
=
1
α
(1 − e
−αgt
). Immediately integrable
Integrating yields
x =
t
α
+
1
α
2
g
e
−αgt
+ D,
where the constant D is determined by the initial condition t =0, x =0:
0 = 0 +
1
α
2
g
· e
0
+ D =⇒ D = −
1
α
2
g
.
Hence the displacement is given by
x =
t
α
+
1
α
2
g
(e
−αgt
−1).
Example 3.3 — Bullet through a Plate 3.3
A bullet is ﬁred perpendicularly into a plate at an initial speed of v
0
=100 m/sec.
When the bullet exits the plate, its speed is v
1
=80 m/sec. It is known that the
thickness of the plate is b=0.1 mand the resistant force of the plate on the bullet is
proportional to the square of the speed of the bullet, i.e., R=βv
2
. Determine the
time T that the bullet takes to pass through the plate.
t =0
v
0
t =T
v
1
βv
2
t , v
x
b
3.2 motion of a particle in a resisting medium 95
Applying Newton’s Second Law to the bullet as shown yields
→ ma =
F : m
dv
dt
= −βv
2
. Variable separable
The general solution is given by
_
−
dv
v
2
=
_
β
m
dt + C =⇒
1
v
= kt + C, k =
β
m
,
where the constant C is determined from the initial condition t =0, v =v
0
:
1
v
0
= k· 0 + C =⇒ C =
1
v
0
.
Hence
1
v
= kt +
1
v
0
=⇒ v =
dx
dt
=
1
kt +
1
v
0
. Immediately integrable (1)
Integrating with respect to t leads to
x =
1
k
ln
_
kt +
1
v
0
_
+ D,
where the constant D is determined from the initial condition t =0, x =0:
0 =
1
k
ln
_
k· 0 +
1
v
0
_
+ D =⇒ D = −
1
k
ln
1
v
0
.
Hence
x =
1
k
ln
_
kt +
1
v
0
_
−
1
k
ln
1
v
0
. (2)
From equation (1),
t = T, v = v
1
:
1
v
1
= kT +
1
v
0
. (3)
From equation (2), t =T, x =b:
b =
1
k
ln
_
kT+
1
v
0
_
−
1
k
ln
1
v
0
=
1
k
ln
1
v
1
−
1
k
ln
1
v
0
=
1
k
ln
v
0
v
1
=⇒ k =
1
b
ln
v
0
v
1
.
Using equation (3),
T =
1
k
_
1
v
1
−
1
v
0
_
= b
1
v
1
−
1
v
0
ln
v
0
v
1
= 0.1 ×
1
80
−
1
100
ln
100
80
= 0.000819 sec.
96 3 applications of ﬁrstorder and simple higherorder equations
Example 3.4 — Object Falling in Air 3.4
An object of mass m falls against air resistance which is proportional to the speed
(i.e., R=βv) and under gravity g.
1. If v
0
and v
E
are the initial and ﬁnal (terminal) speeds, and v is the speed at
time t, show that
v − v
E
v
0
− v
E
= e
−kt
, k =
β
m
.
2. The speed of the object is found to be 30, 40, 45 m/sec at times t =1, 2, and
3 sec, respectively, after starting. Find v
E
and v
0
.
3. At what time will the speed of the object be 49 m/sec?
mg
R=kv
x, v, a
1. The object is subjected to two forces as shown: the downward gravity mg and
the upward air resistance βv. Newton’s Second Law requires
↓ ma =
F : m
dv
dt
= mg − βv =⇒
dv
dt
= g − kv, k =
β
m
.
Noting that g −kv >0, the equation is variable separable and the solution is
_
dv
g −kv
=
_
dt + C =⇒ −
1
k
ln(g −kv) = t + C =⇒ v =
g
k
− Ce
−kt
.
Constant C is determined from the initial condition t =0, v =v
0
:
v
0
=
g
k
− Ce
0
=⇒ C =
g
k
− v
0
.
When t →∞, v =v
E
=⇒ v
E
=
g
k
=⇒ C=v
E
− v
0
. Hence the velocity is given by
v = v
E
− (v
E
−v
0
)e
−kt
=⇒
v−v
E
v
0
−v
E
= e
−kt
. (1)
2. From equation (1),
t = 1, v = 30:
30−v
E
v
0
−v
E
= e
−k
, (2)
3.3 hanging cables 97
t = 2, v = 40:
40−v
E
v
0
−v
E
= e
−2k
, (3)
t = 3, v = 45:
45−v
E
v
0
−v
E
= e
−3k
. (4)
Since e
−k
· e
−3k
= e
−4k
= (e
−2k
)
2
,
Eq(2) × Eq(4) = Eq(3)
2
:
_
30−v
E
v
0
−v
E
__
45−v
E
v
0
−v
E
_
=
_
40−v
E
v
0
−v
E
_
2
,
(30−v
E
)(45−v
E
) = (40−v
E
)
2
=⇒ 1350 − 75v
E
+ v
2
E
= 1600 − 80v
E
+ v
2
E
,
∴ 5v
E
= 250 =⇒ v
E
= 50 m/sec.
Eq(2)
2
Eq(3)
:
_
30−v
E
v
0
−v
E
_
2
40−v
E
v
0
−v
E
=
e
−2k
e
−2k
= 1 =⇒
(30−v
E
)
2
(v
0
−v
E
)(40−v
E
)
= 1,
∴ v
0
=
(30−v
E
)
2
40−v
E
+ v
E
=
(30−50)
2
40−50
+ 50 = 10 m/sec.
3. From equation (2),
k = −ln
¸
¸
¸
¸
30−v
E
v
0
−v
E
¸
¸
¸
¸
= −ln
¸
¸
¸
¸
30−50
10−50
¸
¸
¸
¸
= −ln
1
2
= ln 2,
e
−kt
=
v−v
E
v
0
−v
E
=
49−50
10−50
=
1
40
=⇒ t = −
1
k
ln
1
40
=
ln 40
ln 2
= 5.32 sec.
3.3 Hanging Cables
3.3.1 The Suspension Bridge
A typical suspension bridge consists of cables, piers (towers), anchors, hangers
(suspenders), and deck (stiffening girder) as shown in Figure 3.4. Normally the
selfweights of the cables are negligible compared with the load they carry. The
load on the cables is from the load on the deck, which includes the selfweight of
the deck and trafﬁc load, and is transmitted by the hangers.
Consider a cable supported at two supports A and B as shown in Figure 3.5(a).
The load on the cable is modeled as a distributed load w(x). Set up the Cartesian
coordinate system by placing the origin at the lowest point of the cable.
98 3 applications of ﬁrstorder and simple higherorder equations
Pier Hanger Cable
Deck Anchor
Figure 3.4 A suspension bridge.
w(x)
O
y
A
B
A
B
(a)
(b)
x
w(x)
W(x)
O
H
T(x)
y
x
x
θ
h
B
h
A
L
A
L
B
Figure 3.5 A cable under distributed load.
To establish the governing differential equation, consider the equilibrium of a
segment of cable between 0 and x as shown in Figure 3.5(b). The cable is subjected
to three forces:
❧ the horizontal tension force H at the left end,
❧ the tension force T(x) tangent to the cable at the right end, and
❧ the portion of the distributed load w(x) between 0 and x. It can be replaced
by its resultant W(x) applied at the centroid of the area enclosed by the load
intensity curve w(x) (the shaded area).
3.3 hanging cables 99
The equilibrium of the segment requires that
→
F
x
= 0: T(x) cos θ − H = 0, (1)
↑
F
y
= 0: T(x) sin θ − W(x) = 0, W(x) =
_
x
0
w(x)dx. (2)
Eliminating T(x) from these two equations yields
T(x) sin θ
T(x) cos θ
=
W(x)
H
=⇒ tan θ =
W(x)
H
.
From geometry, one has
tan θ =
dy
dx
=⇒
dy
dx
=
W(x)
H
.
Differentiating with respect to x leads to
d
2
y
dx
2
=
1
H
dW(x)
dx
=
w(x)
H
,
dW(x)
dx
= w(x).
Suppose that the load is uniformly distributed, i.e., w(x) =w. The differential
equation becomes
d
2
y
dx
2
=
w
H
.
Secondorder DE
Immediately integrable
(3)
Since the origin is taken at the lowest point, one has x =0, y =0,
dy
dx
=0.
Integrating equation (3) once yields
dy
dx
=
w
H
x + C,
where the constant C is determined from the initial condition x =0,
dy
dx
=0:
0 = 0 + C =⇒ C = 0.
Integrating again leads to
y =
w
2H
x
2
+ D,
where the constant D is determined from the initial condition x =0, y =0:
0 = 0 + D =⇒ D = 0.
Hence the shape of the cable is a parabola given by
y =
w
2H
x
2
. (4)
100 3 applications of ﬁrstorder and simple higherorder equations
The sags h
A
and h
B
can be determined from equation (4)
when x = −L
A
, y = h
A
: h
A
=
w
2H
L
2
A
=⇒ H =
wL
2
A
2h
A
, (5a)
when x = L
B
, y = h
B
: h
B
=
w
2H
L
2
B
=⇒ H =
wL
2
B
2h
B
. (5b)
From equations (5), one obtains the relationship among L
A
, L
B
, h
A
, and h
B
:
H =
wL
2
A
2h
A
=
wL
2
B
2h
B
=⇒
L
2
A
h
A
=
L
2
B
h
B
.
To determine the tension at any point, use equations (1) and (2)
T(x) cos θ = H, T(x) sin θ = W(x).
Squaring both sides of these two equations and adding them lead to
T
2
cos
2
θ + T
2
sin
2
θ = H
2
+ W
2
(x) =⇒ T
2
= H
2
+ W
2
(x).
Since the load is uniformly distributed, w(x) =w,
W(x) =
_
x
0
w(x)dx = w
_
x
0
dx = wx.
Therefore, the tension at any point is given by
T =
_
H
2
+ W
2
(x) =
_
H
2
+ w
2
x
2
. (6)
The tension T is maximumwhen
¸
¸
x
¸
¸
is maximum. Hence the tension is maximum
at the higher support.
O
y
x
h=sag
2
L
2
L
Figure 3.6 A suspension bridge cable with supports at equal height.
For a suspension bridge cable with supports at equal height h
A
=h
B
=h, one has
L
A
=L
B
=L/2, where L is the span length, as shown in Figure 3.6. Then equations
(5) give the relationship between the sag h and the horizontal tension H at the
lowest point:
h =
wL
2
8H
or H =
wL
2
8h
. (5
)
3.3 hanging cables 101
Example 3.5 — Cable of a Suspension Bridge 3.5
Consider the main cable of a suspension bridge carrying a uniformly distributed
load of intensity w. The two supports of the cable are at the same height. The span
of the cable is L, the sag is h, and the axial rigidity is EA.
1. Derive a formula for the elongation δ of the cable.
2. One of the main cables of the central span of the Golden Gate Bridge has the
following properties: L=1, 280 m, h=143 m, w=200 kN/m, E =200 GPa.
The cable consists of 27,572 parallel wires of diameter 5 mm. Determine the
elongation of this cable.
1. Consider a small segment of cable of length ds as shown. It is subjected to the
axial tension forces: T at the left end and T+dT at the right end.
O
T+dT
T
y
x
x
dx
ds
Using the formula of elongationof anaxially loaded member inMechanics of Solids
δ =
TL
EA
,
where T is the axial force, L is the length of the member, E is the Young’s modulus,
and A is the crosssectional area, the elongation of the cable segment ds is
dδ =
Tds
EA
.
Since the length of the cable segment is
ds =
_
1 +
_
dy
dx
_
2
=
_
1 +
_
w
H
x
_
2
dx, Using equation (4), y =
w
2H
x
2
and the tension is given by equation (6)
T =
_
H
2
+ w
2
x
2
,
one has
dδ =
_
H
2
+ w
2
x
2
EA
_
1 +
_
w
H
x
_
2
dx =
H
EA
_
1 +
_
w
H
x
_
2
_
dx.
102 3 applications of ﬁrstorder and simple higherorder equations
Integrating over the length of the span yields the elongation of the cable
δ = 2
_
L/2
0
H
EA
_
1 +
_
w
H
x
_
2
_
dx
=
2H
EA
_
x +
w
2
3H
2
x
3
_
L/2
0
=
2H
EA
_
L
2
+
w
2
3H
2
·
L
3
8
_
.
Using equation (5
) to express the horizontal tension in terms of the sag h, one
obtains
δ =
2
EA
_
wL
2
8h
_
⎡
⎣
L
2
+
w
2
3
_
wL
2
8h
_
2
·
L
3
8
⎤
⎦
=
wL
3
8hEA
_
1 +
16h
2
3L
2
_
.
2. The crosssectional area of the cable is
A = 27, 572×
_
π
4
×0.005
2
_
= 0.5414 m
2
.
The elongation of the cable is
δ =
wL
3
8hEA
_
1 +
16h
2
3L
2
_
=
200, 000×1, 280
3
8×143×200· 10
9
×0.5414
_
1 +
16×143
2
3×1, 280
2
_
= 3.61 m.
3.3.2 Cable under SelfWeight
There are many applications, such as cloth lines and power transmission cables as
shown in Figure 3.7, in which cables are suspended between two supports under
own weights.
Figure 3.7 Power transmission cables.
Consider a cable suspended between two supports as shown in Figure 3.8(a).
The cable is hung under its own weight. Set up the Cartesian coordinate system by
placing the origin under the lowest point of the cable. The length of the cable s is
measured from the lowest point.
3.3 hanging cables 103
O
y
A
B
A
B
(a)
Lowest point
(b)
x
W(s)
O
H
T(x)
y
x
x
s
s
θ
h
B
h
A
L
A
L
B
Figure 3.8 A cable under selfweight.
To establish the governing differential equation, consider the equilibrium of a
segment of the cable of length s as shown in Figure 3.8(b). The cable segment is
subjected to three forces: the selfweight W(s), the horizontal tension force H at
the left end, and the tension force T tangent to the cable at the right end. The
equilibrium of this cable segment requires
→
F
x
= 0: T(x) cos θ − H = 0, (1)
↑
F
y
= 0: T(x) sin θ − W(s) = 0. (2)
Dividing equation (2) by equation (1) yields
T(x) sin θ
T(x) cos θ
=
W(s)
H
=⇒ tan θ =
W(s)
H
.
Since tan θ =
dy
dx
, one obtains
dy
dx
=
W(s)
H
,
or, after differentiating the equation with respect to x,
d
2
y
dx
2
=
1
H
dW(s)
dx
. (3)
104 3 applications of ﬁrstorder and simple higherorder equations
Since the length of a cable segment is
ds =
_
1 +
_
dy
dx
_
2
dx,
then, if the cable is uniform,
dW(s)
ds
= w = Weight density of the cable per unit length,
and, using the chain rule,
dW(s)
dx
=
dW(s)
ds
ds
dx
= w
_
1 +
_
dy
dx
_
2
,
which leads to
d
2
y
dx
2
=
w
H
_
1 +
_
dy
dx
_
2
. (4)
This is a secondorder differential equation with both x and y absent. It is easier to
solve the equation as the type of y absent. Letting u=
dy
dx
,
du
dx
=
d
2
y
dx
2
, equation
(4) becomes
du
dx
=
w
H
_
1 + u
2
, Variable separable
_
du
_
1+u
2
=
w
H
_
dx+C =⇒ sinh
−1
u=
w
H
x+C,
_
dx
√
a
2
+x
2
= sinh
−1
x
a
∴ u = sinh
_
w
H
x+C
_
.
Some properties of the hyperbolic functions are summarized on page 145.
Constant C is determined from the initial condition x =0,
dy
dx
=0:
0 = sinh
_
w
H
· 0+C
_
=⇒ 0 = sinh C =⇒ C = 0.
Hence
u =
dy
dx
= sinh
_
w
H
x
_
.
Firstorder DE
Immediately integrable
Integrating leads to
y =
H
w
cosh
_
w
H
x
_
+ D.
_
sinh axdx =
1
a
cosh ax
When x =0,
y =
H
w
cosh 0 + D =
H
w
+ D.
3.3 hanging cables 105
To simplify the expression of y, choose the origin such that y =
H
w
when x =0,
which results in D=0.
Hence, the shape of the cable as shown in Figure 3.9, which is called catenary, is
y =
H
w
cosh
_
w
H
x
_
. (5)
O
y
A
H
w
B
Lowest point
x
s
h
B
h
A
L
A
L
B
Figure 3.9 Shape of a cable under selfweight.
The sag of the cable can be easily determined in terms of h
A
and h
B
at x = −L
A
, y =
H
w
+h
A
:
H
w
+ h
A
=
H
w
cosh
_
w(−L
A
)
H
_
,
at x =L
B
, y =
H
w
+h
B
:
H
w
+ h
B
=
H
w
cosh
_
wL
B
H
_
,
∴ h
A
=
H
w
_
cosh
_
wL
A
H
_
−1
_
, h
B
=
H
w
_
cosh
_
wL
B
H
_
−1
_
. (6)
The length of the cable is given by
p = perimeter =
_
L
B
−L
A
ds
dx
dx =
_
L
B
−L
A
_
1 +
_
dy
dx
_
2
dx
=
_
L
B
−L
A
_
1 + sinh
2
_
w
H
x
_
dx =
_
L
B
−L
A
cosh
_
w
H
x
_
dx 1+sinh
2
x = cosh
2
x
=
H
w
sinh
_
w
H
x
_
¸
¸
¸
¸
L
B
−L
A
_
cosh axdx =
1
a
sinh ax
=
H
w
_
sinh
_
wL
B
H
_
+ sinh
_
wL
A
H
_
_
. (7)
To determine the tension at any point, use equations (1) and (2)
Eq(1)
2
+Eq(2)
2
: T
2
cos
2
θ + T
2
sin
2
θ = H
2
+W
2
(s) =⇒ T =
_
H
2
+W
2
(s),
106 3 applications of ﬁrstorder and simple higherorder equations
where s is the length of the cable between the point of interest and the lowest point,
W(s) =ws is the weight of this segment of cable. Hence
T =
_
H
2
+ w
2
s
2
. (8)
The tension T is maximum when s is maximum. Hence the tension is maximum
at the higher support.
For the special case when the two supports A and B are at the same height, one
has L
A
=L
B
=
1
2
L, where L is the span length, h
A
=h
B
=h.
h =
H
w
_
cosh
_
wL
2H
_
− 1
_
. (6
)
The length of the cable is, from equation (7),
p =
2H
w
sinh
_
wL
2H
_
. (7
)
The tensions at the two supports are the same and are maximum given by
T
max
=
_
H
2
+
1
4
w
2
p
2
, s
max
=
1
2
p. (8
)
Example 3.6 — Hanging Cable 3.6
A cable of weight density of 50 N/m is suspended at two supports of equal height.
The supports are 10 m apart and the sag is 2 m. Determine the following:
(1) the horizontal tension at the lowest point;
(2) the tension at the support;
(3) the length of the cable.
The following parameters are known: w=50 N/m, L=10 m, h=2 m.
From equation (6
),
h =
H
w
_
cosh
_
wL
2H
_
− 1
_
=⇒ 2 =
H
50
_
cosh
_
250
H
_
− 1
_
.
This is a transcendental equation and a numerical method is required to determine
its root. For example, use fsolve in Maple (see Example 12.19 for details on using
fsolve for rootﬁnding). It is found that H =327.93 N.
Using equation (7
), the length of the cable is given by
p =
2H
w
sinh
_
wL
2H
_
=
2×327.93
50
sinh
_
50×10
2×327.93
_
= 11.00 m.
The tension at the support is, from equation (8
),
T
max
=
_
H
2
+
1
4
w
2
p
2
=
_
327.93
2
+
1
4
×50
2
×11.00
2
= 427.98 N.
3.3 hanging cables 107
Example 3.7 — Float and Cable 3.7
A spherical ﬂoat used to mark the course for a sailboat race is shown in Figure
3.10(a). A water current from the left to right causes a horizontal drag on the ﬂoat.
The length of the cable between points A and B is 60 m, and the effective mass
density of the cable is 2 kg/m when the buoyancy of the cable is accounted for. If
the effect of the current on the cable can be neglected, determine the tensions at
points A and B.
A
B
O
y
A
C
H
w
B
Lowest point
with zero slope
30 m
(a)
(b)
50 m
x
x
A
y
B
y
A
x
B
Figure 3.10 A ﬂoat and a cable.
The weight density of the cable is w=2×9.8=19.6 N/m. In this problem, the
lowest point with zero slope does not appear between points A and B. To apply
the formulation established in this section, add an imaginary segment of cable CA
as shown in Figure 3.10(b), and place the origin below the lowest point with zero
slope a distance of H/w.
Applying equation (5) to points A and B yields, denoting H
0
=H/w,
point A : y
A
=
H
w
cosh
_
w
H
x
A
_
=⇒ y
A
= H
0
cosh
x
A
H
0
,
point B : y
B
=
H
w
cosh
_
w
H
x
B
_
=⇒ y
A
+50 = H
0
cosh
x
A
+30
H
0
.
Subtracting these two equations leads to
50 = H
0
_
cosh
x
A
+30
H
0
− cosh
x
A
H
0
_
. (∗)
Following the same procedure as in deriving equation (7), the length of the cable is
p =
_
x
B
x
A
_
1 +
_
dy
dx
_
2
dx =
H
w
sinh
_
w
H
x
_
¸
¸
¸
¸
x
B
x
A
= H
0
_
sinh
x
B
H
0
− sinh
x
A
H
0
_
.
108 3 applications of ﬁrstorder and simple higherorder equations
Hence
60 = H
0
_
sinh
x
A
+30
H
0
− sinh
x
A
H
0
_
. (∗∗)
Equations (∗) and (∗∗) give two transcendental equations for two unknowns H
0
and x
A
. The equations have to be solved numerically, e.g. using fsolve in Maple,
to yield
x
A
= 7.95 m, H
0
=
H
w
= 19.14 =⇒ H = H
0
w = 19.14×19.6 = 375.14 N.
The length of curve CA is
s
CA
=
_
x
A
0
_
1 +
_
dy
dx
_
2
dx =
H
w
sinh
_
w
H
x
_
¸
¸
¸
¸
x
A
0
= H
0
sinh
x
A
H
0
= 19.14 sinh
7.95
19.14
= 8.18 m,
and the length of curve CB is
s
CB
= s
CA
+ s
AB
= 8.18 + 60 = 68.18 m.
Using equation (8), the tensions at points A and B are
T
A
=
_
H
2
+w
2
s
2
CA
=
_
375.14
2
+ (19.6×8.18)
2
= 408 N,
T
B
=
_
H
2
+w
2
s
2
CB
=
_
375.14
2
+ (19.6×68.18)
2
= 1, 388 N.
3.4 Electric Circuits
There are three basic passive electric elements: resistors, capacitors, and inductors.
Resistance R is the capacity of materials to impede the ﬂow of current, which is
modeled by a resistor.
Basic Laws
Ohm’s Law: v =i R, or i =
v
R
, where v is the voltage, i is the current.
Kirchhoff’s Current Law (KCL): The algebraic sum of all the currents at any
node in a circuit equals zero.
Kirchhoff’s Voltage Law (KVL): The algebraic sum of all the voltages around
any closed path in a circuit equals zero.
A capacitor is an electrical component consisting of two conductors separated by
an insulator or dielectric material. If the voltage varies with time, the electric ﬁeld
3.4 electric circuits 109
varies with time, which produces a displacement current in the space occupied by
the ﬁeld. The circuit parameter capacitance C relates the displacement current to
the voltage
i(t) = C
dv
C
(t)
dt
, or v
C
(t) =
1
C
_
t
−∞
i(t)dt =
1
C
_
t
t
0
i(t)dt + v
C
(t
0
).
v
C
C i
A capacitor behaves as an open circuit in the presence of a constant voltage.
Voltage cannot change abruptly across the terminals of a capacitor.
An inductor is an electrical component that opposes any change in electrical
current. It is composed of a coil of wire wound around a supporting core. If
the current varies with time, the magnetic ﬁeld varies with time, which induces a
voltage in the conductor linked by the ﬁeld. The circuit parameter inductance L
relates the induced voltage to the current
v
L
(t) = L
di(t)
dt
, or i(t) =
1
L
_
t
−∞
v
L
(t)dt =
1
L
_
t
t
0
v
L
(t)dt + i(t
0
).
v
L
L i
An inductor behaves as a short circuit in the presence of a constant current.
Current cannot change abruptly in an inductor.
Four types of simple circuits (see Figures 3.11 and 3.12), a circuit comprising a
resistor and a capacitor (RC circuit) and a circuit comprising a resistor and inductor
(RL circuit), either in series or parallel connection, all lead to the ﬁrstorder linear
ordinary differential equation of the form
dx
dt
+
1
τ
x = Q(t).
The solution is given by, with P(t) =1/τ,
x(t) = e
−
_
P(t)dt
_
_
Q(t)e
_
P(t)dt
dx + B
_
= Be
−t/τ
+ e
−t/τ
_
Q(s)e
s/τ
ds,
where the constant B canbe determined using the initial condition: x(t) =x
0
when
t =0 =⇒ B=x
0
. Thus, the solution is
x(t) = x
0
e
−t/τ
+ e
−t/τ
_
Q(s)e
s/τ
ds.
110 3 applications of ﬁrstorder and simple higherorder equations
V(t)
R
R C C
i
v
I(t)
(a) Series RC Circuit (b) Parallel RC Circuit
1
Figure 3.11 RC circuits.
V(t)
R
R L L
i
I(t)
v
(a) Series RL Circuit (b) Parallel RL Circuit
1
Figure 3.12 RL circuits.
If Q(t) =Q
0
, the solution becomes
x(t) = Q
0
τ + (x
0
−Q
0
τ)e
−t/τ
.
Series RC Circuit
Referring to Figure 3.11(a), applying Kirchhoff ’s Voltage Law yields
−V(t) + Ri +
1
C
_
t
−∞
i(t)dt = 0.
Differentiating with respect to t gives
R
di
dt
+
1
C
i =
dV(t)
dt
=⇒
di
dt
+
1
RC
i =
1
R
dV(t)
dt
,
in which x(t) =i(t), τ =RC, Q(t) =
1
R
dV(t)
dt
.
Parallel RC Circuit
Referring to Figure 3.11(b), applying Kirchhoff ’s Current Law at node 1 yields
I(t) −
v
R
− C
dv
dt
= 0 =⇒
dv
dt
+
1
RC
v =
I(t)
C
,
in which x(t) =v(t), τ =RC, Q(t) =
I(t)
C
.
Series RL Circuit
Referring to Figure 3.12(a), applying Kirchhoff ’s Voltage Law yields
−V(t) + Ri + L
di
dt
= 0 =⇒
di
dt
+
R
L
i =
V(t)
L
,
3.4 electric circuits 111
in which x(t) =i(t), τ =
L
R
, Q(t) =
V(t)
L
.
Parallel RL Circuit
Referring to Figure 3.12(b), applying Kirchhoff ’s Current Law at node 1 yields
I(t) −
v
R
−
1
L
_
t
−∞
v(t)dt = 0.
Differentiating with respect to t gives
1
R
dv
dt
+
1
L
v =
dI(t)
dt
=⇒
dv
dt
+
R
L
v = R
dI(t)
dt
,
in which x(t) =v(t), τ =
L
R
, Q(t) =R
dI(t)
dt
.
Example 3.8 — FirstOrder Circuit 3.8
For the electric circuit shown in the following ﬁgure, determine v
L
for t >0.
4A
25V
1H 12 20
6
5
t =0
v
L
❧ For t <0, the switch is closed and the inductor behaves as a short circuit.
R
eq
i
2
(0
−
)
v(0
−
)
i(0
−
)
1
i
L
(0
−
)
v(0
−
)
1
t 0
–
4A
25V
1H 12 20
6
5
v
L
4A
25V
5
The three resistors of 12 , 20 , and 6 are in parallel connection and can be
combined as an equivalent resistor
1
R
eq
=
1
12
+
1
20
+
1
6
=⇒ R
eq
=
10
3
.
Applying Kirchhoff ’s Current Law at node 1 yields
4 = i(0
−
) + i
2
(0
−
) =
v(0
−
)
R
eq
+
v(0
−
) − 25
5
=
3v(0
−
)
10
+
v(0
−
)
5
− 5,
112 3 applications of ﬁrstorder and simple higherorder equations
∴ v(0
−
) =18 V =⇒ i
L
(0
−
) =
v(0
−
)
6
= 3 A.
❧ At t =0, the switch is open. Since the current in an inductor cannot change
abruptly, i
L
(0
−
) =i
L
(0
+
) =3 A.
❧ For t >0, the switch is open and the circuit becomes
t 0
+
1H v
L
i
2
v
i
L
i
1
1
25V
5 6
20
It is easy to evaluate that
i
L
=
v−v
L
6
=⇒ v = 6i
L
+ v
L
,
i
1
=
v
20
=
6i
L
+v
L
20
, i
2
=
v−25
5
=
6i
L
+v
L
−25
5
.
Applying Kirchhoff ’s Current Law at node 1 yields
i
1
+ i
2
+ i
L
= 0 =⇒
6i
L
+v
L
20
+
6i
L
+v
L
−25
5
+ i
L
= 0,
v
L
+ 10i
L
= 20 =⇒
di
L
dt
+ 10i
L
= 20. v
L
=L
di
L
dt
=
di
L
dt
With τ =
1
10
, Q
0
=20, i
L
(0
+
) =3, the solution of the differential equation is
i
L
(t) = Q
0
τ +
_
i
L
(0
+
) − Q
0
τ
_
e
−t/τ
= 2 + e
−10t
,
∴ v
L
(t) = 1×
di
L
(t)
dt
= −10e
−10t
(V).
Example 3.9 — FirstOrder Circuit 3.9
For the electric circuit shown in the following ﬁgure, determine i
1
for t >0.
I
0
t =0
C
i
1
R
3
R
2
R
1
3.4 electric circuits 113
❧ For t <0, the switch is open and the current source is disconnected. The
capacitor behaves as an open circuit. Hence i
1
(0
−
) =0 and v
2
(0
−
) =0.
I
0 C
i
1
i
2 v
2
v
1
i
C
i
3
R
3
R
2
R
1
C
i
1
(0
−
)
v
2
(0
−
)
R
3
R
2
R
1
1 2
t 0
–
t 0
+
❧ At t =0, the switch is closed. Since the capacitor voltage cannot change abruptly,
v
2
(0
+
) =v
2
(0
−
) =0. Applying Kirchhoff ’s Current Law at node 1 yields
I
0
= i
1
(0
+
) + i
2
(0
+
) =
v
1
(0
+
)
R
1
+
v
1
(0
+
)−v
2
(0
+
)
R
2
=
v
1
(0
+
)
R
1
+
v
1
(0
+
)
R
2
,
∴ v
1
(0
+
) =
R
1
R
2
R
1
+R
2
I
0
=⇒ i
1
(0
+
) =
v
1
(0
+
)
R
1
=
R
2
R
1
+R
2
I
0
.
❧ For t >0, i
1
=
v
1
R
1
=⇒ v
1
=R
1
i
1
,
i
2
=
v
1
−v
2
R
2
=
R
1
i
1
−v
2
R
2
, i
3
=
v
2
R
3
, i
C
= C
dv
2
dt
.
Applying Kirchhoff ’s Current Law at node 1 gives
I
0
= i
1
+ i
2
=⇒ i
2
= I
0
− i
1
=
R
1
i
1
−v
2
R
2
=⇒ v
2
= (R
1
+R
2
)i
1
− R
2
I
0
.
Applying Kirchhoff ’s Current Law at node 2 leads to
i
2
= i
C
+ i
3
=⇒ I
0
− i
1
= C
dv
2
dt
+
v
2
R
3
,
∴ I
0
− i
1
= C· (R
1
+R
2
)
di
1
dt
+
(R
1
+R
2
)i
1
− R
2
I
0
R
3
,
∴
di
1
dt
+
1
τ
i
1
= Q
0
, τ =
C(R
1
+R
2
)R
3
R
1
+R
2
+R
3
, Q
0
=
R
2
+R
3
C(R
1
+R
2
)R
3
I
0
.
The solution is given by
i
1
(t) = Q
0
τ +
_
i
1
(0
+
) − Q
0
τ
_
e
−t/τ
=
R
2
+R
3
R
1
+R
2
+R
3
I
0
+
_
R
2
R
1
+R
2
−
R
2
+R
3
R
1
+R
2
+R
3
_
I
0
e
−t/τ
.
Remarks: Since these circuits are characterized by firstorder differential equa
tions, they are called firstorder circuits. They consist of resistors and the equiva
lent of one energy storage element, such as capacitors and inductors.
114 3 applications of ﬁrstorder and simple higherorder equations
3.5 Natural Puriﬁcation in a Stream
In this section, the variation of water quality in a stream due to pollution is inves
tigated. The amount of pollutant is considered to be small enough such that the
stream ﬂow is not altered by its presence.
When sewage and wastes are discharged into a stream, the stream water will be
degraded in its physical quality (e.g., odor and color), its chemical contents, and
the type and population of aquatic life. The criterion for determining the quality
of the stream water depends on the uses the water is to serve. For many purposes,
engineers use the concentration of Dissolved Oxygen (DO) and decomposable or
ganic matter in the water as indicators of its quality. The DOmeasures the capacity
of the water to assimilate many polluting materials and to support aquatic life. The
organic matter consumes oxygen in its decomposition. In sewage, the organic mat
ter includes a great variety of compounds, represented by the amount of oxygen
required for its biological decomposition (Biochemical Oxygen Demand, or BOD).
Clean stream water is usually saturated with DO. As sewage is added and ﬂows
in the stream, the DO in the polluted water is consumed as the organic matter
is decomposed. In the meantime, oxygen from the atmosphere dissolves into the
water, as it is now no longer saturated with DO. Finally, the organic matter is
completely decomposed and the stream water becomes saturated with DO again.
This natural process of puriﬁcation takes place within a period of several days. It
is necessary to ascertain the variation of DO and BOD along the ﬂow to determine
the effect of pollution on the stream.
The BOD added to the stream is assumed to spread across the stream over a
distance that is very short in comparison with the length of the stream where
deoxygenation by the BODand reoxygenation by the atmosphere take place, so that
the problem can be considered to be one dimensional with DO and BOD assumed
to be uniform at a crosssection.
To derive the governing equations, consider the mass balance of BOD during dt
in a volume Adx bounded by two crosssections dx apart, as shown in Figure 3.13,
in which A is the crosssectional area of the stream, x is the distance measured
along the stream, and t is time.
Employ the following notations:
Q = the discharge,
b = the concentration of BOD in mass per unit volume of water,
c = the concentration of DO in mass per unit volume of water,
M = the mass of BOD added per unit time per unit discharge along the ﬂow,
N = the mass of oxygen added per unit time per unit discharge along the
3.5 natural puriﬁcation in a stream 115
Q
Q+ dx
Q' dx
dx x
∂Q
∂x
At time t
At time (t+dt)
Figure 3.13 Natural puriﬁcation in a stream.
stream from sources other than the atmosphere (e.g., from
photosynthesis of green plants in the stream),
r
1
= the mass of BOD decomposed per unit volume per unit time,
r
2
= the rate of atmospheric reoxygenation in mass per unit volume per
unit time.
Any difference between inﬂow and outﬂow and between addition and subtrac
tion will cause a change in the mass of BOD contained between these two cross
sections. Thus, during dt,
Qbdt −
_
Q+
∂Q
∂x
dx
__
b+
∂b
∂x
dx
_
dt + Mdxdt − r
1
Adxdt =
∂(bA)
∂t
dxdt.
. ,, . . ,, . . ,, . . ,, . . ,, .
mass
inflow
mass
outflow
added
mass
mass
decomposed
increase mass
involume
Simplifying this equation by dividing (Adxdt) yields
∂b
∂t
+ V
∂b
∂x
= −
Q
b
A
− r
1
+
M
A
, (1)
where V =
Q
A
is the mean velocity at a crosssection, and
Q
=
∂Q
∂x
+
∂A
∂t
is the discharge added per unit length of stream. Similarly, from a mass balance for
the DO, one has
Qc dt −
_
Q+
∂Q
∂x
dx
__
c +
∂c
∂x
dx
_
dt − r
1
Adxdt + r
2
Adxdt + Ndxdt
. ,, . . ,, . . ,, . . ,, . . ,, .
mass
inflow
mass
outflow
mass
consumed
by BOD
mass from
atmosphere
other
added
mass
=
∂(cA)
∂t
dxdt. increase mass involume
116 3 applications of ﬁrstorder and simple higherorder equations
Simplifying the equation by dividing (Adxdt) leads to
∂c
∂t
+ V
∂c
∂x
= −
Q
c
A
− r
1
+ r
2
+
N
A
. (2)
Generally, the rate r
1
of oxygen consumption depends on, among other factors,
the DO concentration c and the BOD concentration b. In practical cases, the
BOD concentration is usually sufﬁciently low, so that r
1
can be assumed to be
proportional to the BOD concentration and independent of the DO concentration
as long as it is greater than a very small value; that is
r
1
= k
1
b, for c >0
+
, (3)
where the coefﬁcient k
1
depends on the composition of the sewage and its tem
perature. For a given sewage, the numerical value of k
1
can be determined in the
laboratory.
The rate r
2
of atmospheric reoxygenation is usually assumed to be proportional
to the DO deﬁcit (c
s
−c):
r
2
= k
2
(c
s
−c), for 0c c
s
, (4)
where c
s
is the saturation concentration of DO, which depends on the water tem
perature. The coefﬁcient k
2
depends on the temperature, the area of the airwater
interface per unit volume of the stream, and the turbulence of the air and water.
With hydrographical data and sources of BOD and DO of the stream, equation
(1) can be solved independently for the BOD distribution b(x, t). The DO distri
bution c(x, t) can then be obtained from equation (2), as r
1
=k
1
b is now a known
function of x and t.
For simplicity of analysis, the steadystate case is considered, in which the vari
ables, such as b(x, t) and c(x, t), do not change with time t. Hence, equations (1)
and (2) become
db
dx
+
_
1
Q
dQ
dx
+
k
1
V
_
b =
M
Q
, (5)
dc
dx
+
_
1
Q
dQ
dx
c +
k
1
V
b −
k
2
V
(c
s
−c)
_
=
N
Q
,
which can be rewritten as
d(c
s
−c)
dx
+
_
1
Q
dQ
dx
+
k
2
V
_
(c
s
−c) =
k
1
V
b +
1
Q
dQ
dx
c
s
+
N
Q
. (6)
In the following, various special cases are studied.
3.5 natural puriﬁcation in a stream 117
1. Consider b(x) as BOD concentration in a stream of constant Q and V. De
termine b(x) for the case with b=b
0
at x =0 and M=0. Determine the steady
distribution of DO along the stream with c =c
0
at x =0 and N =0.
Since M=0, Q and V are constants,
dQ
dx
=0, equation (5) becomes
db
dx
+
k
1
V
b = 0, Variable separable
the solution of which is
_
db
b
= −
_
k
1
V
dx + C =⇒ ln b = −
k
1
V
x + ln D.
Hence
b(x) = De
−k
1
x/V
,
in which the constant D is determined by the initial condition b=b
0
at x =0:
b
0
= De
0
=⇒ D = b
0
=⇒ b(x) = b
0
e
−k
1
x/V
.
For N =0, equation (6) becomes
d(c
s
−c)
dx
+
k
2
V
· (c
s
−c) =
k
1
V
b
0
e
−k
1
x/V
, Linear firstorder DE
in which the dependent variable is (c
s
−c). It is easy to evaluate
P(x) =
k
2
V
, Q(x) =
k
1
V
b
0
e
−k
1
x/V
,
_
P(x)dx =
k
2
V
x, e
_
P(x)dx
= e
k
2
x/V
, e
−
_
P(x)dx
= e
−k
2
x/V
,
_
Q(x)e
_
P(x)dx
dx =
_
k
1
V
b
0
e
−k
1
x/V
· e
k
2
x/V
dx =
k
1
b
0
k
2
−k
1
e
(k
2
−k
1
)x/V
.
The general solution of the differential equation is
(c
s
− c) = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
_
= e
−k
2
x/V
_
k
1
b
0
k
2
−k
1
e
(k
2
−k
1
)x/V
+ C
_
,
in which the constant C is determined by the initial condition c =c
0
at x =0:
(c
s
− c
0
) =
k
1
b
0
k
2
−k
1
+ C =⇒ C = (c
s
−c
0
) −
k
1
b
0
k
2
−k
1
.
Hence
c
s
− c(x) = (c
s
−c
0
)e
−k
2
x/V
+
k
1
b
0
k
2
−k
1
_
e
−k
1
x/V
− e
−k
2
x/V
_
.
118 3 applications of ﬁrstorder and simple higherorder equations
2. Determine the BODdistribution b(x) for the case with Q=Q
0
(1+γ x), constant
V, M and k
1
, and b=b
0
at x =0. Determine the DO distribution in the stream
with a constant N and c =c
0
at x =0. From the solution, ﬁnd the value of c far
downstream for the case of a uniform stream (γ =0).
Since V, M and k
1
are constants, Q=Q
0
(1+γ x) =⇒
dQ
dx
=Q
0
γ , equation (5)
becomes
db
dx
+
_
γ
1+γ x
+
k
1
V
_
b =
M
Q
0
(1+γ x)
, Linear firstorder DE
which is of the form
db
dx
+ P(x) · b = Q(x), where
P(x) =
γ
1+γ x
+
k
1
V
, Q(x) =
M
Q
0
(1+γ x)
,
_
P(x)dx =
_
_
γ
1+γ x
+
k
1
V
_
dx = ln(1+γ x) +
k
1
V
x,
e
_
P(x)dx
= e
ln(1+γ x)
e
k
1
x/V
= (1+γ x)e
k
1
x/V
, e
−
_
P(x)dx
=
1
1+γ x
e
−k
1
x/V
,
_
Q(x)e
_
P(x)dx
dx =
_
M
Q
0
(1+γ x)
(1+γ x)e
k
1
x/V
dx =
MV
Q
0
k
1
e
k
1
x/V
.
The general solution of the differential equation is
b(x) = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
_
=
1
1+γ x
e
−k
1
x/V
_
MV
Q
0
k
1
e
k
1
x/V
+ C
_
=
MV
Q
0
k
1
1
1+γ x
+
C
1+γ x
e
−k
1
x/V
,
in which the constant C is determined by the initial condition b=b
0
at x =0:
b
0
=
MV
Q
0
k
1
+ C =⇒ C = b
0
−
MV
Q
0
k
1
.
Hence
b(x) =
1
1+γ x
_
MV
Q
0
k
1
+
_
b
0
−
MV
Q
0
k
1
_
e
−k
1
x/V
_
=
MV
Q
0
k
1
1
1+γ x
_
1−e
−k
1
x/V
_
+
b
0
1+γ x
e
−k
1
x/V
.
Equation (6) becomes
d(c
s
−c)
dx
+
_
γ
1+γ x
+
k
2
V
_
· (c
s
−c)
3.5 natural puriﬁcation in a stream 119
=
k
1
V
_
MV
Q
0
k
1
1
1+γ x
_
1−e
−k
1
x/V
_
+
b
0
1+γ x
e
−k
1
x/V
_
+
γ c
s
1+γ x
+
N
Q
0
(1+γ x)
,
which is a linear ﬁrstorder differential equation with dependent variable being
(c
s
−c), and
P(x) =
γ
1+γ x
+
k
2
V
, Q(x) =
α
1+γ x
+
β
1+γ x
e
−k
1
x/V
,
where
α =
k
1
V
MV
Q
0
k
1
+ γ c
s
+
N
Q
0
= γ c
s
+
M+N
Q
0
, β =
k
1
b
0
V
−
M
Q
0
.
Referring to the differential equation for b(x), one has
e
_
P(x)dx
= (1+γ x)e
k
2
x/V
, e
−
_
P(x)dx
=
1
1+γ x
e
−k
2
x/V
,
_
Q(x)e
_
P(x)dx
dx =
_
_
α
1+γ x
+
β
1+γ x
e
−k
1
x/V
_
· (1+γ x)e
k
2
x/V
dx
=
_
_
αe
k
2
x/V
+ βe
(k
2
−k
1
)x/V
_
dx =
αV
k
2
e
k
2
x/V
+
βV
k
2
−k
1
e
(k
2
−k
1
)x/V
.
The general solution of the differential equation is
(c
s
− c) = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
_
=
1
1+γ x
e
−k
2
x/V
_
αV
k
2
e
k
2
x/V
+
βV
k
2
−k
1
e
(k
2
−k
1
)x/V
+ C
_
,
in which the constant C is determined by the initial condition c =c
0
at x =0:
(c
s
− c
0
) =
αV
k
2
+
βV
k
2
−k
1
+ C =⇒ C = (c
s
−c
0
) −
αV
k
2
−
βV
k
2
−k
1
.
Hence
c
s
− c(x) =
e
−k
2
x/V
1+γ x
_
(c
s
−c
0
) +
αV
k
2
_
e
k
2
x/V
−1
_
+
βV
k
2
−k
1
_
e
(k
2
−k
1
)x/V
−1
_
_
.
For a uniform stream, γ =0, and for far downstream, x→∞. By taking the limit
as x→∞, one obtains,
lim
x→∞
_
c
s
− c(x)
_
= lim
x→∞
_
(c
s
−c
0
)e
−k
2
x/V
+
αV
k
2
_
1 − e
−k
2
x/V
_
+
βV
k
2
−k
1
_
e
−k
1
x/V
− e
−k
2
x/V
_
_
=
αV
k
2
=⇒ lim
x→∞
c(x) = c
s
−
αV
k
2
.
120 3 applications of ﬁrstorder and simple higherorder equations
Procedure for Solving an Application Problem
1. Establish the governing differential equations based on physical principles
and geometrical properties underlying the problem.
2. Identify the type of these differential equations and then solve them.
3. Determine the arbitrary constants in the general solutions using the initial
or boundary conditions.
3.6 Various Application Problems
Example 3.10 — Ferry Boat 3.10
A ferry boat is crossing a river of width a from point A to point O as shown in the
following ﬁgure. The boat is always aiming toward the destination O. The speed of
the river ﬂow is constant v
R
and the speed of the boat is constant v
B
. Determine
the equation of the path traced by the boat.
v
R
v
B
v
B
cosθ
θ
v
B
sinθ
x x
y
River Flow
y
P(x, y)
A
H a O
Suppose that, at time t, the boat is at point P with coordinates (x, y). The velocity
of the boat has two components: the velocity of the boat v
B
relative to the river ﬂow
(as if the river is not ﬂowing), which is pointing toward the origin O or along line
PO, and the velocity of the river v
R
in the y direction.
Decompose the velocity components v
B
and v
R
in the x and ydirections
v
x
= −v
B
cos θ, v
y
= v
R
− v
B
sin θ.
From OHP, it is easy to see
cos θ =
OH
OP
=
x
_
x
2
+y
2
, sin θ =
PH
OP
=
y
_
x
2
+y
2
.
Hence, the equations of motion are given by
v
x
=
dx
dt
= −v
B
x
_
x
2
+y
2
, v
y
=
dy
dt
= v
R
− v
B
y
_
x
2
+y
2
.
3.6 various application problems 121
Since only the equation between x and y is sought, variable t can be eliminated by
dividing these two equations
dy
dx
=
dy
dt
dx
dt
=
v
R
− v
B
y
_
x
2
+y
2
−v
B
x
_
x
2
+y
2
= −
k
_
x
2
+y
2
− y
x
, k =
v
R
v
B
= −k
_
1+
_
y
x
_
2
+
y
x
. Homogeneous DE
Let u=
y
x
or y =xu,
dy
dx
=u+x
du
dx
. Hence, the equation becomes
u + x
du
dx
= −k
_
1+u
2
+ u,
∴ x
du
dx
= −k
_
1+u
2
. Variable separable
The general solution is
_
du
√
1+u
2
= −k
_
dx
x
+ D =⇒ ln
_
u+
_
1+u
2
_
= −k ln x + ln C,
∴ u +
_
1+u
2
= Cx
−k
.
Replacing u by the original variables yields
y
x
+
_
1+
_
y
x
_
2
= Cx
−k
=⇒
_
x
2
+y
2
= Cx
1−k
− y.
Squaring both sides leads to
x
2
+ y
2
= C
2
x
2(1−k)
− 2Cx
1−k
y + y
2
=⇒ x
2
= C
2
x
2(1−k)
− 2Cx
1−k
y.
The constant C is determined by the initial condition t =0, x =a, y =0:
a
2
= C
2
a
2(1−k)
− 0 =⇒ C = a
k
.
Hence, the equation of the path is
y =
1
2Cx
1−k
_
C
2
x
2(1−k)
− x
2
_
=
1
2
_
a
k
x
1−k
− a
−k
x
1+k
_
,
∴ y =
a
2
_
_
x
a
_
1−k
−
_
x
a
_
1+k
_
.
122 3 applications of ﬁrstorder and simple higherorder equations
Example 3.11 — Bar with Variable CrossSection 3.11
A bar with circular crosssections is supported at the top end and is subjected to a
load of P as shown in Figure 3.14(a). The length of the bar is L. The weight density
of the materials is ρ per unit volume. It is required that the stress at every point is
constant σ
a
. Determine the equation for the crosssection of the bar.
x
y
P P
L
x
x
dx
y
y
P+W(x)
(a) (b) (c)
x
x
y
Figure 3.14 A bar under axial load.
Consider a crosssection at level x as shown in Figure 3.14(b). The corresponding
radius is y. The volume of a circular disk of thickness dx is dV =πy
2
dx. The
volume of the segment of bar between 0 and x is
V(x) =
_
x
0
πy
2
dx,
and the weight of this segment is
W(x) = ρV(x) = ρ
_
x
0
πy
2
dx.
The load applied on crosssection at level x is equal to the sum of the externally
applied load P and the weight of the segment between 0 and x, i.e.,
F(x) = W(x) + P = ρ
_
x
0
πy
2
dx + P.
The normal stress is
σ(x) =
F(x)
A(x)
=
1
πy
2
_
ρ
_
x
0
πy
2
dx + P
_
= σ
a
=⇒ ρ
_
x
0
πy
2
dx + P = σ
a
πy
2
.
Differentiating with respect to x yields
ρπy
2
= σ
a
π · 2y
dy
dx
. Variable separable
3.6 various application problems 123
Since y =0, the equation can be written as
ρ
2σ
a
dx =
1
y
dy,
and the general solution is given by
ρ
2σ
a
_
dx =
_
1
y
dy + C =⇒
ρ
2σ
a
x = ln y + C,
or
y = C exp
_
ρ
2σ
a
x
_
=⇒ A(x) = πy
2
= πC
2
exp
_
ρ
σ
a
x
_
.
The constant C is determined by the initial condition: x =0, W(0) =0, F(0) =P,
σ(0) =
P
A(0)
=
P
πC
2
exp
_
ρ
σ
a
· 0
_ =
P
πC
2
= σ
a
=⇒ C
2
=
P
πσ
a
.
Hence
A(x) = π ·
P
πσ
a
· exp
_
ρ
σ
a
x
_
=
P
σ
a
exp
_
ρ
σ
a
x
_
, 0x L.
Example 3.12 — Chain Moving 3.12
Auniformchain of length L with mass density per unit length ρ is laid on a smooth
horizontal table with an initial hang of length l as shown in Figure 3.15(a). The
chain is released from rest at time t =0. Show that the time it takes for the chain to
leave the table is given by
T =
_
L
g
ln
L+
_
L
2
−l
2
l
.
At time t, the length of the chain hanging off the table is y(t) as shown in Figure
3.15(b). The chain is subjected to a downward force F(t) =(ρy)g, which is the
weight of the segment of the chain hanging off the table. Apply Newton’s Second
Law to the chain
↓ ma =
F : (ρL) ¨ y = (ρ y)g =⇒ ¨ y −
g
L
y = 0,
or
¨ y − k
2
y = 0, k =
_
g
L
.
The initial conditions are t =0, y =l, ˙ y =0.
Remarks: Since the problem is equivalent to the entire chain moving in the
vertical direction under gravity (ρy)g as shown in Figure 3.15(c), the mass m is
for the entire chain, not just the segment that is hanging off the table.
124 3 applications of ﬁrstorder and simple higherorder equations
time t =0
(a)
(b) (c)
time t
l
y(t)
(ρy)g
y, y, y
y
L
Figure 3.15 A chain moving off a smooth table.
The equation of motion is a secondorder differential equation with the inde
pendent variable t absent. Let y be the new independent variable and u=
dy
dt
be
the new dependent variable,
d
2
y
dt
2
= u
du
dy
. Hence
u
du
dy
− k
2
y = 0. Variable separable
The general solution is given by
_
udu =
_
k
2
ydy + C =⇒
1
2
u
2
=
k
2
2
y
2
+ C.
The constant of integration C is determined by the initial condition t =0, y =l,
u= ˙ y =0:
1
2
· 0
2
=
k
2
2
· l
2
+ C =⇒ C = −
k
2
2
l
2
.
Hence,
u
2
= k
2
( y
2
−l
2
) =⇒ u =
dy
dt
= k
_
y
2
−l
2
. Variable separable
The general solution is
_
dy
_
y
2
−l
2
=
_
kdt + D =⇒ ln
_
y+
_
y
2
−l
2
_
= kt + D.
Using the initial condition t =0, y =l, one obtains
ln l = k· 0 + D =⇒ D = ln l.
3.6 various application problems 125
The solution of the equation of motion is
ln
_
y+
_
y
2
−l
2
_
= kt + ln l,
or
t =
1
k
_
ln
_
y+
_
y
2
−l
2
_
−ln l
_
=
1
k
ln
y+
_
y
2
−l
2
l
.
When the chain leaves the table, t =T, y =L:
T =
_
L
g
ln
L+
√
L
2
−l
2
l
.
Example 3.13 — Chain Moving 3.13
One end of a pile of uniform chain falls through a hole in its support and pulls the
remaining links after it as shown. The links, which are initially at rest, acquire the
velocity of the chain suddenly without any frictional resistance or interference from
the support and adjacent links. At t =0, y(t) =0 and v(t) = ˙ y(t) =0. Determine
the length y(t) and the velocity v(t) of the chain.
At time t, the length of the chain hanging off the support is y(t) and the velocity of
the chain is v(t) = ˙ y(t). The chain is subjected to a downward force F(t) =(ρ y)g,
which is the weight of the segment of the chain hanging off the support.
y(t+t)=y+y
Time t+t
Time t
y(t)
(ρy)g
y(t)=v
y(t+t)=v+v
y, y, y
To set up the equation of motion, apply the ImpulseMomentum Principle:
(Momentum at time t) + (Impulse during t) = (Momentum at time t +t),
where
Momentum at time t = (ρ y)v, ρ = mass density of the chain,
Momentum at time t +t =
_
ρ( y+y)
_
(v+v),
Impulse during t =
_
(ρ y)g
_
t.
126 3 applications of ﬁrstorder and simple higherorder equations
Hence,
ρ yv + ρg yt = ρ( y+y)(v+v).
Dividing the equation by t and taking the limit as t →0 result in the equation
of motion
y
dv
dt
+ v
dy
dt
= g y =⇒
d( yv)
dt
= g y.
Noting that v =
dy
dt
=⇒ dt =
dy
v
, and letting V =yv, one has
dV
dy
v
= g y =⇒ vdV = g ydy.
Multiplying the equation by y yields
VdV = g y
2
dy. Variable separable
Integrating both sides gives
1
2
V
2
=
1
3
g y
3
+ C
1
.
Using the initial conditions y =0 and v =0 when t =0, one has C
1
=0. Hence
1
2
V
2
=
1
3
g y
3
=⇒
dy
dt
= v =
_
2g y
3
. Variable separable
_
dy
√
y
=
_
_
2g
3
dt + C
2
=⇒ 2
√
y =
_
2g
3
t + C
2
.
Using the initial conditions again gives C
2
=0. Thus
y(t) =
g
6
t
2
, v(t) =
dy
dt
=
g
3
t.
Example 3.14 — Water Leaking 3.14
Ahemispherical bowl of radius R is ﬁlled with water. There is a small hole of radius
r at the bottom of the convex surface as shown in Figure 3.16(a). Assume that the
velocity of efﬂux of the water when the water level is at height h is v =c
_
2gh,
where c is the discharge coefﬁcient. The volume of the cap of the sphere of height
h, shown as shaded volume in Figure 3.16(a), is given by
V =
π
3
h
2
(3R − h).
Determine the time taken for the bowl to empty.
3.6 various application problems 127
h(t)
2r
R
dh
dV
dU
(a)
(b)
v
.
dt
2r
R
h
Figure 3.16 A hemispherical bowl with a hole.
At time t, the water level is h(t) and the volume of the water is
V =
π
3
h
2
(3R − h).
Considering a small time interval dt, the water level drops dh as shown in Figure
3.16(b). The water lost is
dV =
dV
dh
· dh =
π
3
_
2h(3R−h) + h
2
(−1)
_
dh = π(2Rh−h
2
)dh.
The water lost is leaked from the hole at the bottom. The water level drops, i.e.,
dh<0, which leads to a negative volume change, i.e., dV<0.
Since the velocity of efﬂux of water is v, the amount of water leaked during time
dt is
dU = πr
2
· vdt = πr
2
· c
_
2gh dt,
which is indicated by the small shaded cylinder at the bottom of the bowl.
From the conservation of water volume, dV+dU =0, i.e.,
π(2Rh−h
2
)dh + πr
2
· c
_
2gh dt = 0 =⇒ (2R−h)
√
h dh = −r
2
c
_
2g dt.
The equation is variable separable and the general solution is
_
_
2Rh
1
2
−h
3
2
_
dh = −r
2
c
_
2g
_
dt + D,
2R
h
3
2
3
2
−
h
5
2
5
2
= −r
2
c
_
2g t + D.
_
x
n
dx =
x
n+1
n+1
The constant D is determined from the initial condition t =0, h=R:
4
3
R
5
2
−
2
5
R
5
2
= D =⇒ D =
14
15
R
5
2
.
When the bowl is empty, t =T, h=0:
0 = −r
2
c
_
2g T +
14
15
R
5
2
=⇒ T =
14
15
R
5
2
r
2
c
_
2g
=
14
15c
_
R
r
_
2
_
R
2g
.
128 3 applications of ﬁrstorder and simple higherorder equations
Example 3.15 — Reservoir Pollution 3.15
A reservoir initially contains polluted water of volume V
0
(m
3
) with a pollutant
concentration in percentage being c
0
. In order to reduce c(t), which is the pollutant
concentration in the reservoir at time t, it is arranged to have inﬂowand outﬂowof
water at the rates of Q
in
and Q
out
(m
3
/day), respectively, as shown in the following
ﬁgure. Unfortunately, the inﬂowing water is also polluted, but to a lower extent of
c
in
. Assume that the outﬂowing water is perfectly mixed.
Outflow
Q
out
, c(t)
Inflow
Q
in
, c
in
Volume V(t)
Concentration c(t)
1. Set up the differential equation governing the pollutant c(t).
2. Considering the case with the following parameters
V
0
=500 m
3
, Q
in
=200 m
3
, Q
out
=195 m
3
, c
0
=0.05%, c
in
=0.01%,
ﬁnd the time (in days) it will take to reduce the pollutant concentration to the
acceptable level of 0.02%.
1. At time t,
Volume V(t) = V
0
+ (Q
in
−Q
out
)t,
Pollutant concentration c(t),
Amount of pollutant = V(t)c(t) =
_
V
0
+ (Q
in
−Q
out
)t
_
c.
At time t +t,
Volume V(t +t) = V
0
+ (Q
in
−Q
out
)(t +t),
Pollutant concentration c(t +t) = c(t)+c,
Amount of pollutant = V(t +t)c(t +t)
=
_
V
0
+(Q
in
−Q
out
)(t +t)
_
(c +c),
Inﬂow pollutant = Q
in
t · c
in
,
Outﬂow pollutant = Q
out
t · c.
3.6 various application problems 129
Since
(Amount of pollutant at t +t) = (Amount of pollutant at t)
+
_
(Inﬂow pollutant) − (Outﬂow pollutant)
_
,
∴
_
V
0
+ (Q
in
−Q
out
)(t +t)
_
(c +c) =
_
V
0
+ (Q
in
−Q
out
)t
_
c
+ Q
in
t · c
in
− Q
out
t · c,
expanding yields
_
V
0
+ (Q
in
−Q
out
)t
_
c +
_
V
0
+ (Q
in
−Q
out
)t
_
c + (Q
in
−Q
out
)t · c
+ (Q
in
−Q
out
) t · c =
_
V
0
+ (Q
in
−Q
out
)t
_
c + Q
in
t · c
in
− Q
out
t · c.
Neglecting higher order term t·c, dividing by t, and simplifying lead to
_
V
0
+ (Q
in
−Q
out
)t
_
c
t
+ Q
in
c = Q
in
c
in
.
Taking the limit t →0 results in the differential equation
_
V
0
+ (Q
in
−Q
out
)t
_
dc
dt
+ Q
in
c = Q
in
c
in
.
2. For the parameters
V
0
= 500, Q
in
= 200, Q
out
= 195, c
0
= 0.05, c
in
= 0.01,
the differential equation becomes
_
500 + (200−195)t
_
dc
dt
+ 200c = 200×0.01 =⇒ (100+t)
dc
dt
= 0.4 − 40c.
The equation is variable separable and the general solution is
_
dc
0.4−40c
=
_
dt
100+t
+ C =⇒ −
1
40
ln(0.4−40c) = ln(100+t) + ln D,
∴ (0.4−40c)
−
1
40
= D(100+t). C= ln D
The constant D is determined from the initial condition t =0, c =c
0
=0.05:
(0.4−40c
0
)
−
1
40
= D· 100 =⇒ D =
(0.4−40c
0
)
−
1
40
100
.
The solution becomes
(0.4−40c)
−
1
40
=
(0.4−40c
0
)
−
1
40
100
(100+t) =⇒ t =100
_
_
0.4−40c
0.4−40c
0
_
−
1
40
−1
_
.
When the pollutant concentration is reduced to c =0.02, the time required is
t = 100
_
_
0.4−40×0.02
0.4−40×0.05
_
−
1
40
− 1
_
= 3.53 days.
130 3 applications of ﬁrstorder and simple higherorder equations
Problems
3.1 At a crime scene, a forensic technician found the body temperature of the
victim was 33
◦
C at 6:00 p.m. One hour later, the coroner arrived and found the
body temperature of the victim fell to 31.5
◦
C. The forensic technician determined
that the change in the atmospheric temperature could be modeled satisfactorily as
20e
−0.02t
in the time window of ±3 hours starting at 6:00 p.m. It is known that
the body temperature of a live person is 37
◦
C. When was the victim murdered?
A
NS 3:42 p.m.
3.2 The value of proportionality of cooling of a large workshop is k (1/hr) due to
its ventilating system. The atmospheric temperature ﬂuctuates sinusoidally with a
period of 24 hours, reaching a minimum of 15
◦
C at 2:00 a.m. and a maximum of
35
◦
C at 2:00 p.m. Let t denote the time in hours starting with t =0 at 8:00 a.m.
1. By applying Newton’s Law of Cooling, set up the differential equation gov
erning the temperature of the workshop T(t).
A
NS
dT
dt
+ kT = k
_
25 + 10 sin
πt
12
_
2. Determine the steadystate solution of the differential equation, which is
solution for time t large or the solution due to the atmospherical temperature
change.
A
NS T
steadystate
= 25 +
120k
√
144k
2
+π
2
sin
_
πt
12
− tan
−1
_
π
12k
__
3. If k =0.2 (1/hr), what are the maximum and minimum temperatures that
the workshop will reach?
A
NS T
min
=18.9
◦
C, T
max
=31.1
◦
C
3.3 Suppose that the air resistance on a parachute is proportional to the effec
tive area A of the parachute, which is the area of the parachute projected in the
horizontal plane, and to the square of its velocity v, i.e., R=kAv
2
, where k is a
constant. A parachute of mass m falls with zero initial velocity, i.e., x(t) =0 and
v(t) =0 when t =0.
m
Problems 131
1. Show that the terminal velocity v
T
of the parachute is
v
T
= lim
t→∞
v(t) =
_
mg
kA
.
2. Show that the velocity and the displacement of the mass are given by
v(t) = v
T
tanh
_
g t
v
T
_
, x(t) =
v
2
T
g
ln cosh
_
gt
v
T
_
. tanh z =
e
z
− e
−z
e
z
+ e
−z
3.4 A mass m falls from a height of H =3200 km with zero initial velocity as
shown in the following ﬁgure. The gravity on the mass changes as
F =
mg R
2
(R+H−x)
2
,
where R=6400 kmis the radius of the earth. Neglect the air resistance. Determine
the time it takes for the mass to reach the ground and the velocity of the mass.
t =0, x=0, v=0
t
F
x
H
R
x, v, a
Hint: Use the formulas a =
dv
dt
=
dv
dx
dx
dt
= v
dv
dx
, and
_
_
k−x
x
dx =
_
x(k−x) + k sin
−1
_
x
k
+ C.
A
NS v =
_
2g RH
R+H
= 6.47 km/sec
t =
1
R
_
R+H
2g
_
√
RH + (R+H) sin
−1
_
H
R+H
_
= 1, 141 sec
3.5 An undersea explorer traveling along a straight line in a horizontal direction
is propelled by a constant force T. Suppose the resistant force is R=a+bv
2
, where
0<a<T and b>0 are constants, and v is the velocity of the explorer.
132 3 applications of ﬁrstorder and simple higherorder equations
1. If the explorer is at rest at time t =0, show that the velocity of the explorer is
v(t) = α tanh βt, α =
_
T−a
b
, β =
αb
m
=
1
m
_
b(T−a).
2. Determine the distance x(t) traveled at time t.
A
NS x =
α
β
ln cosh βt
3.6 A boat starting from rest at time t =0 is propelled by a constant force F. The
resistance fromair is proportional tothe velocity, i.e., R
air
= −αv, andthe resistance
fromwater is proportional tothe square of the velocity, i.e., R
water
= −βv
2
, inwhich
α, β are positive constants. Assume that F >αv+βv
2
.
1. Show that the velocity of the boat is given by
v(t) = a tanh
_
βat
m
+ tanh
−1
b
a
_
− b, where a
2
=
F
β
+
α
2
4β
2
, b=
α
2β
.
2. What is the maximum velocity that the boat can reach?
A
NS v
max
=a−b
3.7 A mass m moves up a slope with an initial velocity of 10 m/sec as shown in
the following ﬁgure. Suppose the coefﬁcient of friction between the mass and slope
is 0.25.
30
m
x
,
v
,
a
1. Determine the largest displacement x
max
that the mass can reach and the time
it takes.
A
NS x
max
=7.12 m, t =1.42 sec
2. Determine the time it takes for the mass to return to its original position and
the corresponding velocity.
A
NS v =6.29 m/sec, t =3.69 sec
3.8 A skier skis from rest on a slope with θ =30
◦
from point A at t =0 as shown
in the following ﬁgure. The skier is clocked t
1
=3.61 sec at the 25m checkpoint
and t
2
=5.30 sec at the 50m checkpoint. The length of the slope is AB=L=100
m. The wind resistance is proportional to the speed of the skier, i.e., R=kv.
1. Find the coefﬁcient of wind resistance k and the coefﬁcient of kinetic friction
μ between the snow and the skis.
A
NS k =10.395 N· sec/m, μ=0.041
2. Determine the time t
3
that it takes for the skier toreachthe bottomof the slope
B and the corresponding speed v
3
.
A
NS t
3
=7.92 sec, v
3
=21.20 m/sec
Problems 133
θ
A
L
B
3.9 Auniformchain of length 2L with mass density per unit length ρ is laid on a
rough inclined surface with y =L at t =0 as shown in the following ﬁgure.
The coefﬁcients of static and kinetic friction between the chain and the surface
have inclined the same value μ. The chain starts to drop from rest at time t =0.
Show that the relationship between the velocity of the chain v and y is given by
v =
_
(1+sin θ +μcos θ)g
2L
( y
2
−L
2
) − 2(sin θ +μcos θ)g ( y−L).
Time t =0 Time t
L
L
θ
y(t)
y, y, y
θ
3.10 A uniform chain of length L with mass density per unit length ρ is laid on
a rough inclined surface with y =0 at t =0 as shown in the following ﬁgure.
L−y
y(t)
y, y, y
θ
The coefﬁcients of static and kinetic friction between the chain and the surface
have the same value μ. The chain is release from rest at time t =0. Show that the
relationship between the velocity of the chain v and y is given by
v =
_
2g y
_
(sin θ −μcos θ) +
1−sin θ +μcos θ
2L
y
_
.
134 3 applications of ﬁrstorder and simple higherorder equations
3.11 One end of a pile of uniform chain is hung on a small smooth pulley of
negligible size with y(t) =0 and v(t) = ˙ y(t) =0 when t =0, as shown here.
h
y, y, y
y(t)
The chain starts to fall at time t =0 and pulls the remaining links. The links on
the support, which are initially at rest, acquire the velocity of the chain suddenly
without any resistance or interference. Show that the velocity v as a function of y
is given by
v =
y
2h+y
_
2g
_
h+
y
3
_
.
3.12 Shown in the following ﬁgure is an experimental race car propelled by a
rocket motor. The drag force (air resistance) is given by R=βv
2
. The initial mass
of the car, which includes fuel of mass m
f
, is m
0
. The rocket motor is burning fuel
at the rate of q with an exhaust velocity of u relative to the car. The car is at rest at
t =0. Show that the velocity of the car is given by, for 0
t T,
v(t) = μ·
1 −
_
m
m
0
_
2βμ
q
1 +
_
m
m
0
_
2βμ
q
, m=m
0
−qt, μ
2
=
qu
β
,
T =m
f
/q is the time when the fuel is burnt out.
For m
0
=900 kg, m
f
=450 kg, q=15 kg/sec, u=500 m/sec, β =0.3, what is
the burnout velocity of the car?
A
NS v
f
=555.2 km/hr
Problems 135
3.13 Cable OA supports a structure of length L as shown in the following ﬁgure.
The structure applies a trapezoidally distributed load on the cable through the
hangers. The load intensity is w
0
at the left end and w
1
at the right end. Cable OA
has zero slope at its lowest point O and has sag h. Determine the tensions T
O
=H
at O and T
A
at A.
A
NS H =
L
2
6h
(2w
0
+w
1
), T
A
=
_
H
2
+W(L)
2
, W(L) =
w
0
+w
1
2
L
w
0
w
1
O
y
A
x
h
L
3.14 As a structural engineer, you are asked to design a footbridge across the
Magniﬁcent Gorge, which is 50 m across. The conﬁguration of the bridge is shown
in the following ﬁgure; the lowest point of the cable is 20 m below the left support
and 10 mbelowthe right support. Both supports are anchored on the cliffs. The ef
fective loadof the footbridge is assumedtobe uniformwithanintensity of 500 N/m.
Determine the maximumand minimumtensions in the cable, and the length of the
cable.
A
NS T
min
=H =10723 N, T
max
=T
A
=18151 N, Length = 60.36 m
Lowest
point
h
L
w
A
B
h
3.15 Consider the moving cable AB of a ski lift between two supporting towers as
shown in Figure 3.17. The cable has a mass of 10 kg/m and carries equally spaced
chairs and passengers, which result in an added mass of 20 kg/m when averaged
over the length of the cable. The cable leads horizontally fromthe supporting guide
wheel at A. Determine the tensions in the cable at A and B, and the length of the
cable between A and B.
A
NS T
A
=H =27387 N, T
B
=33267 N, s =64.24 m
136 3 applications of ﬁrstorder and simple higherorder equations
A
h=20 m
L=60 m
B
Figure 3.17 Ski lift.
3.16 For the circuit in Figure 3.18(a), determine v(t) for t >0.
A
NS v(t) =
R
1
V
0
R
1
+R
2
_
1−e
−t/τ
_
, τ =
CR
1
R
2
R
1
+R
2
3.17 For the circuit in Figure 3.18(b), ﬁnd i(t) for t >0.
A
NS i(t) =e
−t/40
(A)
3.18 For the circuit in Figure 3.18(c), ﬁnd i
L
(t) for t >0.
A
NS i
L
(t) =6+e
−7t
(A)
3.19 For the circuit in Figure 3.18(d), ﬁnd i
L
(t) for t >0.
A
NS i
L
(t) =
V
0
R
1
V
0
C
R
2
R
1
v
t =0
1F
0.5i
40
20
40V
i
t =0
(a) (b)
1H
i
L
14A
12 4
4
t =0
i
L
t =0
V
0
L
R
2
R
1
(c) (d)
Figure 3.18 Firstorder circuits.
Problems 137
3.20 Consider b(x) as the concentration of an inert pollutant (k
1
=0) which is
added to the streamat an outfall such that b=b
0
at x =0 and M=0. Due to the in
crease of drainage area, the discharge increases along the streamas Q=Q
0
(1+γ x),
where Q
0
and γ are constants. Determine the BOD distribution b(x) and the DO
distribution c(x) in the stream with c =c
0
at x =0 and N =0.
A
NS b(x) =
b
0
1+γ x
c(x) =
1
1+γ x
_
(k
2
−γ V)c
s
k
2
_
1−e
−k
2
x/V
_
+ c
0
e
−k
2
x/V
+ γ c
s
x
_
3.21 In a streamof constant Q and V, BODis added uniformly along the stream,
starting at x =0 where b=0. Determine the BOD distribution b(x) and the DO
distribution c(x) in the stream with c =c
0
at x =0 and N =0.
A
NS b(x) =
MV
Qk
1
_
1 − e
−k
1
x/V
_
c(x) = c
s
+ (c
0
−c
s
)e
−k
2
x/V
+
MV
Q
_
e
−k
2
x/V
−1
k
2
+
e
−k
1
x/V
−e
−k
2
x/V
k
2
−k
1
_
3.22 A drop of liquid of initial mass of m
0
falls vertically in air from rest. The
liquid evaporates uniformly, losing mass m
1
in unit time. Suppose the resistance
from air is proportional to the velocity of the drop, i.e., R=kv. Show that the
velocity of the drop v(t) is
v(t) =
g
k−m
1
_
(m
0
−m
1
t) − m
0
_
1−
m
1
m
0
t
_
k/m
1
_
, t <
m
0
m
1
.
3.23 As an engineer, you are asked to design a bridge pier with circular horizontal
crosssections as shown in the following ﬁgure.
y
O
12 m
A
A
0
A
1 B
x
The height of the pier is 12 m. The top crosssection is subjected to a uniformly
distributed pressure with a resultant P =3×10
5
N. It is known that the material of
the pier has a density ρ =2.5×10
4
N/m
3
and an allowable pressure P
A
=3×10
5
N/m
2
. Design the bridge pier with a minimumamount of material and include the
weight of the bridge pier in the calculations.
138 3 applications of ﬁrstorder and simple higherorder equations
1. Note that the surface of the pier is obtained by rotating curve AB about the
xaxis. Determine the equation of curve AB.
2. Determine the areas of the top and bottom crosssections A
0
and A
1
.
A
NS y =
1
√
π
exp
_
x
24
_
, A
0
=1 m
2
, A
1
=2.718 m
2
3.24 A tank in the form of a right circular cone of height H, radius R, with
its vertex below the base is ﬁlled with water as shown in the following ﬁgure. A
hole, having a crosssectional area a at the vertex, causes the water to leak out.
Assume that the leaking ﬂow velocity is v(t) =c
_
2g h(t), where c is the discharge
coefﬁcient.
1. Show that the differential equation governing the height of water level h(t) is
h
3
2
dh
dt
= −
acH
2
A
_
2g, A=πR
2
.
2. Show that the time for the cone to empty is T =
2A
5ac
_
H
2g
.
h(t)
H
R
r
3.25 A conical tank with an open top of radius R has a depth H (Figure 3.19)
is initially empty. Water is added at a rate of q (volume per unit time). Water
evaporates fromthe tank at a rate proportional to the surface area with the constant
of proportionality being k. The volume of water in the tank is V =
1
3
πr
2
h, where
h is the depth of the water as shown. Show that the depth of water h(t) is given by
μ
2
ln
¸
¸
¸
¸
μ+h
μ−h
¸
¸
¸
¸
− h = kt, μ
2
=
H
2
q
πkR
2
.
3.26 A water tank in the form of a right circular cylinder of crosssectional area
A and height H is ﬁlled with water at a rate of Q (volume per unit time) as shown
in Figure 3.20. A hole, having a crosssectional area a at the base, causes the water
to leak out. The leaking ﬂow velocity is v(t) =c
_
2g h(t), where c is the discharge
coefﬁcient. The tank is initially empty. Show that the time T to ﬁll the tank is
T =
2A
b
2
_
−b
√
H + Q ln
Q
Q−b
√
H
_
, b = ac
_
2g.
Problems 139
h(t)
H
R
q
r
h(t)
H
A
Q
a
Figure 3.19 Figure 3.20
3.27 Lake Ontario has a volume of approximately V =1600 km
3
. It is heavily
polluted by a certain pollutant with a concentration of c
0
=0.1%. As a part of pol
lution control, the pollutant concentration of the inﬂow is reduced to c
in
=0.02%.
The inﬂow and outﬂow rates are Q
in
=Q
out
=Q=500 km
3
per year. Suppose that
pollutant from the inﬂow is well mixed with the lake water before leaving the lake.
Howlong will it take for the pollutant concentration in the lake to reduce to 0.05%?
A
NS t = 3.14 years
3.28 A mouse Q runs along the positive yaxis at the constant speed v
1
starting
at the origin. A cat P chases the mouse along curve C at the constant speed v
2
starting at point (1, 0) as shown in the following ﬁgure. At any time instance, line
PQ is tangent to curve C. Determine the equation of curve C.
v
1
v
2
P(x, y)
(1, 0)
Q
C
O
x
y
A
NS y = −
1
2
_
x
μ+1
μ+1
+
x
−(μ−1)
μ−1
_
+
μ
μ
2
−1
, μ =
v
1
v
2
= 1
y =
1
4
(1−x
2
) +
1
2
ln x, v
1
= v
2
4
C H A P T E R
Linear Differential Equations
4.1 General Linear Ordinary Differential Equations
In general, an nthorder linear ordinary differential equation is of the form
a
n
(x)
d
n
y
dx
n
+ a
n−1
(x)
d
n−1
y
dx
n−1
+ · · · + a
1
(x)
dy
dx
+ a
0
(x) y = F(x), (1)
in which the dependent variable y and its derivatives of various orders
y,
dy
dx
,
d
2
y
dx
2
, . . . ,
d
n
y
dx
n
appear linearly in the differential equation. The coefﬁcients a
0
(x), a
1
(x), . . . , a
n
(x)
and the righthand side F(x) are functions of x only.
If the coefﬁcients a
0
, a
1
, . . . , a
n
are constants, then equation (1) is a linear
ordinary differential equation with constant coefﬁcients.
The DOperator
The Doperator is deﬁned as
Dy ≡
dy
dx
, Dy is taking firstorder derivative of y w.r.t. x.
D
2
y = D(Dy) =
d
2
y
dx
2
, D
2
y is taking secondorder derivative of y w.r.t. x.
· · · · · ·
D
n
y =
d
n
y
dx
n
, n is a positive integer.
D
n
y is taking nthorder derivative of y
w.r.t. x.
140
4.1 general linear ordinary differential equations 141
Hence the Doperator is a differential operator; applying the Doperator on func
tion f (x) means differentiating f (x) with respect to x, i.e.,
Df (x) =
df (x)
dx
.
Properties of the DOperator
The following properties of the Doperator can be easily veriﬁed:
(1) D
_
y
1
(x)+y
2
(x)
_
=
d
dx
( y
1
+y
2
) =
dy
1
dx
+
dy
2
dx
= Dy
1
+ Dy
2
;
(2) D
_
c y(x)
_
=
d
dx
(c y) = c
dy
dx
= c Dy, c = constant;
(3) D
_
c
1
y
1
(x)+c
2
y
2
(x)
_
= c
1
Dy
1
+ c
2
Dy
2
, c
1
, c
2
= constants.
Hence, D is a linear operator. Using the Doperator, the general nthorder linear
differential equation (1) can be rewritten as
_
a
n
(x)D
n
+a
n−1
(x)D
n−1
+· · · +a
1
(x)D +a
0
(x)
_
y = F(x) =⇒ φ(D) y =F(x),
where φ(D) is an operator given by
φ(D) = a
n
(x)D
n
+ a
n−1
(x)D
n−1
+ · · · + a
1
(x)D + a
0
(x) =
n
r=0
a
r
(x)D
r
.
The operator φ(D) is a linear operator, since
φ(D)( y
1
+y
2
) = φ(D) y
1
+ φ(D) y
2
,
φ(D)(c y) = c φ(D) y, c = constant,
φ(D)(c
1
y
1
+c
2
y
2
) = c
1
φ(D) y
1
+ c
2
φ(D) y
2
, c
1
, c
2
= constants.
Example 4.1 4.1
Rewrite the following differential equations using the Doperator:
(1) 6x
2
d
2
y
dx
2
+ 2x
dy
dx
− 3y = x
3
e
2x
;
(2) 5
d
3
x
dt
3
+ 2
d
2
x
dt
2
−
dx
dt
+ 7x = 3 sin 8t.
(1) (6x
2
D
2
+2xD−3) y = x
3
e
2x
, D≡
d
dx
; The independent variable is x.
(2) (5D
3
+2D
2
−D+7)x =3 sin 8t, D≡
d
dt
. The independent variable is t.
142 4 linear differential equations
Fundamental Theorem
Let y =u(x) be any solution of the differential equation
φ(D) y = F(x), (2)
where φ(D) =a
n
(x)D
n
+a
n−1
(x)D
n−1
+· · · +a
1
(x)D+a
0
(x), and y =v(x) be
a solution of the complementary differential equation, or the homogeneous differ
ential equation,
φ(D) y = 0, (3)
which is obtained by setting the righthand side of equation (2) to zero. Then
y =u(x)+v(x) is also a solution of equation (2).
Proof: Since u(x) and v(x) are solutions of differential equations (2) and (3),
respectively,
φ(D)u(x) = F(x), φ(D)v(x) = 0.
Adding these two equations yields
φ(D)u(x) + φ(D)v(x) = F(x) + 0.
Since φ(D) is a linear operator, one has
φ(D)
_
u(x)+v(x)
_
= φ(D)u(x) + φ(D)v(x).
Hence,
φ(D)
_
u(x)+v(x)
_
= F(x),
i.e., y =u(x)+v(x) is a solution of differential equation (2).
Procedure for Finding the General Solution
1. Find a particular solution y
P
(x) of the original equation (2), i.e.,
φ(D) y
P
= F(x), Particular solution
where the subscript “P” stands for particular solution.
2. Find the general solution y
C
(x) of the complementary differential equation
(3), i.e.,
φ(D) y
C
= 0, Complementary solution
where the subscript “C” stands for complementary solution.
3. Add y
P
(x) and y
C
(x) to obtain the general solution y(x), i.e.,
y(x) = y
C
(x) + y
P
(x). General solution
4.2 complementary solutions 143
In the following sections, various methods for determining complementary so
lutions and particular solutions for linear differential equations with constant coef
ﬁcients are studied in detail.
4.2 Complementary Solutions
The complementary differential equation is obtained by setting the righthand side
of the differential equation to zero, i.e.,
φ(D) y = 0, φ(D) =a
n
D
n
+a
n−1
D
n−1
+ · · · +a
1
D+a
0
,
where a
0
, a
1
, . . . , a
n
are constants.
4.2.1 Characteristic Equation Having Real Distinct Roots
Motivating Example
Consider a linear ﬁrstorder differential equation of the form
dy
dx
+ ay = 0 or (D+a) y = 0, a = constant.
The equation is of the form
dy
dx
+ P(x) · y = Q(x), P(x) =a, Q(x) =0,
and, using the result in Section 2.4.1, the solution is
y = e
−
_
P(x)dx
_
_
Q(x)e
_
P(x)dx
dx + C
_
= Ce
−ax
,
i.e., the solution is of the form e
λx
with λ= −a.
Since the solutionof the linear ﬁrstorder differential equationinthe above example
is y =Ce
−ax
, one is tempted to try a solution of the form y =e
λx
, where λ is a con
stant to be determined, for the general nthorder differential equation φ(D) y =0.
It is easy to verify that
Dy = De
λx
=
d
dx
(e
λx
) = λe
λx
,
D
2
y = D
2
e
λx
= D(De
λx
) = λ
2
e
λx
,
· · · · · ·
D
n
y = D
n
e
λx
= λ
n
e
λx
.
144 4 linear differential equations
Substituting into the differential equation φ(D) y =0 results in
_
a
n
λ
n
+ a
n−1
λ
n−1
+ · · · + a
1
λ + a
0
_
e
λx
= 0.
Since e
λx
=0, one must have
a
n
λ
n
+ a
n−1
λ
n−1
+ · · · + a
1
λ + a
0
= 0. Characteristic equation
This algebraic equation, called the characteristic equation or auxiliary equation,
will give n roots λ
1
, λ
2
, . . . , λ
n
, which are called characteristic numbers and can be
real or complex.
If the roots λ
1
, λ
2
, . . . , λ
n
are distinct, there will be n solutions of the form
e
λ
1
x
, e
λ
2
x
, . . . , e
λ
n
x
.
Since φ(D) y =0 is a linear differential equation, the sum of two solutions is also a
solution. Hence
y = C
1
e
λ
1
x
+ C
2
e
λ
2
x
+ · · · + C
n
e
λ
n
x
is also a solution, where C
1
, C
2
, . . . , C
n
are arbitrary constants.
It is known that the general solution of an nthorder differential equation
φ(D) y =0 should contain n arbitrary constants. Since the equation above con
tains n arbitrary constants, the solution is therefore the general solution of the
complementary equation φ(D) y =0, i.e.,
y
C
= C
1
e
λ
1
x
+ C
2
e
λ
2
x
+ · · · + C
n
e
λ
n
x
,
where λ
1
, λ
2
, . . . , λ
n
are distinct.
Procedure for Finding the Complementary Solution
1. For the nthorder linear differential equation
φ(D) y = F(x) or
_
a
n
D
n
+a
n−1
D
n−1
+ · · · +a
1
D+a
0
_
y = F(x),
set the righthand side to zero to obtain the complementary equation
φ(D) y = 0 or
_
a
n
D
n
+a
n−1
D
n−1
+ · · · +a
1
D+a
0
_
y = 0.
2. Replace D by λ to obtain the characteristic equation
φ(λ) = 0 or a
n
λ
n
+ a
n−1
λ
n−1
+ · · · + a
1
λ + a
0
= 0.
3. Solve the characteristic equation, which is an algebraic equation, to ﬁnd the
characteristic numbers (roots) λ
1
, λ
2
, . . . , λ
n
.
4. Write the complementary solution y
C
using the characteristic numbers λ
1
,
λ
2
, . . . , λ
n
.
4.2 complementary solutions 145
Hyperbolic Functions
The hyperbolic functions are deﬁned as
sinh x =
e
x
− e
−x
2
= −sinh(−x), cosh x =
e
x
+ e
−x
2
= cosh(−x),
tanh x =
sinh x
cosh x
, coth x =
cosh x
sinh x
, cschx =
1
sinh x
, sechx =
1
cosh x
.
–10
–8
–6
–4
–2
0
2
4
6
8
10
–3 –2
–1
3 2 1
cosh x
sinh x
x
Analogous to the trigonometric identities, the hyperbolic functions satisfy many
identities
cosh
2
x − sinh
2
x = 1, sinh 2x = 2 sinh x cosh x,
cosh 2x = cosh
2
x + sinh
2
x = 2 cosh
2
x − 1 = 1 + 2 sinh
2
x,
cosh x + sinh x = e
x
, cosh x − sinh x = e
−x
.
The derivatives and the integrals of hyperbolic sine and cosine functions are
d
dx
(sinh x) = cosh x,
d
dx
(cosh x) = sinh x,
_
sinh xdx = cosh x,
_
cosh xdx = sinh x.
Animportant applicationof hyperbolic functions is inthe complementary solutions
of linear ordinary differential equations. If the characteristic equation has roots
λ= ±β, the corresponding complementary solution is
y
C
= C
1
e
−βx
+ C
2
e
βx
= C
1
(cosh βx−sinh βx) + C
2
(cosh βx+sinh βx)
= (C
1
+C
2
) cosh βx + (C
2
−C
1
) sinh βx,
λ = ±β =⇒ y
C
= Acosh βx + Bsinh βx.
146 4 linear differential equations
Example 4.2 4.2
Solve (D
4
− 13D
2
+ 36) y = 0.
The characteristic equation is φ(λ) =0, i.e.,
λ
4
− 13λ
2
+ 36 = 0,
which is a quadratic equation in λ
2
and can be factorized as
(λ
2
− 4)(λ
2
− 9) = 0.
The roots are λ
2
= 4, 9 =⇒ λ = ±2, ±3. The complementary solution is
y
C
= C
1
e
−3x
+ C
2
e
−2x
+ C
3
e
2x
+ C
4
e
3x
.
The complementary solution can also be written as
y
C
= A
1
cosh 2x + B
1
sinh 2x + A
2
cosh 3x + B
2
sinh 3x.
Example 4.3 4.3
Solve (D
3
+ 4 D
2
+ D − 6) y = 0.
The characteristic equation is φ(λ) =0, i.e.,
λ
3
+ 4λ
2
+ λ − 6 = 0.
By trialanderror, ﬁnd a root of the characteristic equation
1
3
+ 4×1
2
+ 1 − 6 = 0 =⇒ λ=1 is a root =⇒ (λ−1) is a factor.
Use long division to ﬁnd the other factor as follows
λ
2
+ 5λ + 6
λ − 1
¸
¸
¸ λ
3
+ 4λ
2
+ λ − 6
λ
3
− λ
2
(−
5λ
2
+ λ − 6
5λ
2
− 5λ (−
6λ − 6
6λ − 6 (−
0
The characteristic equation becomes
(λ − 1)(λ
2
+ 5λ + 6) = 0 =⇒ (λ − 1)(λ + 2)(λ + 3) = 0.
The three roots are λ = −3, −2, +1, and the complementary solution is
y
C
= C
1
e
−3x
+ C
2
e
−2x
+ C
3
e
x
.
4.2 complementary solutions 147
4.2.2 Characteristic Equation Having Complex Roots
The following example is considered to derive the general result.
Example 4.4 4.4
Solve (D
2
+ 2D + 3) y = 0.
The characteristic equation is φ(λ) =0, i.e., λ
2
+2λ+3=0, and its roots are
λ =
−2 ±
√
2
2
− 4×1×3
2
= −1 ±
√
−2 = −1 ± i
√
2. i =
√
−1
The complementary solution is
y
C
= C
1
e
(−1−i
√
2)x
+ C
2
e
(−1+i
√
2)x
= e
−x
_
C
1
e
−i
√
2x
+ C
2
e
i
√
2x
_
Using Euler’s formula
e
i θ
= cos θ + i sin θ, e
−i θ
= cos θ − i sin θ,
the solution y
C
can be written as
y
C
= e
−x
_
C
1
_
cos
√
2x − i sin
√
2x
_
+ C
2
_
cos
√
2x + i sin
√
2x
__
= e
−x
_
(C
1
+C
2
) cos
√
2x + i (−C
1
+C
2
) sin
√
2x
_
.
Since the differential equationis real, its solutionmust be real. Hence the coefﬁcients
of (cos
√
2x) and (sin
√
2x) must be real:
C
1
+C
2
=2a, i (−C
1
+C
2
) =2b =⇒ −C
1
+C
2
= −i 2b, a, b real constants,
or
C
1
= a + i b, C
2
= a − i b,
i.e., C
1
and C
2
are complex conjugate, C
1
=
¯
C
2
.
The complementary solution becomes
y
C
= e
−x
_
2a cos
√
2x + 2b sin
√
2x
_
= e
−x
_
Acos
√
2x + Bsin
√
2x
_
, A=2a, B=2b,
where A and B are real constants.
In general, if the characteristic equation has a pair of complex roots λ=α±i β,
where α and β are real, then the complementary solution is
λ = α ± i β =⇒ y
C
(x) = e
αx
(Acos βx + Bsin βx),
where A and B are real constants.
148 4 linear differential equations
Example 4.5 4.5
Solve (D
3
+ 2D
2
+ 9D + 18) y = 0.
The characteristic equation is φ(λ) =0, i.e.,
λ
3
+ 2λ
2
+ 9λ + 18 = 0.
Try to ﬁnd a root of the characteristic equation
∵ (−2)
3
+ 2×(−2)
2
+ 9×(−2) + 18 = −8 + 8 − 18 + 18 = 0,
∴ λ= −2 is a root =⇒ (λ+2) is a factor.
Use long division to ﬁnd the other factor
λ
2
+ 9
λ + 2
¸
¸
¸ λ
3
+ 2λ
2
+ 9λ + 18
λ
3
+ 2λ
2
(−
9λ + 18
9λ + 18 (−
0
The characteristic equation becomes
(λ + 2)(λ
2
+ 9) = 0.
The three roots are λ= −2, ±i 3, and the complementary solution is
y
C
= Ce
−2x
+ e
0· x
_
Acos 3x + Bsin 3x
_
= Ce
−2x
+ Acos 3x + Bsin 3x.
Before more challenging problems can be studied, the following formula for
evaluating the nth roots of complex numbers is reviewed.
Review of Complex Numbers
Given that λ
n
=a ± i b, n is a positive integer, it is required to ﬁnd λ. The complex
numbers can be converted from the rectangular form to the polar form:
λ
n
= a ± i b Rectangular form of complex numbers
=
_
a
2
+b
2
_
a
√
a
2
+b
2
± i
b
√
a
2
+b
2
_
. ,, . . ,, . . ,, .
r cos θ sin θ
4.2 complementary solutions 149
= r (cos θ ± i sin θ), r =
_
a
2
+b
2
, θ = tan
−1
b
a
Polar form
= r e
±i θ
. Euler's formula
Since cos θ and sin θ are periodic functions of period 2π,
λ
n
= r
_
cos(2kπ+θ) ± i sin(2kπ+θ)
_
= r e
±i (2kπ+θ)
.
Hence
λ =
n
√
r e
±i (2kπ+θ)/n
,
or
λ =
n
√
r
_
cos
2kπ+θ
n
± i sin
2kπ+θ
n
_
, k = 0, 1, . . . , n−1.
It can be easily shown that when k = pn+q, where p and q are positive integers,
the value of λ repeats that when k =q. Therefore only n values of k =0, 1, . . . ,
n−1 are taken.
Example 4.6 4.6
Solve (D
5
− 1) y = 0.
The characteristic equation is φ(λ) =0, i.e.,
λ
5
− 1 = 0 =⇒ λ
5
= 1 = 1 + i 0 = cos 0 + i sin 0.
The ﬁve roots are given by
λ = cos
2kπ+0
5
+ i sin
2kπ+0
5
, k = 0, 1, . . . , 4,
k =0: λ = cos 0 + i sin 0 = 1,
k =1: λ = cos
2π
5
+ i sin
2π
5
,
k =2: λ = cos
4π
5
+ i sin
4π
5
= cos
_
π −
π
5
_
+ i sin
_
π −
π
5
_
= −cos
π
5
+ i sin
π
5
,
k =3: λ = cos
6π
5
+ i sin
6π
5
= cos
_
π +
π
5
_
+ i sin
_
π +
π
5
_
= −cos
π
5
− i sin
π
5
,
k =4: λ = cos
8π
5
+ i sin
8π
5
= cos
_
2π −
2π
5
_
+ i sin
_
2π −
2π
5
_
= cos
2π
5
− i sin
2π
5
.
150 4 linear differential equations
Hence, the ﬁve roots are
λ = 1, cos
2π
5
± i sin
2π
5
, −cos
π
5
± i sin
π
5
.
The complementary solution is
y
C
= Ce
x
+ exp
_
cos
2π
5
· x
__
A
1
cos
_
sin
2π
5
· x
_
+ B
1
sin
_
sin
2π
5
· x
__
+ exp
_
−cos
π
5
· x
__
A
2
cos
_
sin
π
5
· x
_
+ B
2
sin
_
sin
π
5
· x
__
.
Example 4.7 4.7
Solve (D
4
− 16D
2
+ 100) y = 0.
The characteristic equation is φ(λ) =0, i.e., λ
4
−16λ
2
+100=0, a quadratic equa
tion in λ
2
. Hence
λ
2
=
16 ±
√
16
2
− 4×1×100
2
=
16 ±
√
−144
2
= 8 ± i 6 = 10
_
4
5
± i
3
5
_
= 10
_
cos θ ± i sin θ
_
, cos θ =
4
5
, sin θ =
3
5
.
The roots are given by
λ =
√
10
_
cos
2kπ +θ
2
± i sin
2kπ +θ
2
_
, k = 0, 1,
k =0: λ =
√
10
_
cos
θ
2
± i sin
θ
2
_
,
k =1: λ =
√
10
_
cos
_
π +
θ
2
_
± i sin
_
π +
θ
2
__
=
√
10
_
−cos
θ
2
∓ i sin
θ
2
_
.
The four roots are
λ =
√
10
_
± cos
θ
2
± i sin
θ
2
_
.
Using the halfangle formula, one obtains
cos
θ
2
=
_
1+cos θ
2
=
_
1+
4
5
2
=
3
√
10
,
sin
θ
2
=
_
1−cos θ
2
=
_
1−
4
5
2
=
1
√
10
,
and hence
λ =
√
10
_
±
3
√
10
± i
1
√
10
_
= ±3 ± i 1.
The complementary solution is
y
C
= e
3x
_
A
1
cos x + B
1
sin x
_
+ e
−3x
_
A
2
cos x + B
2
sin x
_
.
4.2 complementary solutions 151
4.2.3 Characteristic Equation Having Repeated Roots
Example 4.8 4.8
Solve (D
2
− 4D + 4) y = 0.
The characteristic equation is φ(λ) =0, i.e.,
λ
2
− 4λ + 4 = 0 =⇒ (λ − 2)
2
= 0 =⇒ λ = 2, 2. A double root
If the complementary solution is written as
y = C
1
e
2x
+ C
2
e
2x
= (C
1
+C
2
)e
2x
= Ce
2x
, C = C
1
+ C
2
,
then there is only one arbitrary constant. But for a secondorder differential equa
tion, the solution must contain two arbitrary constants. The problem is due to the
fact that the two solutions C
1
e
2x
and C
2
e
2x
are linearly dependent.
Hence, one must seek a second linearly independent solution. Try a solution of
the form y(x) = v(x)e
2x
. Since
Dy = D(ve
2x
) = e
2x
Dv + 2ve
2x
,
D
2
y = D(e
2x
Dv + 2ve
2x
) = e
2x
(D
2
v + 4 Dv + 4v),
substituting in the original equation yields
e
2x
_
(D
2
v + 4 Dv + 4v) − 4(Dv + 2v) + 4v
_
= 0 =⇒ e
2x
D
2
v = 0.
Hence v(x) satisﬁes the differential equation D
2
v = 0 or
d
2
v
dx
2
= 0. Integrating
the equation twice leads to
v(x) = C
0
+ C
1
x.
The solution is then
y(x) = v(x)e
2x
= (C
0
+ C
1
x) e
2x
,
. ,, .
polynomial of degree 1
in which there are two arbitrary constants.
It is seen that, when the characteristic equation has a double root λ=2, the
coefﬁcient of e
2x
is a polynomial of degree 1 instead of a constant. In general, the
following results can be obtained.
152 4 linear differential equations
Summary on the Forms of Complementary Solution
Having obtained the roots of the characteristic equation or the characteristic
numbers, the complementary solution can be easily written.
1. If the characteristic equation φ(λ) =0 has a pfold root λ=a, the corre
sponding solution is
λ = a, . . . , a =⇒ y
C
=
_
C
0
+C
1
x+ . . . +C
p−1
x
p−1
_
e
ax
. ,, . . ,, .
p times polynomial of degree p−1
2. If the characteristic equation φ(λ) =0 has a pair of complex roots
λ=α±i β of pfold, the corresponding solution is
λ=α±i β, . . . , α±i β =⇒ y
C
= e
αx
__
A
0
+A
1
x+ . . . +A
p−1
x
p−1
_
cos βx
+
_
B
0
+B
1
x+ . . . +B
p−1
x
p−1
_
sin βx
_
. ,, . . ,, .
p times polynomials of degree p−1
Example 4.9 4.9
Given that the characteristic equations have the following roots, write the comple
mentary solutions:
1. −3±i 2, −1, 0, 2;
2. 3, 3, 3, 0, 0, 1±i 3;
3. 2±i 5, 2±i 5, 1, 1, 1, 1, 3.
1. y
C
(x) = e
−3x
(Acos 2x + Bsin 2x) + C
1
e
−x
+ C
2
+ C
3
e
2x
.
2. y
C
(x) = (C
0
+ C
1
x + C
2
x
2
)e
3x
Corresponding to λ=3, 3, 3
+ D
0
+ D
1
x Corresponding to λ=0, 0
+ e
x
(Acos 3x + Bsin 3x). Corresponding to λ=1±i 3
3. y
C
(x) = e
2x
_
(A
0
+A
1
x) cos 5x+(B
0
+B
1
x) sin 5x
_
Corresponding to
λ=2±i 5, 2±i 5
+ (C
0
+C
1
x+C
2
x
2
+C
3
x
3
)e
x
Corresponding to λ=1, 1, 1, 1
+ De
3x
. Corresponding to λ=3
Example 4.10 4.10
Solve (D
8
+ 18D
6
+ 81D
4
) y = 0.
4.3 particular solutions 153
The characteristic equation is φ(λ) =0, i.e.,
λ
8
+ 18λ
6
+ 81λ
4
= 0,
which can be factorized as λ
4
(λ
2
+9)
2
=0. Hence the roots are
λ = 0, 0, 0, 0, ±i 3, ±i 3,
and the complementary solution is
y
C
(x) = (C
0
+ C
1
x + C
2
x
2
+ C
3
x
3
)e
0 · x
Corresponding to λ=0, 0, 0, 0
+ e
0 · x
_
(A
0
+A
1
x) cos 3x + (B
0
+B
1
x) sin 3x
_
Corresponding to
λ= ±i 3, ±i 3
= C
0
+ C
1
x + C
2
x
2
+ C
3
x
3
+ (A
0
+A
1
x) cos 3x + (B
0
+B
1
x) sin 3x.
Example 4.11 4.11
Solve (D
7
− 3D
6
+ 4D
5
− 4D
4
+ 3D
3
− D
2
) y = 0.
The characteristic equation is φ(λ) =0, i.e.,
λ
7
− 3λ
6
+ 4λ
5
− 4λ
4
+ 3λ
3
− λ
2
= 0.
Factorizing the lefthand side yields
λ
7
− 3λ
6
+ 4λ
5
− 4λ
4
+ 3λ
3
− λ
2
= λ
2
(λ
5
− 3λ
4
+ 4λ
3
− 4λ
2
+ 3λ − 1)
= λ
2
_
(λ
5
− 3λ
4
+ 3λ
3
− λ
2
) + (λ
3
− 3λ
2
+ 3λ − 1)
_
= λ
2
_
λ
2
(λ − 1)
3
+ (λ − 1)
3
_
= λ
2
(λ − 1)
3
(λ
2
+ 1).
The roots are λ = 0, 0, 1, 1, 1, ±i , and the complementary solution is
y
C
(x) = C
0
+ C
1
x Corresponding to λ=0, 0
+ (D
0
+ D
1
x + D
2
x
2
)e
x
Corresponding to λ=1, 1, 1
+ Acos x + Bsin x. Corresponding to λ= ±i
4.3 Particular Solutions
In this section, three methods, i.e., the method of undetermined coefﬁcients, the
Doperator method, and the method of variation of parameters, will be introduced
for ﬁnding particular solutions of linear ordinary differential equations.
4.3.1 Method of Undetermined Coefﬁcients
The method is illustrated through speciﬁc examples.
154 4 linear differential equations
Example 4.12 4.12
Solve (D
2
+ 9) y = 3e
2x
.
The characteristic equation is λ
2
+9 = 0 =⇒ λ = ±i 3. The complementary
solution is
y
C
= Acos 3x + Bsin 3x.
To ﬁnd a particular solution y
P
, look at the righthand side of the differential
equation, i.e., F(x) =3e
2x
. Since the derivatives of an exponential function are
exponential functions, one is tempted to try
y
P
= Ce
2x
=⇒ Dy
P
= 2Ce
2x
, D
2
y
P
= 4Ce
2x
,
where C is a constant to be determined.
Substituting into the equation yields
4Ce
2x
+ 9· Ce
2x
= 3e
2x
=⇒ 13Ce
2x
= 3e
2x
.
Comparing the coefﬁcients of e
2x
leads to
13C = 3 =⇒ C =
3
13
=⇒ y
P
=
3
13
e
2x
.
The general solution is
y(x) = y
C
+ y
P
= Acos 3x + Bsin 3x +
3
13
e
2x
.
Table 4.1 Method of undetermined coefﬁcients.
Corresponding to Righthand Side F(x) Assumed Form of y
P
(1) Polynomial of degree k Polynomial of degree k
(2) e
αx
Ce
αx
(3) sin βx, cos βx Acos βx + Bsin βx
(4) e
αx
_
a
0
+a
1
x+ · · · +a
k
x
k
_
cos βx, e
αx
__
A
0
+A
1
x+ · · · +A
k
x
k
_
cos βx
e
αx
_
b
0
+b
1
x+ · · · +b
k
x
k
_
sin βx +
_
B
0
+B
1
x+ · · · +B
k
x
k
_
sin βx
_
. ,, . . ,, .
Polynomial of degree k Polynomial of degree k
❧ The essence of the methodof undeterminedcoefﬁcients is to assume a formfor a
particular solution, with coefﬁcients to be determined, according to the formof the
righthand side of the differential equation. The coefﬁcients are then determined
by substituting the assumed particular solution into the differential equation. In
general, one uses the results given in Table 4.1 to assume the form of a particular
solution.
4.3 particular solutions 155
Remarks: For Cases (3) and (4), even if the righthand side has only sine or
cosine function, the assumed form of a particular solution must contain both the
sine and cosine functions.
Example 4.13 4.13
Solve (D
2
+ 3D + 2) y = 42e
5x
+ 390 sin 3x + 8x
2
− 2.
The characteristic equation is λ
2
+3λ+2 = 0. Factorizing the lefthand side gives
(λ + 1)(λ + 2) = 0 =⇒ λ = −2, −1.
The complementary solution is
y
C
= C
1
e
−2x
+ C
2
e
−x
.
Use the method of undetermined coefﬁcients to ﬁnd y
P
Corresponding to righthand side F(x) Assumed form for y
P
42e
5x
Ce
5x
390 sin 3x Acos 3x + Bsin 3x
8x
2
− 2 D
2
x
2
+ D
1
x + D
0
Hence, assume a particular solution of the form
y
P
= Ce
5x
+ Acos 3x + Bsin 3x + D
2
x
2
+ D
1
x + D
0
.
Differentiating y
P
yields
Dy
P
= 5Ce
5x
− 3Asin 3x + 3Bcos 3x + 2D
2
x + D
1
,
D
2
y
P
= 25Ce
5x
− 9Acos 3x − 9Bsin 3x + 2D
2
.
Substituting into the differential equation leads to
(D
2
+ 3D + 2) y
P
= (25Ce
5x
− 9Acos 3x − 9Bsin 3x + 2D
2
) D
2
y
P
+ 3(5Ce
5x
− 3Asin 3x + 3Bcos 3x + 2D
2
x + D
1
) 3Dy
P
+ 2(Ce
5x
+ Acos 3x + Bsin 3x + D
2
x
2
+ D
1
x + D
0
) 2y
P
= 42Ce
5x
+ (−7A + 9B) cos 3x + (−9A − 7B) sin 3x
+ 2D
2
x
2
+ (2D
1
+ 6D
2
)x + (2D
0
+ 3D
1
+ 2D
2
)
Collecting
similar terms
= 42e
5x
+ 390 sin 3x + 8x
2
− 2. Righthand side of the DE
156 4 linear differential equations
Equating coefﬁcients of corresponding terms results in
e
5x
: 42C = 42 =⇒ C = 1,
cos 3x :
sin 3x :
−7A + 9B = 0
−9A − 7B = 390
_
=⇒
_
A = −27,
B = −21,
x
2
: 2D
2
= 8 =⇒ D
2
= 4,
x : 2D
1
+ 6D
2
= 0 =⇒ D
1
= −3D
2
= −12,
1: 2D
0
+ 3D
1
+ 2D
2
= −2 =⇒ D
0
=
1
2
(−2−3D
1
−2D
2
) = 13.
Hence, a particular solution is
y
P
= e
5x
− 27 cos 3x − 21 sin 3x + 4x
2
− 12x + 13.
Example 4.14 4.14
Solve (D
2
+ 2D + 3) y = 34e
x
cos 2x + 1331x
2
e
2x
.
The characteristic equation is
λ
2
+2λ+3 = 0 =⇒ λ =
−2 ±
√
2
2
− 4×1×3
2
= −1 ± i
√
2.
The complementary solution is
y
C
= e
−x
(C
1
cos
√
2x + C
2
sin
√
2x).
Use the method of undetermined coefﬁcients to ﬁnd y
P
Corresponding to righthand side F(x) Assumed form for y
P
34e
x
cos 2x e
x
(Acos 2x + Bsin 2x)
1331x
2
e
2x
(D
2
x
2
+ D
1
x + D
0
)e
2x
Hence, assume a particular solution of the form
y
P
= e
x
(Acos 2x + Bsin 2x) + (D
2
x
2
+ D
1
x + D
0
)e
2x
.
Differentiating y
P
with respect to x yields
Dy
P
= e
x
_
(A + 2B) cos 2x + (−2A + B) sin 2x
_
+
_
2D
2
x
2
+ (2D
1
+ 2D
2
)x + (2D
0
+ D
1
)
_
e
2x
,
D
2
y
P
= e
x
_
(−3A + 4B) cos 2x + (−4A − 3B) sin 2x
_
+
_
4D
2
x
2
+ (4D
1
+ 8D
2
)x + (4D
0
+ 4D
1
+ 2D
2
)
_
e
2x
.
4.3 particular solutions 157
Substituting into the differential equation leads to
(D
2
+ 2D + 3) y
P
= e
x
_
(−3A + 4B) cos 2x + (−4A − 3B) sin 2x
_
+
_
4D
2
x
2
+ (4D
1
+ 8D
2
)x + (4D
0
+ 4D
1
+ 2D
2
)
_
e
2x
D
2
y
P
+ 2
_
e
x
_
(A + 2B) cos 2x + (−2A + B) sin 2x
_
+
_
2D
2
x
2
+ (2D
1
+ 2D
2
)x + (2D
0
+ D
1
)
_
e
2x
_
2Dy
P
+ 3
_
e
x
(Acos 2x + Bsin 2x) + (D
2
x
2
+ D
1
x + D
0
)e
2x
_
3y
P
= e
x
_
(2A + 8B) cos 2x + (−8A + 2B) sin 2x
_
Collecting similar terms
+
_
11D
2
x
2
+ (11D
1
+ 12D
2
)x + (11D
0
+ 6D
1
+ 2D
2
)
_
e
2x
= 34e
x
cos 2x + 1331x
2
e
2x
. Righthand side of the DE
Equating coefﬁcients of corresponding terms results in
e
x
cos 2x :
e
x
sin 2x :
2A + 8B = 34
−8A + 2B = 0
_
=⇒
_
A = 1,
B = 4,
x
2
e
2x
: 11D
2
= 1331 =⇒ D
2
= 121,
xe
2x
: 11D
1
+ 12D
2
= 0 =⇒ D
1
= −
12
11
D
2
= −132,
e
2x
: 11D
0
+ 6D
1
+ 2D
2
= 0 =⇒ D
0
= −
1
11
(6D
1
+2D
2
) = 50.
Hence, a particular solution is
y
P
= e
x
(cos 2x + 4 sin 2x) + (121x
2
− 132x + 50)e
2x
.
Example 4.15 4.15
Solve (D
2
+ 8D + 25) y = 4 cos
3
2x.
The characteristic equation is λ
2
+8λ+25 = 0. The roots are
λ =
−8 ±
√
8
2
− 4×1×25
2
= −4 ± i 3.
The complementary solution is
y
C
= e
−4x
(Acos 3x + Bsin 3x).
The righthand side of the equation is converted to the standard form using
trigonometric identities:
4 cos
3
2x = 4 cos 2x cos
2
2x
158 4 linear differential equations
= 2 cos 2x(1 + cos 4x) = 2 cos 2x + 2 cos 2x cos 4x cos
2
A=
1+cos 2A
2
= 2 cos 2x + cos 6x + cos 2x 2 cos Acos B = cos(A+B) + cos(A−B)
= cos 6x + 3 cos 2x.
Use the method of undetermined coefﬁcients to ﬁnd y
P
Corresponding to righthand side F(x) Assumed form for y
P
cos 6x A
1
cos 6x + B
1
sin 6x
3 cos 2x A
2
cos 2x + B
2
sin 2x
Hence, assume a particular solution of the form
y
P
= A
1
cos 6x + B
1
sin 6x + A
2
cos 2x + B
2
sin 2x.
Differentiating y
P
with respect to x yields
Dy
P
= −6A
1
sin 6x + 6B
1
cos 6x − 2A
2
sin 2x + 2B
2
cos 2x
D
2
y
P
= −36A
1
cos 6x − 36B
1
sin 6x − 4A
2
cos 2x − 4B
2
sin 2x.
Substituting into the differential equation leads to
(D
2
+ 8D + 25) y
P
= −36A
1
cos 6x − 36B
1
sin 6x − 4A
2
cos 2x − 4B
2
sin 2x D
2
y
P
+ 8(−6A
1
sin 6x + 6B
1
cos 6x − 2A
2
sin 2x + 2B
2
cos 2x) 8Dy
P
+ 25(A
1
cos 6x + B
1
sin 6x + A
2
cos 2x + B
2
sin 2x) 25y
P
= (−11A
1
+48B
1
) cos 6x + (−48A
1
−11B
1
) sin 6x
+ (21A
2
+16B
2
) cos 2x + (−16A
2
+21B
2
) sin 2x
Collecting
similar terms
= cos 6x + 3 cos 2x. Righthand side of the DE
Equating coefﬁcients of corresponding terms results in
cos 6x :
sin 6x :
−11A
1
+ 48B
1
= 1
−48A
1
− 11B
1
= 0
_
=⇒
⎧
⎨
⎩
A
1
= −
11
2425
,
B
1
=
48
2425
,
cos 2x :
sin 2x :
21A
2
+ 16B
2
= 3
−16A
2
+ 21B
2
= 0
_
=⇒
⎧
⎨
⎩
A
2
=
63
697
,
B
2
=
48
697
.
Hence, a particular solution is
y
P
= −
11
2425
cos 6x +
48
2425
sin 6x +
63
697
cos 2x +
48
697
sin 2x.
4.3 particular solutions 159
Exceptions in the Method of Undetermined Coefﬁcients
The method fails if the righthand side of the differential equation is already con
tained in the complementary solution y
C
.
Example 4.16 4.16
Solve (D
2
+ 2D) y = 4x
2
+ 2x + 3.
The characteristic equation is λ
2
+2λ = 0 =⇒ λ = 0, −2. The complementary
solution is
y
C
= C
0
+ C
1
e
−2x
.
Apply the method of undetermined coefﬁcients to ﬁnd a particular solution y
P
.
Corresponding tothe righthandside 4x
2
+2x+3, the formof y
P
wouldnormally
be assumed as D
2
x
2
+D
1
x+D
0
. However
(D
2
+ 2D)(D
2
x
2
+ D
1
x + D
0
) = 2D
2
+ 2(2D
2
x + D
1
) = 4x
2
+ 2x + 3.
This is because the constant “3” in the righthand side is already contained in the
complementary solution C
0
. Hence, one has to assume the form of a particular
solution to be
y
P
= x(D
2
x
2
+ D
1
x + D
0
).
Differentiating y
P
with respect to x yields
Dy
P
= 3D
2
x
2
+ 2D
1
x + D
0
, D
2
y
P
= 6D
2
x + 2D
1
.
Substituting into the differential equation leads to
(D
2
+ 2D) y
P
= 6D
2
x + 2D
1
+ 2(3D
2
x
2
+ 2D
1
x + D
0
)
= 6D
2
x
2
+ (4D
1
+ 6D
2
)x + (2D
0
+ 2D
1
)
= 4x
2
+ 2x + 3. Righthand side of the DE
Equating coefﬁcients of corresponding terms results in
x
2
: 6D
2
= 4 =⇒ D
2
=
2
3
,
x : 4D
1
+ 6D
2
= 2 =⇒ D
1
=
1
4
(2 − 6D
2
) = −
1
2
,
1: 2D
0
+ 2D
1
= 3 =⇒ D
0
=
1
2
(3 − 2D
1
) = 2.
Hence, a particular solution is
y
P
= x
_
2
3
x
2
−
1
2
x + 2
_
.
160 4 linear differential equations
Exceptions in the Method of Undetermined Coefﬁcients
In general, if any of the normally assumed terms of a particular solution occurs
in the complementary solution, one must multiply these assumed terms by a
power of x which is sufﬁciently high, but not higher, so that none of these
assumed terms occur in the complementary solution.
Example 4.17 4.17
Given the complementary solution y
C
and the righthand side F(x) of the differ
ential equation, specify the form of a particular solution y
P
using the method of
undetermined coefﬁcients.
(1) y
C
= c
1
e
−x
+ c
2
e
3x
+ (d
0
+d
1
x+d
2
x
2
)e
5x
,
F(x) = 3e
−x
+ 6e
2x
− 4e
5x
;
(2) y
C
= e
2x
(a cos 3x + b sin 3x) + c
0
+c
1
x+c
2
x
2
,
F(x) = 5xe
2x
cos 3x + 3x + e
2x
;
(3) y
C
= (c
0
+c
1
x)e
x
+ d
0
+ a sin 2x+b cos 2x,
F(x) = 2xe
x
+ 3x
2
+ cos 3x.
(1)
Normally Assumed Contained in
F(x) Form for y
P
y
C
Modiﬁcation
3e
−x
C
1
e
−x
c
1
e
−x
x · C
1
e
−x
6e
2x
C
2
e
2x
———— ————
−4e
5x
C
3
e
5x
(d
0
+d
1
x+d
2
x
2
)e
5x
x
3
· C
3
e
5x
y
P
= x · C
1
e
−x
+ C
2
e
2x
+ x
3
· C
3
e
5x
.
(2)
Normally Assumed Contained in
F(x) Form for y
P
y
C
Modiﬁcation
5xe
2x
cos 3x e
2x
_
(A
0
+A
1
x) cos 3x e
2x
(a cos 3x x · e
2x
_
(A
0
+A
1
x) cos 3x
+(B
0
+B
1
x) sin 3x
_
+b sin 3x) +(B
0
+B
1
x) sin 3x
_
3x C
0
+C
1
x c
0
+c
1
x+c
2
x
2
x
3
· (C
0
+C
1
x)
e
2x
De
2x
———— ————
y
P
= x · e
2x
_
(A
0
+A
1
x) cos 3x + (B
0
+B
1
x) sin 3x
_
+ x
3
· (C
0
+C
1
x) + De
2x
.
4.3 particular solutions 161
(3)
Normally Assumed Contained in
F(x) Form for y
P
y
C
Modiﬁcation
xe
x
(C
0
+C
1
x)e
x
(c
0
+c
1
x)e
x
x
2
· (C
0
+C
1
x)e
x
3x
2
D
0
+D
1
x+D
2
x
2
d
0
x · (D
0
+D
1
x+D
2
x
2
)
cos 3x Acos 3x + Bsin 3x ———— ————
y
P
= x
2
· (C
0
+C
1
x)e
x
+ x · (D
0
+D
1
x+D
2
x
2
) + Acos 3x + Bsin 3x.
Example 4.18 4.18
Solve (D
4
− 4 D
2
) y = sinh 2x + 2x
2
.
The characteristic equation is φ(λ) =0, i.e.,
λ
4
− 4λ
2
= λ
2
(λ
2
−4) = 0 =⇒ λ = ±2, 0, 0.
The complementary solution is y
C
=c
1
e
2x
+c
2
e
−2x
+d
0
+d
1
x, which can be ex
pressed using hyperbolic functions:
y
C
= a cosh 2x + b sinh 2x + d
0
+ d
1
x.
Remarks: When using the method of undetermined coefficients to find a par
ticular solution, the hyperbolic functions (cosh βx) and (sinh βx) can be treated
similar to the sinusoidal functions (cos βx) and (sin βx).
Normally Assumed Contained in
F(x) Form for y
P
y
C
Modiﬁcation
sinh 2x Acosh 2x+Bsinh 2x a cosh 2x+b sinh 2x x · (Acosh 2x+Bsinh 2x)
2x
2
D
0
+D
1
x+D
2
x
2
d
0
+d
1
x x
2
· (D
0
+D
1
x+D
2
x
2
)
Hence, assume a particular solution of the form
y
P
= x · (Acosh 2x+Bsinh 2x) + x
2
· (D
0
+D
1
x+D
2
x
2
).
Differentiating y
P
with respect to x yields
Dy
P
= (Acosh 2x+Bsinh 2x) + 2x(Asinh 2x+Bcosh 2x)
+ 2D
0
x+3D
1
x
2
+4D
2
x
3
,
D
2
y
P
= 4(Asinh 2x+Bcosh 2x) + 4x(Acosh 2x+Bsinh 2x)
+ 2D
0
+6D
1
x+12D
2
x
2
,
D
3
y
P
= 12(Acosh 2x+Bsinh 2x) + 8x(Asinh 2x+Bcosh 2x) + 6D
1
+24D
2
x,
162 4 linear differential equations
D
4
y
P
= 32(Asinh 2x+Bcosh 2x) + 16x(Acosh 2x+Bsinh 2x) + 24D
2
.
Substituting into the differential equation leads to
(D
4
− 4D
2
) y
P
= 32(Asinh 2x+Bcosh 2x) + 16x(Acosh 2x+Bsinh 2x) + 24D
2
D
4
y
P
− 4
_
4(Asinh 2x+Bcosh 2x) + 4x(Acosh 2x+Bsinh 2x)
+ 2D
0
+6D
1
x+12D
2
x
2
_
4D
2
y
P
= 16(Asinh 2x+Bcosh 2x) + (24D
2
−8D
0
)−24D
1
x−48D
2
x
2
= sinh 2x + 2x
2
. Righthand side of the DE
Equating coefﬁcients of corresponding terms results in
cosh 2x : 16B = 0 =⇒ B = 0,
sinh 2x : 16A = 1 =⇒ A =
1
16
,
x
2
: −48D
2
= 2 =⇒ D
2
= −
1
24
,
x : −24D
1
= 0 =⇒ D
1
= 0,
1: 24D
2
− 8D
0
= 0 =⇒ D
0
= 3D
2
= −
1
8
.
Hence, a particular solution is
y
P
=
x
16
cosh 2x −
x
2
8
−
x
4
24
.
4.3.2 Method of Operators
In Section 4.1, the Doperator is deﬁned as the differential operator, i.e.,
D(·) ≡
d(·)
dx
.
Deﬁne the inverse operator of D, denoted as D
−1
, by (D
−1
D) y ≡ y. Hence
D
−1
(·) =
1
D
(·) ≡
_
(·)dx,
i.e., the inverse operator D
−1
is the integral operator.
Similarly, for differential equation φ(D) y =F(x), deﬁne the inverse operator
φ
−1
(D) by
_
φ
−1
(D)φ(D)
_
y = y or
_
1
φ(D)
φ(D)
_
y = y.
4.3 particular solutions 163
For the operator φ(D) of the form
φ(D) = a
n
D
n
+ a
n−1
D
n−1
+ · · · + a
1
D + a
0
,
where a
0
, a
1
, . . . , a
n
are constant coefﬁcients, a solution of the differential equation
φ(D) y =F(x) can be rewritten using the inverse operator
φ(D) y = F(x) =⇒ y =
1
φ(D)
F(x),
where
φ
−1
(D) =
1
φ(D)
=
1
a
n
D
n
+ a
n−1
D
n−1
+ · · · + a
1
D + a
0
.
The inverse operator φ
−1
(D) has the following properties:
1.
1
φ
1
(D)φ
2
(D)
F(x) =
1
φ
1
(D)
_
1
φ
2
(D)
F(x)
_
=
1
φ
2
(D)
_
1
φ
1
(D)
F(x)
_
;
2.
1
φ(D)
_
F
1
(x) + F
2
(x)
_
=
1
φ(D)
F
1
(x) +
1
φ(D)
F
2
(x).
Remarks: For the given differential equation φ(D) y =F(x), the operator φ(D)
and function F(x) are unique. However, the function φ
−1
(D)F(x) is not uniquely
determined. The difference between any two results is a solution of the comple
mentary equation φ(D) y =0. But this is not important because only one partic
ular solution is to be determined. In fact, one always tries to find a simple result
of φ
−1
(D)F(x).
Theorem 1
1
φ(D)
e
ax
=
1
φ(a)
e
ax
, provided φ(a) = 0. Replace D by a.
☞
If φ(a) =0, use Theorem 4.
Proof: Since De
ax
=ae
ax
, D
2
e
ax
=a
2
e
ax
, . . . , D
n
e
ax
=a
n
e
ax
, then
φ(D)e
ax
= (a
n
D
n
+ a
n−1
D
n−1
+ · · · + a
1
D + a
0
)e
ax
= (a
n
a
n
+ a
n−1
a
n−1
+ · · · + a
1
a + a
0
)e
ax
= φ(a)e
ax
.
Applying φ
−1
(D) on both sides yields:
φ
−1
(D)φ(D)e
ax
= φ
−1
(D)
_
φ(a)e
ax
_
=⇒ e
ax
= φ(a) φ
−1
(D)e
ax
,
∴
1
φ(D)
e
ax
=
1
φ(a)
e
ax
, φ(a) = 0.
164 4 linear differential equations
Theorem 2 (Shift Theorem)
1
φ(D)
_
e
ax
f (x)
_
= e
ax
1
φ(D+a)
f (x).
Move e
ax
out of φ
−1
(D),
shift the operator D by a.
Proof: Consider the differential equation φ(D) y =F(x). Since
(D+a)
_
e
−ax
y
_
= D
_
e
−ax
y
_
+ ae
−ax
y = e
−ax
Dy,
(D+a)
2
_
e
−ax
y
_
= (D+a)
_
e
−ax
Dy
_
= e
−ax
D
2
y,
· · · · · ·
(D+a)
n
_
e
−ax
y
_
= e
−ax
D
n
y,
one obtains
_
a
n
(D+a)
n
+ a
n−1
(D+a)
n−1
+ · · · + a
1
(D+a) + a
0
__
e
−ax
y
_
= e
−ax
(a
n
D
n
+ a
n−1
D
n−1
+ · · · + a
1
D + a
0
) y,
∴ φ(D+a)
_
e
−ax
y
_
= e
−ax
φ(D) y =⇒ y = e
ax
1
φ(D+a)
_
e
−ax
φ(D) y
_
.
Using φ(D) y =e
ax
f (x) and y =φ
−1
(D)
_
e
ax
f (x)
_
leads to
1
φ(D)
_
e
ax
f (x)
_
= e
ax
1
φ(D+a)
f (x).
Example 4.19 4.19
Solve (D
2
+ 5D + 4) y = e
2x
+ x
2
e
−2x
.
The characteristic equation is
λ
2
+ 5λ + 4 = 0 =⇒ (λ+4)(λ+1) = 0 =⇒ λ = −4, −1.
The complementary solution is y
C
= C
1
e
−4x
+ C
2
e
−x
. For a particular solution
y
P
=
1
D
2
+5D+4
(e
2x
+ x
2
e
−2x
).
y
P1
=
1
D
2
+5D+4
e
2x
=
1
2
2
+5×2+4
e
2x
=
1
18
e
2x
. Theorem 1: a=2
y
P2
=
1
D
2
+5D+4
(x
2
e
−2x
) Apply Shift Theoremto
1
φ(D)
_
e
ax
f (x)
_
.
= e
−2x
1
(D−2)
2
+5 (D−2) +4
x
2
Take e
−2x
out of the operator
and shift operator D by −2.
4.3 particular solutions 165
= e
−2x
1
D
2
+D−2
x
2
= −
e
−2x
2
1
1−
D
2
−
D
2
2
x
2
Write the operator in
ascending order of D.
Set the constant to 1.
= −
e
−2x
2
_
1 +
D
2
+
3D
2
4
+ · · ·
_
x
2
Use long division to expand
the operator and stop at D
2
.
= −
e
−2x
2
_
x
2
+ x +
3
2
+ 0 + · · ·
_
D
k
x
2
=0, for k >2
= −
e
−2x
2
_
x
2
+ x +
3
2
_
.
1 +
D
2
+
3D
2
4
Stop at D
2
.
1 −
D
2
−
D
2
2
¸
¸
¸
¸
¸
1
1 −
D
2
−
D
2
2
(−
D
2
+
D
2
2
D
2
−
D
2
4
−
D
3
4
(−
3D
2
4
+
D
3
4
Alternatively, one can use the series expansion
1
1−u
= 1 + u + u
2
+ · · · + u
k
+ · · ·
to expand the operator
1
1−
D
2
−
D
2
2
=
1
1−
_
D
2
+
D
2
2
_
1
1−u
= 1 + u + u
2
+ u
3
+ · · ·
= 1 +
_
D
2
+
D
2
2
_
+
_
D
2
+
D
2
2
_
2
+
_
D
2
+
D
2
2
_
3
+ · · ·
= 1 +
_
D
2
+
D
2
2
_
+
_
D
2
4
+ · · ·
_
+ · · · Keep terms up to D
2
.
= 1 +
D
2
+
3D
2
4
+ · · · .
Hence, a particular solution is
y
P
= y
P1
+ y
P2
=
1
18
e
2x
−
e
−2x
2
_
x
2
+ x +
3
2
_
.
166 4 linear differential equations
Technique for Evaluating a Polynomial
1
a
n
D
n
+a
n−1
D
n−1
+ · · · +a
1
D+a
0
(C
0
+C
1
x+ · · · +C
k
x
k
)
. ,, .
Polynomial of degree k
=
1
a
0
1
1+
a
1
a
0
D+
a
2
a
0
D
2
+ · · · +
a
n
a
0
D
n
(C
0
+C
1
x+ · · · +C
k
x
k
)
. ,, .
Rewrite the operator in ascending order of D.
Set the constant to 1.
=
1
a
0
(1+b
1
D+b
2
D
2
+ · · · +b
k
D
k
+ · · · )(C
0
+C
1
x+ · · · +C
k
x
k
),
. ,, .
Expand the operator and stop at D
k
.
which can be easily evaluated using
Dx
q
=qx
q−1
, D
2
x
q
=q(q−1)x
q−2
, . . . , D
q
x
q
=q(q−1) · · · 1,
D
p
x
q
=0, for p>q.
Expansion of the operator can be obtained using
1. Long division, or
2. Series
1
1−u
= 1 + u + u
2
+ · · · + u
k
+ · · · .
☞
This method of expanding operators using long division or series is applica
ble only when evaluating polynomials.
Special Case:
1
a
n
D
n
+a
n−1
D
n−1
+ · · · +a
1
D+a
0
C
0
=
C
0
a
0
Constant
Example 4.20 4.20
Find a particular solution of (D
2
+ 6D + 9) y = (x
3
+2x)e
−3x
.
Using the method of operators, a particular solution is given by
y
P
=
1
D
2
+6D+9
_
(x
3
+2x)e
−3x
_
Shift Theorem: a= −3.
= e
−3x
1
(D−3)
2
+6 (D−3) +9
(x
3
+2x)
Take e
−3x
out of the operator
and shift operator D by −3.
= e
−3x
1
D
2
(x
3
+2x).
4.3 particular solutions 167
Since D
−1
is the integral operator, D
−2
f (x) means integrating f (x) twice:
x
3
+2x
Integrate
==⇒
1
4
x
4
+ x
2
Integrate
==⇒
1
20
x
5
+
1
3
x
3
.
Hence,
y
P
= e
−3x
_
1
20
x
5
+
1
3
x
3
_
.
Example 4.21 4.21
Find a particular solution of (D
2
− 2D + 2) y = x
2
e
x
cos 2x.
From Euler’s formula
e
i θ
= cos θ +i sin θ =⇒ cos 2x =Re(e
i 2x
) and e
x
cos 2x =Re
_
e
(1+i 2)x
_
.
Applying the method of operators, a particular solution is
y
P
= Re
_
1
D
2
−2D+2
_
x
2
e
(1+i 2) x
_
_
Theorem 2: a=1+i 2. Take e
(1+i 2)x
out of the operator, shift D by (1+i 2).
= Re
_
e
(1+i 2) x
1
_
D+(1+i 2)
_
2
−2
_
D+(1+i 2)
_
+2
x
2
_
= Re
_
e
(1+i 2) x
1
D
2
+i 4D−3
x
2
_
= Re
_
−
1
3
e
(1+i 2) x
1
1 −
_
i 4
3
D+
1
3
D
2
_
x
2
_
Expand the operator
using series. Stop at D
2
.
= Re
_
−
1
3
e
(1+i 2) x
_
1 +
_
i 4
3
D+
1
3
D
2
_
+
_
i 4
3
D+
1
3
D
2
_
2
+ · · ·
_
x
2
_
= Re
_
−
1
3
e
(1+i 2) x
_
x
2
+
i 4
3
Dx
2
−
13
9
D
2
x
2
+ · · ·
_
_
= Re
_
−
1
3
e
x
(cos 2x + i sin 2x)
_
x
2
+
i 8
3
x −
26
9
_
_
Apply Euler's
formula to e
(1+i 2)x
.
= −
1
3
e
x
_
x
2
cos 2x −
8
3
x sin 2x −
26
9
cos 2x
_
.
Expand the product
and take the real part.
Example 4.22 4.22
Find a particular solution of (D
2
+ 3D − 4) y = 6 sin 3x.
Using the method of operators, a particular solution is given by
y
P
=
6
D
2
+3D−4
sin 3x.
168 4 linear differential equations
In order to use Theorem 1, sin 3x must be converted to an exponential function
using Euler’s formula e
i θ
= cos θ +i sin θ =⇒ cos θ =Re(e
i θ
), sin θ =Im(e
i θ
):
y
P
= Im
_
6
D
2
+3D−4
e
i3x
_
=Im
_
6
(i3)
2
+3(i3)−4
e
i3x
_
Theorem 1: a=i3,
replace D by i3.
= Im
_
6
−13+i 9
e
i3x
_
= Im
_
6(−13−i 9)
(−13+i 9)(−13−i 9)
e
i3x
_
Multiply both the numerator and denominator by (−13−i 9).
= Im
_
6(−13−i 9)
13
2
+9
2
(cos 3x+i sin 3x)
_
Expand e
i3x
using Euler's formula.
=
3
125
Im
_
(−13 cos 3x+9 sin 3x) + i (−13 sin 3x−9 cos 3x)
_
. ,, . . ,, .
Real part Imaginary part
= −
3
125
(13 sin 3x+9 cos 3x).
Remarks:
❧ This approach involves manipulations of complex numbers, which could be
tedious. Note that i
2
= −1, i
4
=1, i
6
= −1, . . . . When one deals with only the
even power terms of D, one can avoid complex numbers.
❧ Since cos βx and sin βx are related to e
iβ x
, when applying Theorem 1, a=i β
and a
2
=(i β)
2
= −β
2
. Hence, by dealing with only even power terms of D, a
simpler method can be devised as follows.
y
P
=
6
D
2
+3D−4
sin 3x
Theorem 1: sin 3x =Im(e
i3x
);
replace D
2
by (i 3)
2
= −3
2
.
=
6
(−3
2
) +3D−4
sin 3x =
6
−13+3D
sin 3x
=
6 (−13−3D)
(−13+3D) (−13−3D)
sin 3x
Multiply both the numerator
and denominator by (−13−3D).
=
−6(13+3D)
169−9 D
2
sin 3x
Apply Theorem 1. Replace D
2
in the denominator by −3
2
.
= −
6
169−9(−3
2
)
(13 sin 3x+3 Dsin 3x)
Expand the numerator,
Dsin3x =(sin3x)
=3 cos3x.
= −
3
125
(13 sin 3x+9 cos 3x).
This procedure can be summarized as Theorem 3. Note that φ(D) can always
be written as φ
1
(D
2
)+φ
2
(D
2
)D, e.g.
φ(D) = 3D
3
+ 2D
2
+ D + 1 = (2D
2
+1) + (3D
2
+1)D.
4.3 particular solutions 169
Theorem 3
1
φ(D)
_
sin βx
cos βx
_
=
1
φ
1
(D
2
)+φ
2
(D
2
)D
_
sin βx
cos βx
_
sin βx and cos βx
are related to e
iβx
.
=
1
φ
1
(−β
2
)+φ
2
(−β
2
)D
_
sin βx
cos βx
_
Replace D
2
by (iβ)
2
=−β
2
.
=
_
φ
1
(−β
2
)−φ
2
(−β
2
)D
_
_
φ
1
(−β
2
)+φ
2
(−β
2
)D
_ _
φ
1
(−β
2
)−φ
2
(−β
2
)D
_
_
sin βx
cos βx
_
Multiply both numerator and denominator by
_
φ
1
(−β
2
)−φ
2
(−β
2
)D
_
.
=
φ
1
(−β
2
)−φ
2
(−β
2
)D
_
φ
1
(−β
2
)
_
2
−
_
φ
2
(−β
2
)
_
2
D
2
_
sin βx
cos βx
_
(a−b)(a+b) =a
2
−b
2
=
1
_
φ
1
(−β
2
)
_
2
−
_
φ
2
(−β
2
)
_
2
(−β
2
)
_
φ
1
(−β
2
) sin βx−φ
2
(−β
2
)Dsin βx
φ
1
(−β
2
) cos βx−φ
2
(−β
2
)Dcos βx
_
Replace D
2
in the denominator by −β
2
; expand the numerator.
=
1
_
φ
1
(−β
2
)
_
2
+β
2
_
φ
2
(−β
2
)
_
2
_
φ
1
(−β
2
) sin βx−βφ
2
(−β
2
) cos βx
φ
1
(−β
2
) cos βx+βφ
2
(−β
2
) sin βx
_
.
☞
If
_
φ
1
(−β
2
)
_
2
+β
2
_
φ
2
(−β
2
)
_
2
=0, use Theorem 4.
It is not advisable to memorize the result of Theorem 3, but to treat the theorem as
a technique for ﬁnding a particular solution corresponding to a sinusoidal function.
Example 4.23 4.23
Evaluate y
P
=
1
D
2
−4D+3
_
e
x
(2 sin 3x−3 cos 2x)
_
.
y
P
= e
x
1
(D+1)
2
−4(D+1)+3
(2 sin 3x−3 cos 2x)
Theorem 2: take e
x
out of operator and shift D by +1.
= e
x
1
D
2
−2D
(2 sin 3x−3 cos 2x)
= e
x
_
1
(−3
2
) −2D
(2 sin 3x) Replace D
2
by −3
2
.
+
1
(−2
2
) −2D
(−3 cos 2x)
_
Replace D
2
by −2
2
.
170 4 linear differential equations
= e
x
_
−2 (9−2D)
(9+2D) (9−2D)
sin 3x +
3 (4−2D)
(4+2D) (4−2D)
cos 2x
_
= e
x
_
−2(9−2D)
81−4 D
2
sin 3x +
3(4−2D)
16−4 D
2
cos 2x
_
= e
x
_
−2
81−4 (−3
2
)
(9 sin 3x − 2Dsin 3x)
Replace D
2
by −3
2
,
expand the numerator.
+
3
16−4 (−2
2
)
(4 cos 2x − 2Dcos 2x)
_
Replace D
2
by −2
2
,
expand the numerator.
= e
x
_
−
2
39
(3 sin 3x − 2 cos 3x) +
3
8
(cos 2x + sin 2x)
_
.
Example 4.24 4.24
Solve (D
2
+ 6D + 9) y = 72 sin
4
3x.
The characteristic equation is λ
2
+6λ+9=0 =⇒ (λ+3)
2
=0 =⇒ λ = −3, −3.
The complementary solution is
y
C
= (C
0
+C
1
x)e
−3x
.
The righthand side of the differential equation can be converted to the standard
form using trigonometric identities:
72 sin
4
3x = 72(sin
2
3x)
2
= 72
_
1−cos 6x
2
_
2
= 18(1−2 cos 6x+cos
2
6x)
= 18
_
1−2 cos 6x+
1+cos12x
2
_
= 27−36 cos 6x+9 cos12x.
y
P
=
1
D
2
+6D+ 9
27 Special case of a polynomial: constant
− 36
1
D
2
+6D+9
cos 6x Theorem 3: replace D
2
by −6
2
.
+ 9
1
D
2
+6D+9
cos 12x Theorem 3: replace D
2
by −12
2
.
= 3 − 36
1
(−6
2
)+6D+9
cos 6x + 9
1
(−12
2
)+6D+9
cos 12x
= 3 − 12
(2D+9)
(2D−9)(2D+9)
cos 6x + 3
(2D+45)
(2D−45)(2D+45)
cos 12x
= 3 −
12(2D+9)
4D
2
−81
cos 6x +
3(2D+45)
4D
2
−2025
cos 12x
= 3 −
12
4(−6
2
)−81
(−12 sin 6x+9 cos 6x)
Replace D
2
by −6
2
,
expand the numerator.
4.3 particular solutions 171
+
3
4(−12
2
)−2025
(−24 sin 12x+45 cos 12x)
Replace D
2
by −12
2
,
expand the numerator.
= 3 −
16
25
sin 6x +
12
25
cos 6x +
8
289
sin 12x −
15
289
cos 12x.
Theorem 4
If φ(a) =0, φ
(a) =0, φ
(a) =0, . . . , φ
( p−1)
(a) =0, φ
( p)
(a) =0, i.e., a is a
pfold root of the characteristic equation φ(λ) =0, then
1
φ(D)
e
ax
=
1
φ
( p)
(a)
x
p
e
ax
.
Take pthorder derivative of φ(D);
replace D by a; multiply the result by x
p
.
Proof: Since a is a pfold root of the characteristic equation φ(λ) =0, φ(D) can
be written as φ(D) =(D−a)
p
φ
1
(D), φ
1
(a) =0. Hence
1
φ(D)
e
ax
=
1
(D−a)
p
φ
1
(D)
_
e
ax
· 1
_
Rewrite e
ax
as (e
ax
· 1).
= e
ax
1
D
p
φ
1
(D+a)
(1)
Apply Theorem 2. Take e
ax
out of
the operator and shift D by a.
= e
ax
1
D
p
_
1
φ
1
(D+a)
e
0 · x
_
Rewrite 1 as e
0 · x
.
= e
ax
1
D
p
1
φ
1
(a)
(1). Apply Theorem 1 on the shaded part, a=0.
Since
1
D
is the integral operator,
1
D
p
means integrating 1 with respect to x for p
times, which gives
1
D
p
(1) =
1
p!
x
p
and
1
φ(D)
e
ax
=
1
p! φ
1
(a)
x
p
e
ax
.
On the other hand, differentiating φ(D) yields
φ
(D) = p(D−a)
p−1
φ
1
(D) + (D−a)
p
φ
1
(D),
φ
(D) = p( p−1)(D−a)
p−2
φ
1
(D) +2p(D−a)
p−1
φ
1
(D)+(D−a)
p
φ
1
(D),
φ
(D) = p( p−1)( p−2)(D−a)
p−3
φ
1
(D) + 3p( p−1)(D−a)
p−2
φ
1
(D)
+ 3p(D−a)
p−1
φ
1
(D) + (D−a)
p
φ
1
(D),
. . . . . .
From the shaded terms, it can be easily seen that φ
( p)
(a) =p! φ
1
(a). Hence,
1
φ(D)
e
ax
=
1
φ
( p)
(a)
x
p
e
ax
.
172 4 linear differential equations
Example 4.25 4.25
Evaluate y
P
=
1
(D−2)
3
e
2x
.
Use Theorem 4:
φ(D) = (D−2)
3
, φ(2) = 0,
φ
(D) = 3(D−2)
2
, φ
(2) = 0,
φ
(D) = 6(D−2), φ
(2) = 0,
φ
(D) = 6, φ
(2) = 6 = 0.
∴ y
P
=
1
φ
(2)
x
3
e
2x
=
1
6
x
3
e
2x
.
Example 4.26 4.26
Solve (D
2
+ 4 D + 13) y = e
−2x
sin 3x.
The characteristic equation is λ
2
+4λ+13=0, which gives
λ =
−4 ±
√
4
2
− 4×13
2
= −2 ± i 3.
Hence the complementary solution is y
C
= e
−2x
(Acos 3x + Bsin 3x).
Remarks: Note that the righthand side of the differential equation is contained
in the complementary solution. Using the method of undetermined coefficient,
the assumed form of a particular solution is x · e
−2x
(a cos 3x+b sin 3x).
A particular solution is given by
y
P
=
1
D
2
+4D+13
_
e
−2x
sin 3x
_
= e
−2x
1
(D−2)
2
+4(D−2)+13
sin 3x
Theorem 2: take e
−2x
out of the operator, shift D by −2.
= e
−2x
1
D
2
+9
sin 3x = e
−2x
Im
_
1
D
2
+9
e
i3x
_
.
This can be evaluated using Theorem 4:
φ(D) = D
2
+9, φ(i 3) = (i 3)
2
+9 = 0,
φ
(D) = 2D, φ
(i 3) = 2(i 3) = i 6 = 0.
Hence,
y
P
= e
−2x
Im
_
1
φ
(i 3)
x e
i3x
_
Theorem 4
= e
−2x
Im
_
1
i 6
x (cos 3x + i sin 3x)
_
= e
−2x
Im
_
−
i
6
x(cos 3x + i sin 3x)
_
= −
1
6
xe
−2x
cos 3x.
4.3 particular solutions 173
4.3.3 Method of Variation of Parameters
When the righthand side of a linear ordinary differential equation with constant
coefﬁcients is a combination of polynomials, exponential functions, and sinusoidal
functions in the following form
e
αx
_
(a
0
+ a
1
x + · · · + a
k
x
k
) cos βx + (b
0
+ b
1
x + · · · + b
k
x
k
) sin βx
_
,
the method of undetermined coefﬁcients or the method of Doperator can be applied
to ﬁnd a particular solution. Otherwise, the method of variation of parameters must
be employed, which is the most general method.
The method of variation of parameters is illustrated using speciﬁc examples.
Example 4.27 4.27
Solve the differential equation y
+ y = csc
3
x. (1)
The characteristic equation is λ
2
+1=0, which gives λ= ±i. The complementary
solution is
y
C
= Acos x + Bsin x. (2a)
Differentiating y
C
with respect to x yields
y
C
= −Asin x + Bcos x. (2b)
Note that in equations (2), A and B are constants.
Remarks: In general, for an nthorder linear differential equation, the com
plementary solution y
C
is differentiated (n−1) times to obtain equations for
y
C
, y
C
, . . . , y
(n−1)
C
.
To ﬁnd a particular solution y
P
, vary the constants A and B in equations (2) to
make them functions of x, i.e., A⇒a(x), B⇒b(x). Thus the expressions for y
P
and y
P
are obtained
y
P
= a(x) cos x + b(x) sin x, (3a)
y
P
= −a(x) sin x + b(x) cos x. (3b)
Differentiating equation (3a) with respect to x yields
y
P
= a
(x) cos x − a(x) sin x + b
(x) sin x + b(x) cos x
Differentiate using
the product rule.
= −a(x) sin x + b(x) cos x, Compare with equation (3b).
which leads to
a
(x) cos x + b
(x) sin x = 0. (4a)
174 4 linear differential equations
Differentiating equation (3b) with respect to x and substituting into equation (1)
result in
y
P
+ y
P
=
_
−a
(x) sin x − a(x) cos x + b
(x) cos x − b(x) sin x
_
+
_
a(x) cos x + b(x) sin x
_
= csc
3
x,
which yields
−a
(x) sin x + b
(x) cos x = csc
3
x. (4b)
Equations (4) give two linear algebraic equations for two unknowns a
(x) and b
(x),
which can be solved easily
Eq(4a) × cos x : a
(x) cos
2
x + b
(x) sin x cos x = 0,
Eq(4b) × sin x : −a
(x) sin
2
x + b
(x) sin x cos x = csc
2
x.
Subtracting these two equations leads to
a
(x) = −csc
2
x. sin
2
x+cos
2
x =1. (5a)
Similarly,
Eq(4a) × sin x + Eq(4b) × cos x : b
(x) = csc
3
x cos x. (5b)
Integrating equation (5a) yields
a(x) = −
_
csc
2
xdx = cot x,
and integrating equation (5b) gives
b(x) =
_
csc
3
x cos xdx =
_
d(sin x)
sin
3
x
= −
1
2 sin
2
x
.
The general solution is then given by
y = y
C
+ y
P
= Acos x + Bsin x + a(x) cos x + b(x) sin x
= Acos x + Bsin x + cot x · cos x −
1
2 sin
2
x
· sin x
= Acos x + Bsin x +
cos
2
x
sin x
−
1
2 sin x
.
Example 4.28 4.28
Solve the differential equation y
− 2y
+ y = ln x. (1)
4.3 particular solutions 175
The characteristic equation is λ
2
−2λ+1=0 =⇒ λ=1, 1. The complementary
solution is
y
C
= (C
0
+ C
1
x)e
x
, (2a)
where C
0
and C
1
are constants. Differentiating y
C
with respect to x yields
y
C
= C
1
e
x
+ (C
0
+ C
1
x)e
x
=
_
(C
0
+C
1
) + C
1
x
_
e
x
. (2b)
To ﬁnd a particular solution y
P
, vary the constants C
0
and C
1
in equations (2) to
make them functions of x, i.e., C
0
⇒c
0
(x), C
1
⇒c
1
(x), to obtain the expressions
for y
P
and y
P
y
P
=
_
c
0
(x) + c
1
(x) · x
_
e
x
, (3a)
y
P
=
__
c
0
(x)+c
1
(x)
_
+ c
1
(x) · x
_
e
x
. (3b)
Differentiating equation (3a) with respect to x yields:
y
P
=
_
c
0
(x)+c
1
(x) · x +c
1
(x)
_
e
x
+
_
c
0
(x)+c
1
(x) · x
_
e
x
Differentiate using
the product rule.
=
__
c
0
(x)+c
1
(x)
_
+ c
1
(x) · x
_
e
x
, Compare with equation (3b).
which leads to
c
0
(x) + c
1
(x) · x = 0. (4a)
Differentiating equation (3b) with respect to x and substituting into equation (1)
result in
y
P
− 2y
P
+ y
P
=
__
c
0
(x)+c
1
(x)
_
+ c
1
(x) · x + c
1
(x)
_
e
x
+
__
c
0
(x)+c
1
(x)
_
+ c
1
(x) · x
_
e
x
− 2
__
c
0
(x)+c
1
(x)
_
+ c
1
(x) · x
_
e
x
+
_
c
0
(x) + c
1
(x) · x
_
e
x
=
_
c
0
(x) + c
1
(x) + c
1
(x) · x
_
e
x
= ln x. Righthand side of equation (1)
Hence,
c
0
(x) + c
1
(x) + c
1
(x) · x = e
−x
ln x. (4b)
Equations (4) give two linear algebraic equations for two unknowns c
0
(x) and c
1
(x).
Subtracting equation (4a) from (4b) yields
c
1
(x) = e
−x
ln x. (5a)
From equation (4a), one has
c
0
(x) = −c
1
(x) · x = −xe
−x
ln x. (5b)
176 4 linear differential equations
Integrating equations (5) yields
c
1
(x) =
_
e
−x
ln x dx = −
_
ln x d(e
−x
) Integration by parts
= −
_
e
−x
ln x −
_
e
−x
x
dx
_
.
c
0
(x) = −
_
xe
−x
ln x dx =
_
x ln x d(e
−x
) Integration by parts
= xe
−x
ln x −
_
e
−x
(ln x+1)dx = xe
−x
ln x + e
−x
−
_
e
−x
ln x dx
= xe
−x
ln x + e
−x
+ e
−x
ln x −
_
e
−x
x
dx,
A particular solution is then given by
y
P
=
_
c
0
(x) + c
1
(x) · x
_
e
x
=
_
xe
−x
ln x + e
−x
+ e
−x
ln x −
_
e
−x
x
dx +
_
−e
−x
ln x +
_
e
−x
x
dx
_
· x
_
e
x
= 1 + ln x + e
x
(x−1)
_
e
−x
x
dx.
Example 4.29 4.29
Solve the differential equation y
− y
=
e
x
1+e
x
. (1)
The characteristic equation is λ
3
−λ=0 =⇒ λ(λ+1)(λ−1) =0 =⇒ λ= −1, 1, 0.
The complementary solution is
y
C
= +C
1
e
−x
+ C
2
e
x
+ C
3
, (2a)
where C
1
, C
2
, and C
3
are constants. Differentiating y
C
with respect to x yields
y
C
= −C
1
e
−x
+ C
2
e
x
, (2b)
y
C
= +C
1
e
−x
+ C
2
e
x
. (2c)
Apply the method of variation of parameters and vary the constants C
1
⇒c
1
(x),
C
2
⇒c
2
(x), C
3
⇒c
3
(x) to express a particular solution and its ﬁrst and second
order derivatives as
y
P
= +c
1
(x)e
−x
+ c
2
(x)e
x
+ c
3
(x), (3a)
y
P
= −c
1
(x)e
−x
+ c
2
(x)e
x
, (3b)
y
P
= +c
1
(x)e
−x
+ c
2
(x)e
x
. (3c)
4.3 particular solutions 177
Differentiating equation (3a) with respect to x yields:
y
P
= c
1
(x)e
−x
− c
1
(x)e
−x
+ c
2
(x)e
x
+ c
2
(x)e
x
+ c
3
(x)
= −c
1
(x)e
−x
+ c
2
(x)e
x
, Compare with equation (3b).
which leads to
c
1
(x)e
−x
+ c
2
(x)e
x
+ c
3
(x) = 0. (4a)
Differentiating equation (3b) with respect to x yields:
y
P
= −c
1
(x)e
−x
+ c
1
(x)e
−x
+ c
2
(x)e
x
+ c
2
(x)e
x
= c
1
(x)e
−x
+ c
2
(x)e
x
, Compare with equation (3c).
which leads to
−c
1
(x)e
−x
+ c
2
(x)e
x
= 0. (4b)
Differentiating equation (3c) and substituting into equation (1) result in
y
P
−y
P
=
_
c
1
(x)e
−x
−c
1
(x)e
−x
+ c
2
(x)e
x
+c
2
(x)e
x
_
−
_
−c
1
(x)e
−x
+c
2
(x)e
x
_
=
e
x
1+e
x
, Righthand side of equation (1)
which leads to
c
1
(x)e
−x
+ c
2
(x)e
x
=
e
x
1+e
x
. (4c)
Equations (4) give three linear algebraic equations for three unknowns c
1
(x), c
2
(x),
and c
3
(x), which can be solved using Cramer’s Rule or Gaussian elimination:
Eq(4c) − Eq(4b) : 2c
1
(x)e
−x
=
e
x
1+e
x
=⇒ c
1
(x) =
e
2x
2(1+e
x
)
, (5a)
Eq(4c) + Eq(4b) : 2c
2
(x)e
x
=
e
x
1+e
x
=⇒ c
2
(x) =
1
2(1+e
x
)
, (5b)
Eq(4a) − Eq(4c) : c
3
(x) = −
e
x
1+e
x
. (5c)
Integrating equations (5) yields
c
3
(x) = −
_
e
x
1+e
x
dx = −
_
1
1+e
x
d(1+e
x
) = −ln(1+e
x
).
c
1
(x) =
1
2
_
e
2x
1+e
x
dx =
1
2
_
(1+e
x
) − 1
1+e
x
e
x
dx
=
1
2
_
_
1 −
1
1+e
x
_
e
x
dx =
1
2
_
e
x
−ln(1+e
x
)
_
,
178 4 linear differential equations
c
2
(x) =
1
2
_
1
1+e
x
dx =
1
2
_
e
−x
e
−x
+1
dx = −
1
2
_
1
e
−x
+1
d(e
−x
+1)
= −
1
2
ln(e
−x
+1) =
1
2
_
−ln(1+e
x
) + x
_
,
A particular solution is then given by
y
P
= c
1
(x)e
−x
+ c
2
(x)e
x
+ c
3
(x)
=
1
2
_
e
x
−ln(1+e
x
)
_
· e
−x
+
1
2
_
−ln(1+e
x
) + x
_
· e
x
− ln(1+e
x
)
=
1
2
_
1 + xe
x
− (e
−x
+ e
x
+ 2) ln(1+e
x
)
_
.
4.4 Euler Differential Equations
AnEuler differential equationis a linear ordinary differential equationwithvariable
coefﬁcients of the form
a
n
x
n
d
n
y
dx
n
+ a
n−1
x
n−1
d
n−1
y
dx
n−1
+ · · · + a
1
x
dy
dx
+ a
0
y = f (x),
where a
0
, a
1
, . . . , a
n
are constants. Using the Doperator, D( · ) ≡d( · )/dx, an
Euler differential equation can be written as
_
a
n
x
n
D
n
+ a
n−1
x
n−1
D
n−1
+ · · · + a
1
xD + a
0
_
y = f (x).
To solve the equation, use the substitution x =e
z
to convert it to one with constant
coefﬁcients.
Assuming the independent variable x >0, let x =e
z
or z = ln x and adopt the
script D notation, D( · ) ≡d( · )/dz, to denote differentiation with respect to z.
Hence, using the chain rule,
dy
dx
=
dy
dz
dz
dx
=
dy
dz
1
x
,
dz
dx
=
d
dx
(ln x) =
1
x
which yields
x
dy
dx
=
dy
dz
=⇒ xD(·) = D(·).
Similarly,
d
2
y
dx
2
=
d
dx
_
dy
dx
_
=
d
dx
_
1
x
dy
dz
_
= −
1
x
2
dy
dz
+
1
x
d
dz
_
dy
dz
_
dz
dx
= −
1
x
2
dy
dz
+
1
x
2
d
2
y
dz
2
,
leading to
x
2
d
2
y
dx
2
=
d
2
y
dz
2
−
dy
dz
=⇒ x
2
D
2
(·) = (D
2
−D)(·) = D(D−1)(·).
4.4 euler differential equations 179
It can be proved, by mathematical induction, that
x
n
D
n
(·) = D(D−1)(D−2) · · · (D−n+1)(·), n=positive integer.
Hence, in terms of the script D operator, D( · ) ≡d( · )/dz, an Euler’s equation
becomes an equation with constant coefﬁcients.
For example, consider
_
a
2
x
2
D
2
+ a
1
x D + a
0
_
y = f (x), D( · ) = d( · )/dx.
Applying the variable substitution x =e
z
, the differential equation becomes
_
a
2
D(D−1) + a
1
D + a
0
_
y = f (e
z
), D( · ) = d( · )/dz,
or
_
a
2
D
2
+ (a
1
−a
2
) D + a
0
_
y = f (e
z
).
Example 4.30 4.30
Solve (x
2
D
2
− xD + 2) y = x(ln x)
3
, D( · ) ≡ d( · )/dx.
Letting x =e
z
, z = ln x, and D( · ) ≡d( · )/dz, the differential equation becomes
_
D(D−1) −D + 2
_
y = e
z
z
3
=⇒
_
D
2
− 2 D + 2
_
y = e
z
z
3
.
The characteristic equation is λ
2
−2λ+2=0, which gives
λ =
−(−2) ±
_
(−2)
2
− 4×1×2
2
= 1 ± i .
The complementary solution is
y
C
= e
z
(Acos z + Bsin z).
A particular solution is given by
y
P
=
1
D
2
− 2 D + 2
_
e
z
· z
3
_
Theorem 2: take e
z
out of the operator
and shift operator D by +1.
= e
z
1
(D+1)
2
− 2(D+1) + 2
z
3
= e
z
1
D
2
+ 1
z
3
= e
z
_
1 −D
2
+ (D
2
)
2
− · · ·
_
z
3
Expand the operator in series,
stop at D
3
.
= e
z
(z
3
− 6z).
Hence, the general solution is
y = y
C
+ y
P
= e
z
(Acos z + Bsin z) + e
z
(z
3
− 6z)
= x
_
Acos(ln x) + Bsin(ln x) + (ln x)
3
− 6 ln x
_
.
Change back to
the original variable x.
180 4 linear differential equations
Example 4.31 4.31
Solve (x
3
D
3
+ xD − 1) y = 4x
5
, D( · ) ≡ d( · )/dx.
Let x =e
z
, z = ln x, and D( · ) ≡d( · )/dz. The differential equation becomes
_
D(D−1)(D−2) +D − 1
_
y = 4e
5z
=⇒ (D−1)
3
y = 4e
5z
.
The characteristic equation is (λ−1)
3
=0 =⇒ λ=1, 1, 1. The complementary
solution is
y
C
= (C
0
+ C
1
z + C
2
z
2
)e
z
.
A particular solution is given by
y
P
= 4
1
(D−1)
3
e
5z
= 4
1
(5−1)
3
e
5z
=
e
5z
16
. Theorem 1: a=5
The general solution is
y = y
C
+ y
P
= (C
0
+ C
1
z + C
2
z
2
)e
z
+
1
16
e
5z
=
_
C
0
+ C
1
ln x + C
2
(ln x)
2
_
x +
1
16
x
5
.
Change back to
the original variable x.
4.5 Summary
Consider an nthorder ordinary differential equation with constant coefﬁcients
φ(D) y =F(x), φ(D) =a
n
D
n
+a
n−1
D
n−1
+ · · · +a
1
D+a
0
, D( · ) ≡d( · )/dx.
General solution y(x) = y
C
(x) + y
P
(x).
Complementary Solution
Complementary differential equation: φ(D) y =0. R.H.S. set to 0.
Characteristic equation: φ(λ) = 0 =⇒ characteristic numbers λ
1
, λ
2
, . . . , λ
n
.
❧ φ(λ) =0 has a real root λ=a of pfold
y
C
=
_
C
0
+C
1
x+ . . . +C
p−1
x
p−1
_
e
ax
. ,, .
polynomial of degree p−1
❧ φ(λ) =0 has a pair of complex roots λ=α±i β of pfold
y
C
= e
αx
__
A
0
+A
1
x+ . . . +A
p−1
x
p−1
_
cos βx
+
_
B
0
+B
1
x+ . . . +B
p−1
x
p−1
_
sin βx
_
. ,, .
polynomials of degree p−1
4.5 summary 181
Particular Solution
1. The method of variation of parameters is the most general method that can be
used for any functions on the righthand side.
2. When the righthand side of the differential equation is not of the form
e
αx
_
(a
0
+ a
1
x + · · · + a
k
x
k
) cos βx + (b
0
+ b
1
x + · · · + b
k
x
k
) sin βx
_
,
the method of variation of parameters must be applied.
3. When the righthand side is of this form, the method of undetermined co
efﬁcients, the method of Doperator, or the general method of variation of
parameters can be applied. The method of Doperator is usually the easiest
and the most efﬁcient; hence it is the preferred method.
1. Method of Undetermined Coefﬁcients
Corresponding to Righthand Side F(x) Assumed Form of y
P
(1) Polynomial of degree k Polynomial of degree k
(2) sin βx, cos βx Acos βx + Bsin βx
(3) e
αx
Ce
αx
(4) e
αx
_
a
0
+a
1
x+ · · · +a
k
x
k
_
cos βx, e
αx
__
A
0
+A
1
x+ · · · +A
k
x
k
_
cos βx
e
αx
_
b
0
+b
1
x+ · · · +b
k
x
k
_
sin βx +
_
B
0
+B
1
x+ · · · +B
k
x
k
_
sin βx
_
. ,, . . ,, .
Polynomial of degree k Polynomial of degree k
If a normally assumed term of a particular solution occurs in the complementary
solution, it must be multiplied by a power of x, which is sufﬁciently high but not
higher, so that it does not occur in the complementary solution.
2. Method of DOperator
❧ Polynomial: Used when φ
−1
(D) operates on a polynomial.
1
a
n
D
n
+a
n−1
D
n−1
+ · · · +a
1
D+a
0
(C
0
+C
1
x+ · · · +C
k
x
k
)
. ,, .
Polynomial of degree k
=
1
a
0
1
1+
a
1
a
0
D+
a
2
a
0
D
2
+ · · · +
a
n
a
0
D
n
(C
0
+C
1
x+ · · · +C
k
x
k
)
. ,, .
Rewrite the operator in ascending order of D. Set the constant to 1.
=
1
a
0
(1+b
1
D+b
2
D
2
+ · · · +b
k
D
k
)(C
0
+C
1
x+ · · · +C
k
x
k
).
. ,, .
Expand the operator using series or long division. Stop at D
k
.
182 4 linear differential equations
❧ Shift Theorem (Theorem 2): Used when φ
−1
(D) operates on e
ax
f (x). The
Theorem takes the exponential function e
ax
out of the operation
1
φ(D)
_
e
ax
f (x)
_
= e
ax
1
φ(D+a)
f (x).
Take e
ax
out of the operator
and shift operator D by a.
❧ Theorem1: Used when φ
−1
(D) operates on exponential function e
ax
.
1
φ(D)
e
ax
=
1
φ(a)
e
ax
, φ(a) =0. Replace D by a.
☞
If φ(a) =0, use Theorem 4.
❧ Theorem3: Used when φ
−1
(D) operates on sin βx or cos βx.
1
φ(D)
_
sin βx
cos βx
_
=
1
φ
1
(D
2
)+φ
2
(D
2
)D
_
sin βx
cos βx
_
Replace D
2
by −β
2
.
=
1
φ
1
(−β
2
)+φ
2
(−β
2
)D
_
sin βx
cos βx
_
=
φ
1
(−β
2
)−φ
2
(−β
2
)D
_
φ
1
(−β
2
)
_
2
−
_
φ
2
(−β
2
)
_
2
D
2
_
sin βx
cos βx
_
Replace D
2
by −β
2
.
Expand the numerator.
=
1
_
φ
1
(−β
2
)
_
2
+β
2
_
φ
2
(−β
2
)
_
2
_
φ
1
(−β
2
) sin βx−βφ
2
(−β
2
) cos βx
φ
1
(−β
2
) cos βx+βφ
2
(−β
2
) sin βx
_
.
☞
If
_
φ
1
(−β
2
)
_
2
+β
2
_
φ
2
(−β
2
)
_
2
=0, use Theorem 4.
❧ Theorem4: Used when the denominator is 0 when Theorem 1 or 3 is applied.
1
φ(D)
e
ax
=
1
φ
( p)
(a)
x
p
e
ax
,
φ(a) =φ
(a) = · · · =φ
( p−1)
(a) =0,
φ
( p)
(a) =0.
Euler’s Formula: e
i βx
= cos βx + i sin βx,
cos βx = Re(e
i βx
), sin βx = Im(e
i βx
).
3. Method of Variation of Parameters
1. For an nthorder linear differential equation φ(D) y = F(x), determine the
complementary solution
y
C
= f (x; C
1
, C
2
, . . . , C
n
),
where C
1
, C
2
, . . . , C
n
are n arbitrary constants.
2. Differentiate y
C
with respect to x to yield the equations y
C
, y
C
, . . . , y
(n−1)
C
.
problems 183
3. Vary the constants C
1
, C
2
, . . . , C
n
to make them functions of x, and obtain
n equations y
(k)
P
, k =0, 1, . . . , n−1,
y
(k)
C
(x; C
1
, C
2
, . . . , C
n
)
C
i
⇒c
i
(x)
====⇒
i =1, 2, . . . , n
y
(k)
P
_
x; c
1
(x), c
2
(x), . . . , c
n
(x)
_
.
4. Differentiating y
(k)
P
with respect to x and comparing with the expressions
y
(k+1)
P
, k =0, 1, . . . , n−2, and substituting y
P
and the derivatives into the
original differential equation yield n linear algebraic equations for c
1
(x),
c
2
(x), . . . , c
n
(x).
5. Solve these equations for c
1
(x), c
2
(x), . . . , c
n
(x).
6. Integrate to obtain the functions c
1
(x), c
2
(x), . . . , c
n
(x). A particular solu
tion y
P
_
x; c
1
(x), c
2
(x), . . . , c
n
(x)
_
is then obtained.
Euler Differential Equations
(a
n
x
n
D
n
+ a
n−1
x
n−1
D
n−1
+ · · · + a
1
xD + a
0
) y = f (x), D( · ) ≡d( · )/dx.
Letting x =e
z
or z = ln x, x >0, D( · ) ≡d( · )/dz, then
x
n
D
n
(·) = D(D−1)(D−2) · · · (D−n+1)(·), n=positive integer.
The Euler differential equation is converted to a differential equation with constant
coefﬁcients.
Problems
Complementary Solutions
4.1 (D
3
− 2D
2
+ D − 2) y = 0
A
NS y
C
= Acos x + Bsin x + Ce
2x
4.2 (D
3
+ D
2
+ 9D + 9) y = 0
A
NS y
C
= Acos 3x + Bsin 3x + Ce
−x
4.3 (D
3
+ D
2
− D − 1) y = 0
A
NS y
C
= (C
0
+C
1
x)e
−x
+ Ce
x
4.4 (D
3
+ 8) y = 0
A
NS y
C
= e
x
_
Acos(
√
3x) + Bsin(
√
3x)
_
+ Ce
−2x
4.5 (D
3
− 8) y = 0
A
NS y
C
= e
−x
_
Acos(
√
3x) + Bsin(
√
3x)
_
+ Ce
2x
4.6 (D
4
+ 4) y = 0
A
NS y
C
=e
x
(A
1
cos x+B
1
sin x)+e
−x
(A
2
cos x+B
2
sin x)
4.7 (D
4
+ 18D
2
+ 81) y = 0
A
NS y
C
=(A
0
+A
1
x) cos 3x+(B
0
+B
1
x) sin 3x
4.8 (D
4
− 4D
2
+ 16) y = 0
A
NS y
C
= e
√
3x
(A
1
cos x + B
1
sin x) + e
−
√
3x
(A
2
cos x + B
2
sin x
_
184 4 linear differential equations
4.9 (D
4
−2D
3
+2D
2
−2D+1) y =0
A
NS y
C
=(C
0
+C
1
x)e
x
+Acos x+Bsin x
4.10 (D
4
− 5D
3
+ 5D
2
+ 5D − 6) y = 0
A
NS y
C
= C
1
e
−x
+ C
2
e
x
+ C
3
e
2x
+ C
4
e
3x
4.11 (D
5
− 6D
4
+ 9D
3
) y = 0
A
NS y
C
=C
0
+C
1
x+C
2
x
2
+(D
0
+D
1
x)e
3x
4.12 (D
6
− 64) y = 0
A
NS y
C
= C
1
e
−2x
+ C
2
e
2x
+ e
−x
_
A
1
cos(
√
3x) + B
1
sin(
√
3x)
_
+ e
x
_
A
2
cos(
√
3x) + B
2
sin(
√
3x)
_
Particular Solutions — Method of Undetermined Coefﬁcients
For the following differential equations, specify the form of a particular solution
using the method of undetermined coefﬁcients.
4.13 (D
2
+ 6D + 10) y = 3xe
−3x
− 2e
3x
cos x
4.14 (D
2
− 8D + 17) y = e
4x
(x
2
− 3x sin x)
4.15 (D
2
− 2D + 2) y = (x + e
x
) sin x
4.16 (D
2
+ 4) y = sinh x sin 2x
4.17 (D
2
+ 2D + 2) y = cosh x sin x
4.18 (D
3
+ D) y = sin x + x cos x
4.19 (D
3
− 2D
2
+ 4D − 8) y = e
2x
sin 2x + 2x
2
4.20 (D
3
− 4D
2
+ 3D) y = x
2
+ xe
2x
4.21 (D
4
+ D
2
) y = 7x − 3 cos x
4.22 (D
4
+ 5D
2
+ 4) y = sin x cos 2x
Particular Solutions — DOperator Method
4.23 (D
5
− 3D
3
+ 1) y = 9e
2x
A
NS y
P
= e
2x
4.24 (D − 1)
3
y = 48xe
x
A
NS y
P
= 2x
4
e
x
4.25 (D
3
− 3D) y = 9x
2
A
NS y
P
= −x
3
− 2x
4.26 (D
5
+ 4D
3
) y = 7 + x
A
NS y
P
=
1
96
x
3
(28+x)
4.27 (D
2
− D − 2) y = 36xe
2x
A
NS y
P
= 2e
2x
(3x
2
−2x)
4.28 (D
4
+ 16) y = 64 cos 2x
A
NS y
P
= 2 cos 2x
4.29 (D
4
+ 4D
2
− 1) y = 44 sin 3x
A
NS y
P
= sin 3x
Problems 185
4.30 (D
3
+ D
2
+ 5D + 5) y = 5 cos 2x
A
NS y
P
= 2 sin 2x + cos 2x
4.31 (D
2
+ 3D + 5) y = 5e
−x
sin 2x
A
NS y
P
= −e
−x
(2 cos 2x + sin 2x)
4.32 (D
4
− 1) y = 4e
−x
A
NS y
P
= −xe
−x
4.33 (D
2
+ 4) y = 8 sin
2
x
A
NS y
P
= 1 − x sin 2x
4.34 (D
3
− D
2
+ D − 1) y = 4 sin x
A
NS y
P
= x(cos x − sin x)
4.35 (D
4
− D
2
) y = 2e
x
A
NS y
P
= xe
x
General Solutions
4.36 y
− 4y
+ 4y = (1 + x)e
x
+ 2e
2x
+ 3e
3x
A
NS y = (C
0
+C
1
x)e
2x
+ (x+3)e
x
+ x
2
e
2x
+ 3e
3x
4.37 (D
2
− 2D + 5) y = 4e
x
cos 2x
A
NS y = e
x
(Acos 2x + Bsin 2x) + xe
x
sin 2x
4.38 (D
2
+ 4) y = 4 sin 2x
A
NS y = Acos 2x + Bsin 2x − x cos 2x
4.39 (D
2
− 1) y = 12x
2
e
x
+ 3e
2x
+ 10 cos 3x
A
NS y
P
= C
1
e
−x
+ C
2
e
x
+ e
x
(2x
3
−3x
2
+3x) + e
2x
− cos 3x
4.40 y
+ y = 2 sin x − 3 cos 2x
A
NS y =Acos x+Bsin x−x cos x+cos 2x
4.41 y
− y
= e
x
(10 + x
2
)
A
NS y = C
1
+ C
2
e
x
+ e
x
_
1
3
x
3
−x
2
+12x
_
4.42 (D
2
− 4) y = 96x
2
e
2x
+ 4e
−2x
A
NS y = C
1
e
−2x
+ C
2
e
2x
+ e
2x
(8x
3
− 6x
2
+ 3x) − xe
−2x
4.43 (D
2
+ 2D + 2) y = 5 cos x + 10 sin 2x
A
NS y =e
−x
(Acos x+Bsin x)+cos x+2 sin x−2 cos 2x−sin 2x
4.44 (D
2
− 2D + 2) y = 4x − 2 + 2e
x
sin x
A
NS y = e
x
(Acos x + Bsin x) + 2x + 1 − xe
x
cos x
4.45 (D
2
− 4D + 4) y = 4xe
2x
sin 2x
A
NS y = (C
0
+C
1
x)e
2x
− e
2x
(x sin 2x + cos 2x)
4.46 (D
3
− D
2
+ D − 1) y = 15 sin 2x
A
NS y = Ce
x
+ Acos x + Bsin x + 2 cos 2x + sin 2x
4.47 (D
3
+ 3D
2
− 4) y = 40 sin 2x
A
NS y = (C
0
+ C
1
x)e
−2x
+ C
2
e
x
+ cos 2x − 2 sin 2x
186 4 linear differential equations
4.48 y
− y
+ y
− y = 2e
x
+ 5e
2x
A
NS y = Acos x + Bsin x + Ce
x
+ xe
x
+ e
2x
4.49 (D
3
− 6D
2
+ 11D − 6) y = 10e
x
sin x
A
NS y = C
1
e
x
+ C
2
e
2x
+ C
3
e
3x
+ e
x
(3 sin x − cos x)
4.50 (D
3
− 2D − 4) y = 50(sin x + e
2x
)
A
NS y = Ce
2x
+e
−x
(Acos x+Bsin x)+6 cos x−8 sin x + 5xe
2x
4.51 y
− 3y
+ 4y = 12e
2x
+ 4e
3x
A
NS y = (C
0
+ C
1
x)e
2x
+ C
3
e
−x
+ 2x
2
e
2x
+ e
3x
4.52 (D
4
− 8D
2
+ 16) y = 32e
2x
+ 16x
3
A
NS y = (C
0
+C
1
x)e
−2x
+ (D
0
+D
1
x)e
2x
+ x
2
e
2x
+ x
3
+ 3x
4.53 (D
4
− 18D
2
+ 81) y = 72e
3x
+ 729x
2
A
NS y = (C
0
+ C
1
x)e
−3x
+ (D
0
+ D
1
x)e
3x
+ x
2
e
3x
+ 9x
2
+ 4
Method of Variation of Parameters
4.54 y
− y = x
−1
− 2x
−3
A
NS y = C
1
e
−x
+ C
2
e
x
− x
−1
4.55 y
− y =
1
sinh x
A
NS y =C
1
e
−x
+C
2
e
x
−xe
−x
+sinh x ln
¸
¸
1−e
−2x
¸
¸
4.56 y
− 2y
+ y =
e
x
x
A
NS y =
_
C
0
+ C
1
x + x ln
¸
¸
x
¸
¸
_
e
x
4.57 y
+ 3y
+ 2y = sin e
x
A
NS y = C
1
e
−2x
+ C
2
e
−x
− e
−2x
sin e
x
4.58 y
− 3y
+ 2y = sin e
−x
A
NS y = C
1
e
x
+ C
2
e
2x
− e
2x
sin e
−x
4.59 y
+ y = sec
3
x
A
NS y = Acos x + Bsin x +
1
2
sec x
4.60 y
−y =
_
1−e
2x
_
−
1
2
A
NS y =C
1
e
−x
+C
2
e
x
−
1
2
e
−x
sin
−1
e
x
−
1
2
_
1−e
2x
4.61 y
− y = e
−2x
sin e
−x
A
NS y =C
1
e
−x
+C
2
e
x
−sin e
−x
−e
x
cos e
−x
4.62 y
+ 2y
+ y = 15e
−x
√
x+1
A
NS y =e
−x
_
C
0
+ C
1
x + 4(x+1)
5
2
_
4.63 y
+ 4y = 2 tan x
A
NS y =Acos 2x + Bsin 2x + sin 2x ln
¸
¸
cos x
¸
¸
− x cos 2x
4.64 y
− 2y
+ y =
e
2x
(e
x
+ 1)
2
A
NS y = (C
0
+C
1
x)e
x
+ e
x
ln(1+e
x
)
4.65 y
+ y
=
1
1 + e
x
A
NS y = C
1
+C
2
e
−x
−ln(e
−x
+1)−e
−x
ln(e
x
+1)
Problems 187
Euler Differential Equations
4.66 (x
2
D
2
− xD + 1) y = ln x
A
NS y = (C
0
+C
1
ln x)x + 2 + ln x
4.67 x
2
y
+ 3x y
+ 5y =
5
x
2
ln x
A
NS y = x
−1
_
Acos(2 ln x) + Bsin(2 ln x)
_
+ x
−2
_
2
5
+ ln x
_
4.68 (x
3
D
3
+ 2x
2
D
2
− xD + 1) y = 9x
2
ln x
A
NS y
C
= C
1
x
−1
+ (C
2
+ C
3
ln x)x + (3 ln x−7)x
2
4.69
_
(x−2)
2
D
2
− 3(x−2)D + 4
_
y = x
A
NS y
C
= (x−2)
2
_
C
0
+C
1
ln
¸
¸
x−2
¸
¸
_
+x−
3
2
4.70 x
3
y
+ 3x
2
y
+ x y
− y = x
2
A
NS y = Cx +
1
√
x
_
Acos
_
√
3
2
ln x
_
+ Bsin
_
√
3
2
ln x
__
+
x
2
7
5
C H A P T E R
Applications of
Linear Differential Equations
5.1 Vibration of a Single DegreeofFreedom System
5.1.1 Formulation—Equation of Motion
In this section, the vibration of a single story shear building as shown in Figure
5.1, which is considered as a model of a single degreeoffreedom(DOF) system, is
studied.
A single story shear building consists of a rigid girder with mass m, which is
supported by columns with combined stiffness k. The columns are assumed to
be weightless, inextensible in the axial (vertical) direction, and they can only take
shear forces but not bending moments. In the horizontal direction, the columns
act as a spring of stiffness k. As a result, the girder can only move in the horizontal
direction, and its motion can be described by a single variable x(t); hence the
systemis called a single degreeoffreedom(DOF) system. The number of degrees
offreedom is the total number of variables required to describe the motion of a
system.
x(t)
x
0
(t)
m
Rigid girder
Weightless columns
Ground displacement
k
F(t)
c
Figure 5.1 A singlestory shear building.
188
5.1 vibration of a single degreeoffreedomsystem 189
The combined stiffness k of the columns can be determined as follows. Apply
a horizontal static force P on the girder. If the displacement of the girder is as
shown in Figure 5.2, then the combined stiffness of the columns is k =P/.
P
P
k
k
m
Figure 5.2 Determination of column stiffness.
The internal friction between the girder and the columns is described by a
viscous dashpot damper with damping coefﬁcient c. A dashpot damper is shown
schematically in Figure 5.3 and provides a damping force −c (v
B
−v
A
), where v
A
and v
B
are the velocities of points A and B, respectively, and (v
B
−v
A
) is the
relative velocity between points B and A. The damping force is opposite to the
direction of the relative velocity.
v
A
(t) v
B
(t)
A B c
Figure 5.3 A dashpot damper.
1. Vibration of a Shear Building under Externally Applied Force F(t)
In this case, the girder is subjected to an externally applied force F(t), which can be
a model of wind load. Consider the vibration of the girder; its freebody diagram
is drawn in Figure 5.4.
k
m
c
cx(t)
kx(t)
x(t) F(t)
Figure 5.4 Freebody diagram of the building under externally applied force.
The girder is subjected to the shear force (elastic force) kx(t), the viscous damp
ing force c ˙ x(t), and the externally applied load F(t). The equation of motion is
190 5 applications of linear differential equations
governed by Newton’s Second Law
→ ma =
F =⇒ m¨ x(t) = −kx(t) − c ˙ x(t) + F(t),
or
m¨ x(t) + c ˙ x(t) + kx(t) = F(t). (1)
2. Vibration of a Shear Building under Base Excitation x
0
(t)
In this case, the base (foundation) of the building is subjected to a dynamic dis
placement x
0
(t), which can be a model of an earthquake and is a known function.
The freebody diagram of the girder is shown in Figure 5.5. The shear (elastic)
force and the damping force applied on the girder are given by
Shear force = k· (Relative displacement between girder and base) = k(x−x
0
),
Damping force = c · (Relative velocity between girder and base) = c (˙ x−˙ x
0
).
Newton’s Second Law requires that
→ ma =
F =⇒ m¨ x(t) = −k
_
x(t)−x
0
(t)
_
− c
_
˙ x(t)−˙ x
0
(t)
_
.
k
Reference position
m
c
c(x−x
0
) k(x−x
0
)
x(t)
x
0
(t) y(t)
Figure 5.5 Freebody diagram of the building under base excitation.
Let y(t) =x(t)−x
0
(t), which is the relative displacement between the girder
and the base, be the new dependent variable. Then differentiating with respect
to t results in ˙ y(t) = ˙ x(t)−˙ x
0
(t), ¨ y(t) = ¨ x(t)−¨ x
0
(t). In terms of the relative
displacement y(t), the equation of motion is given by
m( ¨ y + ¨ x
0
) = −ky − c ˙ y,
i.e.,
m¨ y(t) + c ˙ y(t) + ky(t) = −m ¨ x
0
(t), (2)
where ¨ x
0
(t) is the base or ground acceleration. The loading on the girder created
from ground excitation (earthquake) is F(t) = −m¨ x
0
(t), which is proportional to
the mass of the girder and the ground acceleration.
5.1 vibration of a single degreeoffreedomsystem 191
Both systems (1) and (2) are secondorder linear ordinary differential equations
with constant coefﬁcients.
In general, a linear single degreeoffreedom system can be modeled by a me
chanical massdamperspring system. The example of a single story shear building
under externally applied load or base excitation can be described using the follow
ing equivalent massdamperspring system.
1. Single DegreeofFreedom System under Externally Applied Force
m
x(t)
F(t)
k
c
m
x, x, x
F(t)
cx
kx
m
x(t)
F(t)
k
c
Figure 5.6 A massdamperspring system under externally applied force.
2. Single DegreeofFreedom System under Base Excitation
m
x
0
(t) x(t)
k
c
m
x
0
(t)
x(t)
k
c
m
x, x, x
c(x−x
0
)
k(x−x
0
)
Figure 5.7 A massdamperspring system under base excitation.
192 5 applications of linear differential equations
(a)
c
k
m
y
O
x
y
0
(x)
y
0
K
k
U
c
y
Airplane
m
(b) (c)
Figure 5.8 Mathematical modeling of jet engine and landing gear.
In order to obtain the main dynamic characteristics and behavior, many engi
neering systems can be idealized as single degreeoffreedomsystems. For example,
consider the airplane shown in 5.8(a).
❧ Jet engines are supported by the wings of the airplane, which can be modeled
as cantilevers with variable crosssections. Vibration of a jet engine in the
vertical direction can be modeled as a single degreeoffreedom system as
shown in Figure 5.8(b).
❧ The landing gear of the airplane can be modeled as a mass m connected to
the airplane by a spring of stiffness K and a damper of damping coefﬁcient c.
A spring of stiffness k is used to model the forces on the tires. The airplane
moves at a constant speed U on a rough surface with proﬁle y
0
(x). Assuming
that the airplane moves in the horizontal direction only, the landing gear is
modeled as a single degreeoffreedom system as shown in Figure 5.8(c).
5.1 vibration of a single degreeoffreedomsystem 193
5.1.2 Response of a Single DegreeofFreedom System
FromSection 5.1.1, the equation of motion of a single degreeoffreedomsystemis
given by
m¨ x + c ˙ x + kx = F(t),
where F(t) is a known forcing function. It is a secondorder linear differential
equation with constant coefﬁcients, and can be rewritten in the standard form in
the theory of vibration by dividing both sides by m
¨ x +
c
m
˙ x +
k
m
x =
F(t)
m
.
Denoting
k
m
= ω
2
0
, ω
0
= natural circular frequency,
c
m
= 2ζ ω
0
, ζ = nondimensional damping coefﬁcient,
the equation of motion becomes
¨ x + 2ζ ω
0
˙ x + ω
2
0
x =
F(t)
m
.
The solution of the system x(t) consists of two parts: complementary solution and
particular solution.
❧ The complementary solution x
C
(t) is obtained when the righthand side of
the equation is set to zero, i.e., F(t) =0, implying that the system is not
subjected to loading. In the terminology of vibration, the system is in free
(not forced) vibration, and the solution of the equation is the response of free
vibration.
❧ The particular solution x
P
(t) corresponds to the righthand side of the equa
tion or the forcing term F(t), hence the response of forced vibration.
5.1.2.1 Free Vibration—Complementary Solution
The equation of motion is the complementary differential equation given by
¨ x + 2ζ ω
0
˙ x + ω
2
0
x = 0.
The characteristic equation is λ
2
+2ζ ω
0
λ+ω
2
0
=0, which gives
λ =
−2ζ ω
0
±
_
(2ζ ω
0
)
2
− 4×1×ω
2
0
2
= ω
0
_
−ζ ±
_
ζ
2
−1
_
.
Whether the characteristic equation has distinct real roots, double roots, or com
plex roots depends on the value of the nondimensional damping coefﬁcient ζ .
194 5 applications of linear differential equations
Case 1: Underdamped System 0ζ <1
Most engineering structures fall in this category with damping coefﬁcient ζ usually
less than 10%. The roots of the characteristic equation are
λ = ω
0
_
−ζ ± i
_
1−ζ
2
_
= −ζ ω
0
± i ω
d
,
where ω
d
=ω
0
_
1−ζ
2
is the damped natural circular frequency. The comple
mentary solution is
x
C
(t) = e
−ζ ω
0
t
(Acos ω
d
t + Bsin ω
d
t),
where constants A and B are determined fromthe initial conditions x(0) =x
0
and
˙ x(0) =v
0
. Since
˙ x
C
(t) = −ζ ω
0
e
−ζ ω
0
t
(Acos ω
d
t + Bsin ω
d
t)
+ e
−ζ ω
0
t
(−Aω
d
sin ω
d
t + Bω
d
cos ω
d
t)
= e
−ζ ω
0
t
_
(−ζ ω
0
A+ω
d
B) cos ω
d
t + (−ζ ω
0
B−ω
d
A) sin ω
d
t
_
,
substituting in the initial conditions yields
x
C
(0) = A = x
0
,
˙ x
C
(0) = −ζ ω
0
A + ω
d
B = v
0
=⇒ B =
v
0
+ζ ω
0
x
0
ω
d
.
Hence the response of free vibration is
x
C
(t) = e
−ζ ω
0
t
_
x
0
cos ω
d
t +
v
0
+ζ ω
0
x
0
ω
d
sin ω
d
t
_
, 0ζ <1.
Special Case: Undamped System ζ =0, ω
d
=ω
0
The response becomes
x
C
(t) = x
0
cos ω
0
t +
v
0
ω
0
sin ω
0
t
=
_
x
2
0
+
_
v
0
ω
0
_
2
⎡
⎣
x
0
_
x
2
0
+(v
0
/ω
0
)
2
cos ω
0
t +
v
0
/ω
0
_
x
2
0
+(v
0
/ω
0
)
2
sin ω
0
t
⎤
⎦
. ,, . . ,, . . ,, .
a cos ϕ sin ϕ
= a cos(ω
0
t −ϕ),
which is a harmonic function shown in Figure 5.9 with amplitude a and phase
angle ϕ given by
a =
_
x
2
0
+
_
v
0
ω
0
_
2
, ϕ = tan
−1
_
v
0
ω
0
x
0
_
.
5.1 vibration of a single degreeoffreedomsystem 195
x
0
acos(ω
0
t −ϕ)
x
C
(t)
t
Slope=v
0
at t=0
Amplitude a
−a
a
ϕ
Phase angle
Period T=
2π
ω
0
Figure 5.9 Response of undamped free vibration.
The harmonic solution has period T =2π/ω
0
; hence ω
0
is called the natu
ral circular frequency of the system, with unit rad/sec. The natural frequency is
f =ω
0
/(2π) =1/T, in cycles/sec or Hz. The maximum displacement is
max x
C
(t) = a =
_
x
2
0
+
_
v
0
ω
0
_
2
= amplitude of the motion.
Remarks: Using the trigonometric identities
sin(α+β) = sin α cos β + cos α sin β, cos(α−β) = cos α cos β + sin α sin β,
the amplitude of
x(t) = Acos ωt + Bsin ωt
can be determined as follows.
Rewrite the expression of x(t)
x(t) =
_
A
2
+B
2
_
A
_
A
2
+B
2
cos ωt +
B
_
A
2
+B
2
sin ωt
_
. ,, . . ,, .
cos ϕ sin ϕ
=
_
A
2
+B
2
_
cos ωt cos ϕ + sin ωt sin ϕ
_
=
_
A
2
+B
2
cos(ωt −ϕ), tan ϕ =
B
A
, ϕ = tan
−1
B
A
,
which gives the amplitude a=
_
A
2
+B
2
.
Alternatively,
x(t) =
_
A
2
+B
2
_
B
_
A
2
+B
2
sin ωt +
A
_
A
2
+B
2
cos ωt
_
. ,, . . ,, .
cos ψ sin ψ
196 5 applications of linear differential equations
=
_
A
2
+B
2
_
sin ωt cos ψ + cos ωt sin ψ
_
=
_
A
2
+B
2
sin(ωt +ψ), tan ψ =
A
B
, ψ = tan
−1
A
B
,
which also gives the amplitude a=
_
A
2
+B
2
.
Underdamped Free Vibration
The response of damped free vibration can be written as
x
C
(t) = ae
−ζ ω
0
t
cos(ω
d
t −ϕ),
where
a =
_
x
2
0
+
_
v
0
+ζ ω
0
x
0
ω
d
_
2
, ϕ = tan
−1
_
v
0
+ζ ω
0
x
0
ω
d
x
0
_
.
To sketch the response of damped free vibration, compare it with the response of
undamped free vibration. The difference lies in the amplitude: instead of having a
constant amplitude a, the amplitude ae
−ζ ω
0
t
decays exponentially with time. The
following steps are followed in sketching the response x
C
(t) (Figure 5.10).
1. Sketch the sinusoidal function cos(ω
d
t −ϕ), in which ω
d
is the damped
natural circular frequency (rad/sec) and ϕ is the phase angle. The period
is T
d
=2π/ω
d
and f
d
=ω
d
/(2π) =1/T
d
is the damped natural frequency in
cycles/sec or Hz.
2. Sketch the amplitude ae
−ζ ω
0
t
and its mirror image −ae
−ζ ω
0
t
in dashed
lines. These two lines form the envelope of the response.
3. Fit the sinusoidal function cos(ω
d
t −ϕ) inside the envelope to obtain the
response of damped free vibration.
The response of damped free vibration can be used to determine the damping
coefﬁcient ζ as follows:
At time t, the response is
x
C
(t) = e
−ζ ω
0
t
cos(ω
d
t −ϕ).
After a period T
d
, the response becomes
x
C
(t +T
d
) = e
−ζ ω
0
(t +T
d
)
cos
_
ω
d
(t +T
d
)−ϕ
_
= e
−ζ ω
0
(t +T
d
)
cos(ω
d
t −ϕ),
which leads to
x
C
(t +T
d
)
x
C
(t)
=
e
−ζ ω
0
(t +T
d
)
cos(ω
d
t −ϕ)
e
−ζ ω
0
t
cos(ω
d
t −ϕ)
= e
−ζ ω
0
T
d
.
5.1 vibration of a single degreeoffreedomsystem 197
cos(ω
d
t −ϕ)
ae
−ζω
0
t
cos(ω
d
t −ϕ)
x
C
(t)
t
t
−1
−a
−ae
−ζω
0
t
(Envelope)
ae
−ζω
0
t
1
a
ϕ
T
d
=
2π
ω
d
Figure 5.10 Response of underdamped free vibration.
Taking logarithm yields
ln
x
C
(t +T
d
)
x
C
(t)
= −ζ ω
0
T
d
,
in which ζ ω
0
T
d
is called the logarithmic decrement δ. For lightly damped struc
tures with 0<ζ
1,
δ = ζ ω
0
T
d
= ζ ω
0
2π
ω
d
= ζ ω
0
2π
ω
0
_
1−ζ
2
=
2πζ
_
1−ζ
2
≈ 2πζ.
Hence, the nondimensional damping coefﬁcient ζ is given by
ln
x
C
(t +T
d
)
x
C
(t)
= −2πζ =⇒ ζ =
1
2π
ln
x
C
(t)
x
C
(t +T
d
)
.
198 5 applications of linear differential equations
x
C
(t)
t
T
d
T
d
T
d
t
0
t
2
t
1
t
3
Figure 5.11 Determination of nondimensional damping coefﬁcient.
In practice, it is advantageous to select the time t such that x
C
(t) reaches the
maximum value. If t
0
, t
1
, . . . , t
n
are the n+1 consecutive times when x
C
(t) takes
maximum values as shown in Figure 5.11, then
ln
x
C
(t
0
)
x
C
(t
n
)
= ln
x
C
(t
0
)
x
C
(t
0
+nT
d
)
= ln
e
−ζ ω
0
t
0
e
−ζ ω
0
(t
0
+nT
d
)
= nζ ω
0
T
d
= 2nπζ
∴ ζ =
1
2nπ
ln
x
C
(t
0
)
x
C
(t
0
+nT
d
)
.
On the other hand, let
ζ
(i)
=
1
2π
ln
x
C
(t
i
)
x
C
(t
i
+T
d
)
, i = 0, 1, . . . , n−1,
be the n estimated values of the nondimensional damping coefﬁcient estimated
using the response values at times t
i
and t
i
+T
d
. The average of these n values is
given by
¯
ζ =
1
n
n−1
i=0
ζ
(i)
=
1
n
n−1
i=0
1
2π
ln
x
C
(t
i
)
x
C
(t
i
+T
d
)
=
1
2nπ
_
ln
x
C
(t
0
)
x
C
(t
0
+T
d
)
+ ln
x
C
(t
1
)
x
C
(t
1
+T
d
)
+ · · · + ln
x
C
(t
n−1
)
x
C
(t
n−1
+T
d
)
_
=
1
2nπ
ln
_
x
C
(t
0
)
x
C
(t
0
+T
d
)
·
x
C
(t
1
)
x
C
(t
1
+T
d
)
· · ·
x
C
(t
n−1
)
x
C
(t
n−1
+T
d
)
_
x
C
(t
i
+T
d
) =x
C
(t
i+1
), i = 0, 1, . . . , n−1
=
1
2nπ
ln
x
C
(t
0
)
x
C
(t
n
)
.
From the Central Limit Theorem in the theory of probability, it is known that
¯
ζ
approaches the true value of the nondimensional damping coefﬁcient when n→∞.
Hence, by using a larger value of n in the above equation, a better estimation of ζ
is achieved.
5.1 vibration of a single degreeoffreedomsystem 199
Case 2: Critically Damped System ζ =1
When ζ =1, the systemis called critically damped. The characteristic equation has
a double root λ= −ω
0
, −ω
0
. The complementary solution is given by
x
C
(t) = (C
0
+C
1
t)e
−ω
0
t
,
in which the constants C
0
and C
1
are determined from the initial conditions
x(0) =x
0
, ˙ x(0) =v
0
. Since
˙ x
C
(t) = C
1
e
−ω
0
t
− (C
0
+C
1
t)ω
0
e
−ω
0
t
,
one has
x
C
(0) = C
0
= x
0
, ˙ x
C
(0) = C
1
−C
0
ω
0
= v
0
=⇒ C
1
= v
0
+ ω
0
x
0
.
Hence, the response is
x
C
(t) =
_
x
0
+ (v
0
+ω
0
x
0
)t
_
e
−ω
0
t
, ζ =1.
A typical response of the free vibration of a critically damped system is shown in
Figure 5.12, which is not oscillatory and decays exponentially.
x
0
x
C
(t)
t
Slope=v
0
at t=0
Figure 5.12 Response of critically damped free vibration.
Case 3: Overdamped System ζ >1
The system is called overdamped when the nondimensional damping coefﬁcient
ζ >1. The characteristic equationhas twodistinct real roots λ=ω
0
_
−ζ ±
_
ζ
2
−1
_
.
The complementary solution is given by
x
C
(t) = C
1
e
−ω
0
(ζ −
√
ζ
2
−1)t
+ C
2
e
−ω
0
(ζ +
√
ζ
2
−1)t
,
in which the constants C
1
and C
2
are determined from the initial conditions
x(0) =x
0
, ˙ x(0) =v
0
. Since
˙ x
C
(t) = −ω
0
_
C
1
_
ζ −
_
ζ
2
−1
_
e
−ω
0
(ζ −
√
ζ
2
−1)t
+ C
2
_
ζ +
_
ζ
2
−1
_
e
−ω
0
(ζ +
√
ζ
2
−1)t
_
,
200 5 applications of linear differential equations
one obtains
x
C
(0) = C
1
+ C
2
= x
0
,
˙ x
C
(0) = −ω
0
_
C
1
_
ζ −
_
ζ
2
−1
_
+ C
2
_
ζ +
_
ζ
2
−1
_
_
= v
0
=⇒
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
C
1
=
1
2ω
0
_
ζ
2
−1
_
v
0
+
_
ζ +
_
ζ
2
−1
_
ω
0
x
0
_
,
C
2
= −
1
2ω
0
_
ζ
2
−1
_
v
0
+
_
ζ −
_
ζ
2
−1
_
ω
0
x
0
_
.
Hence, the response of free vibration of an overdamped system is
x
C
(t) =
1
2ω
0
_
ζ
2
−1
_
_
v
0
+
_
ζ +
_
ζ
2
−1
_
ω
0
x
0
_
e
−ω
0
(ζ −
√
ζ
2
−1)t
−
_
v
0
+
_
ζ −
_
ζ
2
−1
_
ω
0
x
0
_
e
−ω
0
(ζ +
√
ζ
2
−1)t
_
, ζ >1.
A typical plot of the response is shown in Figure 5.13, which is not oscillatory and
decays exponentially.
x
0
x
C
(t)
t
Slope=v
0
at t=0
Figure 5.13 Response of overdamped free vibration.
5.1.2.2 Forced Vibration—Particular Solution
For an underdamped system with 0<ζ <1, the complementary solution or the
response of free vibration is given by
x
C
(t) = e
−ζ ω
0
t
_
x
0
cos ω
d
t +
v
0
+ζ ω
0
x
0
ω
d
sin ω
d
t
_
,
which decays exponentially and approaches zero as t →∞, as shown in Figure
5.10. Hence the complementary solution is called the transient solution. Because
its value becomes negligible after some time, its effect is small and is not important
in practice.
The particular solution x
P
(t) is associated with the righthand side of the differ
ential equation and hence corresponds to forced vibration. The particular solution
5.1 vibration of a single degreeoffreedomsystem 201
is called the steadystate solution, because it is the solution that persists when time
is large. Suppose F(t) is periodic and of the form F(t) =F
0
sin t. Then the
particular solution satisﬁes
¨ x
P
+ 2ζ ω
0
˙ x
P
+ ω
2
0
x
P
=
F
0
m
sin t,
or, in the Doperator notation,
(D
2
+ 2ζ ω
0
D + ω
2
0
)x
P
=
F
0
m
sin t.
Hence
x
P
=
F
0
m
1
D
2
+ 2ζ ω
0
D + ω
2
0
sin t
=
F
0
m
1
−
2
+ 2ζ ω
0
D + ω
2
0
sin t
Theorem 3 of Chapter 4:
replace D
2
by −
2
.
=
F
0
m
(ω
2
0
−
2
)−2ζ ω
0
D
_
(ω
2
0
−
2
) + 2ζ ω
0
D
__
(ω
2
0
−
2
)−2ζ ω
0
D
_ sin t
=
F
0
m
(ω
2
0
−
2
)−2ζ ω
0
D
(ω
2
0
−
2
)
2
−(2ζ ω
0
D)
2
sin t
=
F
0
m
(ω
2
0
−
2
) sin t −2ζ ω
0
cos t
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
.
Replace D
2
by −
2
in denominator, and
evaluate numerator.
Deﬁning angle ϕ by
cos ϕ =
ω
2
0
−
2
_
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
, sin ϕ =
2ζ ω
0
_
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
,
the response of forced vibration becomes
x
P
(t) =
F
0
m
1
_
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
sin(t −ϕ).
The applied force F(t) is maximum at t =
1
2
π,
3
2
π, . . . ; the response x
P
(t) is
maximum at t −ϕ =
1
2
π,
3
2
π, . . . . Hence the response x
P
(t) lags behind the
forcing by a time ϕ/. The angle ϕ is called a phase angle or phase lag.
The amplitude of the response of forced vibration is
¸
¸
x
P
(t)
¸
¸
max
=
F
0
m
1
_
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
.
202 5 applications of linear differential equations
Denoting
r =
ω
0
=
Excitation frequency
Undamped natural frequency
= Frequency ratio,
one has
¸
¸
x
P
(t)
¸
¸
max
=
F
0
mω
2
0
_
_
1−
_
ω
0
_
2
_
2
+
_
2ζ
ω
0
_
2
=
F
0
mω
2
0
_
(1−r
2
)
2
+(2ζ r)
2
=
F
0
k
1
_
(1−r
2
)
2
+(2ζ r)
2
, ω
2
0
=
k
m
=⇒ k = mω
2
0
.
If dynamic effect is not considered, i.e., if only static terms are considered and the
shaded dynamic terms in the following equation are dropped, one obtains
m¨ x + c ˙ x + kx = F(t) = F
0
sin t =⇒ x
static
=
F
0
k
,
.,,. .,,. .,,. . ,, .
Dynamic Dynamic Static Dynamic
inertia damping elastic variation
force force force
which is the static displacement of the system under the static force F
0
.
Deﬁne
¸
¸
x
P
(t)
¸
¸
max
x
static
=
1
_
(1−r
2
)
2
+(2ζ r)
2
= Dynamic Magniﬁcation Factor.
The Dynamic Magniﬁcation Factor (DMF) is plotted in Figure 5.14 for various
values of the nondimensional damping coefﬁcient ζ . It is one of the most important
quantities describing the dynamic behavior of an underdamped single degreeof
freedom system under sinusoidal excitation.
❧ When r →0, i.e., when the excitation frequency →0, DMF→1. In this
case, the dynamic excitation approaches a static force and the amplitude of
dynamic response approaches the static displacement.
❧ When r →∞, i.e., when the excitation frequency is large compared to
the natural frequency ω
0
, DMF→0, which means that the dynamic response
approaches zero. This result can be understood intuitively as follows: the
system is excited (pushed and pulled) at such a high frequency that it does
not “know” which way to move so that it just “stands” still or there is no
response.
❧ When r ≈1 or the excitation frequency is close to the natural frequency
ω
0
, ≈ω
0
, DMF tends to large values for small damping.
5.1 vibration of a single degreeoffreedomsystem 203
0
1
2
3
4
5
6
7
0.5 1 1.5 2 2.5
ζ=0
ζ=0.1
ζ=0.2
ζ=0.3
ω
0
r=
D
y
n
a
m
i
c
M
a
g
n
i
f
i
c
a
t
i
o
n
F
a
c
t
o
r
(
D
M
F
)
Figure 5.14 Dynamic magniﬁcation factor (DMF).
❧ Maximum value of DMF occurs when d(DMF)/dr =0, or
d(DMF
2
)
d(r
2
)
=
d
dρ
_
1
(1−ρ)
2
+4ζ
2
ρ
_
= −
2(1−ρ)(−1) + 4ζ
2
[(1−ρ)
2
+4ζ
2
ρ]
2
= 0, ρ = r
2
,
ρ − 1 + 2ζ
2
= 0 =⇒ r
2
= 1 − 2ζ
2
,
∴ r =
_
1−2ζ
2
≈ 1 − ζ
2
≈ 1, if ζ
1.
The maximum value of DMF is approximately
DMF
max
≈ DMF
¸
¸
r=1
=
1
_
(1−r
2
)
2
+(2ζ r)
2
¸
¸
¸
¸
¸
r=1
=
1
2ζ
.
Hence, the smaller the damping coefﬁcient, the larger the DMF values or the
amplitudes of dynamic response. When ζ =0 and =ω
0
, a response of
unbound amplitude occurs and the system is in resonance.
Resonance
When ζ =0 and =ω
0
, the equation of motion becomes
¨ x
P
+ ω
2
0
x
P
=
F
0
m
sin ω
0
t,
and the particular solution is given by
x
P
(t) =
F
0
m
1
D
2
+ω
2
0
sin ω
0
t =
F
0
m
Im
_
1
D
2
+ω
2
0
e
i ω
0
t
_
.
204 5 applications of linear differential equations
Applying Theorem 4 of Chapter 4,
φ(D) = D
2
+ ω
2
0
, φ(i ω
0
) = (i ω
2
0
)
2
+ ω
2
0
= 0,
φ
(D) = 2D, φ
(i ω
0
) = 2i ω
0
= 0,
∴ x
P
(t) =
F
0
m
Im
_
1
φ
(i ω
0
)
t e
i ω
0
t
_
Theorem 4 of Chapter 4
=
F
0
m
Im
_
1
2i ω
0
t (cos ω
0
t + i sin ω
0
t)
_
= −
F
0
2mω
0
t cos ω
0
t.
t
q(t)
Figure 5.15 Response of a system in resonance.
To sketch the response, F
0
t/(2mω
0
), which is a straight line, is regarded as the
amplitude of the response. This straight line and its mirror image −F
0
t/(2mω
0
)
form the envelop of the sinusoidal function −cos ω
0
t. Fitting the sinusoidal func
tion −cos ω
0
t inside the envelop results in the response as shown in Figure 5.15.
Hence, when the system is undamped and the excitation frequency is equal to
the natural frequency, the systemis in resonance and the amplitude of the response
of the system grows linearly with time.
Undamped System under Sinusoidal Excitation
For a damped system, the complementary solution or the response of free vibration
(due to initial conditions) decays exponentially to zero; hence its effect diminishes
when time increases. It is therefore important to focus on the particular solution or
the response of forced vibration (due to externally applied forcing).
However, for an undamped system, the effect of response due to free vibration is
signiﬁcant. Furthermore, it interacts with the response due to forced vibration to
produce a response that is different from either the complementary solution or the
particular solution alone.
5.1 vibration of a single degreeoffreedomsystem 205
The equation of motion of an undamped system under sinusoidal excitation is
¨ x + ω
2
0
x =
F
0
m
sin t,
and the complementary solution and a particular solution are
x
C
(t) = Acos ω
0
t + Bsin ω
0
t, x
P
(t) =
F
0
m
sin t
ω
2
0
−
2
.
Hence, the general solution is given by
x(t) = x
C
(t) + x
P
(t) = Acos ω
0
t + Bsin ω
0
t +
F
0
m(ω
2
0
−
2
)
sin t,
where the constants A and B are determined by the initial conditions x(0) =x
0
and ˙ x(0) =v
0
. Since
˙ x(t) = −Aω
0
sin ω
0
t + Bω
0
cos ω
0
t +
F
0
m(ω
2
0
−
2
)
cos t,
one has x(0) =A=x
0
, and
˙ x(0) = Bω
0
+
F
0
m(ω
2
0
−
2
)
= v
0
=⇒ B =
1
ω
0
_
v
0
−
F
0
m(ω
2
0
−
2
)
_
.
The response of the system is
x(t) = x
0
cos ω
0
t +
1
ω
0
_
v
0
−
F
0
m(ω
2
0
−
2
)
_
sin ω
0
t +
F
0
m(ω
2
0
−
2
)
sin t
= x
0
cos ω
0
t +
1
ω
0
_
v
0
+
F
0
(ω
0
−)
m(ω
2
0
−
2
)
_
sin ω
0
t −
F
0
(sin ω
0
t − sin t)
m(ω
2
0
−
2
)
= x
0
cos ω
0
t +
1
ω
0
_
v
0
+
F
0
m(ω
0
+)
_
sin ω
0
t −
F
0
(sin ω
0
t − sin t)
m(ω
2
0
−
2
)
,
. ,, .
X(t)
in which the ﬁrst two terms are sinusoidal functions of frequency ω
0
. The last term
can be written as
X(t) =
F
0
(sin ω
0
t − sin t)
m(ω
2
0
−
2
)
=
2F
0
m(ω
2
0
−
2
)
sin
_
ω
0
−
2
t
_
cos
_
ω
0
+
2
t
_
.
. ,, .
a(t)
sin A−sin B=2 cos
A+B
2
sin
A−B
2
It has been shown that the response due to forced vibration is of large amplitude
when the excitation frequency is close to the natural frequency.
206 5 applications of linear differential equations
t
t
Amplitude a(t)
−a(t)
t
X(t)
2
t
cos
+ω
0
( )
Figure 5.16 Plotting of X(t).
5.1 vibration of a single degreeoffreedomsystem 207
–20
–10
0
10
20
100 200 300 400 500
–40
–20
0
20
40
100 200 300 400 500
–100
–50
0
50
100
100 200 300 400 500
–200
–100
0
100
200
100 200 300 400 500
t
X(t)
t
X(t)
t
X(t)
t
X(t)
ω
0
=1, =0.95
ω
0
=1, =0.98
ω
0
=1, =0.99
ω
0
=1, =0.999
Figure 5.17 Beats.
208 5 applications of linear differential equations
In the following, the behavior of X(t) when the excitation frequency is close
to the natural frequency ω
0
is studied. For simplicity of discussion, let <ω
0
.
The case when >ω
0
can be discussed similarly.
When ≈ω
0
, (ω
0
+)/2≈ω
0
or , and (ω
0
−)/2 is small. To sketch X(t),
consider
a(t) =
2F
0
m(ω
2
0
−
2
)
sin
_
ω
0
−
2
t
_
as the amplitude of the sinusoidal function cos
_
(ω
0
+)t/2
_
. The amplitude a(t)
is itself a sinusoidal function but with a smaller frequency (ω
0
−)/2 or slower
variation. The amplitude of a(t) increases when approaches ω
0
.
As shown in Figure 5.16, one can ﬁrst plot the amplitude a(t) and its mirror
image −a(t); both curves form the envelope. By ﬁtting the sinusoidal function
cos
_
(ω
0
+)t/2
_
inside the envelope, one obtains the response X(t).
Note that a(t) is positive in the 1st, 3rd, 5th, . . . halfperiods and negative
in the 2nd, 4th, 6th, . . . halfperiods. It should be emphasized that when ﬁt
ting the sinusoidal function cos
_
(ω
0
+)t/2
_
inside the envelope, the maximum
and minimum values of cos
_
(ω
0
+)t/2
_
correspond approximately to the lo
cal maximum and minimum values of the resulting X(t) in the 1st, 3rd, 5th, . . .
halfperiods. On the other hand, in the 2nd, 4th, 6th, . . . halfperiods, the maxi
mumvalues of cos
_
(ω
0
+)t/2
_
correspond approximately to the local minimum
values of the resulting X(t) and vice versa.
Since the amplitude of X(t) varies sinusoidally, X(t) is said to be amplitude
modulated. Such functions are also called beats. Beats X(t) are obtained from the
addition of two sinusoidal functions with close frequencies ω
0
and .
Typical results of X(t) are plotted in Figure 5.17 for F
0
=1, m=1, ω
0
=1, and
various values of . It can be seen that the closer the two frequencies ω
0
and ,
the slower the variation and the larger the magnitude of a(t).
When →ω
0
, it takes time t →∞ for the amplitude a(t) to complete a half
period variation; as a result, the variation of the amplitude a(t) seems to approach
a straight line. In fact, it is easy to show that
lim
→ω
0
a(t) =
2F
0
m
lim
→ω
0
sin
_
ω
0
−
2
t
_
ω
2
0
−
2
=
2F
0
m
lim
→ω
0
sin
_
(ω
0
−)
t
2
_
(ω
0
−)
·
1
(ω
0
+)
=
2F
0
m
·
t
2
·
1
2ω
0
=
F
0
2mω
0
t, L'Hospital's Rule
which grows linearly with time t and is the same as the amplitude obtained for the
resonant case =ω
0
.
5.2 electric circuits 209
5.2 Electric Circuits
Series RLC Circuit
A circuit consisting of a resistor R, an inductor L, a capacitor C, and a voltage
source V(t) connected in series, shown in Figure 5.18, is called the series RLC
circuit. Applying Kirchhoff ’s Voltage Law, one has
−V(t) + Ri + L
di
dt
+
1
C
_
t
−∞
i dt = 0.
V(t)
R
C
L
i
m=L
x(t)=i(t)
F(t)=
k=
1
C
c=R
dV(t)
dt
Figure 5.18 Series RLC circuit.
Differentiating with respect to t yields
L
d
2
i
dt
2
+ R
di
dt
+
1
C
i =
dV(t)
dt
,
or, in the standard form,
d
2
i
dt
2
+ 2ζ ω
0
di
dt
+ ω
2
0
i =
1
L
dV(t)
dt
, ω
2
0
=
1
LC
, ζ ω
0
=
R
2L
.
The series RLC circuit is equivalent to a massdamperspring system as shown.
Parallel RLC Circuit
A circuit consisting of a resistor R, an inductor L, a capacitor C, and a current
source I(t) connected in parallel, as shown in Figure 5.19, is called the parallel
RLC circuit. Applying Kirchhoff ’s Current Law at node 1, one has
I(t) = C
dv
dt
+
1
L
_
t
−∞
vdt +
v
R
.
Differentiating with respect to t yields
C
d
2
v
dt
2
+
1
R
dv
dt
+
1
L
v =
dI(t)
dt
,
or, in the standard form,
d
2
v
dt
2
+ 2ζ ω
0
dv
dt
+ ω
2
0
v =
1
C
dI(t)
dt
, ω
2
0
=
1
LC
, ζ ω
0
=
1
2RC
.
The parallel RLC circuit is equivalent to a massdamperspring system as shown.
210 5 applications of linear differential equations
R L
I(t)
v
1
m=C
x(t)=v(t)
F(t)=
k=
1
L
dI(t)
dt
C
c=
1
R
Figure 5.19 Parallel RLC circuit.
Example 5.1 — Automobile Ignition Circuit 5.1
An automobile ignition system is modeled by the circuit shown in the following
ﬁgure. The voltage source V
0
represents the battery and alternator. The resistor R
models the resistance of the wiring, and the ignition coil is modeled by the inductor
L. The capacitor C, known as the condenser, is in parallel with the switch, which is
known as the electronic ignition. The switch has been closed for a long time prior
to t <0
−
. Determine the inductor voltage v
L
for t >0.
V
0
t =0
R
C
L
Spark Plug
Ignition Coil
v
C
i
v
L
For V
0
=12 V, R=4 , C=1 μF, L=8 mH, determine the maximal inductor
voltage and the time when it is reached.
❧ For t <0, the switch is closed, the capacitor behaves as an open circuit and the
inductor behaves as a short circuit as shown. Hence i(0
−
) =V
0
/R, v
C
(0
−
) =0.
V
0
R
v
C
(0
−
) i(0
−
)
v
L
(0
−
)
t 0
–
t 0
+
V
0
R
C
L
Ignition Coil
Mesh
v
C
i
v
L
❧ At t =0, the switch is opened. Since the current in an inductor and the voltage
across a capacitor cannot change abruptly, one has i(0
+
) =i(0
−
) =V
0
/R, v
C
(0
+
)=
v
C
(0
−
) =0. The derivative i
(0
+
) is obtained from v
L
(0
+
), which is determined by
5.2 electric circuits 211
applying Kirchhoff ’s Voltage Law to the mesh at t =0
+
:
−V
0
+ Ri(0
+
) + v
C
(0
+
) + v
L
(0
+
) = 0 =⇒ v
L
(0
+
) = V
0
− Ri(0
+
) = 0,
v
L
(0
+
) = L
di(0
+
)
dt
=⇒ i
(0
+
) =
v
L
(0
+
)
L
= 0.
A mesh is a loopwhich does not contain any other loops within it.
❧ For t >0, applying Kirchhoff ’s Voltage Law to the mesh leads to
−V
0
+ Ri +
1
C
_
t
−∞
i dt + L
di
dt
= 0.
Differentiating with respect to t yields
R
di
dt
+
i
C
+ L
d
2
i
dt
2
= 0 =⇒
d
2
i
dt
2
+ 2ζ ω
0
di
dt
+ ω
2
0
i =0, ω
2
0
=
1
LC
, ζ ω
0
=
R
2L
.
If ζ <1, the system is underdamped and the solution of the differential equation is
i(t) = e
−ζ ωt
_
i(0
+
) cos ω
d
t +
i
(0
+
)+ζ ω
0
i(0
+
)
ω
d
sin ω
d
_
= i(0
+
)e
−ζ ω
0
t
_
cos ω
d
t +
ζ ω
0
ω
d
sin ω
d
t
_
,
where ω
d
=ω
0
_
1−ζ
2
is the damped natural frequency.
The voltage across the inductor is
v
L
(t) = L
di(t)
dt
= −Li(0
+
)
ω
2
0
ω
d
e
−ζ ω
0
t
sin ω
d
t.
For V
0
=12 V, R=4 , C=1 μF=1×10
−6
F, L=8 mH=8×10
−3
H,
ω
0
=
1
√
LC
=
1
√
1×10
−6
×8×10
−3
= 1.118×10
4
,
ζ ω
0
=
R
2L
=
4
2×8×10
−3
= 250, ζ =
250
ω
0
= 0.02236,
ω
d
= ω
0
_
1−ζ
2
≈ ω
0
= 1.118×10
4
, i(0
+
) =
V
0
R
=
12
4
= 3,
∴ v
L
= −8×10
−3
×3×1.118×10
4
e
−250t
sin(1.118×10
4
t)
= −268.32e
−250t
sin(1.118×10
4
t),
which is maximum when 1.118×10
4
t =π/2 or t =1.405×10
−4
sec =140.5 μs
and is given by
v
L,max
(t) = −268.32e
−250×1.405×10
−4
= −259V.
A device known as a transformer is then used to step up the inductor voltage to the
range of 6000 to 10,000 V required to ﬁre the spark plug in a typical automobile.
212 5 applications of linear differential equations
Example 5.2 — SecondOrder Circuit 5.2
For the electric circuit showninthe following ﬁgure, derive the differential equation
governing i(t) for t >0.
t =0 v
C
i
I
2
I
1
C
L
R
2
R
1
For R
1
=6 , R
2
=2 , C=0.04 F, L=1 H, I
1
=I
2
=2A, ﬁnd i(t) and v
C
(t).
❧ For t <0, the switch is open. The inductor behaves as a short circuit and the
capacitor behaves as an open circuit. Hence, i(0
−
) =0, i
1
(0
−
) = −I
1
, i
2
(0
−
) =I
2
,
and
v
R2
(0
−
) =v
R1
(0
−
)+v
C
(0
−
) =⇒ v
C
(0
−
) =R
2
I
2
− R
1
I
1
.
t 0
–
t 0
+
v
i
1
i
1
(0
−
)
i
2
(0
−
)
i(0
−
)
v(0
−
) 1
v
C
v
L
i
I
1
C
L
R
1
v
C
(0
−
)
I
2
I
1
C
L
R
2
R
1
❧ For t =0, the switch is closed, the current source I
2
and the resistor R
2
are
shortcircuited. The circuit becomes as shown in the ﬁgure. Since the current in
an inductor cannot change abruptly, i(0
+
) =i(0
−
) =0. Since the voltage across a
capacitor cannot change abruptly, v
C
(0
−
) =v
C
(0
+
) =R
2
I
2
−R
1
I
1
.
Note that v(0
+
) =v
L
(0
+
)+v
C
(0
+
) =v
L
(0
+
)+R
2
I
2
−R
1
I
1
,
i
1
(0
+
) =
v(0
+
)
R
1
=
v
L
(0
+
)+R
2
I
2
−R
1
I
1
R
1
.
Applying Kirchhoff ’s Current Law at node 1 yields
I
1
+ i(0
+
) + i
1
(0
+
) = 0 =⇒ I
1
+ 0 +
v
L
(0
+
)+R
2
I
2
−R
1
I
1
R
1
= 0,
∴ V
0
(0
+
) = −R
2
I
2
= L
di(0
+
)
dt
=⇒ i
(0
+
) = −
R
2
I
2
L
.
5.3 vibration of a vehicle passing a speed bump 213
❧ For t >0,
i
1
=
V
R
1
, v = v
L
+ v
C
= L
di
dt
+
1
C
_
t
−∞
i dt.
Applying Kirchhoff ’s Current Law at node 1 leads to
I
1
+ i + i
1
= 0 =⇒ I
1
+ i +
1
R
1
_
L
di
dt
+
1
C
_
t
−∞
i dt
_
= 0.
Differentiating with respect to t yields
d
2
i
dt
2
+
R
1
L
di
dt
+
1
CL
i = 0, i(0
+
) = 0, i
(0
+
) = −
R
2
I
2
L
.
For R
1
=6 , R
2
=2 , C=0.04 F, L=1 H, I
1
=I
2
=2A, the equation becomes
d
2
i
dt
2
+ 6
di
dt
+ 25i = 0, i(0
+
) = 0, i
(0
+
) = −4,
ω
0
=5, 2ζ ω
0
=6 =⇒ ζ ω
0
=3, ζ =
3
5
<1, ω
d
=ω
_
1−ζ
2
=4.
The system is underdamped and the solution is given by
i(t) = e
−ζ ω
0
t
_
i(0
+
) cos ω
d
t +
i
(0
+
)+ζ ω
0
i(0
+
)
ω
d
sin ω
d
t
_
= −e
−3t
sin 4t.
The voltage across the capacitor is, with v
C
(0
+
) = −8,
v
C
=
1
C
_
t
0
i(t)dt + v
C
(0
+
) = 25
_
t
0
−e
−3t
sin 4t − 8
= −12 + e
−3t
(3 sin 4t + 4 cos 4t) (V).
Remarks: Circuits consisting of resistors and the equivalent of two energy
storage elements, such as capacitors and inductors, are called secondorder cir
cuits, because they are characterized by secondorder linear ordinary differential
equations. The governing equations are of the same formas that of a single degree
offreedom system. As a result, the solutions and the behavior of secondorder
circuits are the same as those of a single degreeoffreedom system.
5.3 Vibration of a Vehicle Passing a Speed Bump
As an application of single degreeoffreedom system, the vibration of a vehicle
passing a speed bump is studied in this section.
The vehicle is modeled by a damped single degreeoffreedomsystemwith mass
m, spring stiffness k, and damping coefﬁcient c as shown in Figure 5.20. The
vehicle has been moving at a constant speed U on a smooth surface.
214 5 applications of linear differential equations
y
0
(x)
x
b
h
k
U
t=0
m
c
y
O
Figure 5.20 A vehicle passing a speed bump.
y
0
k
U
m
c
y
m
y, y, y
k(y −y
0
) c(y −y
0
)
Figure 5.21 Freebody diagram of a vehicle passing a speed bump.
At time t =0, the vehicle reaches a speed bump with a proﬁle of a halfsine curve
y
0
(x) =h sin(πx/b), 0<x<b. The absolute displacement of the mass is described
by y(t). Determine the response of the vehicle in terms of the relative displacement
z(t) = y(t)−y
0
(t) for t >0.
To set up the equation of motion of the vehicle, consider the freebody diagram
of the mass m as shown in Figure 5.21. Newton’s Second Law requires that
↑ ma =
F : m ¨ y = −c ( ˙ y − ˙ y
0
) − k( y − y
0
).
Letting the relative displacement of the mass be z(t) = y(t)−y
0
(t), the equation of
motion becomes
m(¨ z + ¨ y
0
) = −c ˙ z − kz =⇒ m¨ z + c ˙ z + kz = −m ¨ y
0
. (1)
Phase 1: On the Speed Bump 0t T, T =b/U
Since the vehicle is moving at the constant speed U, one has x =Ut. The speed
bump is of the halfsine shape
y
0
(x) = h sin
_
π
b
x
_
, 0x b =⇒ y
0
(t) = h sin
_
πU
b
t
_
, 0t T.
5.3 vibration of a vehicle passing a speed bump 215
The equation of motion can be written as
m¨ z +c ˙ z +kz = mh
2
sin t =⇒ (D
2
+2ζ ω
0
D+ω
2
0
)z = h
2
sin t, (2)
where
ω
0
=
_
k
m
, 2ζ ω
0
=
c
m
, =
πU
b
.
The characteristic equation is λ
2
+2ζ ω
0
λ+ω
2
0
=0, which gives, assuming ζ <1,
λ = −ζ ω
0
± ω
0
_
ζ
2
−1 = −ζ ω
0
± i ω
d
, ω
d
= ω
0
_
1−ζ
2
.
The complementary solution is
z
C
(t) = e
−ζ ω
0
t
(A
1
cos ω
d
t + B
1
sin ω
d
t).
A particular solution can be obtained using the Doperator method
z
P
(t) =
1
D
2
+2ζ ω
0
D+ω
2
0
_
h
2
sin t
_
= h
2
1
−
2
+2ζ ω
0
D+ω
2
0
sin t
Theorem 3 of Chapter 4:
replace D
2
by −
2
.
= h
2
(ω
2
0
−
2
)−2ζ ω
0
D
(ω
2
0
−
2
)
2
−(2ζ ω
0
D)
2
sin t
=
h
2
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
_
(ω
2
0
−
2
) sin t − 2ζ ω
0
cos t
_
Replace D
2
by −
2
in denominator and evaluate numerator.
= a sin(t −ϕ),
where
a =
h
2
_
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
, ϕ = tan
−1
2ζ ω
0
ω
2
0
−
2
.
Use Asin θ −Bcos θ =
_
A
2
+B
2
sin(θ −ϕ), ϕ = tan
−1
(B/A).
The general solution is, for 0t T,
z(t) = z
C
(t) + z
P
(t) = e
−ζ ω
0
t
(A
1
cos ω
d
t + B
1
sin ω
d
t) + a sin(t −ϕ), (3)
where the constants are determined by the initial conditions z(0) and ˙ z(0). Since
the vehicle has been traveling on a smooth surface, y(0) =0 and ˙ y(0) =0; hence
z(0) =0 and ˙ z(0) =0. Since
˙ z(t) = −ζ ω
0
e
−ζ ω
0
t
(A
1
cos ω
d
t + B
1
sin ω
d
t)
+ e
−ζ ω
0
t
(−A
1
ω
d
sin ω
d
t + B
1
ω
d
cos ω
d
t) + acos(t −ϕ), (4)
216 5 applications of linear differential equations
one has
z(0) = A
1
+ a sin(−ϕ) = 0 =⇒ A
1
= a sin ϕ,
˙ z(0) = −ζ ω
0
A
1
+ B
1
ω
d
+ acos(−ϕ) = 0
=⇒ B
1
=
1
ω
d
(ζ ω
0
A−acos ϕ) =
a
ω
d
(ζ ω
0
sin ϕ−cos ϕ).
At time t =T, T =b/U, the relative displacement and velocity of the vehicle are
z(T) and ˙ z(T), which can be obtained by evaluating equations (3) and (4) at t =T.
Phase 2: Passed the Speed Bump t T, T =b/U
For time t T, T =b/U, the vehicle has gone over the speed bump and the surface
is smooth again with y
0
(t) =0. The equation of motion becomes
m¨ z + c ˙ z + kz = 0 =⇒ (D
2
+ 2ζ ω
0
D + ω
2
0
)z = 0, t T. (5)
The vehicle is not forced and the response is due to free vibration with initial
conditions at time t =T.
The complementary solution, which is also the general solution, is
z(t) = e
−ζ ω
0
t
(A
2
cos ω
d
t + B
2
sin ω
d
t), t T,
˙ z(t) = −ζ ω
0
e
−ζ ω
0
t
(A
2
cos ω
d
t + B
2
sin ω
d
t)
+ e
−ζ ω
0
t
(−A
2
ω
d
sin ω
d
t + B
2
ω
d
cos ω
d
t),
where constants A
2
and B
2
are determined from the initial conditions z(T) and
˙ z(T) obtained in Phase 1:
z(T) = e
−ζ ω
0
T
_
A
2
cos(ω
d
T) + B
2
sin(ω
d
T)
_
,
˙ z(T) = e
−ζ ω
0
T
__
−ζ ω
0
cos(ω
d
T) − ω
d
sin(ω
d
T)
_
A
2
+
_
−ζ ω
0
sin(ω
d
T) + ω
d
cos(ω
d
T)
_
B
2
_
,
or
α
11
A
2
+ α
12
B
2
= β
1
, α
21
A
2
+ α
22
B
2
= β
2
,
where
α
11
= cos(ω
d
T), α
21
= −ζ ω
0
cos(ω
d
T) − ω
d
sin(ω
d
T),
α
12
= sin(ω
d
T), α
22
= −ζ ω
0
sin(ω
d
T) + ω
d
cos(ω
d
T),
β
1
= e
ζ ω
0
T
z(T), β
2
= e
ζ ω
0
T
˙ z(T).
Using Gaussian elimination or Cramer’s Rule, it is easy to solve for A
2
and B
2
as
A
2
=
α
22
β
1
− α
12
β
2
α
11
α
22
− α
21
α
12
, B
2
=
α
11
β
2
− α
21
β
1
α
11
α
22
− α
21
α
12
.
See page 307 for a brief review on Cramer's Rule.
5.3 vibration of a vehicle passing a speed bump 217
Numerical Results
As a numerical example, use the following parameters
f
0
=3 Hz =⇒ ω
0
=2π f
0
=6π rad/sec, ζ =0.1,
U =1.8 km/hr =0.5 m/sec, b =0.5 m, h =0.1 m, T =b/U =1 sec.
When 0t 1 sec, the vehicle is on the speed bump; whereas when t 1 sec, the
vehicle has passed over the speed bump. It is easy to evaluate that
ω
d
=18.755 rad/sec, a =0.0028555 m, ϕ =0.034272 rad,
A
1
=0.000097844 m, B
1
= −0.00046819 m,
∴ z(t) =e
−1.8850t
_
0.000097844 cos(18.755t) − 0.00046819 sin(18.755t)
_
+ 0.0028555 sin(3.1416t −0.0343) (m), 0t 1.
At t =1 sec, z(1) =0.00011934 m, ˙ z(1) = −0.010307 m/sec. In the second phase,
t 1,
α
11
= 0.99554, α
12
= −0.094344,
α
21
= −0.10712 rad/sec, α
12
= 18.849 rad/sec,
β
1
= 0.00078599 m, β
2
= −0.067883 m/sec,
A
2
= 0.00044846 m, B
2
= −0.0035988 m,
∴ z(t) =e
−1.8850t
_
0.00044846 cos(18.755t) − 0.0035988 sin(18.755t)
_
(m).
The relative displacement response z(t) is shown in Figure 5.22 for ζ =0.1 and
0.01. It can be seen that the response decays rapidly for large values of ζ .
–0.001
0
0.001
0.002
0.003
0.5 1 1.5 2 2.5 3
t (sec)
z (m)
ζ=0.01
ζ=0.1
Figure 5.22 Response of a vehicle passing a speed bump.
218 5 applications of linear differential equations
5.4 BeamColumns
A beamcolumn is a structural member subjected simultaneously to axial load and
bending moments producedby lateral forces or eccentricity of the axial load. Beam
columns are found in many engineering structures. For example, a water tower as
shown in Figure 5.23 is a beamcolumn; the supporting column is subjected to the
axial load due to the weight of the water tank and the lateral load due to wind.
W
Wind
Load
Water Tower BeamColumn
x
y
w(x)=w
EI, L
Figure 5.23 Water tower.
x
EI, L
y(x)
y
w(x)
w(x)
P
A
P
V+V
B
A
C
M+M
M
V
P
B
P
y
y+y
x
x
x
Figure 5.24 Beamcolumn.
Consider a beamcolumn supported at two ends A and B as shown in Figure
5.24, which is subjected to the axial load P and the lateral distributed load w(x).
The ﬂexural rigidity and the length of the beamcolumn are EI and L, respectively.
Study the equilibrium of a segment of the beamcolumn of length x, with its
freebody diagram as shown. The shear force V(x) at x is changed to V(x)+V
at x+x; the bending moment M(x) at x is changed to M(x)+M at x+x.
5.4 beamcolumns 219
For small x, the distributed load is approximated by the uniformly distributed
load of intensity w(x) over the beam segment.
Summing up the forces in the y direction and taking the limit as x→0 yield
↓
F
y
=0:
_
V(x)+V
_
− V(x) + w(x)x = 0 =⇒
dV
dx
= −w(x).
Summing up the moments about point C, which is the midpoint of the beam
segment, gives
M
C
=0:
_
M(x)+M
_
− M(x) + V(x) ·
x
2
+
_
V(x)+V
_
·
x
2
+ P· y = 0 =⇒
dM
dx
+ P
dy
dx
+ V =0.
Table 5.1 Boundary conditions.
x=l
P P
Pinned End:
Deﬂection = 0 =⇒ y(l) = 0
Moment = 0 =⇒ y
(l) = 0
x=l
P P
Clamped End:
Deﬂection = 0 =⇒ y(l) = 0
Slope = 0 =⇒ y
(l) = 0
x=l
P P
Free End:
Moment = 0 =⇒ y
(l) = 0
Shear Force = 0 =⇒ y
(l) + α
2
y
(l) = 0
x=l
P P
Sliding End:
Slope = 0 =⇒ y
(l) = 0
Shear Force = 0 =⇒ y
(l) + α
2
y
(l) = 0
=⇒ y
(l) = 0
Eliminating V between these two equations leads to
d
2
M
dx
2
+ P
d
2
y
dx
2
= w(x).
Using the momentcurvature relationship
M(x) = EI y
(x)
220 5 applications of linear differential equations
results in a fourthorder linear ordinary differential equation governing the lateral
deﬂection y(x)
d
4
y
dx
4
+ α
2
d
2
y
dx
2
=
w(x)
EI
, α
2
=
P
EI
.
Note that the shear force becomes
V(x) = −
dM
dx
− P
dy
dx
= −EI y
(x) − Py
(x) = −EI
_
y
(x) + α
2
y
(x)
_
.
The characteristic equation is λ
4
+α
2
λ
2
=0 =⇒ λ =0, 0, ±i α. Hence, the
general solution is
y(x) = C
0
+ C
1
x + Acos αx + Bsin αx + y
P
(x),
where y
P
(x) is a particular solution which depends on w(x). The constants
A, B, C
0
, and C
1
are determined using the boundary conditions at ends x =0
and x =L. Some typical boundary conditions are listed in Table 6.1.
Example 5.3 — Water Tower 5.3
Consider the water tower shown in Figure 5.23, which is modeled as a cantilever
beam with P =W and w(x) =w. Determine the lateral deﬂection y(x).
The differential equation governing the lateral deﬂection y(x) becomes
d
4
y
dx
4
+ α
2
d
2
y
dx
2
=
w
EI
, α
2
=
W
EI
.
Using the method of Doperator, a particular solution is
y
P
(x) =
1
D
4
+α
2
D
2
_
w
EI
_
=
1
α
2
D
2
1
1+
D
2
α
2
_
w
EI
_
=
w
W
1
D
2
(1) =
wx
2
2W
.
The general solution is then given by
y(x) = C
0
+ C
1
x + Acos αx + Bsin αx +
wx
2
2W
,
where A, B, C
0
, and C
1
are determined from the boundary conditions:
x = 0 (clamped): y(0) =0, y
(0) =0,
x = L (free): y
(L) =0, y
(L)+α
2
y
(L) =0.
Since
y
(x) = C
1
− Aα sin αx + Bα cos αx +
wx
W
,
5.4 beamcolumns 221
y
(x) = −Aα
2
cos αx − Bα
2
sin αx +
w
W
,
y
(x) = Aα
3
sin αx − Bα
3
cos αx,
one has
y
(L)+α
2
y
(L) =0:
(Aα
3
sin αx−Bα
3
cos αx)+α
2
_
C
1
−Aα sin αx+Bα cos αx+
wx
W
_
=0,
∴ C
1
= −
wL
W
,
y
(0) =C
1
+Bα =0 =⇒ B = −
C
1
α
=
wL
αW
,
y
(L) = −Aα
2
cos αL − Bα
2
sin αL +
w
W
=0 =⇒ A=
w(1−αL sin αL)
α
2
W cos αL
,
y(0) =C
0
+A=0 =⇒ C
0
= −A= −
w(1−αL sin αL)
α
2
W cos αL
.
Hence, the general solution becomes
y(x) =
w(1−αL sin αL)
α
2
W cos αL
(cos αx−1) +
wL
αW
(sin αx−αx) +
wx
2
2W
.
Remarks: When the lateral load w(x) includes distributed load over only a
portion of the beamcolumn or concentrated loads, it can be better expressed
using the Heaviside step function or the Dirac delta function. The differential
equation can be easily solved using the Laplace transform as illustrated in Section
6.6.2.
Example 5.4 — Buckling of a Column 5.4
Consider the simply supported column AB subjected to axial compressive load. A
rotational spring of stiffness κ provides resistance to rotation of end B. Set up the
buckling equation for the column.
Since the lateral load w(x) =0, the differential equation governing the lateral de
ﬂection y(x) becomes
d
4
y
dx
4
+ α
2
d
2
y
dx
2
= 0, α
2
=
P
EI
.
The solution is
y(x) = C
0
+ C
1
x + Acos αx + Bsin αx,
where A, B, C
0
, and C
1
are determined from the boundary conditions:
222 5 applications of linear differential equations
x
y
x
y
P
A
A
B
B
κ
θ
L
=y
(L)
M
L
=
–
κθ
L
EI, L
support A, x = 0: y(0) =0, y
(0) =0,
support B, x = L: y(L) =0, M(L) =EI y
(L) = −κ y
(L).
Since
y
(x) = C
1
− Aα sin αx + Bα cos αx,
y
(x) = −Aα
2
cos αx − Bα
2
sin αx,
one has
y
(0) = −Aα
2
= 0 =⇒ A = 0,
y(0) = C
0
+ A = 0 =⇒ C
0
= −A = 0,
y(L) = C
1
L + Bsin αL = 0 =⇒ C
1
= −
sin αL
L
B,
EI y
(L) = −κ y
(L) =⇒ EI(−Bα
2
sin αL) = −κ (C
1
+Bα cos αL),
_
κ
_
−
sin αL
L
+ α cos αL
_
− EI α
2
sin αL
_
B = 0.
When the column buckles, it has nonzero deﬂection y(x), which means B=0.
Note that sin αL=0, otherwise, B=0. Hence, the buckling equation is given by
κ
_
−
sin αL
L
+ α cos αL
_
− EI α
2
sin αL = 0,
∴
κL
EI
(αL cot αL−1) − (αL)
2
= 0, α
2
=
P
EI
.
5.5 various application problems 223
5.5 Various Application Problems
Example 5.5 — Jet Engine Vibration 5.5
As shown in Figure 5.8, jet engines are supported by the wings of the airplane. To
study the horizontal motion of a jet engine, it is modeled as a rigid body supported
by an elastic beam. The mass of the engine is m and the moment of inertia about its
centroidal axis C is J. The elastic beamis further modeled as a massless bar hinged
at A, with the rotational spring κ providing restoring moment equal to κθ, where
θ is the angle between the bar and the vertical line as shown in Figure 5.25.
For small rotations, i.e.,
¸
¸
θ
¸
¸
1, set up the equation of motion for the jet engine
in term of θ. Find the natural frequency of oscillation.
A
A
mg
m, J
R
Ax
R
Ay
κθ
κ
C
θ
J
A
θ
L
Figure 5.25 Horizontal vibration of a jet engine.
The systemrotates about hinge A. The moment of inertia of the jet engine about its
centroidal axis C is J. Using the Parallel Axis Theorem, the moment of inertia of
the jet engine about axis A is
J
A
= J + mL
2
.
Draw the freebody diagram of the jet engine and the supporting bar as shown.
The jet engine is subjected to gravity mg. Remove the hinge at A and replace it
by two reaction force components R
Ax
and R
Ay
. Since the bar rotates an angle θ
counterclockwise, the rotational spring provides a clockwise restoring moment κθ.
Since the angular acceleration of the system is
¨
θ counterclockwise, the inertia
moment is J
A
¨
θ clockwise.
Applying D’Alembert’s Principle, the freebody as shown in Figure 5.25 is in
dynamic equilibrium. Hence,
M
A
=0: J
A
¨
θ + κθ + mg · L sin θ = 0.
224 5 applications of linear differential equations
For small rotations
¸
¸
θ
¸
¸
1, sin θ ≈θ, the equation of motion is
(J +mL
2
)
¨
θ + (κ +mgL)θ = 0.
Since
¨
θ +
κ +mgL
J +mL
2
θ = 0 =⇒
¨
θ + ω
2
0
θ = 0,
the natural circular frequency ω
0
of oscillation is given by
ω
0
=
_
κ +mgL
J +mL
2
.
Example 5.6 — Piston Vibration 5.6
Oil enters a cylinder as shown in the following ﬁgure through a constriction such
that the ﬂow rate is Q=α( p
i
−p
o
), where p
i
is the supply pressure, p
o
is the
pressure in the cylinder, and α is a constant. The cylinder contains a piston of mass
m and area A backed by a spring of stiffness k.
1. Assume that the oil is incompressible, there is no leakage past the piston, and
the inertia of the oil is neglected. Set up the equation of motion for the piston
displacement x.
2. If there is a sinusoidal variation in p
i
of the form p
i
(t) =P
0
+P
1
sin t,
where P
0
, P
1
, and are constants, determine the steadystate displacement
x
P
(t).
x(t)
m m
kx
k
p
i
p
o
p
o
A
Area A
x, x, x
1. To set up the equation of motion, consider the freebody of the piston as shown.
The piston is subjected to two forces: the force due to the internal oil pressure p
o
A
and the spring force kx. Newton’s Second Law requires that
→ ma =
F : m¨ x = p
o
A − kx.
Since the oil is incompressible, in time t,
Inﬂow = Qt = α( p
i
−p
o
)t = Ax,
5.5 various application problems 225
where x is the displacement of the piston displaced by the oil inﬂow. Solving for
p
o
yields
p
o
= p
i
−
A
α
x
t
.
Taking the limit as t →0 leads to
p
o
= p
i
−
A
α
dx
dt
.
Hence, the equation of motion becomes
m¨ x =
_
p
i
−
A
α
˙ x
_
A − kx =⇒ m¨ x +
A
2
α
˙ x + kx = p
i
A,
∴ ¨ x +
A
2
αm
˙ x +
k
m
x =
A
m
p
i
.
2. Since p
i
(t) =P
0
+P
1
sin t, using the Doperator, the equation of motion can
be written as
(D
2
+ cD + ω
2
0
)x =
A
m
(P
0
+P
1
sin t), c =
A
2
αm
, ω
2
0
=
k
m
.
A particular solution is given by
x
P
(t) =
A
m
1
D
2
+cD+ω
2
0
(P
0
+P
1
sin t)
=
AP
0
mω
2
0
+
AP
1
m
1
−
2
+cD+ω
2
0
sin t,
Theorem 3 of Chapter 4:
replace D
2
by −
2
.
=
AP
0
mω
2
0
+
AP
1
m
(ω
2
0
−
2
)−cD
_
(ω
2
0
−
2
)+cD
__
(ω
2
0
−
2
)−cD
_ sin t
=
AP
0
mω
2
0
+
AP
1
m
(ω
2
0
−
2
)−cD
(ω
2
0
−
2
)
2
−c
2
D
2
sin t
=
AP
0
mω
2
0
+
AP
1
m
(ω
2
0
−
2
) sin t −ccos t
(ω
2
0
−
2
)
2
+ c
2
2
Replace D
2
by −
2
and
evaluate the numerator.
=
AP
0
mω
2
0
+
AP
1
m
sin(t −ϕ)
_
(ω
2
0
−
2
)
2
+ c
2
2
, ϕ = tan
−1
c
ω
2
0
−
2
.
Use Asin θ −Bcos θ =
_
A
2
+B
2
sin(θ −ϕ), ϕ = tan
−1
(B/A).
226 5 applications of linear differential equations
Example 5.7 — Single DegreeofFreedom System 5.7
The single degreeoffreedomsystemdescribedby x(t), as showninFigure 5.26(a),
is subjected to a sinusoidal load F(t) =F
0
sin t. Assume that the mass m, the
spring stiffnesses k
1
and k
2
, the damping coefﬁcient c, and F
0
and are known.
Determine the steadystate amplitude of the response of x
P
(t).
m
c
c
A
k
1
k
2
k
2
F
0
sint
k
1
(a)
(b)
y
k
1
y
k
2
(y−x) k
2
(x−y)
cy
x(t)
A
m
F
0
sint
x, x, x
Figure 5.26 A vibrating system.
Introduce a displacement y(t) at A as shown in Figure 5.26(b). Consider the free
body of A. The extension of spring k
1
is y and the compression of spring k
2
is
y−x. Body A is subjected to three forces: spring force k
1
y, damping force c ˙ y, and
spring force k
2
( y−x). Newton’s Second Law requires
→ m
A
¨ y =
F : m
A
¨ y = −k
1
y − c ˙ y − k
2
( y−x).
Since the mass of A is zero, i.e., m
A
=0, one has
x =
(k
1
+k
2
) y + c ˙ y
k
2
. (1)
Consider the freebody of mass m. The extension of spring k
2
is x−y. The mass
is subjected to two forces: spring force k
2
(x−y) and the externally applied load
F
0
sin t. Applying Newton’s Second Law gives
→ m¨ x =
F : m¨ x = F
0
sin t − k
2
(x−y).
Substituting equation (1) yields the equation of motion
m
(k
1
+k
2
) ¨ y + c
...
y
k
2
= F
0
sin t − k
2
_
(k
1
+k
2
) y + c ˙ y
k
2
− y
_
,
5.5 various application problems 227
c m
k
2
...
y +
m(k
1
+k
2
)
k
2
¨ y + c ˙ y + k
1
y = F
0
sin t,
or, using the Doperator
_
c m
k
2
D
3
+
m(k
1
+k
2
)
k
2
D
2
+ cD + k
1
_
y = F
0
sin t.
A particular solution is given by
y
P
=
1
c m
k
2
D
3
+
m(k
1
+k
2
)
k
2
D
2
+ cD + k
1
F
0
sin t
= k
2
F
0
1
_
m(k
1
+k
2
)D
2
+k
1
k
2
_
+ c(mD
2
+k
2
)D
sin t
= k
2
F
0
1
_
−m(k
1
+k
2
)
2
+k
1
k
2
_
+ c(−m
2
+k
2
)D
sin t
Theorem 3 of Chapter 4: replace D
2
by −
2
.
= k
2
F
0
_
k
1
k
2
−m(k
1
+k
2
)
2
_
−c(k
2
−m
2
)D
_
k
1
k
2
−m(k
1
+k
2
)
2
_
2
−c
2
(k
2
−m
2
)
2
D
2
sin t
= k
2
F
0
_
k
1
k
2
−m(k
1
+k
2
)
2
_
sin t −c(k
2
−m
2
) cos t
_
k
1
k
2
−m(k
1
+k
2
)
2
_
2
+c
2
(k
2
−m
2
)
2
2
Replace D
2
by −
2
and evaluate the numerator.
= k
2
F
0
_
_
k
1
k
2
−m(k
1
+k
2
)
2
_
2
+c
2
2
(k
2
−m
2
)
2
sin(t −ϕ)
_
k
1
k
2
−m(k
1
+k
2
)
2
_
2
+c
2
2
(k
2
−m
2
)
2
Use Asin θ −Bcos θ =
_
A
2
+B
2
sin(θ −ϕ), ϕ = tan
−1
(B/A).
= k
2
a sin(t −ϕ),
where
a =
F
0
_
_
k
1
k
2
−m(k
1
+k
2
)
2
_
2
+c
2
2
(k
2
−m
2
)
2
.
Using equation (1), one obtains
x
P
(t) =
(k
1
+k
2
) y
P
+ c ˙ y
P
k
2
=
(k
1
+k
2
) · k
2
a sin(t −ϕ) + c · k
2
acos(t −ϕ)
k
2
= a
_
(k
1
+k
2
) sin(t −ϕ) + ccos(t −ϕ)
_
.
The amplitude of x
P
(t) is x =Asin θ ±Bcos θ =⇒ x
amplitude
=
_
A
2
+B
2
x
P, amplitude
= a
_
(k
1
+k
2
)
2
+ c
2
2
.
228 5 applications of linear differential equations
Example 5.8 — Flywheel Vibration 5.8
A pair of uniform parallel bars AB of length L and together of mass m are hinged
at A and supported at B by a spring of stiffness k as shown. The bars carry a
uniform ﬂywheel D of mass M and radius r supported in bearings, with AD=l.
When the system is in static equilibrium, AB is horizontal. The ﬂywheel rotates
at angular velocity and has a small eccentricity e, i.e., DG=e, where G is the
center of gravity of the ﬂywheel. Determine the natural frequency of vibration of
the system and the total vertical movement of B (assuming no resonance).
l
A
A
H
B
k
B
R
Ax
R
Ay
C
L
M
.
e
2
D
D
e
G
t
θ
y(t)
ky
J
A
θ
The moment of inertia of the bars AB about hinge A is
J
Bars
A
=
1
3
mL
2
.
The moment of inertia of the ﬂywheel about its axis of rotation D is
J
Flywheel
D
=
1
2
Mr
2
.
Using the Parallel Axis Theorem, the moment of inertia of the ﬂywheel about hinge
A is
J
Flywheel
A
= J
Flywheel
D
+ Ml
2
=
1
2
Mr
2
+ Ml
2
.
Hence, the moment of inertia of the system about hinge A is
J
A
= J
Bars
A
+ J
Flywheel
A
=
1
3
mL
2
+
1
2
M(r
2
+2l
2
).
Suppose bars AB has a small angular rotation θ(t) about hinge A as shown.
The upward vertical displacement of end B is y(t) =Lθ(t), θ
1. Hence, the
downward spring force applied on bars AB at B is ky =kLθ.
When the ﬂywheel rotates at angular velocity , the center of gravity G moves
ona circle of radius e withangular velocity , resulting ina centrifugal acceleration
e
2
. Hence, the centrifugal force is M·e
2
. The moment of the centrifugal force
about point A is (Me
2
)×AH, where AH is the moment arm given by
AH = AD· sin ∠ADH ≈ l sin t. ∠ADH = t −θ ≈ t, for θ
1
5.5 various application problems 229
Consider the freebody diagram of the bars AB and the ﬂywheel. The inertia
moment of the system about point A is J
A
¨
θ, where
¨
θ is the angular acceleration of
the system about point A. From D’Alembert’s Principle, the system is in dynamic
equilibrium under the inertia moment J
A
¨
θ and the externally applied forces, i.e.,
the spring force kLθ, the centrifugal force Me
2
, and the two components R
Ax
,
R
Ay
of the reaction force at hinge A. Summing up the moments about A yields
M
A
= 0: −J
A
¨
θ + Me
2
· l sin t − kLθ · L = 0,
∴ J
A
¨
θ + kL
2
θ = Mel
2
sin t =⇒
¨
θ + ω
2
0
θ =
Mel
2
J
A
sin t,
where ω
0
is the natural circular frequency of the system given by
ω
2
0
=
kL
2
J
A
=
kL
2
1
3
mL
2
+
1
2
M(r
2
+2l
2
)
.
The response of the forced vibration is, assuming no resonance, i.e., =ω
0
,
θ
P
(t) =
1
D
2
+ω
2
0
_
Mel
2
J
A
sin t
_
=
Mel
2
J
A
1
ω
2
0
−
2
sin t.
Theorem 3 of Chapter 4:
replace D
2
by −
2
.
The amplitude of the forced vibration is
a =
Mel
2
J
A
¸
¸
ω
2
0
−
2
¸
¸
,
and the total vertical movement of B is 2a.
Example 5.9 — Displacement Meter 5.9
The following ﬁgure shows the conﬁguration of a displacement meter used for mea
suring the vibration of the structure upon which the meter is mounted. The struc
ture undergoes vertical displacement, which may be modeled as y
0
=a
0
sin t
with a
0
and to be measured by the displacement meter. The structure excites
the massspringdamper system of the displacement meter. The displacement of
tip D of rod AD is recorded on the rotating drum. The record shows that the
steadystate displacement of tip D has a peaktopeak amplitude of 2a and the
distance between two adjacent peaks is d. The rotating drum has a radius r and
rotates at a constant speed of v rpm. Determine the amplitude a
0
and the circular
frequency of the displacement of the structure.
230 5 applications of linear differential equations
m
c
A
B C
D
k
y
0
y
d
r
2a
θ
L
1
L
2
L
3
m
c
k
y
0
y
2
y
1
L
1
L
2
L
3
m
c
k
A
A
B
B
C
C
D
D
y
y
m(y
2
+y
0
)
c y
1
k y
1
R
Ax
R
Ay
To establish the differential equation governing the displacement y, consider the
freebody diagram of rod AD. D’Alembert’s Principle is applied to set up the
equation of motion.
Since the relative displacement of tip D is y, the relative displacements at B and
C can be determined using similar triangles
y
1
=
L
1
L
3
y, y
2
=
L
2
L
3
y.
The extension of the spring is y
1
, resulting in the spring force ky
1
applied on
rod AD at point B. Similarly, the damping force applied on rod AD at point B
is c ˙ y
1
. The inertia force applied at point C depends on the absolute acceleration
of point C, i.e., ¨ y
2
+ ¨ y
0
; hence, the inertia force is m( ¨ y
2
+ ¨ y
0
), opposite to the
direction of the absolute acceleration. Because the rod is supported by a hinge at
5.5 various application problems 231
A, two reaction force components R
Ax
and R
Ay
are applied at A when the hinge is
removed.
Rod AD is in dynamic equilibrium under the spring force ky
1
, the damping
force c ˙ y
1
, the inertia force m( ¨ y
2
+ ¨ y
0
), and the support reaction forces R
Ax
, R
Ay
.
Summing up the moments about point A yields:
M
A
= 0: m( ¨ y
2
+ ¨ y
0
) · L
2
+ (c ˙ y
1
+ ky
1
) · L
1
= 0,
or
m·
L
2
L
3
¨ y · L
2
+
_
c ·
L
1
L
3
˙ y + k·
L
1
L
3
y
_
· L
1
= −m¨ y
0
· L
2
.
Noting that y
0
=a
0
sin t, one obtains
mL
2
2
¨ y + cL
2
1
˙ y + kL
2
1
y = a
0
mL
2
L
3
2
sin t,
∴ (MD
2
+ CD + K) y = a
s
sin t,
where
M = mL
2
2
, C = cL
2
1
, K = kL
2
1
, a
s
= a
0
mL
2
L
3
2
.
The steadystate response is given by
y
P
(t) = a
s
1
MD
2
+ CD + K
sin t
= a
s
1
−M
2
+ CD + K
sin t,
Theorem 3 of Chapter 4:
replace D
2
by −
2
.
= a
s
(K−M
2
) − CD
_
(K−M
2
) + CD
_ _
(K−M
2
) − CD
_ sin t
= a
s
(K−M
2
) − CD
(K−M
2
)
2
− C
2
D
2
sin t
= a
s
(K−M
2
) sin t − Ccos t
(K−M
2
)
2
+ C
2
2
,
Replace D
2
by −
2
and
evaluate the numerator.
=
a
s
_
(K−M
2
)
2
+ C
2
2
sin(t −ϕ).
Use Asin θ −Bcos θ =
_
A
2
+B
2
sin(θ −ϕ), ϕ = tan
−1
(B/A).
Hence, the amplitude of the steadystate displacement of tip D is
a =
a
s
_
(K−M
2
)
2
+ C
2
2
=
a
0
mL
2
L
3
2
_
(kL
2
1
−mL
2
2
2
)
2
+ (cL
2
1
)
2
,
which leads to
a
0
=
_
(kL
2
1
−mL
2
2
2
)
2
+ (cL
2
1
)
2
mL
2
L
3
2
a.
232 5 applications of linear differential equations
On the record paper, the distance d between two adjacent peaks is measured in
length, which needs to be changed to time to yield the period T of the response.
Since the drum rotates at a speed of v rpm, i.e., it rotates an angle of 2πv in 60
seconds, hence the time T it takes to rotate an angle θ, as shown in the ﬁgure, is
given by
T
60
=
θ
2πv
=⇒ T =
30θ
πv
.
Furthermore, since d =rθ, which is the arc length corresponding to angle θ, one
has
T =
30
πv
d
r
=
30d
πrv
.
The frequency of vibration of tip D is
f =
1
T
=
πrv
30d
.
Since the steadystate response and the excitation have the same frequency, one
obtains
= 2πf =
π
2
rv
15d
.
Problems
5.1 Acircular cylinder of radius r andmass m is supportedby a spring of stiffness
k and partially submerges in a liquid of density γ . Suppose that, during vibration,
the cylinder does not completely submerge in the liquid. Set up the equation
of motion of the cylinder for the oscillation about the equilibrium position and
determine the period of the oscillation.
A
NS m ¨ y + (k+γ πr
2
) y = 0, T = 2π
_
m
k+γ πr
2
k
m
y, y, y
Problems 233
5.2 A cylinder of radius r, height h, and mass m ﬂoats with its axis vertical in a
liquid of density ρ as shown in the following ﬁgure.
☞
Archimedes' Principle: An object partially or totally submerged in a fluid
is buoyed up by a force equal to the weight of the fluid displaced.
h
Liquid Level
Equilibrium
Position
x(t)
r
1. Set up the differential equation governing the displacement x(t), measured
relative to the equilibrium position, and determine the period of oscillation.
2. If the cylinder is set into oscillation by being pushed down a displacement x
0
at t =0 and then released, determine the response x(t).
A
NS m¨ x + ρπr
2
x = 0, T =
2
r
_
πm
ρ
; x(t) = x
0
cos ω
0
t, ω
0
= r
_
ρπ
m
5.3 A cube of mass m is immersed in a liquid as shown. The length of each side
of the cube is L. At time t =0, the top surface of the cube is leveled with the surface
of the liquid due to buoyancy. The cube is lifted by a constant force F. Show that
the time T when the bottom surface is leveled with the liquid surface is given by
T =
_
L
g
cos
−1
_
1−
mg
F
_
.
y, y, y
m
L
L
F
t =0
Time t
t =T
F
234 5 applications of linear differential equations
5.4 A mass m is dropped with zero initial velocity from a height of h above a
spring of stiffness k as shown in the following ﬁgure. Determine the maximum
compression of the spring and the duration between the time when the mass con
tacts the spring and the time when the spring reaches maximum compression.
A
NS y
max
=
_
mg
k
_
2h+
mg
k
_
+
mg
k
, T =
_
m
k
_
π
2
+−tan
−1
_
mg
2hk
_
m
k
h
5.5 A uniform chain of length L with mass density per unit length ρ is laid on a
rough horizontal table with an initial hang of length l, i.e., y =l at t =0 as shown
in the following ﬁgure. The coefﬁcients of static and kinetic friction between the
chain and the surface have the same value μ. The chain is released fromrest at time
t =0 and it starts sliding off the table if (1+μ)l >μL. Show that the time T it
takes for the chain to leave the table is
T =
_
L
(1+μ)g
cosh
−1
_
L
(1+μ)l −μL
_
.
L−y
y(t)
y, y, y
5.6 A uniform chain of length L with mass density per unit length ρ is laid on a
smooth inclined surface with y =0 at t =0 as shown in the following ﬁgure. The
chain is released from rest at time t =0. Show that the time T it takes for the chain
to leave the surface is
T =
_
L
(1−sin θ)g
cosh
−1
_
1
sin θ
_
.
Problems 235
L−y
y(t)
y, y, y
θ
5.7 Auniformchain of length L with mass density per unit length ρ is hung on a
small smooth pulley with y(t) =l when t =0, l >L/2, as shown in the Figure 5.27.
The chain is released from rest at time t =0. Show that the time T it takes for the
chain to leave the pulley is
T =
_
L
2g
cosh
−1
_
L
2l −L
_
.
y, y, y
y(t)
L–y(t)
A
B
O
L
r
k
1
k
2
a
b
Figure 5.27 Figure 5.28
5.8 A pendulum as shown in Figure 5.28 consists of a uniform solid sphere of
radius r and mass m connected by a weightless bar to hinge O. The bar is further
constrained by two linear springs of stiffnesses k
1
and k
2
at A and B, respectively.
It is known that the moment of inertia of a solid sphere of radius r and mass m
about its diameter is
2
5
mr
2
. Show that the equation of motion governing the angle
of rotation of the pendulum about O and the natural period of oscillation of the
pendulum are given by
m
_
2
5
r
2
+L
2
_
¨
θ + (k
1
a
2
+k
2
b
2
+mg L)θ = 0, T = 2π
¸
¸
¸
_
m
_
2
5
r
2
+L
2
_
k
1
a
2
+k
2
b
2
+mg L
.
236 5 applications of linear differential equations
5.9 A mass m is attached to the end C of a massless rod AC as shown in the
following ﬁgure. The rod is hinged at one end A and supported by a spring of
stiffness k at the middle B. A dashpot damper having a damping coefﬁcient c is
attached at the middle. A sinusoidal load F sin t is applied at end C.
Fsint
L L
x
A
C B
m
c
k
1. Show that the equation of motion governing displacement x(t) of end C is
4m¨ x + c ˙ x + kx = 4F sin t.
2. Show that the natural circular frequency ω
d
of the damped free vibration of
the system is given by
ω
d
= ω
0
_
1−
c
2
16km
, ω
0
=
1
2
_
k
m
.
5.10 Amassless rod is hinged at one end A and supported by a spring of stiffness
k at the other end D as shown in the following ﬁgure. Amass m is attached at
1
3
of
the length fromthe hinge and a dashpot damper having a damping coefﬁcient c is
attached at
2
3
of the length from the hinge. A sinusoidal load F sin t is applied at
end D.
L
Fsint
L
x
A B C
D
m
L
c k
1. Show that the equation of motion governing displacement x(t) of end D is
m¨ x + 4c ˙ x + 9kx = 9F sin t.
2. Show that the natural circular frequency ω
d
of the damped free vibration of
the system is given by
ω
d
= ω
0
_
1 −
4c
2
9km
, ω
0
= 3
_
k
m
.
Problems 237
5.11 Adamped single degreeoffreedomsystemis shown in the following ﬁgure.
The displacement of the mass M is described by x(t). The excitation is provided
by x
0
(t) =a sin t.
M
K
2
c
2
x
0
(t)
x(t)
K
1
c
1
1. Show that the equation of motion governing the displacement of the mass M
is given by
¨ x + 2ζ ω
0
˙ x + ω
2
0
x = α sin t + β cos t,
where
ω
0
=
_
K
1
+K
2
M
, 2ζ ω
0
=
c
1
+c
2
M
, α =
aK
1
M
, β =
ac
1
M
.
2. Determine the amplitude of the steadystate response x
P
(t).
A
NS
_
_
α(ω
2
0
−
2
)+2ζ ω
0
β
_
2
+
_
β(ω
2
0
−
2
)−2ζ ω
0
α
_
2
(ω
2
0
−
2
)
2
+ (2ζ ω
0
)
2
5.12 The single degreeoffreedom system shown in the following ﬁgure is sub
jected to dynamic force F(t) =F
0
sin t .
m
k
x(t)
k
c
F(t)
1. Set up the equation of motion in terms of x(t) and determine the damped
natural circular frequency.
2. Determine the steadystate response of the system x
P
(t).
A
NS m¨ x + c ˙ x + 2kx = F
0
sin t, ω
d
=
_
2k
m
_
1−
c
2
8km
_
x
P
(t) =
F
0
_
(2k−m
2
) sin t − ccos t
_
(2k−m
2
)
2
+c
2
2
238 5 applications of linear differential equations
5.13 The single degreeoffreedom system shown in the following ﬁgure is sub
jected to dynamic displacement x
0
(t) =a sin t at point A.
x
0
(t)=a sint
x(t)
c
A
k
1
k
2
m
1. Set up the equation of motion in terms of of x(t).
2. If the system is lightly damped, determine the steadystate response of the
system x
P
(t).
A
NS m¨ x + c ˙ x + (k
1
+k
2
)x = ak
2
sin t
x
P
(t) = ak
2
(k
1
+k
2
−m
2
) sin t − ccos t
(k
1
+k
2
−m
2
)
2
+c
2
2
5.14 A precision instrument having a mass of m=400 kg is to be mounted on a
ﬂoor. It is known that the ﬂoor vibrates vertically with a peaktopeak amplitude
of 2 mm and frequency of 5 Hz. To reduce the effect of vibration of the ﬂoor on
the instrument, four identical springs are placed underneath the instrument. If the
peaktopeak amplitude of vibration of the instrument is to be limited to less than
0.2 mm, determine the stiffness of each spring. Neglect damping.
A
NS k =8.97 kN/m
5.15 The single degreeoffreedom system, shown in the following ﬁgure, is
subjected to a sinusoidal load F(t) =F
0
sin t at point A. Assume that the mass m,
the spring stiffnesses k
1
and k
2
, and F
0
and are known. The system is at rest
when t =0.
m
A
k
1
k
2
F(t)=F
0
sint
x(t) y(t)
1. Show that the differential equation governing the displacement of the mass
x(t) is
¨ x + ω
2
0
x = f sin t, ω
0
=
_
k
1
k
2
m(k
1
+k
2
)
, f =
k
2
m(k
1
+k
2
)
F
0
.
Problems 239
2. For the case =ω
0
, determine the response of the system x(t).
3. For the case =ω
0
, determine the response of the system x(t).
A
NS 2. =ω
0
: x(t) =
f
ω
2
0
−
2
_
−
ω
0
sin ω
0
t + sin t
_
;
3. =ω
0
: x(t) = −
f
2ω
2
0
_
−sin ω
0
t + ω
0
t cos ω
0
t
_
5.16 A vehicle is modeled by a damped single degreeoffreedom system with
mass M, spring stiffness K, and damping coefﬁcient c as shown in the following
ﬁgure. The absolute displacement of the mass M is described by y(t). The vehicle
is moving at a constant speed U on a wavy surface with proﬁle y
0
(x) =μsin x.
At time t =0, the vehicle is at x =0.
K
U
O
x
M
c
y
y
0
(x)
y
0
1. Show that the equation of motion governing the relative displacement of the
vehicle given by z(t) =y(t)−y
0
(t) is
¨ z + 2ζ ω
0
˙ z + ω
2
0
z = μ
2
U
2
sin(Ut), ω
0
=
_
K
M
, 2ζ ω
0
=
c
M
.
2. Determine the amplitude of the steadystate response z(t), which is a partic
ular solution of the equation of motion.
3. Assuming that the damping coefﬁcient c =0, determine the speed U at which
resonance occurs.
A
NS
μ
2
U
2
_
(ω
2
0
−
2
U
2
)
2
+ (2ζ ω
0
U)
2
; U =
1
_
K
M
5.17 The landing gear of an airplane as shown in Figure 5.8 can be modeled as a
mass connected to the airplane by a spring of stiffness K and a damper of damping
coefﬁcient c. A spring of stiffness k is used to model the forces on the tires. The
airplane lands at time t =0 with x =0 and moves at a constant speed U on a
wavy surface with proﬁle y
0
(x) =μsin x. Assuming that the airplane moves in
240 5 applications of linear differential equations
the horizontal direction only, determine the steadystate response of the absolute
displacement y(t) of the mass m.
A
NS y(t) =
kμsin(Ut −ϕ)
_
(K+k−m
2
U
2
)
2
+ (c U)
2
, ϕ = tan
−1
c U
K+k−m
2
U
2
5.18 In Section 5.1, it is derived that the equation of motion of a single story
shear building under the base excitation x
0
(t) is given by
m ¨ y(t) + c ˙ y(t) + ky = −m¨ x
0
(t),
or
¨ y(t) + 2ζ ω
0
˙ y(t) + ω
2
0
y = −¨ x
0
(t),
where
ω
2
0
=
k
m
, 2ζ ω
0
=
c
m
,
and y(t) =x(t)−x
0
(t) is the relative displacement between the girder and the base.
x(t)
x
0
(t)=a sint
m
Rigid girder
Weightless columns k
c
For x
0
(t) =a sin t, determine the Dynamic Magniﬁcation Factor (DMF) de
ﬁned as
DMF =
¸
¸
y
P
(t)
¸
¸
max
¸
¸
x
0
(t)
¸
¸
max
,
where y
P
(t) is the steadystate response of the relative displacement or the par
ticular solution due to the base excitation. Plot DMF versus the frequency ratio
r =/ω
0
for ζ =0, 0.1, 0.2, and 0.3.
A
NS DMF =
r
2
_
(1−r
2
)
2
+ (2ζ r)
2
, r =
ω
0
5.19 Consider the undamped single degreeoffreedom system with m=10 kg,
k =1 kN/m. The system is subjected to a dynamic load F(t) as shown in the
following ﬁgure. The system is at rest at time t =0.
Determine the analytical expression of the displacement as a function of time up
to t =10 sec.
A
NS x(t) = 0.02(10t − sin 10t), 0t 5
x(t) = 1.005 cos 10(t −5) + 0.0007 sin 10(t −5), 5t 10
Problems 241
F(t) (kN)
0
1
t (sec)
5 10
m
x(t)
F(t)
k
5.20 The following ﬁgure shows the conﬁguration of a displacement meter used
for measuring the vibration of the structure that the meter is mounted on. The
structure undergoes vertical displacement a
0
sin t and excites the massspring
damper systemof the displacement meter. The displacement of the mass is recorded
on the rotating drum. It is known that m=1 kg, k =1000 N/m, c =5 N· sec/m,
and the steadystate record on the rotating drum shows a sinusoidal function with
frequency of 5 Hz andpeaktopeak amplitude of 50 mm. Determine the amplitude
a
0
and the frequency f =/(2π) of the displacement of the structure.
A
NS a
0
=4.0 mm, f =5 Hz
m
c
k
a
0
sint
2a
5.21 For the circuit shown in Figure 5.29(a), the switch has been at position a for
a long time prior to t =0
−
. At t =0, the switch is moved to position b. Determine
i(t) for t >0.
A
NS i(t) =(3−9t)e
−5t
(A)
5.22 For the circuit shown in Figure 5.29(b), the switch has been at position a for
a long time prior to t =0
−
. At t =0, the switch is moved to position b. Show that
the differential equation governing v
C
(t) for t >0 is
d
2
v
C
dt
2
+
R
L
dv
C
dt
+
1
LC
v
C
=
V(t)
LC
, v
C
(0
+
) = −RI
0
,
dv
C
(0
+
)
dt
=0.
For R=6 , C=
1
25
F, L=1 H, I
0
=1A, V(t) =39 sin 2t (V), determine v
C
(t)
for t >0.
A
NS v
C
(t) =7e
−3t
(2 cos 4t −sin 4t)+35 sin 2t −20 cos 2t (V)
242 5 applications of linear differential equations
t =0
i
a
b
4A
2H
0.02F
14
2
6
12V
t =0
t =0
V(t)
R
a
b
L
C
I
0
v
C
(a) (b)
t =0
R
2
R
1
C
1
C
2
V(t)
v
C
t =0
L
C
R
2
R
1
V
0
I(t)
v
C
(c) (d)
Figure 5.29 Secondorder circuits.
5.23 For the circuit shown in Figure 5.29(c), show that the differential equation
governing v
C
(t) for t >0 is
R
1
C
1
R
2
C
2
d
2
v
C
dt
2
+ (R
1
C
1
+R
1
C
2
+R
2
C
2
)
dv
C
dt
+ v
C
= R
1
C
1
dV(t)
dt
,
with the initial conditions given by v
C
(0
+
) =0,
dv
C
(0
+
)
dt
=
V(0
+
)
R
2
C
2
.
For R
1
=1 , R
2
=2 , C
1
=2 F, C
2
=1 F, V(t) =12e
−t
(V), determine v
C
(t) for
t >0.
A
NS v
C
(t) = −
8
3
e
−
t
4
+
8
3
(1+3t)e
−t
(V)
5.24 For the circuit shown in Figure 5.29(d), show that the differential equation
governing v
C
(t) for t >0 is
d
2
v
C
dt
2
+
R
1
+R
2
L
dv
C
dt
+
1
LC
v
C
=
V
0
+R
2
I(t)
LC
, v
C
(0
+
) =V
0
,
dv
C
(0
+
)
dt
=0.
For R
1
=R
2
=5 , C=0.2 F, L=5 H, V
0
=12V, I(t) =2 sin t (A), determine
v
C
(t) for t >0.
A
NS v
C
(t) =5(1+t)e
−t
+ 12 − 5 cos t (V)
Problems 243
5.25 Consider column AB clamped at the base and pinsupported at the top by
an elastic spring of stiffness k. Show that the buckling equation for the column is
kL
3
EI
_
tan(αL) − (αL)
_
+ (αL)
3
= 0, α
2
=
P
EI
.
x
y
P
A
B
k
EI, L
5.26 Consider the beamcolumn shown in the following ﬁgure. Determine the
lateral deﬂection y(x).
A
NS y(x) = −
w
P
_
Lx
6
_
1−
x
2
L
2
_
+
1
α
2
_
x
L
−
sin αx
sin αL
_
_
, α
2
=
P
EI
.
EI, L
x
L
y
x P
P
w(x)= w
w
6
C H A P T E R
The Laplace Transform
and Its Applications
The Laplace transform is one of the most important integral transforms. Because
of a number of special properties, it is very useful in studying linear differential
equations.
Applying the Laplace transform to a linear differential equation with constant
coefﬁcients converts it into a linear algebraic equation, which can be easily solved.
The solution of the differential equation can then be obtained by determining
the inverse Laplace transform. Furthermore, the method of Laplace transform is
preferable and advantageous in solving linear ordinary differential equations with
the righthand side functions involving discontinuous and impulse functions.
In this chapter, Laplace transform and its properties are introduced and applied
to solve linear differential equations.
6.1 The Laplace Transform
Deﬁnition — Laplace Transform
Let f (t), t >0, be a given function. The Laplace transform F(s) of function f (t)
is deﬁned by
F(s) = L
_
f (t)
_
=
_
∞
0
e
−st
f (t)dt, s >0.
The integral inthe Laplace transformis improper because of the unboundedinterval
of integration and is given by
_
∞
0
e
−st
f (t)dt = lim
M→∞
_
M
0
e
−st
f (t)dt.
244
6.1 the laplace transform 245
There are tables of Laplace transforms for various functions f (t), similar to
tables of integrals. Some frequently used Laplace transforms are listed in Table of
Laplace Transforms (Appendix A.4).
Example 6.1 6.1
Determine the Laplace transforms of the following functions
1. f (t) = 1; 2. f (t) = e
at
; 3. f (t) =
_
sin ωt
cos ωt
_
.
1. F(s) = L
_
1
_
=
_
∞
0
e
−st
· 1dt = −
1
s
e
−st
¸
¸
¸
∞
t=0
=
1
s
, s >0.
2. F(s) = L
_
e
at
_
=
_
∞
0
e
−st
· e
at
dt =
_
∞
0
e
−(s−a)t
dt
= −
1
s −a
e
−(s−a)t
¸
¸
¸
∞
t=0
=
1
s −a
, s >a.
3. Note Euler’s formula e
i ωt
= cos ωt +i sin ωt. To determine the Laplace trans
forms of cos ωt and sin ωt, consider the Laplace transform of e
i ωt
:
L
_
e
i ωt
_
=
_
∞
0
e
−st
· e
i ωt
dt =
_
∞
0
e
−(s−i ω)t
dt = −
1
s −i ω
e
−(s−i ω)t
¸
¸
¸
∞
t=0
=
1
s −i ω
=
s + i ω
(s −i ω)(s +i ω)
=
s + i ω
s
2
+ ω
2
, s >0.
Since cos ωt =Re(e
i ωt
) and sin ωt =Im(e
i ωt
), one has
L
_
cos ωt
_
= Re
_
L
_
e
i ωt
__
=
s
s
2
+ ω
2
,
L
_
sin ωt
_
= Im
_
L
_
e
i ωt
__
=
ω
s
2
+ ω
2
.
Properties of the Laplace Transform
1. Note that
L
_
c
1
f
1
(t) + c
2
f
2
(t)
_
=
_
∞
0
e
−st
_
c
1
f
1
(t) + c
2
f
2
(t)
_
dt, c
1
and c
2
are constants
= c
1
_
∞
0
e
−st
f
1
(t)dt + c
2
_
∞
0
e
−st
f
2
(t)dt
= c
1
L
_
f
1
(t)
_
+ c
2
L
_
f
2
(t)
_
∴ The Laplace transform L{ · } is a linear operator.
246 6 the laplace transformand its applications
2. Laplace Transformof Derivatives
L
_
f
(t)
_
=
_
∞
0
e
−st
f
(t)dt =
_
∞
0
e
−st
d
_
f (t)
_
= e
−st
f (t)
¸
¸
¸
∞
t=0
−
_
∞
0
f (t)
_
−s e
−st
_
dt Integration by parts
= −f (0) + s
_
∞
0
e
−st
f (t)dt,
L
_
f
(t)
_
= s F(s) − f (0).
Using this result, Laplace transforms of higherorder derivatives can be derived:
L
_
f
(t)
_
= s L
_
f
(t)
_
− f
(0), f
(t) =
_
f
(t)
_
= s
_
s F(s) − f (0)
_
− f
(0),
L
_
f
(t)
_
= s
2
F(s) − s f (0) − f
(0).
In general, one obtains
L
_
f
(n)
(t)
_
= s
n
F(s) − s
n−1
f (0) − s
n−2
f
(0) − · · · − s f
(n−2)
(0) − f
(n−1)
(0).
Remarks: The Laplace transformof the nthorder derivative f
(n)
(t) of function
f (t) is reduced to s
n
F(s), along with terms of the form s
i
f
(n−i−1)
(0), i =0, 1, . . . ,
n−1.
3. Property of Shifting
Given the Laplace transform F(s) =
_
∞
0
e
−st
f (t)dt, replacing s by s −a leads to
F(s −a) =
_
∞
0
e
−(s−a)t
f (t)dt =
_
∞
0
e
−st
_
e
at
f (t)
_
dt = L
_
e
at
f (t)
_
,
L
_
e
at
f (t)
_
= F(s −a) = L
_
f (t)
_
¸
¸
¸
s→(s−a)
.
The Laplace transform of f (t) multiplied by e
at
is equal to F(s −a), with s shifted
by −a.
4. Property of Differentiation
Since F(s) =
_
∞
0
e
−st
f (t)dt, differentiating with respect to s gives
F
(s) =
_
∞
0
(−t)e
−st
f (t)dt = −
_
∞
0
e
−st
_
t f (t)
_
dt,
6.1 the laplace transform 247
L
_
t f (t)
_
= −F
(s).
Differentiating with respect to s again leads to
F
(s) = −
_
∞
0
(−t)e
−st
_
t f (t)
_
dt =
_
∞
0
e
−st
_
t
2
f (t)
_
dt,
L
_
t
2
f (t)
_
= F
(s).
Following this procedure, one obtains
L
_
t
n
f (t)
_
= (−1)
n
F
(n)
(s) = (−1)
n
d
n
F(s)
ds
n
, n = 1, 2, . . . .
Example 6.2 6.2
Determine the Laplace transform of f (t) =t
n
, n=0, 1, . . . .
Solution 1: Using the formula of Laplace transform of derivative
L
_
f
(t)
_
= s L
_
f (t)
_
− f (0), with f (t) = t
n
,
one obtains
L
_
n t
n−1
_
= s L
_
t
n
_
− 0
n
=⇒ L
_
t
n
_
=
n
s
L
_
t
n−1
_
.
Hence, for n=1, 2, . . . ,
n = 1: L
_
t
_
=
1
s
L
_
1
_
=
1
s
2
, L
_
1
_
=
1
s
n = 2: L
_
t
2
_
=
2
s
L
_
t
_
=
2
s
·
1
s
2
=
2· 1
s
3
,
n = 3: L
_
t
3
_
=
3
s
L
_
t
2
_
=
3
s
·
2· 1
s
3
=
3· 2· 1
s
4
,
· · · · · ·
in general,
L
_
t
n
_
=
n!
s
n+1
, n = 0, 1, . . . . 0! =1
Solution 2: Use the property of differentiation
L
_
t
n
f (t)
_
= (−1)
n
d
n
F(s)
ds
n
, n = 1, 2, . . . ,
with f (t) =1 and F(s) =
1
s
. Note that
d
ds
_
1
s
_
= −
1
s
2
,
d
2
ds
2
_
1
s
_
=
d
ds
_
−
1
s
2
_
=
1· 2
s
3
=
2!
s
3
,
248 6 the laplace transformand its applications
d
3
ds
3
_
1
s
_
=
d
ds
_
−
2!
s
3
_
= −
3· 2!
s
4
= −
3!
s
4
,
· · · · · ·
d
n
ds
n
_
1
s
_
= (−1)
n
n!
s
n+1
.
Hence
n = 1: L
_
t
_
= L
_
t · 1
_
= −
d
ds
_
1
s
_
=
1
s
2
,
n = 2: L
_
t
2
_
= L
_
t
2
· 1
_
=
d
2
ds
2
_
1
s
_
=
2!
s
3
,
· · · · · ·
n: L
_
t
n
_
= L
_
t
n
· 1
_
= (−1)
n
d
n
ds
n
_
1
s
_
=
n!
s
n+1
.
Example 6.3 6.3
Evaluate L
_
e
at
cos ωt
_
and L
_
e
at
sin ωt
_
.
Note that
L
_
cos ωt
_
=
s
s
2
+ ω
2
, L
_
sin ωt
_
=
ω
s
2
+ ω
2
.
Applying the property of shifting L
_
e
at
f (t)
_
=F(s −a), one obtains
L
_
e
at
cos ωt
_
=
s
s
2
+ ω
2
¸
¸
¸
¸
s→(s−a)
=
s −a
(s −a)
2
+ ω
2
,
L
_
e
at
sin ωt
_
=
ω
s
2
+ ω
2
¸
¸
¸
¸
s→(s−a)
=
ω
(s −a)
2
+ ω
2
.
Example 6.4 6.4
Evaluate L
_
t sin t
_
and L
_
t
2
sin t
_
.
L
_
sin ωt
_
=
ω
s
2
+ ω
2
=⇒ L
_
sin t
_
=
1
s
2
+ 1
= F(s).
Apply the property of differentiation L
_
t
n
f (t)
_
=(−1)
n
F
(n)
(s), n=1, 2, . . . .
dF(s)
ds
=
d
ds
_
1
s
2
+ 1
_
= −
2s
(s
2
+ 1)
2
,
d
2
F(s)
ds
2
=
d
ds
_
dF(s)
ds
_
=
d
ds
_
−
2s
(s
2
+ 1)
2
_
=
6s
2
− 2
(s
2
+ 1)
3
,
∴ L
_
t sin t
_
= −
dF(s)
ds
=
2s
(s
2
+ 1)
2
, L
_
t
2
sin t
_
=
d
2
F(s)
ds
2
=
6s
2
− 2
(s
2
+ 1)
3
.
6.2 the heaviside step function 249
Example 6.5 6.5
Evaluate L
_
t e
at
sin ωt
_
.
Using the property of shifting L
_
e
at
f (t)
_
=F(s −a) =L
_
f (t)
_
¸
¸
¸
s→(s−a)
leads to
L
_
e
at
sin ωt
_
= L
_
sin ωt
_
¸
¸
¸
s→(s−a)
=
ω
s
2
+ ω
2
¸
¸
¸
¸
s→(s−a)
=
ω
(s −a)
2
+ ω
2
.
Applying the property of differentiation L
_
t f (t)
_
= −
d
ds
L
_
f (t)
_
L
_
t · e
at
sin ωt
_
= −
d
ds
L
_
e
at
sin ωt
_
= −
d
ds
_
ω
(s −a)
2
+ ω
2
_
=
2ω(s −a)
_
(s −a)
2
+ ω
2
_
2
.
6.2 The Heaviside Step Function
The Heaviside step function is deﬁned by
H(t −a) =
_
0, t <a,
1, t >a,
where a is a real number, and is shown in Figure 6.1.
t
a
H(t −a)
1
0
Figure 6.1 Heaviside step function.
The Laplace transform of H(t −a) is given by
L
_
H(t −a)
_
=
_
∞
0
e
−st
H(t −a)dt
=
_
a
0
e
−st
H(t −a)dt +
_
∞
a
e
−st
H(t −a)dt
=
_
a
0
e
−st
· 0dt +
_
∞
a
e
−st
· 1dt = −
1
s
e
−st
¸
¸
¸
¸
∞
t=a
=
1
s
e
−as
,
250 6 the laplace transformand its applications
L
_
H(t −a)
_
=
1
s
e
−as
, s >0,
and, for s >a0,
L
_
f (t −a)H(t −a)
_
=
_
∞
0
e
−st
f (t −a)H(t −a)dt
=
_
∞
−a
e
−s(τ+a)
f (τ)H(τ)dτ, t −a = τ,
= e
−as
_
∞
0
e
−sτ
f (τ)dτ = e
−as
F(s),
L
_
f (t −a)H(t −a)
_
= e
−as
F(s), s >a0.
The Heaviside step function is very useful in dealing with functions with discon
tinuities or piecewise smooth functions. The following are some examples:
(1)
f (t) =
_
f
1
(t), t <t
0
,
0, t >t
0
,
= f
1
(t)
_
1−H(t −t
0
)
_
;
t
t
0
f (t)
f
1
(t)
(2)
f (t) =
_
0, t <t
0
,
f
2
(t), t >t
0
,
= f
2
(t)H(t −t
0
);
t
t
0
f (t)
f
2
(t)
(3)
f (t) =
_
f
1
(t), t <t
0
,
f
2
(t), t >t
0
,
=
⎧
⎨
⎩
f
1
(t) +
_
f
2
(t)−f
1
(t)
_
· 0 , t <t
0
,
f
1
(t) +
_
f
2
(t)−f
1
(t)
_
· 1 , t >t
0
,
= f
1
(t) +
_
f
2
(t)−f
1
(t)
_
H(t −t
0
)
= f
1
(t)
_
1−H(t −t
0
)
_
+ f
2
(t)H(t −t
0
).
t
t
0
f (t)
f
1
(t)
f
2
(t)
Remarks: Since Case (3) is the combination of Cases (1) and (2), the result
obtained reflects this observation.
6.2 the heaviside step function 251
If function f (t) has nonzero values only in the range of a<x<b as shown in the
following ﬁgure, than it can be expressed as
f (t) =
⎧
⎪
⎪
⎨
⎪
⎪
⎩
0, t <a,
g(x), a<t <b,
0, t >b,
= g(x)
_
H(t −a) − H(t −b)
_
,
t
a
b
f (t)
g(t)
since
H(t −a) − H(t −b) =
⎧
⎪
⎨
⎪
⎩
0, t <a,
1, a<t <b,
0, t >b.
As a generalization, function f (t) of the following form can be easily written in
terms of the Heaviside step function
f (t) =
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
0, t <t
0
,
f
1
(t), t
0
<t <t
1
,
f
2
(t), t
1
<t <t
2
,
· · · · · ·
f
n
(t), t
n−1
<t <t
n
,
0, t >t
n
,
= f
1
(t)
_
H(t −t
0
) − H(t −t
1
)
_
+ f
2
(t)
_
H(t −t
1
) − H(t −t
2
)
_
+ · · · + f
n
(t)
_
H(t −t
n−1
) − H(t −t
n
)
_
.
Example 6.6 6.6
Express the following functions in terms of the Heaviside function
1.
f (t) =
⎧
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎩
0, t <0,
1, 0<t <1,
2, 1<t <2,
0, t >2;
2.
f (t) =
⎧
⎪
⎨
⎪
⎩
t, t < −1,
t
2
, −1<t <1,
t
3
, t >1.
1.
f (t) = 1 ·
_
H(t −0) − H(t −1)
_
f (t) =1, 0<t <1
+ 2 ·
_
H(t −1) − H(t −2)
_
f (t) =2, 1<t <2
= H(t) + H(t −1) − 2 H(t −2);
252 6 the laplace transformand its applications
2. f (t) = t ·
_
1 − H(t +1)
_
f (t) =t, t < −1
+ t
2
·
_
H(t +1) − H(t −1)
_
f (t) =t
2
, −1<t <1
+ t
3
· H(t −1) f (t) =t
3
, t >1
= t + (t
2
−t) H(t +1) + (t
3
−t
2
) H(t −1).
Laplace transforms of functions involving the Heaviside step function can be
determined from the deﬁnition by direct integration or using the properties of the
Laplace transform.
Example 6.7 6.7
Evaluate L
_
t H(t −2)
_
.
Solution 1: Using the deﬁnition of Laplace transform
L
_
t H(t −2)
_
=
_
∞
0
e
−st
· t H(t −2)dt =
_
∞
2
e
−st
t dt
= −
1
s
_
∞
2
t d(e
−st
), Integration by parts
= −
1
s
_
t e
−st
¸
¸
¸
∞
t=2
−
_
∞
2
e
−st
dt
_
= −
1
s
_
−2e
−2s
+
1
s
e
−st
¸
¸
¸
∞
t=2
_
= −
1
s
_
−2e
−2s
−
1
s
e
−2s
_
=
e
−2s
s
_
2 +
1
s
_
.
Solution 2: Using the following results
L
_
H(t −a)
_
=
1
s
e
−as
, L
_
t f (t)
_
= −F
(s)
leads to
L
_
t H(t −2)
_
= −
d
ds
L
_
H(t −2)
_
= −
d
ds
_
1
s
e
−2s
_
= −
_
−
1
s
2
e
−2s
+
1
s
e
−2s
(−2)
_
=
e
−2s
s
_
2 +
1
s
_
.
Solution 3: To use the formula L
_
f (t −a)H(t −a)
_
=e
−as
F(s) in evaluating
L
_
f (t)H(t −a)
_
, one must rewrite f (t) as f
_
(t −a)+a
_
=g(t −a).
L
_
t H(t −2)
_
= L
__
(t −2)+2
_
H(t −2)
_
= L
_
(t −2)H(t −2)
_
+ 2L
_
H(t −2)
_
= e
−2s
L
_
t
_
+ 2 ·
1
s
e
−2s
= e
−2s
_
1
s
2
+
2
s
_
.
6.2 the heaviside step function 253
Example 6.8 6.8
Evaluate L
_
e
2t
H(t) − e
−3t
H(t −4)
_
.
Three methods are used to evaluate L
_
e
at
H(t −b)
_
.
Method 1: By deﬁnition
L
_
e
at
H(t −b)
_
=
_
∞
0
e
−st
· e
at
H(t −b)dt, b>0
=
_
∞
b
e
−(s−a)t
dt, H(t −b) =0, for t <b
= −
1
s −a
e
−(s−a)t
¸
¸
¸
∞
t=b
=
e
−b(s−a)
s −a
, s >a.
Method 2: Using the following results
L
_
H(t −b)
_
=
1
s
e
−bs
, L
_
e
at
f (t)
_
= L
_
f (t)
_
¸
¸
¸
s→(s−a)
leads to
L
_
e
at
H(t −b)
_
= L
_
H(t −b)
_
¸
¸
¸
s→(s−a)
=
1
s
e
−bs
¸
¸
¸
s→(s−a)
=
e
−b(s−a)
s −a
.
Method 3: Using the result L
_
f (t −a)H(t −a)
_
= e
−as
F(s) results in
L
_
e
at
H(t −b)
_
= L
_
e
a(t−b)+ab
H(t −b)
_
= e
ab
L
_
e
a(t −b)
H(t −b)
_
= e
ab
· e
−bs
L
_
e
at
_
= e
−b(s−a)
1
s −a
.
Hence,
L
_
e
2t
H(t) − e
−3t
H(t −4)
_
=
e
−0·(s−2)
s −2
−
e
−4·(s+3)
s +3
=
1
s −2
−
e
−4(s+3)
s +3
.
Example 6.9 6.9
Evaluate F(s) = L
_
cos 2t H
_
t −
π
8
_
+ (9t
2
+2t −1) H(t −2)
_
.
F(s) = L
_
cos
_
2
_
t −
π
8
_
+
π
4
_
H
_
t −
π
8
_
+
_
9(t
2
−4t +4)+38t −37
_
H(t −2)
_
= L
_
_
cos 2
_
t −
π
8
_
cos
π
4
−sin 2
_
t −
π
8
_
sin
π
4
_
H
_
t −
π
8
_
+
_
9(t −2)
2
+38(t −2)+39
_
H(t −2)
_
=
1
√
2
e
−
π
8
s
L
_
cos 2t −sin 2t
_
+ e
−2s
L
_
9t
2
+38t +39
_
254 6 the laplace transformand its applications
=
1
√
2
e
−
π
8
s
_
s
s
2
+2
2
−
2
s
2
+2
2
_
+ e
−2s
_
9·
2!
s
3
+ 38·
1
s
2
+ 39·
1
s
_
= e
−
π
8
s
s −2
√
2(s
2
+4)
+ e
−2s
39s
2
+38s +18
s
3
.
6.3 Impulse Functions and the Dirac Delta Function
Impulse functions have wide applications in modeling of physical phenomena. For
example, consider an elastic ball of mass m moving at velocity v
0
toward a rigid
wall as shown in Figure 6.2.
t
0
< t < t
0
+ε
In contact
t
0
t
0
+ε
v
0
v
1
Figure 6.2 An elastic ball colliding with a rigid wall.
At time t
0
, the ball collides with the wall; the wall exerts a force f (t) on the ball
over a short period of time ε. During this time, the ball is in contact with the wall
and the velocity of the ball reduces from v
0
to 0 and then changes its direction,
ﬁnally leaving the wall with velocity −v
1
.
t
Area
t
0
t
0
+ε
f (t)
Figure 6.3 Contact force between the elastic ball and rigid wall.
The force f (t) on the ball, shown in Figure 6.3, depends on the contact between
the elastic ball and the rigid wall. The force f (t) is negative because it is opposite to
the direction of the initial velocity v
0
.
The area under the force curve is called the impulse I, i.e.,
I =
_
t
0
+ε
t
0
f (t)dt.
The ImpulseMomentum Principle states that the change in momentum of mass m
is equal to the total impulse on m, i.e., m(−v
1
) − mv
0
= I.
6.3 impulse functions and the dirac delta function 255
As a mathematical idealization, consider an impulse function f (t) over a time
interval t
0
<t <t
0
+ε with constant amplitude I/ε as shown in Figure 6.4, such
that the area under the function f (t), or the impulse, is I.
t
t
0
t
0
+ε
1
ε
1
f (t)
ε
1
I
(a) (b) (c)
Area=I
t
t
0
t
0
+ε
2
ε
2
f (t)
ε
2
I
Area=I
t
t
0
f (t)
Area=I
ε→0
Figure 6.4 The Dirac delta function.
The impulse function f (t) can be expressed in terms of the Heaviside step
function
f (t) =
I
ε
_
H(t −t
0
) − H[t −(t
0
+ε)]
_
.
When ε decreases as depicted in Figure 6.4(a) and Figure 6.4(b), the width of the
time interval over which the impulse is deﬁned decreases and the amplitude I/ε
of the function increases, while keeping the area under the function constant. For
I =1, the limiting function as ε→0, i.e.,
lim
ε→0
f (t) = lim
ε→0
H(t −t
0
) − H[t −(t
0
+ε)]
ε
,
is called the unit impulse function or the Dirac delta function (Figure 6.4(c)) and is
denoted by δ(t −t
0
).
Properties of the Dirac Delta Function
1. δ(t −a) = 0, if t =a;
2. δ(t −a) → +∞, as t →a;
3.
_
a+α
a−α
δ(t −a)dt = 1, α >0;
4. Shifting Property If g(t) is any function,
_
a+α
a−α
g(t)δ(t −a)dt = g(a), α >0.
5.
_
t
−∞
δ(t −a)dt = H(t −a) =
_
0, t <a,
1, t >a,
=⇒
dH(t −a)
dt
=δ(t −a).
256 6 the laplace transformand its applications
6. L
_
δ(t −a)
_
=
_
∞
0
e
−st
· δ(t −a)dt = e
−as
, a>0. Shifting property
7. L
_
f (t)δ(t −a)
_
=
_
∞
0
e
−st
· f (t)δ(t −a)dt = e
−as
f (a), a>0.
δ(t −a) is not a function in the usual sense; it is known as a generalized function.
Example 6.10 6.10
Evaluate F(s) = L
_
cos πt δ(t +1)+2 cos
πt
6
δ(t −2)+(t
3
−2t
2
+5)e
4t
δ(t −3)
_
.
Note that, if a>0,
δ(t +a) = 0, for t 0 =⇒ L
_
f (t)δ(t +a)
_
=
_
∞
0
e
−st
· f (t)δ(t +a)dt = 0,
and
L
_
f (t)δ(t −a)
_
=
_
∞
0
e
−st
· f (t)δ(t −a)dt = e
−as
f (a).
Hence
F(s) = 0 + e
−2s
· 2 cos
_
π
6
· 2
_
+ e
−3s
· (3
3
−2· 3
2
+5)e
4 · 3
= e
−2s
+ 14e
12−3s
.
Applications of the Dirac Delta Function
Consider a distributed load of intensity w(x) over a length of width ε as shown in
Figure 6.5(a).
x
a
Area=W
ε
w(x)
(a) (b)
ε→0
x
a
W
Figure 6.5 Distributed and concentrated loads.
The resultant force of the distributed load is
W =
_
a+
ε
2
a−
ε
2
w(x)dx.
When ε→0, the distributed load approaches a concentrated load W, as shown in
Figure 6.5(b). In terms of the Dirac delta function, the load can be expressed as
w(x) =Wδ(x−a), so that
_
a+
ε
2
a−
ε
2
w(x)dx = W
_
a+
ε
2
a−
ε
2
δ(x−a)dx = W.
6.4 the inverse laplace transform 257
As an example, consider a beam under two concentrated loads W
1
and W
2
,
applied at x =a
1
and x =a
2
, respectively, and a uniformly distributed load of
intensity w
0
over b
1
<x<b
2
as shown in the following ﬁgure.
W
1
w
0
a
1 b
1
b
2
a
2
W
2
x
y
The deﬂection y(x) of the beam is governed by the differential equation
d
2
dx
2
_
EI(x)
d
2
y
dx
2
_
= w(x),
where EI(x) is the ﬂexural rigidity of the beam, and the loading is given by
w(x) = w
0
_
H(x−b
1
)−H(x−b
2
)
_
+ W
1
δ(x−a
1
) + W
2
δ(x−a
2
).
6.4 The Inverse Laplace Transform
Given the Laplace transform F(s) of function f (t), F(s) =L
_
f (t)
_
, the inverse
Laplace transform is f (t) =L
−1
_
F(s)
_
.
Inverse Laplace transforms of frequently used functions are listed in Table of
Inverse Laplace Transforms (Appendix A.5). Hence, to determine the inverse
Laplace transform L
−1
_
F(s)
_
, F(s) has to be recast into a combination of the
known functions so that the formulas in the Table of Inverse Laplace Transforms
can be applied.
Properties of the Inverse Laplace Transform
Corresponding to the properties of the Laplace transform, the following properties
of the inverse Laplace transform can be readily obtained.
1. L
_
·
_
is a linear operator =⇒ L
−1
_
·
_
is a linear operator;
2. L
_
e
at
f (t)
_
= F(s −a) =⇒ L
−1
_
F(s −a)
_
= e
at
L
−1
_
F(s)
_
= e
at
f (t);
3. L
_
t
n
f (t)
_
= (−1)
n
F
(n)
(s) =⇒ L
−1
_
F
(n)
(s)
_
= (−1)
n
t
n
f (t);
4. L
_
f (t −a)H(t −a)
_
=e
−as
F(s −a) =⇒ L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a).
258 6 the laplace transformand its applications
Example 6.11 6.11
Evaluate L
−1
_
s
(s −2)
5
_
.
∵ F(s) =
s
(s −2)
5
=
(s −2) + 2
(s −2)
5
=
1
(s −2)
4
+
2
(s −2)
5
,
∴ f (t) = e
2t
L
−1
_
1
s
4
_
+ 2e
2t
L
−1
_
1
s
5
_
, L
−1
_
F(s −a)
_
=e
at
L
−1
_
F(s)
_
= e
2t
·
t
3
3!
+ 2e
2t
·
t
4
4!
, L
−1
_
1
s
n
_
=
t
n−1
(n−1)!
=
1
12
e
2t
t
3
(2 + t).
Example 6.12 6.12
Evaluate L
−1
_
1 + e
−3s
s
4
_
.
F(s) =
1 + e
−3s
s
4
=
1
s
4
+ e
−3s
1
s
4
.
Note that
L
−1
_
1
s
n
_
=
t
n−1
(n−1)!
, L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a).
Hence,
f (t) = L
−1
_
F(s)
_
=
t
3
3!
+
(t −3)
3
3!
H(t −3) =
1
6
_
t
3
+ (t −3)
3
H(t −3)
_
.
Convolution Integral
Theorem — Convolution Integral
If L
−1
_
F(s)
_
= f (t) and L
−1
_
G(s)
_
=g(t), then
L
−1
_
F(s)G(s)
_
=
_
t
0
f (u) g(t −u)du =
_
t
0
g(u) f (t −u)du =
_
f ∗ g
_
(t),
in which the integral is known as a Convolution Integral.
Example 6.13 6.13
Evaluate L
−1
_
s
(s
2
+4)
2
_
.
6.4 the inverse laplace transform 259
Note that
L
−1
_
2
s
2
+2
2
_
= sin 2t, L
−1
_
s
s
2
+2
2
_
= cos 2t.
Using convolution integral, one has
L
−1
_
s
(s
2
+4)
2
_
=
1
2
L
−1
_
2
s
2
+2
2
·
s
s
2
+2
2
_
=
1
2
sin 2t ∗ cos 2t
=
1
2
_
t
0
sin 2u cos 2(t −u)du, sin Acos B=
1
2
_
sin(A+B)+sin(A−B)
_
=
1
4
_
t
0
_
sin 2t + sin(4u−2t)
_
du =
1
4
_
u sin 2t −
1
4
cos(4u−2t)
_
t
u=0
=
1
4
_
t sin 2t −
1
4
cos 2t +
1
4
cos(−2t)
_
=
1
4
t sin 2t.
Partial Fractions
Partial fraction decomposition is an essential step in solving ordinary differential
equations using the method of Laplace transform. Some important aspects of
partial fractions are brieﬂy reviewed.
Consider a fraction N(x)/D(x), where N(x) and D(x) are polynomials in x of
degrees n
N
and n
D
, respectively.
❧ If n
N
n
D
, i.e., the degree of the numerator N(x) is greater than or equal
to the degree of the denominator, the fraction can be simpliﬁed using long
division to yield
N(x)
D(x)
= P(x) +
N
1
(x)
D(x)
,
where P(x) and N
1
(x) are both polynomials in x, and the degree of N
1
(x) is
less than that of D(x). Hence, without loss of generality, the case for n
N
<n
D
is considered in the following.
❧ Completely factorize the denominator D(x) into factors of the form
(αx+β)
m
and (ax
2
+bx+c)
n
,
where ax
2
+bx+c is an unfactorable quadratic.
For each factor of the form (αx+β)
m
, the partial fraction decomposition
includes the following m terms
A
m
(αx+β)
m
+
A
m−1
(αx+β)
m−1
+ · · · +
A
1
(αx+β)
.
260 6 the laplace transformand its applications
For each factor of the form (ax
2
+bx+c)
n
, the partial fraction decomposi
tion includes the following n terms
B
n
x+C
n
(ax
2
+bx+c)
n
+
B
n−1
x+C
n−1
(ax
2
+bx+c)
n−1
+ · · · +
B
1
x+C
1
(ax
2
+bx+c)
.
For example,
5x
3
+2x+7
2x
8
+7x
7
−10x
5
−6x
4
−x
3
=
5x
3
+2x+7
x
3
(2x−1)(x
2
+2x−1)
2
=
A
3
x
3
+
A
2
x
2
+
A
1
x
+
B
2x−1
+
C
2
x+D
2
(x
2
+2x−1)
2
+
C
1
x+D
1
x
2
+2x−1
.
Summing up the righthand side, the numerator of the resulting fraction is a
polynomial of degree 7. Comparing the coefﬁcients of the numerators leads
to a system of eight linear algebraic equations for the unknown constants.
❧ The CoverUp Method
Suppose D(x) has a factor (x−a)
m
, then the partial fraction decomposition
can be written as
N(x)
D(x)
=
N(x)
(x−a)
m
D
1
(x)
=
A
m
(x−a)
m
+
A
m−1
(x−a)
m−1
+ · · · +
A
1
(x−a)
+
N
1
(x)
D
1
(x)
,
in which D
1
(x) does not have (x−a) as a factor. Multiplying both sides of
the equation by (x−a)
m
yields
N(x)
D
1
(x)
= A
m
+ A
m−1
(x−a) + · · · + A
1
(x−a)
m−1
+
N
1
(x)
D
1
(x)
(x−a)
m
.
Setting x =a gives the value of A
m
A
m
=
N(x)
D
1
(x)
¸
¸
¸
¸
x=a
.
This result can be restated as follows: to ﬁnd A
m
, “coverup” (remove) the
term (x−a)
m
and set x =a:
A
m
=
N(x)
×(x−a)
m
D
1
(x)
¸
¸
¸
¸
¸
x=a
.
✍
The coverup method works only for the highest power of repeated
linear factor.
For example,
4
(x−1)
2
(x+1)
=
A
2
(x−1)
2
+
A
1
x−1
+
B
x+1
.
6.4 the inverse laplace transform 261
To ﬁnd A
2
, coverup (x−1)
2
and set x =1:
A
2
=
4
x+1
¸
¸
¸
¸
x=1
= 2.
To ﬁnd B, coverup (x+1) and set x = −1:
B =
4
(x−1)
2
¸
¸
¸
¸
x=−1
= 1.
A
1
cannot be determined using the coverup method. But since A
2
and B
are known, A
1
can be found by substituting any numerical value (other than
1 and −1) for x in the partial fraction decomposition equation. For instance,
setting x =0:
4
(0−1)
2
(0+1)
=
2
(0−1)
2
+
A
1
0−1
+
1
0+1
=⇒ A
1
= −1.
Example 6.14 6.14
Evaluate L
−1
_
8
(s −1)(s
2
+2s +5)
_
.
Solution 1: Using partial fractions
F(s) =
8
(s −1)(s
2
+2s +5)
=
A
s −1
+
Bs +C
s
2
+2s +5
=
A(s
2
+2s +5) + (Bs +C)(s −1)
(s −1)(s
2
+2s +5)
=
(A+B)s
2
+(2A−B+C)s +(5A−C)
(s −1)(s
2
+2s +5)
.
To ﬁnd A, coverup (s −1) and set s =1:
A =
8
s
2
+2s +5
¸
¸
¸
¸
s=1
=
8
1+2+5
= 1.
Hence, comparing the coefﬁcients of the numerators leads to
s
2
: A + B = 0 =⇒ B = −A = −1,
s : 2A − B + C = 0 =⇒ C = B − 2A = −1 − 2· 1 = −3,
1: 5A − C = 8. Use this equation as a check: 5· 1−(−3) =8.
F(s) =
1
s −1
−
s +3
s
2
+2s +5
=
1
s −1
−
(s +1) + 2
(s +1)
2
+ 2
2
=
1
s −1
−
(s +1)
(s +1)
2
+ 2
2
−
2
(s +1)
2
+ 2
2
.
Using the property of shifting L
−1
{F(s −a)} =e
at
f (t) along with the results
L
−1
_
1
s
_
= 1, L
−1
_
s
s
2
+ω
2
_
= cos ωt, L
−1
_
ω
s
2
+ω
2
_
= sin ωt
262 6 the laplace transformand its applications
yields
f (t) = L
−1
{F(s)} = e
t
− e
−t
cos 2t − e
−t
sin 2t.
Solution 2: Using convolution integral
f (t) = L
−1
_
8
(s −1)(s
2
+2s +5)
_
= 4L
−1
_
1
s −1
·
2
(s +1)
2
+ 2
2
_
= 4 e
t
∗
_
e
−t
sin 2t
_
= 4
_
t
0
e
−u
sin 2u · e
t−u
du
= 4e
t
_
t
0
e
−2u
sin 2udu = 4e
t
_
e
−2u
(−2)
2
+2
2
(−2 sin 2u − 2 cos 2u)
_
t
0
= 4e
t
_
e
−2t
8
(−2 sin 2t − 2 cos 2t) −
1
8
(−2)
_
= −e
−t
(sin 2t + cos 2t) + e
t
.
Example 6.15 6.15
Evaluate L
−1
_
s +1
(s
2
+1)(s
2
+9)
_
.
Solution 1: Using partial fractions
F(s) =
s +1
(s
2
+1)(s
2
+9)
=
As +B
s
2
+1
+
Cs +D
s
2
+9
=
(As +B)(s
2
+9)+(Cs +D)(s
2
+1)
(s
2
+1)(s
2
+9)
=
(A+C)s
3
+ (B+D)s
2
+ (9A+C)s + (9B+D)
(s
2
+1)(s
2
+9)
.
Hence, comparing the coefﬁcients of the numerators leads to
s
3
: A + C = 0, (1)
s
2
: B + D = 0, (2)
s : 9A + C = 1, (3)
1: 9B + D = 1. (4)
∴ Eqn (3) − Eqn (1) : 8A = 1 =⇒ A =
1
8
,
Eqn (4) − Eqn (2) : 8B = 1 =⇒ B =
1
8
,
Eqn (1) : C = −A = −
1
8
,
Eqn (2) : D = −B = −
1
8
.
∴ F(s) =
1
8
_
s +1
s
2
+1
−
s +1
s
2
+9
_
=
1
8
_
s
s
2
+1
2
+
1
s
2
+1
2
−
s
s
2
+3
2
−
1
3
·
3
s
2
+3
2
_
,
6.5 solving differential equations using the laplace transform 263
f (t) = L
−1
{F(s)} =
1
8
_
cos t + sin t − cos 3t −
1
3
sin 3t
_
.
Solution 2: Using convolution integral
L
−1
_
s +1
(s
2
+1)(s
2
+9)
_
= L
−1
_
s
(s
2
+1)(s
2
+9)
_
+L
−1
_
1
(s
2
+1)(s
2
+9)
_
=
1
3
L
−1
_
s
s
2
+1
2
·
3
s
2
+3
2
_
+
1
3
L
−1
_
1
s
2
+1
2
·
3
s
2
+3
2
_
=
1
3
_
cos t ∗ sin 3t + sin t ∗ sin 3t
_
=
1
3
_
t
0
(cos u+sin u) sin 3(t −u)du
=
1
3
_
t
0
_
1
2
_
sin(3t −2u)−sin(4u−3t)
_
−
1
2
_
cos(3t −2u)−cos(4u−3t)
_
_
du
=
1
6
_
1
2
cos(3t −2u)+
1
4
cos(4u−3t)+
1
2
sin(3t −2u)+
1
4
sin(4u−3t)
_
t
u=0
=
1
8
_
cos t + sin t − cos 3t −
1
3
sin 3t
_
.
6.5 Solving Differential Equations Using the Laplace
Transform
Consider an nthorder linear differential equation with constant coefﬁcients
a
n
y
(n)
(t) + a
n−1
y
(n−1)
(t) + · · · + a
1
y
(t) + a
0
y(t) = f (t).
Taking the Laplace transform on both sides of the equation and noting that
L
_
y(t)
_
=Y(s),
L
_
y
(t)
_
= sY(s) − y(0),
L
_
y
(t)
_
= s
2
Y(s) − s y(0) − y
(0),
· · · · · ·
L
_
y
(n)
(t)
_
= s
n
Y(s) − s
n−1
y(0) − s
n−2
y
(0) − · · · − s y
(n−2)
(0) − y
(n−1)
(0),
lead to an algebraic equation for Y(s)
a
n
_
s
n
Y(s) −
n
i=1
s
n−i
y
(i−1)
(0)
_
+ a
n−1
_
s
n−1
Y(s) −
n−1
i=1
s
n−1−i
y
(i−1)
(0)
_
+ · · ·
+ a
1
_
sY(s) − y(0)
_
+ a
0
Y(s) = L
_
f (t)
_
.
264 6 the laplace transformand its applications
Solving for Y(s) yields
Y(s) =
L
_
f (t)
_
+
n
k=1
k
i=1
a
k
y
(i−1)
(0)s
k−i
n
i=0
a
i
s
i
.
Taking the inverse transform y(t) =L
−1
_
Y(s)
_
results in the solution of the dif
ferential equation.
Example 6.16 6.16
Solve y
− 8 y
+ 25 y = e
4t
sin 3t.
Let Y(s) =L
_
y(t)
_
. Using the property of shifting, L
_
e
at
f (t)
_
=F(s −a), one
has
L
_
e
4t
sin 3t
_
= L
_
sin 3t
_
¸
¸
¸
s →s−4
=
3
s
2
+ 3
2
¸
¸
¸
¸
s →s−4
=
3
(s −4)
2
+ 3
2
.
Taking the Laplace transform of both sides of the differential equation yields
_
s
2
Y(s) − s y(0) − y
(0)
_
− 8
_
sY(s) − y(0)
_
+ 25Y(s) =
3
(s −4)
2
+ 3
2
,
or, denoting y(0) = y
0
, y
(0) =v
0
,
(s
2
− 8s + 25)Y(s) = y
0
s + (v
0
−8y
0
) +
3
(s −4)
2
+ 3
2
.
Solving for Y(s) leads to
Y(s) =
y
0
(s −4)
(s −4)
2
+ 3
2
+
v
0
−4y
0
(s −4)
2
+ 3
2
+
3
_
(s −4)
2
+ 3
2
_
2
.
Taking the inverse Laplace transform and using the property of shifting
L
−1
_
F(s −a)
_
= e
at
L
−1
_
F(s)
_
, L
−1
_
1
(s
2
+a
2
)
2
_
=
1
2a
3
(sin at −at cos at),
one obtains
y(t) = L
−1
_
Y(s)
_
= e
4t
L
_
y
0
·
s
s
2
+ 3
2
+
v
0
−4y
0
3
·
3
s
2
+ 3
2
+ 3·
1
(s
2
+ 3
2
)
2
_
= e
4t
_
y
0
cos 3t +
v
0
−4y
0
3
sin 3t +
3
2· 3
3
(sin 3t − 3t cos 3t)
_
= e
4t
_
y
0
cos 3t +
_
v
0
−4y
0
3
+
1
18
_
sin 3t −
t
6
cos 3t
_
.
6.5 solving differential equations using the laplace transform 265
Denoting
A = y
0
, B =
v
0
− 4y
0
3
+
1
18
,
the general solution of the differential equation can be written as
y(t) = e
4t
_
Acos 3t + Bsin 3t −
t
6
cos 3t
_
.
Example 6.17 6.17
Solve y
+ 9 y = 18 sin 3t H(t −π), y(0) = 1, y
(0) = 0.
Let Y(s) =L
_
y(t)
_
. Taking the Laplace transform of both sides of the differential
equation yields
_
s
2
Y(s) − s y(0) − y
(0)
_
+ 9Y(s) = 18L
_
sin 3t H(t −π)
_
,
where
L
_
sin 3t H(t−π)
_
=L
_
sin
_
3(t−π)+3π
_
H(t−π)
_
sin(3π +θ) =−sin θ
= −L
_
sin 3(t−π)H(t−π)
_
L
_
f (t−a)H(t−a)
_
=e
−as
L
_
f (t)
_
= −e
−πs
L
_
sin 3t
_
= −e
−πs
3
s
2
+3
2
.
Hence, solving for Y(s) leads to
Y(s) =
s
s
2
+9
− 54e
−πs
1
(s
2
+9)
2
.
Noting that
L
−1
_
1
(s
2
+3
2
)
2
_
=
1
2· 3
3
(sin 3t − 3t cos 3t),
and using the property of shifting L
−1
_
e
as
F(s)
_
= f (t −a)H(t −a), one has
54L
−1
_
e
−πs
1
(s
2
+3
2
)
2
_
=
_
sin 3(t −π) − 3(t −π) cos 3(t −π)
_
H(t −π)
=
_
−sin3t + 3(t −π) cos 3t
_
H(t −π).
sin(θ −3π) = −sin(3π −θ) = −sin θ, cos(θ −3π) = cos(3π −θ) = −cos θ
The solution of the differential equation is
y(t) = L
−1
_
s
s
2
+3
2
_
− 54L
−1
_
e
−πs
1
(s
2
+3
2
)
2
_
= cos 3t +
_
sin 3t − 3(t −π) cos 3t
_
H(t −π).
266 6 the laplace transformand its applications
Example 6.18 6.18
Solve y
−y
+4 y
−4 y =40(t
2
+t +1)H(t −2), y(0) =5, y
(0) =0, y
(0) =10.
Let Y(s) =L
_
y(t)
_
. Taking the Laplace transform of both sides of the differential
equation yields
_
s
3
Y(s) − s
2
y(0) − s y
(0) − y
(0)
_
−
_
s
2
Y(s) − s y(0) − y
(0)
_
+ 4
_
sY(s) − y(0)
_
− 4Y(s) = L
_
40(t
2
+t +1)H(t −2)
_
,
where, using L
_
f (t −a)H(t −a)
_
=e
−as
L
_
f (t)
_
,
L
_
40(t
2
+t +1)H(t −2)
_
= 40L
__
(t
2
−4t +4)+5t −3
_
H(t −2)
_
= 40L
__
(t −2)
2
+5(t −2)+7
_
H(t −2)
_
= 40e
−2s
L
_
t
2
+5t +7
_
= 40e
−2s
_
2!
s
3
+ 5·
1!
s
2
+ 7·
1
s
_
L
_
t
n
_
=
n!
s
n+1
= e
−2s
40(7s
2
+5s +2)
s
3
.
Solving for Y(s) gives
Y(s) =
5s
2
−5s +30
s
3
−s
2
+4s −4
+ e
−2s
40(7s
2
+5s +2)
s
3
(s
3
−s
2
+4s −4)
.
Using partial fractions, one has
5s
2
−5s +30
(s −1)(s
2
+4)
=
A
s −1
+
Bs +C
s
2
+4
=
(A+B)s
2
+(−B+C)s +(4A−C)
(s −1)(s
2
+4)
To ﬁnd A, coverup (s −1) and set s =1
A =
5s
2
−5s +30
(s
2
+4)
¸
¸
¸
¸
s=1
=
5−5+30
1+4
= 6.
Comparing the coefﬁcients of the numerators leads to
s
2
: A + B = 5 =⇒ B = 5 − A = 5 − 6 = −1,
s : −B + C = −5 =⇒ C = B − 5 = −1 − 5 = −6,
1: 4A − C = 30. Use this equation as a check: 4· 6−(−6) =30.
Hence,
L
−1
_
5s
2
−5s +30
(s −1)(s
2
+4)
_
= L
−1
_
6
s −1
+
−s −6
s
2
+4
_
= L
−1
_
6·
1
s −1
−
s
s
2
+2
2
− 3·
2
s
2
+2
2
_
= 6e
t
− cos 2t − 3 sin 2t.
6.5 solving differential equations using the laplace transform 267
Similarly, using partial fractions, one has
40(7s
2
+5s +2)
s
3
(s −1)(s
2
+4)
=
A
3
s
3
+
A
2
s
2
+
A
1
s
+
B
s −1
+
Cs +D
s
2
+4
.
Comparing the coefﬁcients of the numerators leads to
s
5
: A
1
+ B + C = 0, (1)
s
4
: −A
1
+ A
2
− C + D = 0, (2)
s
3
: 4A
1
− A
2
+ A
3
+ 4B − D = 0, (3)
s
2
: −4A
1
+ 4A
2
− A
3
= 280, (4)
s : −4A
2
+ 4A
3
= 200, (5)
1: −4A
3
= 80, (6)
from Eqn (6) : A
3
= −20,
from Eqn (5) : A
2
= A
3
−50 = −70,
from Eqn (4) : A
1
=
1
4
(4A
2
−A
3
−280) = −135,
to ﬁnd B, coveringup (s −1) and setting s =1:
B =
40(7s
2
+5s +2)
s
3
(s
2
+4)
¸
¸
¸
¸
s=1
=
40(7+5+2)
1· 5
= 112,
from Eqn (1) : C = −A
1
− B = −(−135) − 112 = 23,
from Eqn (6) : D = A
1
− A
2
+ C = −135 − (−70) + 23 = −42.
Hence,
L
−1
_
40(7s
2
+5s +2)
s
3
(s −1)(s
2
+4)
_
= L
−1
_
−
20
s
3
−
70
s
2
−
135
s
+
112
s −1
+
23s −42
s
2
+2
2
_
= −135 − 70t − 10t
2
+ 112e
t
+ 23 cos 2t − 21 sin 2t.
Applying the property of shifting L
−1
_
e
as
F(s)
_
= f (t −a)H(t −a), one obtains
the solution of the differential equation
y(t) = L
−1
_
5s
2
−5s +30
s
3
−s
2
+4s −4
_
+L
−1
_
e
−2s
40(7s
2
+5s +2)
s
3
(s
3
−s
2
+4s −4)
_
= 6e
t
− cos 2t − 3 sin 2t +
_
−135 − 70(t −2) − 10(t −2)
2
+ 112e
(t −2)
+ 23 cos 2(t −2) − 21 sin 2(t −2)
_
H(t −2)
= 6e
t
− cos 2t − 3 sin 2t +
_
−10t
2
− 30t − 35 + 112e
t −2
+ 23 cos(2t −4) − 21 sin(2t −4)
_
H(t −2).
268 6 the laplace transformand its applications
Remarks:
❧ Using the method of Laplace transform to solve linear ordinary equations is
a general approach, as long as the Laplace transform of the righthand side
function f (t) and the resulting inverse Laplace transform Y(s) can be readily
obtained.
❧ The Laplace transforms of the Heaviside step function and the Dirac delta
function are both continuous functions. As a result, the method of Laplace
transform offers great advantage in dealing with differential equations
involving the Heaviside step function and the Dirac function.
❧ The determination of inverse Laplace transforms is in general a tedious task.
As a result, for a linear differential equation with continuous righthand side
function f (t), the methods presented in Chapter 4 are usually more efficient
than the method of Laplace transform, especially for higherorder differential
equations.
6.6 Applications of the Laplace Transform
6.6.1 Response of a Single DegreeofFreedom System
Consider a single degreeoffreedomsystemas showninFigure 6.6 under externally
applied force f (t). The initial conditions of the system at time t =0 are x(0) =x
0
and ˙ x(0) =v
0
.
m
x(t)
f (t)
k
c
Figure 6.6 A single degreeoffreedom system.
As derived in Section 5.1, the equation of motion of the system is
m¨ x + c ˙ x + kx = f (t), x(0) = x
0
, ˙ x(0) = v
0
.
Dividing the equation by m and rewriting it in the standard form yield
¨ x + 2ζ ω
0
˙ x + ω
2
0
x =
1
m
f (t), ω
2
0
=
k
m
, 2ζ ω
0
=
c
m
.
Applying the Laplace transform, L
_
x(t)
_
=X(s), one obtains
_
s
2
X(s) − s x(0) − ˙ x(0)
_
+ 2ζ ω
0
_
s X(s) − x(0)
_
+ ω
2
0
X(s) =
1
m
L
_
f (t)
_
.
6.6 applications of the laplace transform 269
Solving for X(s) leads to
X(s) =
x
0
s + (v
0
+2ζ ω
0
x
0
)
s
2
+ 2ζ ω
0
s + ω
2
0
+
1
m
·
L
_
f (t)
_
s
2
+ 2ζ ω
0
s + ω
2
0
,
. ,, . . ,, .
X
Free
(s), free vibration X
Forced
(s), forcedvibration
in which the ﬁrst term is the Laplace transform of the response of free vibration
x
Free
(t), due to the initial conditions x(0) and ˙ x(0); whereas the second term is
the Laplace transform of the response of forced vibration x
Forced
(t), due to the
externally applied force f (t).
Free Vibration
The response of free vibration is given by
x
Free
(t) = L
−1
_
X
Free
(s)
_
= L
−1
_
x
0
s + (v
0
+2ζ ω
0
x
0
)
s
2
+ 2ζ ω
0
s + ω
2
0
_
.
❧ For 0ζ <1, ω
d
=ω
0
_
1−ζ
2
,
x
Free
(t) = L
−1
_
x
0
(s +ζ ω
0
) + (v
0
+ζ ω
0
x
0
)
(s +ζ ω
0
)
2
+ ω
2
0
(1−ζ
2
)
_
= e
−ζ ω
0
t
L
−1
_
x
0
s + (v
0
+ζ ω
0
x
0
)
s
2
+ ω
2
0
(1−ζ
2
)
_
, L
−1
_
F(s −a)
_
=e
at
L
−1
_
F(s)
_
= e
−ζ ω
0
t
_
x
0
L
−1
_
s
s
2
+ ω
2
d
_
+
v
0
+ζ ω
0
x
0
ω
d
L
−1
_
ω
d
s
2
+ ω
2
d
__
,
x
Free
(t) = e
−ζ ω
0
t
_
x
0
cos ω
d
t +
v
0
+ζ ω
0
x
0
ω
d
sin ω
d
t
_
, 0ζ <1.
❧ For ζ =1,
x
Free
(t) = L
−1
_
x
0
s + (v
0
+2ω
0
x
0
)
s
2
+ 2ω
0
s + ω
2
0
_
= L
−1
_
x
0
(s +ω
0
) + (v
0
+ω
0
x
0
)
(s +ω
0
)
2
_
= e
−ω
0
t
L
−1
_
x
0
s + (v
0
+ω
0
x
0
)
s
2
_
= e
−ω
0
t
_
x
0
L
−1
_
1
s
_
+ (v
0
+ω
0
x
0
)L
−1
_
1
s
2
__
,
x
Free
(t) = e
−ω
0
t
_
x
0
+ (v
0
+ω
0
x
0
)t
_
, ζ =1.
❧ For ζ >1,
x
Free
(t) = L
−1
_
x
0
(s +ζ ω
0
) + (v
0
+ζ ω
0
x
0
)
(s +ζ ω
0
)
2
−ω
2
0
(ζ
2
−1)
_
270 6 the laplace transformand its applications
= e
−ζ ω
0
t
L
−1
_
x
0
s + (v
0
+ζ ω
0
x
0
)
s
2
− ω
2
0
(ζ
2
−1)
_
Using partial fractions
= e
−ζ ω
0
t
·
1
2
L
−1
_
x
0
+
v
0
+ζ ω
0
x
0
ω
0
_
ζ
2
−1
s − ω
0
_
ζ
2
−1
+
x
0
−
v
0
+ζ ω
0
x
0
ω
0
_
ζ
2
−1
s + ω
0
_
ζ
2
−1
_
= e
−ζ ω
0
t
·
1
2
_
_
x
0
+
v
0
+ζ ω
0
x
0
ω
0
_
ζ
2
−1
_
e
ω
0
√
ζ
2
−1 t
+
_
x
0
−
v
0
+ζ ω
0
x
0
ω
0
_
ζ
2
−1
_
e
−ω
0
√
ζ
2
−1 t
_
,
x
Free
(t) =
1
2ω
0
_
ζ
2
−1
_
_
v
0
+ω
0
_
ζ +
_
ζ
2
−1
_
x
0
_
e
−ω
0
(ζ −
√
ζ
2
−1)t
−
_
v
0
+ω
0
_
ζ −
_
ζ
2
−1
_
x
0
_
e
−ω
0
(ζ +
√
ζ
2
−1)t
_
, ζ >1.
Remarks: These results are the same as those obtained in Section 5.1.2.1 and
correspond to the complementary solutions.
Forced Vibration
Some examples are studied in the following to illustrate the determination of re
sponse of forced vibration using the Laplace transform.
Example 6.19 — Single DOF System under Sinusoidal Excitation 6.19
Determine x
Forced
(t) of a single degreeoffreedom system with 0ζ <1 under
external excitations f (t) =msin t and f (t) =mcos t.
For f (t) =msin t, one has
X
Forced
(s) = X
sin
(s) =
L
_
sin t
_
s
2
+ 2ζ ω
0
s + ω
2
0
=
(s
2
+2ζ ω
0
s +ω
2
0
)(s
2
+
2
)
.
Applying partial fractions
(s
2
+2ζ ω
0
s +ω
2
0
)(s
2
+
2
)
=
As +B
s
2
+2ζ ω
0
s +ω
2
0
+
Cs +D
s
2
+
2
.
Summing up the righthand side and comparing the coefﬁcients of the numerators,
one obtains
s
3
: A + C = 0, (1)
s
2
: B + 2ζ ω
0
C + D = 0, (2)
s :
2
A + ω
2
C + 2ζ ω
0
D = 0, (3)
1:
2
B + ω
2
0
D = , (4)
from Eqn (1) : A = −C,
6.6 applications of the laplace transform 271
Eqn (2)×
2
− Eqn (4) : D =
+2ζ ω
0
2
C
ω
0
−
2
.
Substituting into equation (3) yields
C = −
2ζ ω
0
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
= −A =⇒ D =
(ω
2
0
−
2
)
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
.
From equation (4),
B =
1
2
(−ω
2
D) =
_
(2ζ ω
0
)
2
+
2
−ω
2
0
_
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
.
Hence,
X
sin
(s) =
A· (s +ζ ω
0
)
(s +ζ ω
0
)
2
+ω
2
d
+
B−ζ ω
0
A
ω
d
·
ω
d
(s +ζ ω
0
)
2
+ω
2
d
+
C· s
s
2
+
2
+
D
·
s
2
+
2
.
Taking inverse Laplace transform results in
x
sin
(t) = L
−1
_
X
sin
(s)
_
= e
−ζ ω
0
t
_
Acos ω
d
t +
B−ζ ω
0
A
sin
ω
d
t
_
+ C cos t +
D
sin t.
x
sin
(t) = L
−1
_
L
_
sin t
_
s
2
+ 2ζ ω
0
s + ω
2
0
_
=
1
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
_
−2ζ ω
0
cos t + (ω
2
0
−
2
) sin t
+ e
−ζ ω
0
t
_
2ζ ω
0
cos ω
d
t +
(2ζ
2
ω
2
0
+
2
−ω
2
0
)
ω
d
sin ω
d
t
_
_
.
Similarly, for f (t) =mcos t, one has
X
Forced
(s) = X
cos
(s) =
L
_
cos t
_
s
2
+ 2ζ ω
0
s + ω
2
0
=
s
(s
2
+2ζ ω
0
s +ω
2
0
)(s
2
+
2
)
,
x
cos
(t) = L
−1
_
L
_
cos t
_
s
2
+ 2ζ ω
0
s + ω
2
0
_
=
1
(ω
2
0
−
2
)
2
+(2ζ ω
0
)
2
_
(ω
2
0
−
2
) cos t + 2ζ ω
0
sin t
+ e
−ζ ω
0
t
_
−(ω
2
0
−
2
) cos ω
d
t −
ζ ω
0
(ω
2
0
+
2
)
ω
d
sin ω
d
t
_
_
.
272 6 the laplace transformand its applications
Remarks: The results x
sin
(t) and x
cos
(t) are very useful for vibration problems
of a single degreeoffreedomsystemunder loads that can be expressed in terms of
sinusoidal functions, inwhichthe response of the forcedvibration canbe expressed
in terms of x
sin
(t) and x
cos
(t). An example using these functions is presented in
the following.
Example 6.20 — Vibration of a Vehicle Passing a Speed Bump 6.20
In Section 5.3, the vibration of a vehicle passing a speed bump is studied. The
equation of motion is solved and the response is obtained separately for two time
durations: vehicle on the speed bump and passed the speed bump. Solve this
problem again using the Laplace transform.
Since the speed bump occurs for 0x b or 0t T, T =b/U, it can be more
easily expressed using the Heaviside step function:
y
0
(x) = h sin
_
π
b
x
_
_
1−H(x−b)
_
, or y
0
(t) = h sin
_
πU
b
t
_
_
1−H(t −T)
_
.
Referring to Section 5.3, the equation of motion for the relative displacement
z(t) = y(t)−y
0
(t) becomes, for t 0,
m¨ z + c ˙ z + kz = mh
2
sin t
_
1−H(t −T)
_
, =
πU
b
,
or, in the standard form as in Section 6.6.1,
¨ z + 2ζ ω
0
˙ z + ω
2
0
z =
1
m
f (t), 0<ζ <1,
where f (t) =mh
2
sin t
_
1−H(t −T)
_
.
Since z(0) =˙ z(0) =0, the response of free vibration is z
Free
(t) =0.
The Laplace transform Z
Forced
(s) of the response of forced vibration is
Z
Forced
(s) =
1
m
·
L
_
f (t)
_
s
2
+ 2ζ ω
0
s + ω
2
0
,
where, using the property L
_
f (t −a)H(t −a) =e
−as
F(s)
_
,
L
_
f (t)
_
= mh
2
L
_
sin t
_
1−H(t −T)
__
,
L
_
sin t
_
1−H(t −T)
__
= L
_
sin t
_
−L
_
sin
_
(t −T)+T
_
H(t −T)
_
= L
_
sint
_
−L
__
sin (t −T) cos T + cos (t −T) sin T
_
H(t −T)
_
= L
_
sin t
_
− cos T · e
−Ts
L
_
sin t
_
− sin T · e
−Ts
L
_
cos t
_
=
_
1−cos T · e
−Ts
_
L
_
sin t
_
− sin T · e
−Ts
L
_
cos t
_
.
6.6 applications of the laplace transform 273
Referring to Section 6.6.1, in terms of X
sin
(s) and X
cos
(s), the Laplace transform
of the response of forced vibration is
Z
Forced
(s) = h
2
_
_
1−cos T · e
−Ts
_ L
_
sin t
_
s
2
+ 2ζ ω
0
s + ω
2
0
− sin T · e
−Ts
L
_
cos t
_
s
2
+ 2ζ ω
0
s + ω
2
0
_
= h
2
_
X
sin
(s) −
_
cos T · e
−Ts
X
sin
(s) + sin T · e
−Ts
X
cos
(s)
__
.
Taking inverse Laplace transform and using L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a),
the response of forced vibration is, in terms of x
sin
(t) and x
cos
(t),
z
Forced
= h
2
_
x
sin
(t) −
_
cos T · x
sin
(t −T) + sin T · x
cos
(t −T)
_
H(t −T)
_
.
Remarks: Usingthe Heaviside stepfunction, the speedbump andthe flat surface
canbe expressedas a compact analytical equation. Applyingthe methodof Laplace
transform, the equation of motion with the loading involving the Heaviside step
function can easily be solved. Both the solution procedure and the expression of
the response are much simpler than those presented in Section 5.3.
Example 6.21 — Single DOF System under Blast Force 6.21
Determine x
Forced
(t) with 0<ζ <1 and f (t) being a model of a blast force, which
can be expressed as f (t) =
f
0
T
(T−t)
_
1 − H(t −T)
_
.
t
f (t)
f
0
0 T
T
(t −T)
Applying the formula L
_
f (t −a)H(t −a)
_
=e
−as
L
_
f (t)
_
, one has
L
_
f (t)
_
=
f
0
T
L
_
T − t + (t −T)H(t −T)
_
=
f
0
T
_
T
s
−
1
s
2
+
1
s
2
e
−Ts
_
,
∴ X
Forced
(s) =
1
m
·
L
_
f (t)
_
s
2
+ 2ζ ω
0
s + ω
2
0
=
f
0
mT
_
T ·
1
s (s
2
+ 2ζ ω
0
s + ω
2
0
)
+
_
−1 + e
−Ts
_
·
1
s
2
(s
2
+ 2ζ ω
0
s + ω
2
0
)
_
.
. ,, . . ,, .
Y
1
(s) Y
2
(s)
274 6 the laplace transformand its applications
Applying partial fractions to Y
1
(s)
Y
1
(s) =
1
s (s
2
+ 2ζ ω
0
s + ω
2
0
)
=
A
1
s
+
B
1
s + C
1
s
2
+ 2ζ ω
0
s + ω
2
0
=
(A
1
+B
1
)s
2
+ (2ζ ω
0
A
1
+C
1
)s + ω
2
0
A
1
s (s
2
+ 2ζ ω
0
s + ω
2
0
)
,
and comparing the coefﬁcients of the numerators, one obtains
1: ω
2
0
A
1
= 1 =⇒ A
1
=
1
ω
2
0
,
s : 2ζ ω
0
A
1
+ C
1
= 0 =⇒ C
1
= −2ζ ω
0
A
1
= −
2ζ
ω
0
,
s
2
: A
1
+ B
1
= 0 =⇒ B
1
= −A
1
= −
1
ω
2
0
.
Hence,
Y
1
(s) =
1
ω
2
0
·
1
s
+
_
−
1
ω
2
0
(s +ζ ω
0
) +
ζ
ω
0
_
−
2ζ
ω
0
(s +ζ ω
0
)
2
+ ω
2
d
, ω
d
=ω
0
_
1−ζ
2
,
=
1
ω
2
0
·
1
s
−
1
ω
2
0
·
(s +ζ ω
0
)
(s +ζ ω
0
)
2
+ ω
2
d
−
ζ
ω
0
ω
d
·
ω
d
(s +ζ ω
0
)
2
+ ω
2
d
,
η
1
(t) = L
−1
_
Y
1
(s)
_
=
1
ω
2
0
−
1
ω
2
0
· e
−ζ ω
0
t
cos ω
d
t −
ζ
ω
0
ω
d
· e
−ζ ω
0
t
sin ω
d
t,
η
1
(t) = L
−1
_
1
s (s
2
+ 2ζ ω
0
s + ω
2
0
)
_
=
1
ω
2
0
−
e
−ζ ω
0
t
ω
0
_
1
ω
0
cos ω
d
t +
ζ
ω
d
sin ω
d
t
_
.
Similarly, applying partial fractions to Y
2
(s)
Y
2
(s) =
1
s
2
(s
2
+ 2ζ ω
0
s + ω
2
0
)
=
A
2
s + B
2
s
2
+
C
2
s + D
2
s
2
+ 2ζ ω
0
s + ω
2
0
=
(A
2
+C
2
)s
3
+ (2ζ ω
0
A
2
+B
2
+D
2
)s
2
+ (ω
2
0
A
2
+2ζ ω
0
B
2
)s + ω
2
0
B
2
s
2
(s
2
+ 2ζ ω
0
s + ω
2
0
)
,
and comparing the coefﬁcients of the numerators, one obtains
1: ω
2
0
B
2
= 1 =⇒ B
2
=
1
ω
2
0
,
s : ω
2
0
A
2
+ 2ζ ω
0
B
2
= 0 =⇒ A
2
= −
2ζ
ω
0
B
2
= −
2ζ
ω
3
0
,
6.6 applications of the laplace transform 275
s
2
: 2ζ ω
0
A
2
+ B
2
+ D
2
= 0 =⇒ D
2
= −2ζ ω
0
A
2
− B
2
=
4ζ
2
−1
ω
2
0
,
s
3
: A
2
+ C
2
= 0 =⇒ C
2
= −A
2
=
2ζ
ω
3
0
.
Hence,
Y
2
(s) = −
2ζ
ω
3
0
·
1
s
+
1
ω
2
0
·
1
s
2
+
_
2ζ
ω
3
0
(s +ζ ω
0
) −
2ζ
2
ω
2
0
_
+
4ζ
2
−1
ω
2
0
(s +ζ ω
0
)
2
+ ω
2
d
= −
2ζ
ω
3
0
·
1
s
+
1
ω
2
0
·
1
s
2
+
2ζ
ω
3
0
·
(s +ζ ω
0
)
(s +ζ ω
0
)
2
+ω
2
d
+
2ζ
2
−1
ω
2
0
ω
d
·
ω
d
(s +ζ ω
0
)
2
+ω
2
d
,
η
2
(t) = L
−1
_
Y
2
(s)
_
= −
2ζ
ω
3
0
+
1
ω
2
0
· t +
2ζ
ω
3
0
· e
−ζ ω
0
t
cos ω
d
t +
2ζ
2
−1
ω
2
0
ω
d
· e
−ζ ω
0
t
sin ω
d
t,
η
2
(t) = L
−1
_
1
s
2
(s
2
+ 2ζ ω
0
s + ω
2
0
)
_
=
ω
0
t −2ζ
ω
3
0
+
e
−ζ ω
0
t
ω
2
0
_
2ζ
ω
0
cos ω
d
t +
2ζ
2
−1
ω
d
sin ω
d
t
_
.
Noting that L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a), one obtains the response of the
forced vibration in terms of functions η
1
(t) and η
2
(t)
x
Forced
(t) = L
−1
_
X
Forced
(s)
_
=
f
0
mT
_
Tη
1
(t) − η
2
(t) + η
2
(t −T)H(t −T)
_
.
Remarks: The inverse Laplace transforms η
1
(t) and η
2
(t) are very useful for
vibration problems of a single degreeoffreedom system under piecewise linear
loads, in which the response of the forced vibration can be expressed in terms of
η
1
(t) and η
2
(t).
6.6.2 Other Applications
Example 6.22 — Electric Circuit 6.22
Consider the circuit shown in the following ﬁgure. The switch is moved from
position a to b at t =0. Derive the differential equation governing i(t) for t >0.
The voltage source V(t) gives impulses of amplitude 1Vperiodically withperiod
1 sec as shown. Given that R
1
=R
2
=2 , V
0
=4V, L=64 H, α =1, determine
i(t) for the following three cases:
(1) C=4 F; (2) C=2 F; (3) C=
7
8
F.
276 6 the laplace transformand its applications
V(t)H(t)
a b
L
v
1
V
0
C
R
2
R
1
αv
1
t =0
i
t (sec)
V(t) (V)
1 1
0 1 2
1
3
❧ For t <0, the switch is at position a and V(t)H(t) =0. The inductor behaves as
a short circuit and the capacitor behaves as an open circuit.
Applying Kirchhoff ’s Voltage Law on the bigger mesh gives
−v
2
(0
−
) + v
1
(0
−
) + V
0
+ v
L
(0
−
) = 0 =⇒ −R
2
_
−i(0
−
)
_
+R
1
i(0
−
)+V
0
=0,
∴ i(0
−
) = −
V
0
R
1
+R
2
, v
1
(0
−
) = R
2
i(0
−
) = −
R
1
R
1
+R
2
V
0
.
Applying Kirchhoff ’s Voltage Law on the right mesh yields
−αv
1
(0
−
) − v
C
(0
−
) + v
1
(0
−
) + V
0
= 0,
∴ v
C
(0
−
) = V
0
+ (1−α)v
1
(0
−
) =
αR
1
+R
2
R
1
+R
2
V
0
.
V(t)
L
v
1
C
R
2
R
1
αv
1
i
L
v
1
(0
−
) v
2
(0
−
) v
2
v
L
(0
−
)
V
0
C
R
2
R
1
αv
1
(0
−
)
i(0
−
)
v
C
(0
−
)
v
L
v
C
i
C
i
2
i
2
(0
−
)
t 0
–
t 0
+
1
❧ At t =0, the switch is moved from position a to b. Since the current in an
inductor cannot change abruptly,
i(0
+
) = i(0
−
) = −
V
0
R
1
+R
2
.
6.6 applications of the laplace transform 277
Since the voltage across a capacitor cannot change abruptly,
v
C
(0
+
) = v
C
(0
−
) =
αR
1
+R
2
R
1
+R
2
V
0
.
Applying Kirchhoff ’s Voltage Law on the right mesh yields
−αv
1
(0
+
) − v
C
(0
+
) + v
1
(0
+
) + v
L
(0
+
) = 0.
Since i(0
+
) =i(0
−
), one has v
1
(0
+
) =v
1
(0
−
). Hence
(1−α)v
1
(0
+
)−v
C
(0
+
) = −V
0
, v
L
(0
+
) =L
di(0
+
)
dt
=⇒ i
(0
+
) =
V
0
L
.
❧ For t >0, applying Kirchhoff ’s Voltage Law on the right mesh
−αv
1
− v
C
+ v
1
+ v
L
= 0,
∴ v
C
= (1−α)v
1
+ v
L
=⇒ i
C
= C
dv
C
dt
= C
_
(1−α)
dv
1
dt
+
dv
L
dt
_
.
Applying Kirchhoff ’s Voltage Law on the bigger mesh leads to
−V(t) − v
2
+ v
1
+ v
L
=0 =⇒ v
2
= −V(t) + v
1
+ v
L
, i
2
=
v
2
R
2
.
Noting that v
1
=R
1
i, v
L
=L
di
dt
, one has
i
C
= C
_
(1−α)R
1
di
dt
+ L
d
2
i
dt
2
_
, i
2
=
1
R
2
_
−V(t) + R
1
i + L
di
dt
_
.
Applying Kirchhoff ’s Current Law at node 1 yields i +i
2
+i
C
=0. Hence
i +
1
R
2
_
−V(t) + R
1
i + L
di
dt
_
+ C
_
(1−α)R
1
di
dt
+ L
d
2
i
dt
2
_
,
∴ R
2
CL
d
2
i
dt
2
+
_
L+C(1−α)R
1
R
2
_
di
dt
+ (R
1
+R
2
)i = V(t),
with initial conditions
i(0
+
) = −
V
0
R
1
+R
2
, i
(0
+
) =
V
0
L
.
The differential equation is of the standard form
d
2
i
dt
2
+ 2ζ ω
0
di
dt
+ ω
2
0
i =
1
m
f (t),
where
ω
0
=
_
R
1
+R
2
R
2
CL
, ζ ω
0
=
L+C(1−α)R
1
R
2
2R
2
CL
, m = R
2
CL, f (t) = V(t).
278 6 the laplace transformand its applications
Remarks: As discussed in Chapter 5, a secondorder circuit is equivalent to a
single degreeoffreedom system. Thus, the results obtained for a single degree
offreedom system can be applied to a secondorder circuit.
The solution is i(t) =i
Free
(t)+i
Forced
(t), where i
Free
(t) is the complementary so
lution or the response of “free” vibration.
Using the Dirac delta function, the forcing function f (t) =V(t) is
f (t) = V(t) = 1 · δ(t −1) + 1 · δ(t −2) + 1 · δ(t −3) + · · · =
∞
n=1
δ(t −n).
Applying the Laplace transform
L
_
δ(t −a)
_
= e
−as
=⇒ L
_
f (t)
_
=
∞
n=1
L
_
δ(t −n)
_
=
∞
n=1
e
−ns
.
Case 1. Underdamped System
For R
1
=R
2
=2 , V
0
=4V, L=64 H, α =1, C=4 F, the differential equation
becomes
d
2
i
dt
2
+
1
8
di
dt
+
1
128
i =
1
512
V(t), i(0
+
) = −1, i
(0
+
) =
1
16
,
which is of the standard form with m=512, f (t) =V(t), and
ω
0
=
1
8
√
2
, 2ζ ω
0
=
1
8
=⇒ ζ =
1
√
2
<1, ω
d
=ω
0
_
1−ζ
2
=
1
16
.
The system is underdamped and
i
Free
(t) = e
−ζ ω
0
t
_
i(0
+
) cos ω
d
t +
i
(0
+
)+ζ ω
0
i(0
+
)
ω
d
sin ω
d
t
_
= −e
−
t
16
cos
t
16
.
The Laplace transform of the “forced” response I
Forced
(s) =L
_
i
Forced
(t)
_
is
I
Forced
(s) =
1
m
·
L
_
f (t)
_
s
2
+ 2ζ ω
0
s + ω
2
0
=
1
512
∞
n=1
e
−ns
1
s
2
+
1
8
s +
1
128
=
1
512
∞
n=1
e
−ns
1
_
1
16
_
·
_
1
16
_
_
s +
1
16
_
2
+
_
1
16
_
2
.
Taking the inverse Laplace transform, one has
i
Forced
(t) =
1
32
∞
n=1
e
n
16
L
−1
_
e
−n(s +
1
16
)
_
1
16
_
_
s +
1
16
_
2
+
_
1
16
_
2
_
L
−1
_
F(s −a)
_
=e
at
L
_
F(s)
_
6.6 applications of the laplace transform 279
=
1
32
∞
n=1
e
n
16
· e
−
t
16
L
−1
_
e
−ns
_
1
16
_
s
2
+
_
1
16
_
2
_
L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a)
=
1
32
∞
n=1
e
−
t−n
16
sin
t −n
16
H(t −n). L
−1
_
ω
s
2
+ω
2
_
= sin ωt
Hence, the current i(t) is
i(t) = i
Free
(t) + i
Forced
(t) =−e
−
t
16
cos
t
16
+
1
32
∞
n=1
e
−
t−n
16
sin
t −n
16
H(t −n) (A).
Case 2. Critically Damped System
For R
1
=R
2
=2 , V
0
=4V, L=64 H, α =1, C=2 F, the differential equation
becomes
d
2
i
dt
2
+
1
4
di
dt
+
1
64
i =
1
256
V(t), i(0
+
) = −1, i
(0
+
) =
1
16
,
which is of the standard form with m=256, f (t) =V(t), and
ω
0
=
1
8
, 2ζ ω
0
=
1
4
=⇒ ζ =1.
The system is critically damped and
i
Free
(t) =
_
i(0
+
) +
_
i
(0
+
)+ω
0
i(0
+
)
_
t
_
e
−ω
0
t
= −
_
1+
t
16
_
e
−
t
8
.
The Laplace transform of the “forced” response I
Forced
(s) =L
_
i
Forced
(t)
_
is
I
Forced
(s) =
1
m
·
L
_
f (t)
_
s
2
+ 2ζ ω
0
s + ω
2
0
=
1
256
∞
n=1
e
−ns
1
s
2
+
1
4
s +
1
64
.
Take the inverse Laplace transform, one has
i
Forced
(t) =
1
256
∞
n=1
e
n
8
L
−1
_
e
−n(s +
1
8
) 1
_
s +
1
8
_
2
_
L
−1
_
F(s −a)
_
=e
at
L
_
F(s)
_
=
1
256
∞
n=1
e
n
8
· e
−
t
8
L
−1
_
e
−ns
1
s
2
_
L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a)
=
1
256
∞
n=1
e
−
t−n
8
(t −n) H(t −n). L
−1
_
1
s
2
_
=t
Hence, the current i(t) is
i(t) = i
Free
(t) + i
Forced
(t) =−
_
1+
t
16
_
e
−
t
8
+
1
256
∞
n=1
e
−
t−n
8
(t−n) H(t−n) (A).
280 6 the laplace transformand its applications
Case 3. Overdamped System
For R
1
=R
2
=2 , V
0
=4V, L=64 H, α =1, C=
7
8
F, the differential equation
becomes
d
2
i
dt
2
+
4
7
di
dt
+
1
28
i =
1
112
V(t), i(0
+
) = −1, i
(0
+
) =
1
16
,
which is of the standard form with m=112, f (t) =V(t), and
ω
0
=
1
2
√
7
, 2ζ ω
0
=
4
7
=⇒ ζ =
4
√
7
>1, ω
0
_
ζ
2
−1 =
3
14
.
The system is overdamped and
i
Free
(t) =
1
2ω
0
_
ζ
2
−1
_
_
i
(0
+
)+ω
0
_
ζ +
_
ζ
2
−1
_
i(0
+
)
_
e
−ω
0
(ζ −
√
ζ
2
−1)t
−
_
i
(0
+
)+ω
0
_
ζ −
_
ζ
2
−1
_
i(0
+
)
_
e
−ω
0
(ζ +
√
ζ
2
−1)t
_
=
1
48
e
−
t
2
−
49
48
e
−
t
14
.
The Laplace transform of the “forced” response I
Forced
(s) =L
_
i
Forced
(t)
_
is
I
Forced
(s) =
1
m
·
L
_
f (t)
_
s
2
+ 2ζ ω
0
s + ω
2
0
=
1
112
∞
n=1
e
−ns
1
s
2
+
4
7
s +
1
28
.
Take the inverse Laplace transform, one has
i
Forced
(t) =
1
112
∞
n=1
e
2n
7
L
−1
_
e
−n(s +
2
7
) 1
_
s +
2
7
_
2
−
_
3
14
_
2
_
L
−1
_
F(s −a)
_
=e
at
L
_
F(s)
_
=
1
112
∞
n=1
e
2n
7
· e
−
2t
7
L
−1
_
e
−ns
1
s
2
−
_
3
14
_
2
_
L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a)
=
1
112
∞
n=1
e
−
2(t−n)
7
1
_
3
14
_
sinh
3(t −n)
14
H(t −n) L
−1
_
1
s
2
−a
2
_
=
sinh at
a
=
1
24
∞
n=1
e
−
2(t−n)
7
·
1
2
_
e
3(t−n)
14
− e
−
3(t−n)
14
_
H(t −n) sinh at =
e
at
−e
−at
2
=
1
48
∞
n=1
_
e
−
t−n
14
− e
−
t−n
2
_
H(t −n).
Hence, the current i(t) is
i(t) =i
Free
(t)+i
Forced
(t) =
1
48
e
−
t
2
−
49
48
e
−
t
14
+
1
48
∞
n=1
_
e
−
t−n
14
−e
−
t−n
2
_
H(t−n) (A).
6.6 applications of the laplace transform 281
Example 6.23 — BeamColumn 6.23
Consider the beamcolumn shown in the following ﬁgure. Determine the lateral
deﬂection y(x).
a
EI, L
w
b
y
x
W
P
P
Using the Heaviside step function and the Dirac delta function, the lateral load can
be expresses as
w(x) = w
_
1−H(x−a)
_
+ Wδ(x−b).
Following the formulation in Section 5.4, the differential equation becomes
d
4
y
dx
4
+ α
2
d
2
y
dx
2
= ˆ w
_
1−H(x−a)
_
+
ˆ
Wδ(x−b), α
2
=
P
EI
, ˆ w =
w
EI
,
ˆ
W =
W
EI
.
Since the left end is a hinge support and the right end is a sliding support, the
boundary conditions are
at x =0: deﬂection = 0 =⇒ y(0) =0,
bending moment = 0 =⇒ y
(0) =0,
at x =L: slope = 0 =⇒ y
(L) =0,
shear force = 0 =⇒ V(L) = −EI y
(L)−Py
(L) =0
=⇒ y
(L) =0.
Applying the Laplace transform Y(s) =L
_
y(x)
_
, one has
_
s
4
Y(s) − s
3
y(0) − s
2
y
(0) − s y
(0) − y
(0)
_
+ α
2
_
s
2
Y(s) − s y(0) − y
(0)
_
=
ˆ w
s
(1−e
−as
) +
ˆ
We
−bs
.
Since y(0) = y
(0) =0, solving for Y(s) leads to
Y(s) =
y
(0)
s
2
+ α
2
+
_
y
(0)+α
2
y
(0)
_
+
ˆ
We
−bs
s
2
(s
2
+α
2
)
+
ˆ w
s
3
(s
2
+α
2
)
(1−e
−as
).
Applying partial fractions
1
s
3
(s
2
+α
2
)
=
A
s
3
+
B
s
2
+
C
s
+
Ds +E
s
2
+α
2
.
282 6 the laplace transformand its applications
Summing up the righthand side and comparing the coefﬁcients of the numerators,
one obtains
1: Aα
2
= 1 =⇒ A =
1
α
2
,
s : Bα
2
= 0 =⇒ B = 0,
s
2
: A + Cα
2
= 0 =⇒ C = −
A
α
2
= −
1
α
4
,
s
3
: B + E = 0 =⇒ E = −B = 0,
s
4
: C + D = 0 =⇒ D = −C = −
1
α
4
.
Hence,
Y(s) =
y
(0)
α
α
s
2
+ α
2
+
_
1
s
2
−
1
s
2
+α
2
_
_
y
(0)+α
2
y
(0)
_
+
ˆ
We
−bs
α
2
+ ˆ w
_
1
α
2
s
3
−
1
α
4
s
+
1
α
4
s
s
2
+α
2
_
(1−e
−as
).
Taking inverse Laplace transform results in
y(x) = L
−1
_
Y(s)
_
= y
(0) · x +
_
x−
1
α
sin αx
_
y
(0)
α
2
+
_
(x−b)−
1
α
sin α(x−b)
_
ˆ
W
α
2
H(x−b) + ˆ w
_
x
2
2α
2
−
1
α
4
+
1
α
4
cos αx
_
− ˆ w
_
(x−a)
2
2α
2
−
1
α
4
+
1
α
4
cos α(x−a)
_
H(x−a),
in which y
(0) and y
(0) are determined from the boundary conditions at x =L.
At the right end of the beamcolumn, x =L, i.e., x >a, x >b, giving H(x−a) =1,
H(x−b) =1. The lateral deﬂection is simpliﬁed as
y(x) = y
(0) · x +
_
x−
1
α
sin αx
_
y
(0)
α
2
+
_
(x−b)−
1
α
sin α(x−b)
_
ˆ
W
α
2
+
ˆ w
_
α
2
(2ax−a
2
)+2 cos αx−2 cos α(x−a)
_
2α
4
.
Differentiating with respect to x yields
y
(x) = y
(0) +
1−cos αx
α
2
y
(0) +
ˆ
W
_
1−cos α(x−b)
_
α
2
+
ˆ w
_
αa−sin αx+sin α(x−a)
_
α
3
,
y
(x) = cos αx · y
(0) +
ˆ
W cos α(x−b) +
ˆ w
_
sin αx−sin α(x−a)
_
α
.
6.6 applications of the laplace transform 283
Denoting
W
1
=
1
α
2
_
ˆ
W
_
1−cos α(L−b)
_
+
ˆ w
α
_
αa−sin αL+sin α(L−a)
_
_
,
W
2
=
ˆ
W cos α(L−b) +
ˆ w
α
_
sin αL−sin α(L−a)
_
,
and applying the boundary conditions at x =L lead to
y
(L) = y
(0) +
1−cos αx
α
2
y
(0) + W
1
= 0,
y
(L) = cos αL· y
(0) + W
2
= 0,
which results in
y
(0) = −W
1
+
W
2
(1−cos αL)
α
2
cos αL
, y
(0) = −
W
2
cos αL
.
6.6.3 Beams on Elastic Foundation
Structures that can be modeled as beams placed on elastic foundation are found
in many engineering applications, for example, railroad tracks, foundation beams
and retaining walls of buildings, and underground infrastructures (Figure 6.7).
Networks of beams such as those used in ﬂoor systems for ships, buildings, bridges,
shells of revolution such as those used in pressure vessels, boilers, containers, and
largespan reinforced concrete halls and domes can also be analyzed using the
theory of beams on elastic foundation.
Figure 6.7 Examples of beams on elastic foundation.
Winkler’s model of an elastic foundation assumes that the deﬂection y at any
point on the surface of the foundation is proportional to the stress σ at that point,
i.e., σ =k
0
y, where k
0
is called the modulus of the foundation with dimension
[force/length
3
].
In the study of beams on elastic foundation, let p be the intensity per unit length
of the distributed load on the foundation along the length of the beam, i.e., p=σb,
where b is the width of the beam. Hence, as shown in Figure 6.8(a) and from
Winkler’s assumption, p=ky, where k =k
0
b with dimension [force/length
2
].
284 6 the laplace transformand its applications
p
(a) (b)
k
0
x
y p(x)=ky(x)
Surface
Rigid Base
w(x)
Figure 6.8 Winkler’s model of elastic foundation.
For a beam on an elastic foundation under the action of a distributed load w(x)
as shown in Figure 6.8(b), the ﬂexural deﬂection y(x) is governed by the equation
EI
d
4
y
dx
4
= w(x) − p(x),
where EI is the ﬂexural rigidity of the beam. Substituting p(x) =ky(x) into the
equation leads to a fourthorder linear ordinary differential equation
d
4
y
dx
4
+ 4β
4
y =
w(x)
EI
, 4β
4
=
k
EI
.
The constants in the solution of the differential equation are determined from the
boundary conditions of the beam, which are given by the end supports of the beam.
Some typical boundary conditions are listed in Table 6.1.
Table 6.1 Boundary conditions.
x=l
Pinned End:
Deﬂection = 0 =⇒ y(l) = 0
Moment = 0 =⇒ y
(l) = 0
x=l
Clamped End:
Deﬂection = 0 =⇒ y(l) = 0
Slope = 0 =⇒ y
(l) = 0
x=l
Free End:
Moment = 0 =⇒ y
(l) = 0
Shear Force = 0 =⇒ y
(l) = 0
x=l
Sliding End:
Slope = 0 =⇒ y
(l) = 0
Shear Force = 0 =⇒ y
(l) = 0
6.6 applications of the laplace transform 285
Example 6.24 — Beams on Elastic Foundation 6.24
Determine the deﬂection of a beam free at both ends under a trapezoidally dis
tributed load as shown in the following ﬁgure.
a
k
EI, L
y
x
b
w
2
w
1
Using the Heaviside step function, the distributed load can be expressed as
w(x) =
_
w
1
+
w
2
−w
1
b−a
(x−a)
_
_
H(x−a)−H(x−b)
_
.
The differential equation becomes
d
4
y
dx
4
+ 4β
4
y =
w(x)
EI
=
_
ˆ w
1
+ ¯ w(x−a)
_ _
H(x−a)−H(x−b)
_
,
where
ˆ w
1
=
w
1
EI
, ˆ w
2
=
w
2
EI
, ¯ w =
w
2
−w
1
(b−a)EI
=
ˆ w
2
− ˆ w
1
b−a
,
Since both ends are free, the boundary conditions are
y
(0) = y
(0) = y
(L) = y
(L) = 0.
Applying the Laplace transform gives
L
_
_
ˆ w
1
+ ¯ w(x−a)
_ _
H(x−a)−H(x−b)
_
_
= L
_
_
ˆ w
1
+ ¯ w(x−a)
_
H(x−a)
_
−L
_
_
ˆ w
1
+ ¯ w
_
(x−b)+(b−a)
__
H(x−b)
_
= L
_
_
ˆ w
1
+ ¯ w(x−a)
_
H(x−a)
_
−L
_
_
ˆ w
2
+ ¯ w(x−b)
_
H(x−b)
_
= e
−as
L
_
ˆ w
1
+ ¯ wx
_
−e
−bs
L
_
ˆ w
2
+ ¯ wx
_
L
_
f (x−a) H(x−a)
_
=e
−as
F(s)
= e
−as
_
ˆ w
1
s
+
¯ w
s
2
_
− e
−bs
_
ˆ w
2
s
−
¯ w
s
2
_
, L
_
1
_
=
1
s
, L
_
x
_
=
1
s
2
and the differential equation becomes
_
s
4
Y(s) − s
3
y(0) − s
2
y
(0) − s y
(0) − y
(0)
_
+ 4β
4
Y(s)
=
1
s
_
ˆ w
1
e
−as
− ˆ w
2
e
−bs
_
+
1
s
2
¯ w
_
e
−as
−e
−bs
_
.
286 6 the laplace transformand its applications
Since y
(0) = y
(0) =0, solving for Y(s) leads to
Y(s) = y(0)
s
3
s
4
+ 4β
4
+ y
(0)
s
2
s
4
+ 4β
4
+
1
s
·
1
s
4
+ 4β
4
_
ˆ w
1
e
−as
− ˆ w
2
e
−bs
_
+
1
s
2
·
1
s
4
+ 4β
4
¯ w
_
e
−as
−e
−bs
_
.
Using the notations and formulas in Table 6.2 and using the property of shifting
L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a), one obtains the deﬂection of the beam
y(x) = L
−1
_
Y(s)
_
= y(0) · φ
3
(x) + y
(0) · φ
2
(x)
+ ˆ w
1
1−φ
3
(x−a)
4β
4
H(x−a) − ˆ w
2
1−φ
3
(x−b)
4β
4
H(x−b)
+ ¯ w
_
(x−a)−φ
2
(x−a)
4β
4
H(x−a) −
(x−b)−φ
2
(x−b)
4β
4
H(x−b)
_
.
For x >b>a, H(x−b) =H(x−a) =1, and y(x) is simpliﬁed as
y(x) = y(0) · φ
3
(x) + y
(0) · φ
2
(x) +
1
4β
4
_
ˆ w
2
φ
3
(x−b) − ˆ w
1
φ
3
(x−a)
+ ¯ w
_
φ
2
(x−b) − φ
2
(x−a)
_
_
. Noting ˆ w
1
− ˆ w
2
+ ¯ w(b−a) =0
Differentiating y(x) with respect to x three times yields
y
(x) = y(0) · φ
3
(x) + y
(0) · φ
2
(x)
+
1
4β
4
_
ˆ w
2
φ
3
(x−b) − ˆ w
1
φ
3
(x−a) + ¯ w
_
φ
2
(x−b) − φ
2
(x−a)
_
_
= −4β
4
_
y(0) · φ
1
(x) + y
(0) · φ
0
(x)
_
−
_
ˆ w
2
φ
1
(x−b) − ˆ w
1
φ
1
(x−a) + ¯ w
_
φ
0
(x−b) − φ
0
(x−a)
_
_
,
y
(x) = −4β
4
_
y(0) · φ
1
(x) + y
(0) · φ
0
(x)
_
−
_
ˆ w
2
φ
1
(x−b) − ˆ w
1
φ
1
(x−a) + ¯ w
_
φ
0
(x−b) − φ
0
(x−a)
_
_
= −4β
4
_
y(0) · φ
2
(x) + y
(0) · φ
1
(x)
_
−
_
ˆ w
2
φ
2
(x−b) − ˆ w
1
φ
2
(x−a) + ¯ w
_
φ
1
(x−b) − φ
1
(x−a)
_
_
.
Using the boundary conditions at x =L leads to
y
(L) = −4β
4
_
y(0) · φ
1
(L) + y
(0) · φ
0
(L)
_
−
_
ˆ w
2
φ
1
(L−b) − ˆ w
1
φ
1
(L−a) + ¯ w
_
φ
0
(L−b) − φ
0
(L−a)
_
_
= 0,
6.6 applications of the laplace transform 287
Table 6.2 Useful formulas of inverse Laplace transforms for beams on elastic foundation.
φ
0
(x) = L
−1
_
1
s
4
+ 4β
4
_
=
1
4β
3
(sinβx coshβx − cosβx sinhβx),
φ
1
(x) = L
−1
_
s
s
4
+ 4β
4
_
=
1
2β
2
sinβx sinhβx,
φ
2
(x) = L
−1
_
s
2
s
4
+ 4β
4
_
=
1
2β
(sinβx coshβx + cosβx sinhβx),
φ
3
(x) = L
−1
_
s
3
s
4
+ 4β
4
_
= cosβx coshβx,
L
−1
_
1
s
·
1
s
4
+ 4β
4
_
=
1 − φ
3
(x)
4β
4
,
L
−1
_
1
s
2
·
1
s
4
+ 4β
4
_
=
x − φ
2
(x)
4β
4
,
L
−1
_
1
s
3
·
1
s
4
+ 4β
4
_
=
x
2
− 2φ
1
(x)
8β
4
,
L
−1
_
1
s
4
·
1
s
4
+ 4β
4
_
=
x
3
− 6φ
0
(x)
24β
4
,
φ
0
(x) =
1
2β
2
sinβx sinhβx = φ
1
(x),
φ
1
(x) =
1
2β
(sinβx coshβx + cosβx sinhβx) = φ
2
(x),
φ
2
(x) = cosβx coshβx = φ
3
(x),
φ
3
(x) = −β(sinβx coshβx − cosβx sinhβx) = −4β
4
φ
0
(x),
φ
0
(x) = φ
1
(x) = φ
2
(x),
φ
1
(x) = φ
2
(x) = φ
3
(x),
φ
2
(x) = φ
3
(x) = −4β
4
φ
0
(x),
φ
3
(x) = −4β
4
φ
0
(x) = −4β
4
φ
1
(x),
φ
0
(x) = φ
2
(x) = φ
3
(x),
φ
1
(x) = φ
3
(x) = −4β
4
φ
0
(x),
φ
2
(x) = −4β
4
φ
0
(x) = −4β
4
φ
1
(x),
φ
3
(x) = −4β
4
φ
1
(x) = −4β
4
φ
2
(x).
288 6 the laplace transformand its applications
y
(L) = −4β
4
_
y(0) · φ
2
(L) + y
(0) · φ
1
(L)
_
−
_
ˆ w
2
φ
2
(L−b) − ˆ w
1
φ
2
(L−a) + ¯ w
_
φ
1
(L−b) − φ
1
(L−a)
_
_
= 0,
which results in two algebraic equations for two unknowns y(0) and y
(0)
_
φ
1
(L) · y(0) + φ
0
(L) · y
(0) = α
2
,
φ
2
(L) · y(0) + φ
1
(L) · y
(0) = α
3
,
α
2
= −
1
4β
4
_
ˆ w
2
φ
1
(L−b) − ˆ w
1
φ
1
(L−a) + ¯ w
_
φ
0
(L−b) − φ
0
(L−a)
_
_
,
α
3
= −
1
4β
4
_
ˆ w
2
φ
2
(L−b) − ˆ w
1
φ
2
(L−a) + ¯ w
_
φ
1
(L−b) − φ
1
(L−a)
_
_
,
and gives
y(0) =
α
2
φ
1
(L) − α
3
φ
0
(L)
φ
2
1
(L) − φ
0
(L)φ
2
(L)
, y
(0) =
α
3
φ
1
(L) − α
2
φ
2
(L)
φ
2
1
(L) − φ
0
(L)φ
2
(L)
.
Example 6.25 — Beams on Elastic Foundation 6.25
Determine the deﬂection of a beam free at both ends under a concentrated load as
shown in the following ﬁgure.
a
k
EI, L
W
y
x
Using the Dirac delta function, the concentrated load can be expressed as
w(x) = Wδ(x−a).
The differential equation becomes
d
4
y
dx
4
+ 4β
4
y =
w(x)
EI
=
¯
W δ(x−a),
¯
W =
W
EI
.
Since both ends of the beam are pinned, the boundary conditions are
y(0) = y
(0) = y(L) = y
(L) = 0.
Applying the Laplace transform Y(s) =L
_
y(x)
_
and L
_
δ(x−a)
_
=e
−as
gives
_
s
4
Y(s) − s
3
y(0) − s
2
y
(0) − s y
(0) − y
(0)
_
+ 4β
4
Y(s) =
¯
W e
−as
.
6.7 summary 289
Since y(0) = y
(0) =0, solving for Y(s) leads to
Y(s) = y
(0)
s
2
s
4
+ 4β
4
+ y
(0)
1
s
4
+ 4β
4
+
¯
W
1
s
4
+ 4β
4
e
−as
.
Employing the results and notations as in the previous example, one obtains the
deﬂection of the beam
y(x) = L
−1
_
Y(s)
_
= y
(0) · φ
2
(x) + y
(0) · φ
0
(x) +
¯
W φ
0
(x−a)H(x−a),
in which the unknowns y
(0) and y
(0) are determined from the boundary con
ditions y(L) = y
(L) =0. For x >a, the deﬂection y(x) is simpliﬁed as
y(x) = y
(0) · φ
2
(x) + y
(0) · φ
0
(x) +
¯
W φ
0
(x−a).
Differentiating y(x) with respect to x twice times yields
y
(x) = y
(0) · φ
2
(x) + y
(0) · φ
0
(x) +
¯
W φ
0
(x−a)
= −4β
4
y
(0) · φ
0
(x) + y
(0) · φ
2
(x) +
¯
W φ
2
(x−a).
Applying the boundary conditions at x =L gives
y(L) = y
(0) · φ
2
(L) + y
(0) · φ
0
(L) +
¯
W φ
0
(L−a) = 0,
y
(L) = −4β
4
y
(0) · φ
0
(L) + y
(0) · φ
2
(L) +
¯
W φ
2
(L−a) = 0,
which leads to two algebraic equations for two unknowns y
(0) and y
(0)
φ
2
(L) · y
(0) + φ
0
(L) · y
(0) = α
0
, α
0
= −
¯
W φ
0
(L−a),
−4β
4
φ
0
(L) · y
(0) + φ
2
(L) · y
(0) = α
2
, α
2
= −
¯
W φ
2
(L−a),
and results in
y
(0) =
α
0
φ
2
(L) − α
2
φ
0
(L)
φ
2
2
(L) + 4β
4
φ
2
0
(L)
, y
(0) =
α
2
φ
2
(L) + 4β
4
α
0
φ
0
(L)
φ
2
2
(L) + 4β
4
φ
2
0
(L)
.
6.7 Summary
❧ The Laplace transform F(s) of function f (t) is deﬁned as
F(s) = L
_
f (t)
_
=
_
∞
0
e
−st
f (t)dt, s >0.
Some important properties of Laplace transform and inverse Laplace trans
form are listed in Table 6.3.
290 6 the laplace transformand its applications
Table 6.3 Properties of Laplace transform and inverse Laplace transform.
L
f (t)
= F(s) L
−1
F(s)
= f (t)
1. Linear operator
L
_
f (t)
_
is a linear operator L
−1
_
F(s)
_
is a linear operator
2. Property of shifting
L
_
e
at
f (t)
_
= F(s −a) L
−1
_
F(s −a)
_
= e
at
f (t)
3. Property of differentiation
L
_
t
n
f (t)
_
= (−1)
n
d
n
F(s)
ds
n
L
−1
_
d
n
F(s)
ds
n
_
= (−1)
n
t
n
f (t)
4. Property of integration
L
_
f (t)
t
n
_
=
_
∞
s
· · ·
_
∞
s
F(s)(ds)
n
L
−1
__
∞
s
· · ·
_
∞
s
F(s)(ds)
n
_
=
f (t)
t
n
5. Laplace transformof integrals
L
__
t
0
· · ·
_
t
0
f (u)(du)
n
_
=
F(s)
s
n
L
−1
_
F(s)
s
n
_
=
_
t
0
· · ·
_
t
0
f (u)(du)
n
6. Convolution integral
L
__
f ∗ g
_
(t)
_
= F(s)G(s) L
−1
_
F(s)G(s)
_
=
_
f ∗ g
_
(t)
= L
__
t
0
f (u) g(t −u)du
_
=
_
t
0
f (u) g(t −u)du
= L
__
t
0
g(u) f (t −u)du
_
=
_
t
0
g(u) f (t −u)du
7. Heaviside function
L
_
f (t −a)H(t −a)
_
= e
−as
F(s) L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a)
8. Dirac delta function
L
_
f (t)δ(t −a)
_
= e
−as
f (a) L
−1
_
e
−as
_
= δ(t −a)
9. Laplace transformof derivatives
L
_
f
(t)
_
= s F(s) − f (0)
L
_
f
(t)
_
= s
2
F(s) − s f (0) − f
(0)
· · · · · ·
L
_
f
(n)
(t)
_
= s
n
F(s) − s
n−1
f (0) − s
n−2
f
(0) − · · · − s f
(n−2)
− f
(n−1)
(0)
problems 291
❧ The Heaviside step function deﬁned as
H(t −a) =
_
0, t <a,
1, t >a,
is very useful in describing piecewise smooth functions by combining the
following results
f (t)
_
1 − H(t −a)
_
=
_
f (t), t <a,
0, t >a,
f (t)
_
H(t −a) − H(t −b)
_
=
⎧
⎪
⎪
⎨
⎪
⎪
⎩
0, t <a,
f (t), a<t <b,
0, t >b,
f (t) H(t −a) =
_
0, t <a,
f (t), t >a.
❧ The Dirac delta function δ(t −a) is a mathematical idealization of impulse
functions. It is useful in modeling impulse functions, such as concentrated or
point loads.
❧ Applying the Laplace transform to an nthorder linear differential equation
with constant coefﬁcients
a
n
y
(n)
(t) + a
n−1
y
(n−1)
(t) + · · · + a
1
y
(t) + a
0
y(t) = f (t)
converts it into a linear algebraic equation for Y(s) =L
_
y(t)
_
, which can
easily be solved. The solution of the differential equation can be obtained by
determining the inverse Laplace transform y(t) =L
−1
_
Y(s)
_
.
❧ The method of Laplace transform is preferable and advantageous in solving
linear ordinary differential equations with the righthand side functions f (t)
involving the Heaviside step function and the Dirac delta function.
Problems
Evaluate the Laplace transform of the following functions.
6.1 f (t) = 4t
3
− 2t
2
+ 5
A
NS F(s) =
24−4s +5s
3
s
4
6.2 f (t) = 3 sin 2t − 4 cos 5t
A
NS F(s) =
6
s
2
+4
−
4s
s
2
+25
6.3 f (t) = e
−2t
(4 cos 3t + 5 sin 3t)
A
NS F(s) =
4s +23
s
2
+4s +13
292 6 the laplace transformand its applications
6.4 f (t) = 3 cosh 6t + 8 sinh 3t
A
NS F(s) =
3s
s
2
−36
+
24
s
2
−9
6.5 f (t) = 3t cos 2t + t
2
e
t
A
NS F(s) =
3(s
2
−4)
(s
2
+4)
2
+
2
(s −1)
3
6.6 f (t) = t cosh 2t + t
2
sin 5t + t
3
A
NS F(s) =
s
2
+4
(s
2
−4)
2
+
10(3s
2
−25)
(s
2
+25)
3
+
6
s
4
6.7 f (t) = 7e
−5t
cos 2t + 9 sinh
2
2t
A
NS F(s) =
7(s +5)
(s +5)
2
+4
+
72
s (s
2
−16)
6.8 f (t) =
_
0, t <π,
sin t, t >π.
A
NS F(s) = −
e
−πs
s
2
+1
6.9 f (t) =
_
0, t <1,
4t
2
+3t −8, t >1.
A
NS F(s) = e
−s
_
8
s
3
+
11
s
2
−
1
s
_
6.10 f (t) =
⎧
⎪
⎨
⎪
⎩
0, t <1,
t
2
−1, 1<t <2,
0, t >2.
A
NS F(s) =
2(s +1)e
−s
−(3s
2
+4s +2)e
−2s
s
3
6.11 f (t) =
_
sin t, t <π,
4 sin 3t, t >π.
A
NS F(s) =
1
s
2
+1
−
e
−πs
(11s
2
+3)
(s
2
+1)(s
2
+9)
6.12 f (t) =
⎧
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎩
2t, 0<t <2,
2 + t, 2<t <4,
10 − t, 4<t <10,
0, t >10.
A
NS F(s) =
2−e
−2s
−2e
−4s
+ e
−10s
s
2
6.13 f (t) = t
3
δ(t −2) + 3 cos 5t δ(t −π)
A
NS F(s) = 8e
−2s
− 3e
−πs
6.14 f (t) = sinh 4t δ(t +2) + e
2t
δ(t −1) + t
2
e
−3t
δ(t −2) + cos πt δ(t −3)
A
NS F(s) = e
2−s
+ 4e
−6−2s
− e
−3s
Express the following periodic functions using the Heaviside function or the Dirac
function and evaluate the Laplace transform.
6.15
t
a
T 2T 3T 4T 5T
f (t)
−a
0
A
NS F(s) =
a
s
_
1 + 2
∞
n=1
(−1)
n
e
−nTs
_
Problems 293
6.16
t
a
T 2T 3T 4T 5T
f (t)
0
A
NS F(s) =
a
Ts
2
−
a
s
∞
n=0
e
−(n+1)Ts
6.17
t
a
a sin
T
πt
T 2T 3T 4T 5T
f (t)
0
A
NS F(s) =
aTπ
T
2
s
2
+ π
2
∞
n=0
e
−nTs
6.18
t
f (t)
I
−I −I
2T
T
0
3T
I
−I
4T
5T
I
A
NS F(s) = I ·
∞
n=0
(−1)
n
e
−nTs
Evaluate the inverse Laplace transform of the following functions.
6.19 F(s) =
s
(s +1)
3
A
NS f (t) =
_
t −
1
2
t
2
_
e
−t
6.20 F(s) =
4(2s +1)
s
2
−2s −3
A
NS f (t) = 7e
3t
+ e
−t
6.21 F(s) =
3s +2
s
2
+6s +10
A
NS f (t) = e
−3t
(3 cos t −7 sin t)
6.22 F(s) =
3s
2
+2s −1
s
2
−5s +6
A
NS f (t) = 3δ(t) + 32e
3t
− 15e
2t
6.23 F(s) =
30
(s
2
+1)(s
2
−9)
A
NS f (t) = −3 sin t + sinh 3t
6.24 F(s) =
13s
(s
2
−4)(s
2
+9)
A
NS f (t) = −cos 3t + cosh 2t
6.25 F(s) =
40s
(s +1)(s +2)(s
2
−9)
A
NS f (t) = 10e
−3t
− 16e
−2t
+ 5e
−t
+ e
3t
294 6 the laplace transformand its applications
6.26 F(s) =
2
s
3
(s
2
+1)
A
NS f (t) = 2 cos t + t
2
− 2
6.27 F(s) =
s
(s +1)(s +2)
3
A
NS f (t) = (t
2
+t +1)e
−2t
− e
−t
6.28 F(s) =
8
(s −1)(s +1)
2
(s
2
+1)
A
NS f (t) =e
t
−(2t +3)e
−t
+2 cos t −2 sin t
6.29 F(s) =
162
s
3
(s
2
−9)
A
NS f (t) = −9t
2
− 2 + 2 cosh 3t
Evaluate the inverse Laplace transformof the following functions using convolution
integral.
6.30 Y(s) =
1
s(s
2
+a
2
)
2
A
NS y(t) =
1
2a
4
(2−2 cos at −at sin at)
6.31 Y(s) =
1
s
2
(s
2
+a
2
)
2
A
NS y(t) =
1
2a
5
(2at +at cos at −3 sin at)
6.32 Y(s) =
4
s(s
2
+4s +4)
A
NS y(t) = 1−(2t +1)e
−2t
6.33 Y(s) =
16
s
3
(s
2
+4s +4)
A
NS y(t) = −(2t +3)e
−2t
+2t
2
−4t +3
6.34 Y(s) =
6
s(s
2
+4s +3)
A
NS y(t) = −3e
−t
+e
−3t
+2
6.35 Y(s) =
5
s(s
2
+4s +5)
A
NS y(t) = −e
−2t
(2 sin t +cos t)+1
Solve the following differential equations.
6.36 y
+ 4 y
+ 3 y = 60 cos 3t, y(0) = 1, y
(0) = −1
A
NS y(t) = 5e
−3t
− 2e
−t
−2 cos 3t + 4 sin 3t
6.37 y
+y
−2 y =9e
−2t
, y(0) =3, y
(0) = −6
A
NS y(t) =e
t
−(3t −2)e
−2t
6.38 y
− y
− 2 y = 2t
2
+ 1, y(0) = 6, y
(0) = 2
A
NS y(t) = 5e
−t
+ 3e
2t
− t
2
+ t − 2
6.39 y
+ 4 y = 8 sin 2t, y(0) = 1, y
(0) = 4
A
NS y(t) = (−2t +1) cos 2t + 3 sin 2t
6.40 y
− 2 y
+ y = 4e
−t
+ 2e
t
, y(0) = −1, y
(0) = 2
A
NS y(t) = e
−t
+ (t
2
+5t −2)e
t
6.41 y
− 2 y
+ 2 y = 8e
−t
sin t, y(0) = 1, y
(0) = −1
A
NS y(t) = −e
t
sin t + e
−t
(cos t +sin t)
Problems 295
6.42 y
− 2 y
+ 5 y = 8e
t
sin 2t, y(0) = 1, y
(0) = −1
A
NS y(t) = −(2t −1)e
t
cos 2t
6.43 y
+ y
− 2 y = 54t e
−2t
, y(0) = 6, y
(0) = 0
A
NS y(t) = −(9t
2
+6t)e
−2t
+ 6e
t
6.44 y
− y
− 2 y = 9e
2t
H(t −1), y(0) = 6, y
(0) = 0
A
NS y(t) = 4e
−t
+ 2e
2t
+
_
(3t −4)e
2t
+ e
3−t
_
H(t −1)
6.45 y
+ 2 y
+ y = 2 sin t H(t −π), y(0) = 1, y
(0) = 0
A
NS y(t) = (t +1)e
−t
−
_
cos t + (t +1−π)e
π−t
_
H(t −π)
6.46 y
+ 4 y = 8 sin 2t H(t −π), y(0) = 0, y
(0) = 2
A
NS y(t) = sin 2t +
_
2(π −t) cos 2t + sin 2t
_
H(t −π)
6.47 y
+ 4 y = 8(t
2
+t −1)H(t −2), y(0) = 1, y
(0) = 2
A
NS y(t) = sin 2t +cos 2t +
_
2t
2
+2t −3−9 cos(2t −4)−5 sin(2t −4)
_
H(t −2)
6.48 y
− 3 y
+ 2 y = e
t
H(t −2), y(0) = 1, y
(0) = 2
A
NS y(t) = e
2t
+
_
(1−t)e
t
+ e
2t−2
_
H(t −2)
6.49 y
− 5 y
+ 6 y = δ(t −2), y(0) = −1, y
(0) = 1
A
NS y(t) = −4e
2t
+ 3e
3t
+
_
e
3(t−2)
− e
2(t−2)
_
H(t −2)
6.50 y
+ 4 y = 4H(t −π) + 2δ(t −π), y(0) = −1, y
(0) = 2
A
NS y(t) = sin 2t − cos 2t + (1 + sin 2t − cos 2t)H(t −π)
6.51 y
− y
+ 4 y
− 4 y = 10e
−t
, y(0) = 5, y
(0) = −2, y
(0) = 0
A
NS y(t) = −e
−t
+ 5e
t
+ cos 2t − 4 sin 2t
6.52 y
−5 y
+4 y =120e
3t
H(t −1), y(0) =15, y
(0) = −6, y
(0) =0,
y
(0) =0
A
NS y(t) = 6e
t
+14e
−t
−2e
2t
−3e
−2t
+ (10e
t+2
−5e
−t+4
−10e
2t+1
+2e
−2t+5
+3e
3t
)H(t −1)
6.53 y
+ 3 y
− 4 y = 40t
2
H(t −2), y(0) = y
(0) = y
(0) = y
(0) = 0
A
NS y(t) =
_
−10t
2
−15+40e
t−2
+8e
2−t
+7 cos(2t −4)+4 sin(2t −4)
_
H(t−2)
6.54 y
+ 4 y = (2t
2
+ t + 1)δ(t −1), y(0) =1, y
(0) =−2, y
(0) =0,
y
(0) =0
A
NS y(t) = e
−t
cos t − sin t cosh t +
_
sin(t −1) cosh(t −1)
− cos(t −1) sinh(t −1)
_
H(t −1)
6.55 Determine the lateral deﬂection y(x) of the beamcolumn as shown.
296 6 the laplace transformand its applications
a
EI, L
y
x P
P
W
A
NS W
0
=
W
_
α(L−a)−sin α(L−a)
_
α
3
EI
, W
1
=
W
_
1−cos α(L−a)
_
α
2
EI
,
y(0) =
W
1
(1−cos αL)
α sin αL
−W
0
, y
(0) = −
W
1
α
sin αL
,
y(x) = y(0) +
1−cos αx
α
2
y
(0) +
W
_
α(x−a)−sin α(x−a)
_
H(x−a)
α
3
EI
.
For the single degreeoffreedomsystemshown in Figure 6.6, determine the forced
vibration response x
Forced
(t) due to the externally applied load f (t) shown. The
system is assumed to be underdamped, i.e., 0<ζ <1.
6.56
t
f (t)
0 T
f
0
A
NS x
Forced
(t) =
f
0
m
_
η
1
(t) − η
1
(t −T)H(t −T)
_
6.57
t
f (t)
2T 3T 0 T
f
0
A
NS x
Forced
(t) =
f
0
mT
_
η
2
(t) − η
2
(t −T)H(t −T)
−η
2
(t −2T)H(t −2T) + η
2
(t −3T)H(t −3T)
_
6.58
t
T 2T 3T 4T 5T
f (t)
f
0
0
A
NS x
Forced
(t) =
f
0
m
∞
n=0
_
η
1
(t −2nT)H(t −2nT)
−η
1
_
t −(2n+1)T
_
H
_
t −(2n+1)T
_
_
Problems 297
6.59
t
T 2T 3T 4T 5T
f (t)
0
f
0
A
NS x
Forced
(t) =
f
0
mT
_
η
2
(t)−2
∞
n=1
_
η
2
_
t −(2n−1)T
_
H
_
t −(2n−1)T
_
−η
2
(t −2nT)H(t −2nT)
_
_
6.60 For the circuit shown in Figure 6.9(a), the current source I(t) is
I(t) = I
0
H(−t) + I
1
(t)H(t).
Show that the differential equation governing i(t) is
di
dt
+
L+R
1
R
2
C
R
1
LC
i +
R
1
+R
2
R
1
LC
di
dt
=
I
1
(t)
LC
, i(0
+
) =
R
1
I
0
R
1
+R
2
,
i(0
+
)
dt
= 0.
For R
1
=1 , R
2
=8 , C=
1
4
F, L=4 H, I(t) =13 sin 2t
_
H(t)−H(t −π)
_
(A),
I
0
=0 A, determine i(t) for t >0.
A
NS i(t) =
_
12
13
+2t
_
e
−3t
−
_
12
13
+2(t −π)
_
e
−3(t −π)
H(t −π)
+
1
13
(5 sin 2t −12 cos 2t)
_
1−H(t −π)
_
(A)
L C
R
1
I(t)
i
L
C
2
C
1
R R
2
V(t)
i
(a) (b)
Figure 6.9 Electric circuits.
6.61 For the circuit shown in Figure 6.9(b), the voltage source V(t) is
V(t) = V
0
H(−t) + V
1
(t)H(t).
Show that the differential equation governing i(t) is
d
3
i
dt
3
+
C
1
+C
2
RC
1
C
2
d
2
i
dt
2
+
1
LC
2
di
dt
+
1
RLC
1
C
2
i =
1
L
d
2
V
1
(t)
dt
2
+
1
RLC
1
dV
1
(t)
dt
,
298 6 the laplace transformand its applications
with the initial condtions
i(0
+
) = 0,
di(0
+
)
dt
=
V(0
+
)−V
0
L
,
d
2
i(0
+
)
dt
2
=
1
L
_
dV(0
+
)
dt
−
V(0
+
)−V
0
RC
2
_
.
For R=8 , C
1
=
1
4
F, C
2
=
1
20
F, L=5 H, V(t) =10H(−t)+10e
−2t
H(t) (V),
determine i(t) for t >0.
A
NS i(t) = −3e
−2t
+ (2+cos t −3 sin t)e
−t
(A)
6.62 Determine the deﬂection of a beam pinned at both ends under a uniformly
distributed load as shown.
a
k
EI, L
w
b
y
x
A
NS y(x) = y
(0) · φ
2
(x) + y
(0) · φ
0
(x)
+
w
4EIβ
4
_
_
1−φ
3
(x−a)
_
H(x−a) −
_
1−φ
3
(x−b)
_
H(x−b)
_
,
y
(0) =
α
0
φ
2
(L)−α
2
φ
0
(L)
φ
2
2
(L)+4β
4
φ
2
0
(L)
, y
(0) =
α
2
φ
2
(L)+4β
4
α
0
φ
0
(L)
φ
2
2
(L)+4β
4
φ
2
0
(L)
,
α
0
=
w
4EIβ
4
_
φ
3
(L−a)−φ
3
(L−b)
_
, α
2
= −
w
EI
_
φ
1
(L−a)−φ
1
(L−b)
_
.
6.63 Determine the deﬂection of a beam clamped at both ends under a concen
trated load as shown.
a
k
EI, L
y
x
W
A
NS y(x) = y
(0) · φ
1
(x) + y
(0) · φ
0
(x) +
ˆ
W φ
0
(x−a)H(x−a),
ˆ
W =
W
EI
y
(0) =
α
0
φ
1
(L) − α
1
φ
0
(L)
φ
2
1
(L) − φ
0
(L)φ
2
(L)
, y
(0) =
α
1
φ
1
(L) − α
0
φ
2
(L)
φ
2
1
(L) − φ
0
(L)φ
2
(L)
,
α
0
= −
ˆ
Wφ
0
(L−a), α
1
= −
ˆ
Wφ
1
(L−a).
Problems 299
6.64 Determine the deﬂection of a freeclamped beam under a triangularly dis
tributed load as shown.
a k
EI, L
y
x
b
w
A
NS y(x) = y(0) · φ
3
(x) + y
(0) · φ
2
(x)+ ˆ w
_
_
(x−a)−φ
2
(x−a)
_
H(x−a)
−
_
(x−a)−φ
2
(x−b)−(b−a)φ
3
(x−b)
_
H(x−b)
_
,
ˆ w =
w
4(b−a)EIβ
4
,
y(0) =
α
0
φ
3
(L) − α
1
φ
2
(L)
φ
2
3
(L) + 4β
4
φ
0
(L)φ
2
(L)
, y
(0) =
α
1
φ
3
(L) + 4β
4
α
0
φ
0
(L)
φ
2
3
(L) + 4β
4
φ
0
(L)φ
2
(L)
,
α
0
= − ˆ w
_
φ
2
(L−b) − φ
2
(L−a) + (b−a)φ
3
(L−b)
_
,
α
1
= − ˆ w
_
φ
3
(L−b) − φ
3
(L−a) − 4β
4
(b−a)φ
0
(L−b)
_
.
6.65 Determine the deﬂection of a slidingclamped beam under a triangularly
distributed load as shown.
a
k
EI, L
y
x
w
A
NS y(x) = y(0) · φ
3
(x) + y
(0) · φ
1
(x) − ˆ w
_
x − a − φ
2
(x) + aφ
3
(x)
−
_
(x−a) − φ
2
(x−a)
_
H(x−a)
_
, ˆ w =
w
4aEIβ
4
,
y(0) =
α
0
φ
2
(L)−α
1
φ
1
(L)
φ
1
(L)φ
3
(L)+4β
4
φ
0
(L)φ
1
(L)
, y
(0) =
α
1
φ
3
(L)+4β
4
α
0
φ
0
(L)
φ
1
(L)φ
3
(L)+4β
4
φ
0
(L)φ
1
(L)
,
α
0
= ˆ w
_
−φ
2
(L) + aφ
3
(L) + φ
2
(L−a)
_
,
α
1
= ˆ w
_
−φ
3
(L) − 4aβ
4
φ
0
(L) + φ
3
(L−a)
_
.
7
C H A P T E R
Systems of
Linear Differential Equations
7.1 Introduction
When a system is described by one independent variable and more than one de
pendent variable, the governing equations may be a system of ordinary differential
equations. In the following, two motivating examples will be studied to illustrate
how systems of differential equations arise in practice.
Example 7.1 — Particle Moving in a Plane 7.1
Derive the governing equations of motion of a particle with mass m moving in a
plane in both rectangular and polar coordinate systems.
(a) (b)
y
y
x
x
F
y
(x, y, t)
F
x
(x, y, t)
y
y
x
x
r
O O
A
A
ˆr
ˆ
F
θ
(r,θ, t)
θ
F
r
(r,θ, t)
θ
ˆ
j
ˆ
i
Figure 7.1 A particle moving in a plane.
1. Rectangular Coordinates: The particle moves along a trajectory in the xy
plane as shown in Figure 7.1(a); at time t, its coordinates are (x, y). It is subjected
300
7.1 introduction 301
to externally applied forces F
x
(x, y, t) in the xdirection and F
y
(x, y, t) in the
ydirection.
The equations of motion are given by Newton’s Second Law
in the xdirection: m¨ x = F
x
(x, y, t),
in the ydirection: m¨ y = F
y
(x, y, t),
which is a system of two secondorder ordinary differential equations.
2. Polar Coordinates: At time t, the particle is at point A with polar coordinates
(r, θ) as shown in Figure 7.1(b). Since x =r cos θ, y =r sin θ, the position vector
is given by
r =x
ˆ
i + y
ˆ
j = r (cos θ
ˆ
i + sin θ
ˆ
j) = r ˆ r,
where ˆ r = cos θ
ˆ
i +sin θ
ˆ
j is the unit vector in the direction OA or r. The unit
vector normal to ˆ r, denoted as
ˆ
θ, is
ˆ
θ = −sin θ
ˆ
i +cos θ
ˆ
j.
The velocity and the acceleration of the particle are
v = ˙ r = ˙ x
ˆ
i + ˙ y
ˆ
j, a = ¨ r = ¨ x
ˆ
i + ¨ y
ˆ
j.
Since, differentiating x and y with respect to time t yields
˙ x = ˙ r cos θ − r sin θ ·
˙
θ, ˙ y = ˙ r sin θ + r cos θ ·
˙
θ,
¨ x = (¨ r cos θ − ˙ r sin θ ·
˙
θ) − (˙ r sin θ ·
˙
θ − r cos θ ·
˙
θ
2
− r sin θ ·
¨
θ)
= ¨ r cos θ − 2˙ r
˙
θ sin θ − r
˙
θ
2
cos θ − r
¨
θ sin θ,
¨ y = (¨ r sin θ + ˙ r cos θ ·
˙
θ) + (˙ r cos θ ·
˙
θ − r sin θ ·
˙
θ
2
+ r cos θ ·
¨
θ)
= ¨ r sin θ + 2˙ r
˙
θ cos θ − r
˙
θ
2
sin θ + r
¨
θ cos θ,
the acceleration vector becomes
a =
_
¨ r cos θ − 2˙ r
˙
θ sin θ − r
˙
θ
2
cos θ − r
¨
θ sin θ
_
ˆ
i
+
_
¨ r sin θ + 2˙ r
˙
θ cos θ − r
˙
θ
2
sin θ + r
¨
θ cos θ
_
ˆ
j
= (¨ r − r
˙
θ
2
) (cos θ
ˆ
i + sin θ
ˆ
j) + (2˙ r
˙
θ + r
¨
θ) (−sin θ
ˆ
i + cos θ
ˆ
j)
= (¨ r − r
˙
θ
2
) ˆ r + (2˙ r
˙
θ + r
¨
θ)
ˆ
θ = a
r
ˆ r + a
θ
ˆ
θ,
where
a
r
= ¨ r − r
˙
θ
2
, a
θ
= 2˙ r
˙
θ + r
¨
θ
are the radial and angular accelerations of the particle in the ˆ r and
ˆ
θ directions,
respectively. Hence, the equations of motion are, using Newton’s Second Law,
in the ˆ r direction: ma
r
=
F
r
=⇒ m(¨ r − r
˙
θ
2
) = F
r
(r, θ, t),
in the
ˆ
θ direction: ma
θ
=
F
θ
=⇒ m(2˙ r
˙
θ + r
¨
θ) = F
θ
(r, θ, t),
which is a system of two secondorder ordinary differential equations.
302 7 systems of linear differential equations
Example 7.2 — Vibration of Multiple Story Shear Building 7.2
Derive the equations of motion of an nstory shear building as shown in Figure
7.2(a).
(a) (b)
x
n−1
x
n
x
2
x
1
F
n−1
(t)
F
n
(t)
F
r
(t)
F
2
(t)
F
1
(t)
m
n−1
m
n
m
r m
2
m
1
c
n
c
2
c
1
k
n
x
r+1
x
r
k
r+1
(x
r+1
−x
r
)
k
r
(x
r
−x
r−1
)
x
r−1
c
r
(x
r
−x
r−1
)
c
r
k
r
k
2
k
1
c
r+1
(x
r+1
−x
r
)
c
r+1
k
r+1
Figure 7.2 An nstory shear building.
The rth ﬂoor, r =1, 2, . . . , n, is assumed to be rigid with mass m
r
and is subjected
to externally applied load F
r
(t). The combined stiffness of the columns connecting
the (r −1)th and the rth ﬂoors is k
r
, and the damping coefﬁcient of the dashpot
damper, due to internal friction, between the (r −1)th and the rth ﬂoors is c
r
. The
displacement of the rth ﬂoor is described by x
r
(t).
Consider the motion of the rth ﬂoor, whose freebody diagram is shown in
Figure 7.2(b). Note that the columns between the (r −1)th and the rth ﬂoors
behave as a spring with stiffness k
r
.
Remarks: To determine the shear force applied on the rth floor by the columns
between the (r −1)th and rth floors, stand on the (r −1)th floor to observe the
motion of the rth floor. The rth floor is seen to move toward the right with a
relative displacement of x
r
−x
r−1
; hence the columns will try to pull the rth floor
back to the left, exerting a shear (spring) force of k
r
(x
r
−x
r−1
) toward the left.
Similarly, the damping force is c
r
(˙ x
r
−˙ x
r−1
).
On the other hand, to determine the shear force applied on the rth floor by
the columns between the rth and (r +1)th floors, stand on the (r +1)th floor to
observe the motion of the rth floor. The rth floor is seen to move toward the left
with a relative displacement of x
r+1
−x
r
; thus the columns will try to pull the rth
floor back to the right, resulting in a shear force of k
r+1
(x
r+1
−x
r
) toward the
right. The damping force is c
r+1
(˙ x
r+1
−˙ x
r
).
7.1 introduction 303
Applying Newton’s Second Law, the equation of motion of the rth ﬂoor, r =1,
2, . . . , n, is
m
r
¨ x
r
= F
r
(t) + k
r+1
(x
r+1
−x
r
) + c
r+1
(˙ x
r+1
−˙ x
r
) − k
r
(x
r
−x
r−1
) − c
r
(˙ x
r
−˙ x
r−1
),
or
m
r
¨ x
r
−c
r
˙ x
r−1
+(c
r
+c
r+1
)˙ x
r
−c
r+1
˙ x
r+1
−k
r
x
r−1
+(k
r
+k
r+1
)x
r
−k
r+1
x
r+1
= F
r
(t),
where x
0
=0, c
n+1
=0, and k
n+1
=0.
In the matrix form, the equations of motion of the nstory shear building can be
written as
M¨ x + C ˙ x + Kx = F(t),
where
x =
_
x
1
, x
2
, . . . , x
n
_
T
, F(t) =
_
F
1
(t), F
2
(t), . . . , F
n
(t)
_
T
are the displacement and load vectors, respectively. M, C, and K are the mass,
damping, and stiffness matrices, respectively, given by
M = diag
_
m
1
, m
2
, . . . , m
n
_
,
C =
⎡
⎢
⎢
⎢
⎢
⎢
⎣
c
1
+c
2
−c
2
−c
2
c
2
+c
3
−c
3
.
.
.
.
.
.
.
.
.
−c
n−1
c
n−1
+c
n
−c
n
−c
n
c
n
⎤
⎥
⎥
⎥
⎥
⎥
⎦
,
K =
⎡
⎢
⎢
⎢
⎢
⎢
⎣
k
1
+k
2
−k
2
−k
2
k
2
+k
3
−k
3
.
.
.
.
.
.
.
.
.
−k
n−1
k
n−1
+k
n
−k
n
−k
n
k
n
⎤
⎥
⎥
⎥
⎥
⎥
⎦
.
Hence, the motion of an nstory shear building is governed by a system of n
secondorder linear ordinary differential equations. The systemis equivalent to the
massspringdamper system shown in Figure 7.3.
x
1
(t) x
2
(t) x
n−1
(t) x
n
(t)
F
1
(t) F
2
(t) F
n−1
(t) F
n
(t)
c
1
m
1
k
1
c
2
m
2
k
2
c
n
m
n−1
m
n
k
n
Figure 7.3 An equivalent massspringdamper system.
304 7 systems of linear differential equations
7.2 The Method of Operator
Inthis section, the methodof Doperator is employedtoobtainthe complementary
and particular solutions of systems of linear ordinary differential equations.
7.2.1 Complementary Solutions
Consider a system of two linear ordinary differential equations
φ
11
(D) x
1
+ φ
12
(D) x
2
= 0, D(·) ≡ d(·)/dt, (1a)
φ
21
(D) x
1
+ φ
22
(D) x
2
= 0, (1b)
where φ
ij
(D), i, j =1, 2, are polynomials of D with constant coefﬁcients.
To ﬁnd x
1
, eliminate x
2
as follows:
operate φ
22
(D) on (1a): φ
22
(D) φ
11
(D) x
1
+ φ
22
(D) φ
12
(D) x
2
= 0, (2a)
operate φ
12
(D) on (1b): φ
12
(D) φ
21
(D) x
1
+ φ
12
(D) φ
22
(D) x
2
= 0. (2b)
Subtracting equation (2b) from (2a) yields
_
φ
22
(D) φ
11
(D) − φ
12
(D) φ
21
(D)
_
x
1
= 0. (3)
Similarly, to ﬁnd x
2
, eliminate x
1
as follows:
operate φ
21
(D) on (1a): φ
21
(D) φ
11
(D) x
1
+ φ
21
(D) φ
12
(D) x
2
= 0, (4a)
operate φ
11
(D) on (1b): φ
11
(D) φ
21
(D) x
1
+ φ
11
(D) φ
22
(D) x
2
= 0. (4b)
Subtracting equation (4a) from (4b) yields
_
φ
11
(D) φ
22
(D) − φ
21
(D) φ
12
(D)
_
x
2
= 0. (5)
It can be seen that equation (3) for x
1
and equation (5) for x
2
have the same form
φ(D) x
1
= 0, φ(D) x
2
= 0,
where
φ(D) = φ
11
(D) φ
22
(D) − φ
12
(D) φ
21
(D) =
¸
¸
¸
¸
¸
φ
11
(D) φ
12
(D)
φ
21
(D) φ
22
(D)
¸
¸
¸
¸
¸
,
which is the determinant of the coefﬁcient matrix of system (1) of differential
equations. Instead of a matrix of constant coefﬁcients, as for systems of linear
algebraic equations, the matrix here is a matrix of operators.
☞
When evaluating the determinants, operators must precede functions.
7.2 the method of operator 305
In general, for a system of n linear ordinary differential equations
φ
11
(D) x
1
+ φ
12
(D) x
2
+ · · · + φ
1n
(D) x
n
= 0,
φ
21
(D) x
1
+ φ
22
(D) x
2
+ · · · + φ
2n
(D) x
n
= 0,
· · · · · ·
φ
n1
(D) x
1
+ φ
n2
(D) x
2
+ · · · + φ
nn
(D) x
n
= 0,
(6)
the differential equations for x
1
, x
2
, . . . , x
n
have the same form
φ(D) x
1
= 0, φ(D) x
2
= 0, . . . , φ(D) x
n
= 0,
where φ(D) is the determinant of the coefﬁcient matrix
φ(D) =
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
φ
11
(D) φ
12
(D) · · · φ
1n
(D)
φ
21
(D) φ
22
(D) · · · φ
2n
(D)
.
.
.
.
.
.
· · ·
.
.
.
φ
n1
(D) φ
n2
(D) · · · φ
nn
(D)
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
.
Hence, the unknowns x
1
, x
2
, . . . , x
n
all have the same characteristic equation
φ(λ) =0 and, as a result, the same form of complementary solutions.
The complementary solutions of system (6) contain arbitrary constants, the
number of which is the degree of polynomial of φ(D). It is likely that the comple
mentary solutions x
1C
, x
2C
, . . . , x
nC
, written using the roots of the characteristic
equation φ(λ) =0, will contain more constants. The extra constants can be elim
inated by substituting the solutions into any one of the original equations in
system (6).
Example 7.3 7.3
Solve
dx
dt
− 3x − 6y = 0, 3x +
dy
dt
+ 3y = 0.
Using the Doperator, D(·) ≡d(·)/dt, the differential equations become
(D−3) x − 6y = 0, (1a)
3x + (D+3) y = 0. (1b)
The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
¸
¸
¸
D−3 −6
3 D+3
¸
¸
¸
¸
¸
= (D−3)(D+3) + 18 = D
2
+ 9.
The characteristic equation is φ(λ) =λ
2
+9=0 =⇒ λ= ±i 3. The complementary
solutions of x and y have the same form and are given by
x
C
= A
1
cos 3t + B
1
sin 3t, y
C
= A
2
cos 3t + B
2
sin 3t. (2)
306 7 systems of linear differential equations
which contain four arbitrary constants. However, since φ(D) is a polynomial of
degree 2 in D, the complementary solutions should contain only two arbitrary
constants.
Substitute solutions (2) into equation (1a) to eliminate the two extra constants
(D−3)x
C
− 6y
C
=
_
(−3A
1
sin 3t + 3B
1
cos 3t) − 3(A
1
cos 3t + B
1
sin 3t)
_
− 6(A
2
cos 3t + B
2
sin 3t)
= (−3A
1
+3B
1
−6A
2
) cos 3t + (−3A
1
−3B
1
−6B
2
) sin 3t
= 0. (3)
Since cos 3t and sin 3t are linearly independent, equation (3) implies that the
coefﬁcients of cos 3t and sin 3t are zero:
−3A
1
+ 3B
1
− 6A
2
= 0 =⇒ A
2
= −
1
2
(A
1
−B
1
),
−3A
1
− 3B
1
− 6B
2
= 0 =⇒ B
2
= −
1
2
(A
1
+B
1
).
Hence, the complementary solutions are
x
C
= A
1
cos 3t + B
1
sin 3t, y
C
= −
1
2
(A
1
−B
1
) cos 3t −
1
2
(A
1
+B
1
) sin 3t.
Example 7.4 7.4
Solve (D
2
+3D+2)x + (D+1) y = 0, D(·) ≡d(·)/dt, (1a)
(D+2)x + (D−1) y = 0. (1b)
The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
¸
¸
¸
D
2
+3D+2 D+1
D+2 D−1
¸
¸
¸
¸
¸
=
¸
¸
¸
¸
¸
¸
(D+1) (D+2) D+1
D+2 D−1
¸
¸
¸
¸
¸
¸
= (D+1)(D+2)
¸
¸
¸
¸
¸
1 1
1 D−1
¸
¸
¸
¸
¸
Take (D+1) and (D+2) out.
= (D+1)(D+2)(D−2).
Hence, the characteristic equation is
φ(λ) = (λ+1)(λ+2)(λ−2) = 0 =⇒ λ = −1, −2, 2.
The complementary solutions of x and y have the same form and are given by
x
C
= C
1
e
−t
+ C
2
e
−2t
+ C
3
e
2t
, y
C
= D
1
e
−t
+ D
2
e
−2t
+ D
3
e
2t
. (2)
Since φ(D) is a polynomial of degree 3 in D, the complementary solutions should
contain only three arbitrary constants. The three extra constants in solutions (2)
7.2 the method of operator 307
can be eliminated by substituting them into either equation (1a) or (1b). Since
equation (1b) is simpler, substituting solutions (2) into equation (1b) yields
(D+2)x
C
+ (D−1) y
C
=
_
(−C
1
e
−t
− 2C
2
e
−2t
+ 2C
3
e
2t
) + 2(C
1
e
−t
+ C
2
e
−2t
+ C
3
e
2t
)
_
+
_
(−D
1
e
−t
− 2D
2
e
−2t
+ 2D
3
e
2t
) − (D
1
e
−t
+ D
2
e
−2t
+ D
3
e
2t
)
_
= (C
1
−2D
1
)e
−t
− 3D
2
e
−2t
+ (4C
3
+D
3
)e
2t
= 0. (3)
Since e
−t
, e
−2t
, and e
2t
are linearly independent, each coefﬁcient must be zero:
C
1
− 2D
1
= 0 =⇒ D
1
= −
1
2
C
1
,
−3D
2
= 0 =⇒ D
2
= 0,
4C
3
+ D
3
= 0 =⇒ D
3
= −4C
3
.
Hence, the complementary solutions are
x
C
= C
1
e
−t
+ C
2
e
−2t
+ C
3
e
2t
, y
C
=
1
2
C
1
e
−t
− 4C
3
e
2t
.
7.2.2 Particular Solutions
Review of Cramer’s Rule
For the following system of n linear algebraic equations
a
11
x
1
+ a
12
x
2
+ · · · + a
1n
x
n
= b
1
,
a
21
x
1
+ a
22
x
2
+ · · · + a
2n
x
n
= b
2
,
· · · · · ·
a
n1
x
1
+ a
n2
x
2
+ · · · + a
nn
x
n
= b
n
,
the solutions are given by
x
i
=
i
, i = 1, 2, . . . , n,
where is the determinant of coefﬁcient matrix,
i
is the determinant of the
coefﬁcient matrix with the ith column replaced by the righthand side vector, i.e.,
=
¸
¸
¸
¸
¸
¸
¸
¸
¸
a
11
a
12
· · · a
1n
a
21
a
22
· · · a
2n
.
.
.
.
.
.
· · ·
.
.
.
a
n1
a
n2
· · · a
nn
¸
¸
¸
¸
¸
¸
¸
¸
¸
,
i
=
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
a
11
· · · a
1, i−1
b
1
a
1, i+1
· · · a
1n
a
21
· · · a
2, i−1
b
2
a
2, i+1
· · · a
2n
.
.
.
· · ·
.
.
.
· · ·
.
.
.
· · ·
.
.
.
a
n1
· · · a
n, i−1
b
n
a
n, i+1
· · · a
nn
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
.
.,,.
ith column
308 7 systems of linear differential equations
For a system of linear ordinary differential equations
φ
11
(D) x
1
+ φ
12
(D) x
2
+ · · · + φ
1n
(D) x
n
= f
1
(t),
φ
21
(D) x
1
+ φ
22
(D) x
2
+ · · · + φ
2n
(D) x
n
= f
2
(t),
· · · · · ·
φ
n1
(D) x
1
+ φ
n2
(D) x
2
+ · · · + φ
nn
(D) x
n
= f
n
(t),
where D(·) ≡d(·)/dt, a particular solution is given by, using Cramer’s Rule,
x
iP
(t) =
i
(t)
φ(D)
, i = 1, 2, . . . , n,
where φ(D) is the determinant of the coefﬁcient matrix as studied in the previous
section for complementary solution,
i
(t) is φ(D) with the ith column being
replaced by the righthand side vector of functions, i.e.,
i
(t) =
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
φ
11
(D) · · · φ
1, i−1
(D) f
1
(t) φ
1, i+1
(D) · · · φ
1n
(D)
φ
21
(D) · · · φ
2, i−1
(D) f
2
(t) φ
2, i+1
(D) · · · φ
2n
(D)
.
.
.
· · ·
.
.
.
.
.
.
.
.
.
· · ·
.
.
.
φ
n1
(D) · · · φ
n, i−1
(D) f
n
(t) φ
n, i+1
(D) · · · φ
nn
(D)
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
.
☞
It should be emphasized that, since the elements of the determinant are
operators andfunctions, operators must precede functions whenevaluating
determinants. Furthermore, since
i
(t) , i =1, 2, . . . , n, are functions, when
determining x
iP
, φ
−1
(D) should precede
i
(t).
Example 7.5 7.5
Solve (D−3)x − 6y = 0, D(·) ≡d(·)/dt, (1a)
3x + (D+3) y = 18te
−3t
. (1b)
This system and the system in Example 7.3 have the righthand sides. Hence, they
have the same complementary solutions given by
x
C
= A
1
cos 3t + B
1
sin 3t, (2a)
y
C
= A
2
cos 3t + B
2
sin 3t. (2b)
A particular solution is given by
x
(t) =
¸
¸
¸
¸
¸
0 −6
18te
−3t
D+3
¸
¸
¸
¸
¸
= 108te
−3t
,
7.2 the method of operator 309
y
(t) =
¸
¸
¸
¸
¸
D−3 0
3 18te
−3t
¸
¸
¸
¸
¸
=18(e
−3t
−3te
−3t
)−3· 18te
−3t
=18e
−3t
−108te
−3t
,
x
P
=
x
(t)
φ(D)
=
108
D
2
+9
(te
−3t
) = 108e
−3t
1
(D−3)
2
+9
t
Theorem 2 in
Section 4.3.2
= 108e
−3t
1
D
2
−6D+18
t = 6e
−3t
1
1 −
_
1
3
D−
1
18
D
2
_
t
= 6e
−3t
_
1 +
_
1
3
D−
1
18
D
2
_
+ · · ·
_
t
Expand the operator in series;
stop at D.
= 2e
−3t
(3t +1), (3a)
y
P
=
y
(t)
φ(D)
=
18
D
2
+9
(e
−3t
) −
108
D
2
+9
(te
−3t
)
=
18
(−3)
2
+9
e
−3t
− 2e
−3t
(3t +1)
Apply Theorem 1 in Section 4.3.2
for the first term. Use result of x
P
for the second term.
= −e
−3t
(6t +1). (3b)
The general solutions are
x = x
C
+ x
P
= A
1
cos 3t + B
1
sin 3t + 2(3t +1)e
−3t
, (4a)
y = y
C
+ y
P
= A
2
cos 3t + B
2
sin 3t − (6t +1)e
−3t
. (4b)
Since φ(D) is a polynomial of degree 2 in D, the general solutions should
contain only two arbitrary constants. The two extra constants can be eliminated by
substituting equations (4) into either (1a) or (1b).
Substitute solutions (4) into equation (1b) to eliminate the two extra constants
3x + (D+3) y
= 3
_
A
1
cos 3t + B
1
sin 3t + 2(3t +1)e
−3t
_
3x
+
_
−3A
2
sin 3t + 3B
2
cos 3t − 6e
−3t
+ 3(6t +1)e
−3t
_
Dy
+ 3
_
A
2
cos 3t + B
2
sin 3t − (6t +1)e
−3t
_
3y
= 3(A
1
+A
2
+B
2
) cos 3t + 3(B
1
−A
2
+B
2
) sin 3t + 18te
−3t
= 18te
−3t
. Righthand side of equation (1b)
Comparing the coefﬁcients of similar terms gives
cos 3t : A
1
+ A
2
+ B
2
= 0,
sin 3t : B
1
− A
2
+ B
2
= 0.
310 7 systems of linear differential equations
Since the purpose is to eliminate two arbitrary constants, one can express any two
constants in terms of the other two from these equations. Hence
A
1
= −(A
2
+ B
2
), B
1
= A
2
− B
2
,
or
A
2
= −
1
2
(A
1
− B
1
), B
2
= −
1
2
(A
1
+ B
1
).
The general solutions become
x = −(A
2
+B
2
) cos 3t + (A
2
−B
2
) sin 3t + 2(3t +1)e
−3t
,
y = A
2
cos 3t + B
2
sin 3t − (6t +1)e
−3t
.
Remarks: In the solution above, the general procedure is followed to illustrate
all the steps in solving systems of linear differential equations using the method of
operators. However, for this example, it can be solved more easily as follows.
Having obtained the complementary and particular solutions for x, i.e., x
C
in
(2a) and x
P
in (3a), or the general solution x in (4a), the general solution y can be
determined from equation (1a)
y =
1
6
(D−3)x =
1
6
_
_
−3A
1
sin 3t + 3B
1
cos 3t + 6e
−t
− 3· 2(3t +1)e
−3t
_
−3
_
A
1
cos 3t + B
1
sin 3t + 2(3t +1)e
−3t
_
_
= −
1
2
(A
1
−B
1
) cos 3t −
1
2
(A
1
+B
1
) sin 3t − (6t +1)e
−3t
.
Remarks: It is important and efficient to exploit the differential equations to
devise an easy way to solve the problem.
Example 7.6 7.6
Solve (D−3)x − 6y = 2 cos 3t, D(·) ≡d(·)/dt, (1a)
3x + (D+3) y = 2 sin 3t. (1b)
This system and the system in Example 7.3 have the righthand sides. Hence, they
have the same complementary solutions given by
x
C
= A
1
cos 3t + B
1
sin 3t, y
C
= A
2
cos 3t + B
2
sin 3t.
A particular solution is given by
x
(t) =
¸
¸
¸
¸
¸
2 cos 3t −6
2 sin 3t D+3
¸
¸
¸
¸
¸
= 2(D+3) cos 3t + 12 sin 3t = 6 cos 3t + 6 sin 3t,
y
(t) =
¸
¸
¸
¸
¸
D−3 2 cos 3t
3 2 sin 3t
¸
¸
¸
¸
¸
= 2(D−3) sin 3t − 6 cos 3t = −6 sin 3t,
7.2 the method of operator 311
x
P
=
x
(t)
φ(D)
=
1
D
2
+9
(6 cos 3t + 6 sin 3t),
y
P
=
y
(t)
φ(D)
=
1
D
2
+9
(−6 sin 3t).
Theorem 3 fails in Section 4.3.2 fails when evaluating x
P
and y
P
. Hence, Theorem
4 in Section 4.3.2 must be applied:
∵ φ(D) = D
2
+ 9, φ(i 3) = 0,
φ
(D) = 2D, φ
(i 3) = i 6,
1
D
2
+9
(e
i3t
) =
1
φ
(i 3)
te
i3t
=
1
i 6
t (cos 3t + i sin 3t) =
t
6
sin 3t − i
t
6
cos 3t,
∴ x
P
= 6Re
_
1
D
2
+9
e
i3t
_
+ 6Im
_
1
D
2
+9
e
i3t
_
= 6·
t
6
sin 3t + 6
_
−
t
6
cos 3t
_
= t (sin 3t − cos 3t),
y
P
= −6Im
_
1
D
2
+9
e
i3t
_
= −6
_
−
t
6
cos 3t
_
= t cos 3t.
The general solutions are
x = x
C
+ x
P
= A
1
cos 3t + B
1
sin 3t + t (sin 3t − cos 3t),
y = y
C
+ y
P
= A
2
cos 3t + B
2
sin 3t + t cos 3t.
Substitute the general solutions into equation (1a) to eliminate the extra constants
(D−3)x − 6y
=
_
−3A
1
sin 3t + 3B
1
cos 3t + (sin 3t −cos 3t) + t (3 cos 3t + 3 sin 3t)
_
− 3
_
A
1
cos 3t + B
1
sin 3t + t (sin 3t −cos 3t)
_
− 6
_
A
2
cos 3t + B
2
sin 3t + t cos 3t
_
= (3B
1
−3A
1
−6A
2
−1) cos 3t + (−3A
1
−3B
1
−6B
2
+1) sin 3t
= 2 cos 3t. Righthand side of equation (1a)
Comparing the coefﬁcients of similar terms gives
cos 3t : 3B
1
−3A
1
−6A
2
−1 = 2 =⇒ A
2
= −
1
2
(A
1
−B
1
) −
1
2
,
sin 3t : −3A
1
−3B
1
−6B
2
+1 = 0 =⇒ B
2
= −
1
2
(A
1
+B
1
) +
1
6
.
The general solutions become
x = A
1
cos 3t + B
1
sin 3t + t (sin 3t − cos 3t),
312 7 systems of linear differential equations
y =
_
−
1
2
(A
1
−B
1
) −
1
2
_
cos 3t +
_
−
1
2
(A
1
+B
1
) +
1
6
_
sin 3t + t cos 3t.
Remarks: Substitute the general solutions into one of the original differential
equations to eliminate the extra constants in the complementary solutions. This
also serves as a check for the particular solutions obtained.
Example 7.7 7.7
Solve (D
2
−3D)x − (D−2) y = 14t + 7, D(·) ≡d(·)/dt, (1a)
(D−3)x + Dy = 1. (1b)
The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
¸
¸
¸
D
2
−3D −(D−2)
D−3 D
¸
¸
¸
¸
¸
= D
2
(D−3) + (D−2)(D−3)
= (D−3)(D
2
+D−2) = (D−3)(D+2)(D−1) = D
3
−2D
2
−5D+6.
The characteristic equation is φ(λ) =(λ+2)(λ−1)(λ−3) =0 =⇒ λ= −2, 1, 3.
The complementary solutions are
x
C
= C
1
e
−2t
+ C
2
e
t
+ C
3
e
3t
, y
C
= D
1
e
−2t
+ D
2
e
t
+ D
3
e
3t
.
A particular solution is given by
x
(t) =
¸
¸
¸
¸
¸
14t +7 −(D−2)
1 D
¸
¸
¸
¸
¸
= D(14t +7) + (D−2)(1) = 12,
y
(t) =
¸
¸
¸
¸
¸
D
2
−3D 14t +7
D−3 1
¸
¸
¸
¸
¸
= (D
2
−3D)(1) − (D−3)(14t +7) = 42t +7,
x
P
=
x
(t)
φ(D)
=
1
6 −5D−2D
2
+D
3
( 12 ) =
12
6
= 2,
Special case of polynomial: constant
y
P
=
y
(t)
φ(D)
=
1
6−5D−2D
2
+D
3
(42t +7)
=
1
6
_
1 −
_
5
6
D+
1
3
D
2
−
1
6
D
3
__
(42t +7)
=
1
6
_
1+
_
5
6
D+ · · ·
_
+ · · ·
_
(42t +7)
Expand the operator
in series; stop at D.
=
1
6
_
(42t +7) +
5
6
(42)
_
= 7t + 7.
Hence, the general solutions are
x = x
C
+ x
P
= C
1
e
−2t
+ C
2
e
t
+ C
3
e
3t
+ 2,
7.2 the method of operator 313
y = y
C
+ y
P
= D
1
e
−2t
+ D
2
e
t
+ D
3
e
3t
+ 7t + 7.
Since φ(D) is a polynomial of degree 3 in D, the general solutions should
contain three arbitrary constants. Substitute the solutions into equation (1b) to
eliminate the three extra constants
(D−3)x + Dy = (−2C
1
e
−2t
+ C
2
e
t
+ 3C
3
e
3t
)
− 3(C
1
e
−2t
+ C
2
e
t
+ C
3
e
3t
+ 2) + (−2D
1
e
−2t
+ D
2
e
t
+ 3D
3
e
3t
+ 7)
= (−5C
1
−2D
1
)e
−2t
+ (−2C
2
+D
2
)e
t
+ 3D
3
e
3t
+ 1
= 1. Righthand side of equation (1b)
Comparing the coefﬁcients of similar terms yields
e
−2t
: −5C
1
− 2D
1
= 0 =⇒ D
1
= −
5
2
C
1
,
e
t
: −2C
2
+ D
2
= 0 =⇒ D
2
= 2C
2
,
e
3t
: 3D
3
= 0 =⇒ D
3
= 0.
Hence, the general solutions become
x = x
C
+ x
P
= C
1
e
−2t
+ C
2
e
t
+ C
3
e
3t
+ 2,
y = y
C
+ y
P
= −
5
2
C
1
e
−2t
+ 2C
2
e
t
+ 7t + 7.
Example 7.8 7.8
Solve Dy
1
− y
2
= 0, D(·) ≡d(·)/dx, (1a)
Dy
2
− y
3
= 0, (1b)
6y
1
+ 11y
2
+ (D+6) y
3
= 2e
−x
. (1c)
The determinant of the coefﬁcient matrix is
φ(D) = det
D
E
E
−1
L
L
L
0
J
J
J
J
J D −1
0 D
L
L
L
−1
J
J
J
0
E
E
E
D
6 11 D+6 6 11
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
To evaluate the determinant,
add first two columns at right.
= D
2
(D+6) + 6 + 11D = D
3
+ 6D
2
+ 11D + 6.
The characteristic equation is φ(λ) = λ
3
+ 6λ
2
+ 11λ + 6 = 0. Since
(−1)
3
+ 6(−1)
2
+ 11(−1) + 6 = −1 + 6 − 11 + 6 = 0,
λ= −1 is a root or (λ+1) is a factor. The other factor can be determined using
long division and is obtained as (λ+1)(λ
2
+5λ+6) =0. Therefore
(λ+1)(λ+2)(λ+3) = 0 =⇒ λ = −1, −2, −3.
314 7 systems of linear differential equations
The complementary solution for y
1
is given by
y
1C
= C
1
e
−x
+ C
2
e
−2x
+ C
3
e
−3x
.
A particular solution for y
1
is given by
y1
(x) = det
0 −1 0
0 D −1
2e
−x
11 D+6
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
= 2e
−x
¸
¸
¸
¸
¸
−1 0
D −1
¸
¸
¸
¸
¸
= 2e
−x
,
y
1P
=
y1
(x)
φ(D)
=
1
D
3
+6D
2
+11D+6
(2e
−x
).
Apply Theorem 4 in Section 4.3.2:
∵ φ(D) = D
3
+6D
2
+11D+6, φ(−1) = 0,
φ
(D) = 3D
2
+12D+11, φ
(−1) = 2,
∴ y
1P
=
2
φ
(−1)
xe
−x
= xe
−x
.
Hence, the general solution of y
1
is
y
1
= y
1C
+ y
1P
= C
1
e
−x
+ C
2
e
−2x
+ C
3
e
−3x
+ xe
−x
.
Having obtained y
1
, the general solutions of y
2
and y
3
can be determined from
equations (1a) and (1b), respectively,
y
2
= Dy
1
= −C
1
e
−x
− 2C
2
e
−2x
− 3C
3
e
−3x
+ (1−x)e
−x
,
y
3
= Dy
2
= C
1
e
−x
+ 4C
2
e
−2x
+ 9C
3
e
−3x
− (2−x)e
−x
.
Method of Variation of Parameters
As for linear ordinary differential equations, when the righthand sides of a system
of linear ordinary differential equations are not of the form
e
αt
_
(a
0
+a
1
t + · · · +a
k
t
k
) cos βt + (b
0
+b
1
t + · · · +b
k
t
k
) sin βt
_
,
the method of variation of parameters has to be used to obtain particular solutions.
Example 7.9 7.9
Solve Dx − y = tan
2
t + 1, D(·) ≡d(·)/dt, (1a)
x + Dy = tan t. (1b)
First determine the complementary solutions for the complementary equations
Dx − y = 0, x + Dy = 0. (2)
7.2 the method of operator 315
The characteristic numbers are easily obtained
φ(D) =
¸
¸
¸
¸
D −1
1 D
¸
¸
¸
¸
= D
2
+ 1 =⇒ φ(λ) = λ
2
+ 1 = 0 =⇒ λ= ±i .
The complementary solution of x is
x
C
= Acos t + Bsin t. (3a)
From the ﬁrst equation of (2), the complementary solution of y is
y
C
= Dx
C
= −Asin t + Bcos t. (3b)
The method of variation of parameters is then applied to ﬁnd particular solutions.
Vary the parameters, i.e., make constants A and B functions of t, A⇒a(t),
B⇒b(t), and one has
x
P
= a(t) cos t + b(t) sin t. y
P
= −a(t) sin t + b(t) cos t. (4)
Substituting equations (4) into (1) yields
Dx
P
− y
P
=
_
a
(t) cos t − a(t) sin t + b
(t) sin t + b(t) cos t
_
−
_
−a(t) sin t + b(t) cos t
_
= a
(t) cos t + b
(t) sin t = tan
2
t + 1, (5a)
x
P
+ Dy
P
=
_
a(t) cos t + b(t) sin t
_
+
_
−a
(t) sin t − a(t) cos t + b
(t) cos t − b(t) sin t
_
= −a
(t) sin t + b
(t) cos t = tan t. (5b)
Equations (5a) and (5b) give two linear algebraic equations for two unknowns a
(t)
and b
(t), which can be solved using Gaussian elimination or Cramer’s Rule. To
ﬁnd a
(t), eliminate b
(t) as follows:
Eq(5a) × cos t : a
(t) cos
2
t + b
(t) sin t cos t = (tan
2
t + 1) cos t, (6a)
Eq(5b) × sin t : −a
(t) sin
2
t + b
(t) cos t sin t = tan t sin t, (6b)
Eq(6a) − Eq(6b): a
(t) = tan
2
t cos t + cos t − tan t sin t = cos t. (7a)
From equation (5b), one obtains
b
(t) =
tan t + a
(t) sin t
cos t
=
tan t + cos t sin t
cos t
=
sin t
cos
2
t
+ sin t. (7b)
Integrating equations (7) leads to
a(t) =
_
cos t dt = sin t,
316 7 systems of linear differential equations
b(t) =
_
sin t
cos
2
t
dt +
_
sin t dt = −
_
1
cos
2
t
d(cos t) − cos t =
1
cos t
− cos t.
Hence, the general solutions are, using equations (3) and (4),
x = x
C
+ x
P
= Acos t + Bsin t + sin t cos t +
_
1
cos t
− cos t
_
sin t
= Acos t + Bsin t + tan t,
y = y
C
+ y
P
= −Asin t + Bcos t − sin t sin t +
_
1
cos t
− cos t
_
cos t
= −Asin t + Bcos t.
Example 7.10 7.10
Solve (D
2
+3D+2)x + (D+1) y = 0, D(·) ≡d(·)/dt, (1a)
(D+2)x + (D−1) y =
8
e
2t
+1
. (1b)
First determine the solutions for the complementary equations
(D
2
+3D+2)x + (D+1) y = 0, (1a)
(D+2)x + (D−1) y = 0. (1b
)
The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
¸
¸
¸
D
2
+3D+2 D+1
D+2 D−1
¸
¸
¸
¸
¸
= D
3
+D
2
−4D−4 = (D+1)(D
2
−4).
The characteristic equation is φ(λ) =(λ+1)(λ
2
−4) =0 =⇒ λ= −1, ±2. The
complementary solutions are
x
C
= C
1
e
−t
+ C
2
e
2t
+ C
3
e
−2t
, y
C
= D
1
e
−t
+ D
2
e
2t
+ D
3
e
−2t
.
Since φ(D) is a polynomial of degree 3 in D, the complementary solutions should
contain three arbitrary constants. Substitute the complementary solutions into
equation (1b
) to eliminate the extra constants
(D+2)x
C
+ (D−1) y
C
= (C
1
−2D
1
)e
−t
+ (4C
2
+D
2
)e
2t
− 3D
3
e
−2t
= 0,
which leads to
C
1
−2D
1
= 0 =⇒ C
1
= 2D
1
,
4C
2
−D
2
= 0 =⇒ D
2
= −4C
2
,
D
3
= 0.
7.2 the method of operator 317
Hence
x
C
= 2D
1
e
−t
+ C
2
e
2t
+ C
3
e
−2t
, (2a)
y
C
= D
1
e
−t
− 4C
2
e
2t
. (2b)
Since equation (1a) is a secondorder equation in x, a third equation is needed
from differentiating x
C
with respect to t
x
C
= −2D
1
e
−t
+ 2C
2
e
2t
− 2C
3
e
−2t
. (2c)
Apply the method of variation of parameters, i.e., make D
1
⇒c
1
(t), C
2
⇒c
2
(t),
C
3
⇒c
3
(t) in equations (2) to yield
x
P
= 2c
1
(t)e
−t
+ c
2
(t)e
2t
+ c
3
(t)e
−2t
, (3a)
y
P
= c
1
(t)e
−t
− 4c
2
(t)e
2t
, (3b)
x
P
= −2c
1
(t)e
−t
+ 2c
2
(t)e
2t
− 2c
3
(t)e
−2t
. (3c)
Differentiating equation (3a) with respect to t and comparing with (3c) lead to
x
P
= 2c
1
(t)e
−t
− 2c
1
(t)e
−t
+ c
2
(t)e
2t
+ 2c
2
(t)e
2t
+ c
3
(t)e
−2t
− 2c
3
(t)e
−2t
= −2c
1
(t)e
−t
+ 2c
2
(t)e
2t
− 2c
3
(t)e
−2t
, Equation (3c)
which gives
c
1
(t)e
−t
+ c
2
(t)e
2t
+ c
3
(t)e
−2t
= 0. (4a)
Substituting equations (3) into (1a) yields
(D
2
+3D+2)x
P
+ (D+1) y
P
= −c
1
(t)e
−t
− 2c
2
(t)e
2t
− 2c
3
(t)e
−2t
= 0, (4b)
and into (1b) results in
(D+2)x
P
+ (D−1) y
P
= c
1
(t)e
−t
− 4c
2
(t)e
2t
=
8
e
2t
+1
. (4c)
Equations (4) provide three linear algebraic equations for three unknowns c
1
(t),
c
2
(t), and c
3
(t), which can be solved using Cramer’s Rule:
=
¸
¸
¸
¸
¸
¸
¸
¸
e
−t
e
2t
e
−2t
−e
−t
−2e
2t
−2e
−2t
e
−t
−4e
2t
0
¸
¸
¸
¸
¸
¸
¸
¸
= e
−t
· e
2t
· e
−2t
¸
¸
¸
¸
¸
¸
¸
¸
1 1 1
−1 −2 −2
1 −4 0
¸
¸
¸
¸
¸
¸
¸
¸
= −4e
−t
,
1
= det
0
e
2t
e
−2t
0 −2e
2t
−2e
−2t
8
e
2t
+1
−4e
2t
0
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
=
8
e
2t
+1
· e
2t
· e
−2t
¸
¸
¸
¸
¸
1 1
−2 −2
¸
¸
¸
¸
¸
= 0,
318 7 systems of linear differential equations
2
= det
e
−t
0
e
−2t
−e
−t
0 −2e
−2t
e
−t
8
e
2t
+1
0
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
= −
8
e
2t
+1
· e
−t
· e
−2t
¸
¸
¸
¸
¸
1 1
−1 −2
¸
¸
¸
¸
¸
=
8e
−3t
e
2t
+1
,
3
= det
e
−t
e
2t
0
−e
−t
−2e
2t
0
e
−t
−4e
2t
8
e
2t
+1
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
=
8
e
2t
+1
· e
−t
· e
2t
¸
¸
¸
¸
¸
1 1
−1 −2
¸
¸
¸
¸
¸
= −
8e
t
e
2t
+1
,
c
1
(t) =
1
= 0, c
2
(t) =
2
= −
2e
−2t
e
2t
+1
, c
3
(t) =
3
=
2e
2t
e
2t
+1
.
Integrating c
1
(t), c
2
(t), and c
3
(t) result in
c
1
(t) = 0,
c
2
(t) = −
_
2e
−2t
e
2t
+1
dt =
_
e
−2t
1+e
−2t
d(e
−2t
)
=
_
_
1−
1
1+e
−2t
_
d(e
−2t
) = e
−2t
−ln(1+e
−2t
) = e
−2t
−ln(e
2t
+1)+2t,
c
3
(t) =
_
2e
2t
e
2t
+1
dt =
_
1
e
2t
+1
d(e
2t
) = ln(e
2t
+1).
Hence, the particular solutions are
x
P
= 2c
1
(t)e
−t
+ c
2
(t)e
2t
+ c
3
(t)e
−2t
=
_
e
−2t
− ln(e
2t
+1) + 2t
_
· e
2t
+ ln(e
2t
+1) · e
−2t
= 1 + (e
−2t
−e
2t
) ln(e
2t
+1) + 2t e
2t
,
y
P
= c
1
(t)e
−t
− 4c
2
(t)e
2t
= −4
_
e
−2t
− ln(e
2t
+1) + 2t
_
· e
2t
= −4
_
1 − e
2t
ln(e
2t
+1) + 2t e
2t
_
.
The general solutions are
x = x
C
+ x
P
= 2C
1
e
−t
+ C
2
e
2t
+ C
3
e
−2t
+ 1 + (e
−2t
−e
2t
) ln(e
2t
+1) + 2te
2t
,
y = y
C
+ y
P
= C
1
e
−t
− 4C
2
e
2t
− 4
_
1−e
2t
ln(e
2t
+1)+2te
2t
_
.
7.3 The Method of Laplace Transform
The procedure for solving systems of linear ordinary differential equations is very
straightforward. Consider a system of n ordinary differential equations for n
unknown functions x
i
(t), i =1, 2, . . . , n.
7.3 the method of laplace transform 319
❧ Take Laplace transformof bothsides of the equations, with X
i
(s) =L
_
x
i
(t)
_
,
i =1, 2, . . . , n.
❧ It results in a system of n algebraic equations for the n unknown Laplace
transforms X
i
(s), i =1, 2, . . . , n, which can be solved using Gaussian elimi
nation or Cramer’s Rule.
❧ The solutions of the system of linear differential equations are obtained by
taking inverse Laplace transform x
i
(t) =L
−1
_
X
i
(s)
_
, i =1, 2, . . . , n.
Example 7.11 7.11
Solve
dx
dt
− 3x − 6y = 0,
3x +
dy
dt
+ 3y = 18te
−3t
, x(0) = x
0
, y(0) = y
0
.
Let X(s) =L
_
x(t)
_
and Y(s) =L
_
y(t)
_
. Taking the Laplace transform of both
sides of the differential equations yields
_
s X(s) − x(0)
_
− 3X(s) − 6Y(s) = 0,
3X(s) +
_
sY(s) − y(0)
_
+ 3Y(s) = L
_
18te
−3t
_
,
where, using L
_
e
at
f (t)
_
=L
_
f (t)
_
¸
¸
¸
s→s−a
,
L
_
18te
−3t
_
= 18L
_
t
_
¸
¸
¸
s→s+3
= 18 ·
1
s
2
¸
¸
¸
¸
s→s+3
=
18
(s +3)
2
.
These give two algebraic equations for X(s) and Y(s)
(s −3)X(s) − 6Y(s) = x
0
, 3X(s) + (s +3)Y(s) = y
0
+
18
(s +3)
2
,
which can be solved using Gaussian elimination or Cramer’s Rule:
=
¸
¸
¸
¸
¸
s −3 −6
3 s +3
¸
¸
¸
¸
¸
= (s −3)(s +3) + 18 = s
2
+ 9,
X
=
¸
¸
¸
¸
¸
¸
¸
x
0
−6
y
0
+
18
(s +3)
2
s +3
¸
¸
¸
¸
¸
¸
¸
= x
0
(s +3) + 6
_
y
0
+
18
(s +3)
2
_
,
Y
=
¸
¸
¸
¸
¸
¸
¸
s −3 x
0
3 y
0
+
18
(s +3)
2
¸
¸
¸
¸
¸
¸
¸
= (s −3)
_
y
0
+
18
(s +3)
2
_
− 3x
0
,
∴ X(s) =
X
=
3(x
0
+2y
0
)
s
2
+9
+
x
0
s
s
2
+9
+
108
(s +3)
2
(s
2
+9)
,
320 7 systems of linear differential equations
Y(s) =
Y
=
−3(x
0
+y
0
)
s
2
+9
+
y
0
s
s
2
+9
+
18(s −3)
(s +3)
2
(s
2
+9)
.
Using partial fractions, one has
108
(s +3)
2
(s
2
+9)
=
A
2
(s +3)
2
+
A
1
s +3
+
Bs +C
s
2
+9
.
To ﬁnd A
2
, coverup (s +3)
2
and set s = −3:
A
2
=
108
(s
2
+9)
¸
¸
¸
¸
s=−3
=
108
(−3)
2
+9
= 6.
Comparing the coefﬁcients of the numerators leads to
s
3
: A
1
+ B = 0, (1)
s
2
: 3A
1
+ A
2
+ 6B + C = 0, (2)
s : 9A
1
+ 9B + 6C = 0, (3)
1: −27A
1
+ 9A
2
+ 9C = 108, (4)
Eqn (3) − 9×Eqn (1) : C = 0,
from Eqn (1) : B = −A
1
,
from Eqn (2) : 3A
1
+ 6 + 6(−A
1
) + 0 = 0 =⇒ A
1
= 2, B = −2.
Hence
L
−1
_
108
(s +3)
2
(s
2
+9)
_
= L
−1
_
6
(s +3)
2
+
2
s +3
−
2s
s
2
+3
2
_
= 2e
−3t
+ 6t e
−3t
− 2 cos 3t.
Similarly,
L
−1
_
18(s −3)
(s +3)
2
(s
2
+9)
_
= L
−1
_
−
6
(s +3)
2
−
1
s +3
+
s +3
s
2
+3
2
_
= −e
−3t
− 6t e
−3t
+ cos 3t + sin 3t.
The solutions of the differential equations are
x(t) = L
−1
_
X(s)
_
=L
−1
_
(x
0
+2y
0
) ·
3
s
2
+3
2
+ x
0
·
s
s
2
+3
2
+
108
(s +3)
2
(s
2
+9)
_
= (x
0
+2y
0
) sin 3t + x
0
cos 3t + 2e
−3t
+ 6t e
−3t
− 2 cos 3t
= (x
0
+2y
0
) sin 3t + (x
0
−2) cos 3t + 2(1+3t)e
−3t
,
y(t) = L
−1
_
Y(s)
_
=L
−1
_
−(x
0
+y
0
) ·
3
s
2
+3
2
+ y
0
·
s
s
2
+3
2
+
18(s −3)
(s +3)
2
(s
2
+9)
_
= −(x
0
+y
0
) sin 3t + y
0
cos 3t − e
−3t
− 6t e
−3t
+ cos 3t + sin 3t
= (1−x
0
−y
0
) sin 3t + (1+y
0
) cos 3t − (1+6t)e
−3t
.
7.3 the method of laplace transform 321
Example 7.12 7.12
Solve
dx
dt
+ 2x + 2
dy
dt
+ 5y = 0,
dx
dt
+ 3
dy
dt
+ y = 10 sin 2t H(t −π), x(0) = x
0
, y(0) = y
0
.
Let X(s) =L
_
x(t)
_
and Y(s) =L
_
y(t)
_
. Taking the Laplace transform of both
sides of the differential equations yields
_
s X(s) − x(0)
_
+ 2X(s) + 2
_
sY(s) − y(0)
_
+ 5Y(s) = 0,
_
s X(s) − x(0)
_
+ 3
_
sY(s) − y(0)
_
+Y(s) = 10L
_
sin 2t H(t −π)
_
,
where
L
_
sin 2t H(t −π)
_
= L
_
sin 2
_
(t −π)+π
_
H(t −π)
_
= L
_
sin 2(t −π)H(t −π)
_
L
_
f (t −a)H(t −a)
_
=e
−as
L
_
f (t)
_
= e
−πs
L
_
sin 2t
_
= e
−πs
2
s
2
+2
2
.
These two equations lead to two algebraic equations for X(s) and Y(s)
(s −2)X(s) + (2s +5)Y(s) = x
0
+2y
0
,
s X(s) + (3s +1)Y(s) = x
0
+ 3y
0
+ e
−πs
20
s
2
+4
,
which can be solved using Cramer’s Rule:
=
¸
¸
¸
¸
¸
s +2 2s +5
s 3s +1
¸
¸
¸
¸
¸
= (s +2)(3s +1) − s (2s +5) = s
2
+ 2s + 2,
X
=
¸
¸
¸
¸
¸
¸
¸
x
0
+2y
0
2s +5
x
0
+3y
0
+e
−πs
20
s
2
+4
3s +1
¸
¸
¸
¸
¸
¸
¸
= x
0
s − 4x
0
− 13y
0
− e
−πs
20(2s +5)
s
2
+4
,
Y
=
¸
¸
¸
¸
¸
¸
¸
s +2 x
0
+2y
0
s x
0
+3y
0
+e
−πs
20
s
2
+4
¸
¸
¸
¸
¸
¸
¸
= y
0
s + 2x
0
+ 6y
0
+ e
−πs
20(s +2)
s
2
+4
,
X(s) =
X
=
x
0
(s +1) − 5x
0
− 13y
0
(s +1)
2
+1
2
− e
−πs
20(2s +5)
(s
2
+4)(s
2
+2s +2)
,
Y(s) =
Y
=
y
0
(s +1) + 2x
0
+ 5y
0
(s +1)
2
+1
2
+ e
−πs
20(s +2)
(s
2
+4)(s
2
+2s +2)
.
322 7 systems of linear differential equations
Using partial fractions, one has
−
20(2s +5)
(s
2
+4)(s
2
+2s +2)
=
As +B
s
2
+4
+
Cs +D
s
2
+2s +2
.
Comparing the coefﬁcients of the numerators leads to
s
3
: A + C = 0 =⇒ C = −A, (1)
s
2
: 2A + B + D = 0, (2)
s : 2A + 2B + 4C = −40, (3)
1: 2B + 4D = −100 =⇒ B = −2D − 50. (4)
Substituting equations (1) and (4) into (2) and (3) gives
_
2A + (−2D−50) + D = 0
2A + 2(−2D−50) + 4(−A) = −40
=⇒
_
2A − D = 50,
−2A − 4D = 60,
which can easily be solved to yield
A = 14, D = −22 =⇒ B = −6, C = −14.
Hence
L
−1
_
−
20(2s +5)
(s
2
+4)(s
2
+2s +2)
_
= L
−1
_
14s −6
s
2
+4
−
14s +22
s
2
+2s +2
_
= L
−1
_
14·
s
s
2
+2
2
− 3·
2
s
2
+2
2
−
14(s +1)+8
(s +1)
2
+1
_
= 14 cos 2t − 3 sin 2t − (14 cos t + 8 sin t)e
−t
.
Similarly,
L
−1
_
20(s +2)
(s
2
+4)(s
2
+2s +2)
_
= L
−1
_
−6s +4
s
2
+4
+
6s +8
s
2
+2s +2
_
= −6 cos 2t + 2 sin 2t + (6 cos t + 2 sin t)e
−t
.
The solutions of the differential equations are
x(t) = L
−1
_
X(s)
_
=
_
x
0
cos t − (5x
0
+13y
0
) sin t
_
e
−t
+
_
14 cos 2(t −π) − 3 sin 2(t −π)
−
_
14 cos(t −π) + 8 sin(t −π)
_
e
−(t−π)
_
H(t −π)
=
_
x
0
cos t − (5x
0
+13y
0
) sin t
_
e
−t
+
_
14 cos 2t − 3 sin 2t + (14 cos t + 8 sin t)e
−t +π
_
H(t −π),
7.3 the method of laplace transform 323
y(t) = L
−1
_
Y(s)
_
=
_
y
0
cos t + (2x
0
+5y
0
) sin t
_
e
−t
+
_
−6 cos 2(t −π) + 2 sin 2(t −π)
+
_
6 cos(t −π) + 2 sin(t −π)
_
e
−(t−π)
_
H(t −π)
=
_
y
0
cos t + (2x
0
+5y
0
) sin t
_
e
−t
+
_
−6 cos 2t + 2 sin 2t − (6 cos t + 2 sin t)e
−t +π
_
H(t −π).
Example 7.13 7.13
Solve
d
2
x
dt
2
− 3
dx
dt
−
dy
dt
+ 2y = 60t H(t −1),
dx
dt
− 3x +
dy
dt
= 0, x(0) = 5, y(0) = 0, x
(0) = 15.
Let X(s) =L
_
x(t)
_
and Y(s) =L
_
y(t)
_
. Taking the Laplace transform of both
sides of the differential equations yields
_
s
2
X(s) − s x(0) − x
(0)
_
− 3
_
s X(s) − x(0)
_
−
_
sY(s) − y(0)
_
+ 2Y(s) = 60L
_
t H(t −1)
_
,
_
s X(s) − x(0)
_
− 3X(s) +
_
sY(s) − y(0)
_
= 0,
where
L
_
t H(t −1)
_
= L
__
(t −1)+1
_
H(t −1)
_
= e
−s
L
_
t +1
_
= e
−s
_
1
s
2
+
1
s
_
.
These two equations lead to two algebraic equations for X(s) and Y(s)
(s
2
−3s)X(s) − (s −2)Y(s) = 5s +
60e
−s
(s +1)
s
2
,
(s −3)X(s) + sY(s) = 5,
which can be solved using Cramer’s Rule:
=
¸
¸
¸
¸
¸
s (s −3) −(s −2)
(s −3) s
¸
¸
¸
¸
¸
= (s −3)
¸
¸
¸
¸
¸
s −(s −2)
1 s
¸
¸
¸
¸
¸
= (s −3)(s
2
+s −2) = (s −3)(s −1)(s +2),
X
=
¸
¸
¸
¸
¸
¸
5s +
60e
−s
(s +1)
s
2
−(s −2)
5 s
¸
¸
¸
¸
¸
¸
= 5s
2
+
60e
−s
(s +1)
s
+ 5s − 10,
Y
=
¸
¸
¸
¸
¸
¸
s (s −3) 5s +
60e
−s
(s +1)
s
2
(s −3) 5
¸
¸
¸
¸
¸
¸
= (s −3)
_
−
60e
−s
(s +1)
s
2
_
,
324 7 systems of linear differential equations
∴ X(s) =
X
=
5s
2
+5s − 10
(s −3)(s −1)(s +2)
+
60(s +1)
s(s −3)(s −1)(s +2)
e
−s
=
5
s −3
+
60(s +1)
s(s −3)(s −1)(s +2)
e
−s
,
Y(s) =
Y
= −
60(s +1)
s
2
(s −1)(s +2)
e
−s
.
Using partial fractions, one has
60(s +1)
s(s −3)(s −1)(s +2)
=
A
1
s
+
B
1
s −3
+
C
1
s −1
+
D
1
s +2
.
To ﬁnd A
1
, coverup s and set s =0:
A
1
=
60(s +1)
(s −3)(s −1)(s +2)
¸
¸
¸
¸
s=0
=
60(1)
(−3)(−1)(2)
= 10.
Similarly,
B
1
=
60(s +1)
s(s −1)(s +2)
¸
¸
¸
¸
s=3
=
60(4)
(3)(2)(5)
= 8,
C
1
=
60(s +1)
s(s −3)(s +2)
¸
¸
¸
¸
s=1
=
60(2)
(1)(−2)(3)
= −20,
D
1
=
60(s +1)
s(s −3)(s −1)
¸
¸
¸
¸
s=−2
=
60(−1)
(−2)(−5)(−3)
= 2.
Again, using partial fractions, one has
−60(s +1)
s
2
(s −1)(s +2)
=
A
2
s
+
B
2
s
2
+
C
2
s −1
+
D
2
s +2
.
To ﬁnd B
2
, coverup s
2
and set s =0:
B
2
=
−60(s +1)
(s −1)(s +2)
¸
¸
¸
¸
s=0
=
−60(1)
(−1)(2)
= 30.
Similarly,
C
2
=
−60(s +1)
s
2
(s +2)
¸
¸
¸
¸
s=1
=
−60(2)
(1)(3)
= −40,
D
2
=
−60(s +1)
s
2
(s −2)
¸
¸
¸
¸
s=−2
=
−60(−2)
(4)(−3)
= −5.
To ﬁnd A
2
, set s = −1:
0 =
A
2
−1
+
30
1
+
−40
−2
+
−5
1
=⇒ A
2
= 45.
7.4 the matrix method 325
Hence, taking the inverse Laplace transform, the solutions are
x(t) = L
−1
_
X(s)
_
= L
−1
_
5
s −3
+ e
−s
_
10
s
+
8
s −3
−
20
s −1
+
2
s +2
_
_
= 5e
3t
+
_
10 + 8e
3t
− 20e
t
+ 2e
−2t
_
t→t−1
H(t −1)
= 5e
3t
+
_
10 + 8e
3(t−1)
− 20e
t−1
+ 2e
−2(t−1)
_
H(t −1),
y(t) = L
−1
_
Y(s)
_
= L
−1
_
e
−s
_
45
s
+
30
s
2
−
40
s −1
−
5
s +2
_
_
=
_
45 + 30t − 40e
t
− 5e
−2t
_
t→t−1
H(t −1)
=
_
15 + 30t − 40e
t−1
− 5e
−2(t−1)
_
H(t −1).
7.4 The Matrix Method
Any linear ordinary differential equation or system of linear ordinary differential
equations can be written as a system of ﬁrstorder linear ordinary differential
equations. For example, consider the secondorder differential equation
x
+ 2ζ ω
0
x
+ ω
2
0
x = a sin t.
Denoting x =x
1
, x
=x
2
, the differential equation becomes
x
2
+ 2ζ ω
0
x
2
+ ω
2
0
x
1
= a sin t.
Noting that x
1
=x
2
, one obtains
_
x
1
x
2
_
=
_
x
2
−2ζ ω
0
x
2
−ω
2
0
x
1
+a sin t
_
=
_
0 1
−ω
2
0
−2ζ ω
0
__
x
1
x
2
_
+
_
0
a sin t
_
,
which is a system of two ﬁrstorder differential equations. Similarly, consider a
system of differential equations
x
+ 2x
+ x − y
= 2 sin 3t, 4x
+ 3x − y
+ 5y
− y = e
−t
cos 3t.
Letting x =x
1
, x
=x
2
, x
=x
3
, y =x
4
, y =x
5
, the differential equations become
x
3
+ 2x
2
+ x
1
− x
5
= 2 sin 3t, 4x
2
+ 3x
1
− x
5
+ 5x
5
− x
4
= e
−t
cos 3t,
one obtains
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎩
x
1
x
2
x
3
x
4
x
5
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎭
=
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎩
x
2
x
3
−x
1
− 2x
2
+ x
5
+ 2 sin 3t
x
5
3x
1
+ 4x
2
− x
4
+ 5x
5
− e
−t
cos 3t
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎭
,
326 7 systems of linear differential equations
or, in the matrix form,
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎩
x
1
x
2
x
3
x
4
x
5
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎭
=
⎡
⎢
⎢
⎢
⎢
⎢
⎢
⎣
0 1 0 0 0
0 0 1 0 0
−1 −2 0 0 1
0 0 0 0 1
3 4 0 −1 5
⎤
⎥
⎥
⎥
⎥
⎥
⎥
⎦
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎩
x
1
x
2
x
3
x
4
x
5
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎭
+
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎩
0
0
2 sin 3t
0
−e
−t
cos 3t
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎭
,
which is a system of ﬁve ﬁrstorder differential equations.
Hence, without loss of generality, consider a systemof ndimensional ﬁrstorder
linear ordinary differential equations with constant coefﬁcients of the form
x
(t) = Ax(t) + f (t),
where x(t) =
_
x
1
(t), x
2
(t), . . . , x
n
(t)
_
T
, f (t) =
_
f
1
(t), f
2
(t), . . . , f
n
(t)
_
T
, and A
is an n×n matrix with constant entries.
7.4.1 Complementary Solutions
First consider the ndimensional homogeneous system with f (t) =0, i.e.,
x
(t) = Ax(t). (1)
Seek a solution of the form x(t) =e
λt
v, where v is a constant vector. Substituting
into equation (1) yields λe
λt
v =Ae
λt
v. Since, e
λt
=0, one obtains
(A − λI)v = 0, (2)
where I is the n×n identity matrix, with1’s onthe maindiagonal and0’s elsewhere.
Equation (2) is a system of homogeneous linear algebraic equations. To have
nonzero solutions for v, the determinant of the coefﬁcient matrix must be zero, i.e.,
det(A − λI) = 0, (3)
which leads to the characteristic equation, a polynomial equation in λ of degree n.
Distinct Eigenvalues
The n solutions λ
1
, λ
2
, . . . , λ
n
of the characteristic equation (3) are called the
eigenvalues of A. Suppose the eigenvalues λ
1
, λ
2
, . . . , λ
n
are distinct real numbers.
Anonzero solution v
k
of system (2) with λ=λ
k
, i.e.,
(A − λ
k
I)v
k
= 0, k = 1, 2, . . . , n, (4)
is called an eigenvector corresponding to eigenvalue λ
k
.
7.4 the matrix method 327
From linear algebra it is well known that, if the eigenvalues λ
1
, λ
2
, . . . , λ
n
are
distinct, the corresponding eigenvectors v
1
, v
2
, . . . , v
n
are linearly independent.
Hence, with n eigenvalueeigenvector pairs λ
k
, v
k
, k =1, 2, . . . , n, there are n
linearly indenpendent solutions for system (2): e
λ
1
t
v
1
, e
λ
2
t
v
2
, . . . , e
λ
n
t
v
n
.
Distinct Eigenvalues
Suppose that matrix A of the homogeneous system x
(t) =Ax(t) has distinct
eigenvalues λ
1
, λ
2
, . . . , λ
n
with corresponding eigenvectors v
1
, v
2
, . . . , v
n
. Then
the complementary solution of the homogeneous system is
x(t) = C
1
e
λ
1
t
v
1
+ C
2
e
λ
2
t
v
2
+ · · · + C
n
e
λ
n
t
v
n
,
where C
1
, C
2
, . . . , C
n
are constants.
The n×n matrix
X(t) =
_
e
λ
1
t
v
1
, e
λ
2
t
v
2
, . . . , e
λ
n
t
v
n
_
,
whose columns are n linearly independent solutions of the homogeneous system,
is called a fundamental matrix for x
(t) =Ax(t).
Using the fundamental matrix, the complementary solution can be written as
x
(t) = Ax(t) =⇒ x(t) = X(t)C, C=
_
C
1
, C
2
, . . . , C
n
_
T
.
For the homogeneous system x
(t) =Ax(t) with the initial condition x(t
0
) =x
0
,
one has x(t
0
) =X(t
0
)C=x
0
=⇒ C=X
−1
(t
0
)x
0
,
x
(t) = Ax(t), x(t
0
) = x
0
=⇒ x(t) = X(t) X
−1
(t
0
) x
0
.
Example 7.14 7.14
Solve x
1
− x
2
− 6x
2
= 0, ( · )
= d( · )/dt, (1)
x
1
+ 2x
2
− 3x
1
= 0. (2)
Solve equations (1) and (2) for x
1
and x
2
2×Eqn (1) + Eqn (2) : 3x
1
− 12x
2
− 3x
1
= 0 =⇒ x
1
= x
1
+ 4x
2
,
Eqn (2) − Eqn (1) : 3x
2
− 3x
1
+ 6x
2
= 0 =⇒ x
2
= x
1
− 2x
2
,
which can be written in the matrix form as
x
(t) = Ax(t), x(t) =
_
x
1
x
2
_
, A =
_
1 4
1 −2
_
.
328 7 systems of linear differential equations
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
1−λ 4
1 −2−λ
¸
¸
¸
¸
¸
= λ
2
+ λ − 6 = (λ+3)(λ−2) = 0,
and the two eigenvalues are λ
1
= −3, λ
2
=2. The corresponding eigenvectors are
obtained as follows.
(1) λ=λ
1
= −3:
(A−λ
1
I)v
1
=
_
4 4
1 1
__
v
11
v
21
_
= 0 =⇒ v
11
+ v
21
= 0,
taking v
21
= −1, then v
11
= −v
21
=1 =⇒ v
1
=
_
v
11
v
21
_
=
_
1
−1
_
.
(2) λ=λ
2
=2:
(A−λ
2
I)v
2
=
_
−1 4
1 −4
__
v
12
v
22
_
= 0 =⇒ v
12
− 4v
22
= 0,
taking v
22
=1, then v
12
=4v
22
=4 =⇒ v
2
=
_
v
12
v
22
_
=
_
4
1
_
.
Hence, the complementary solution is
x(t) = C
1
e
λ
1
t
v
1
+ C
2
e
λ
2
t
v
2
= C
1
e
−3t
_
1
−1
_
+ C
2
e
2t
_
4
1
_
,
or
x
1
(t) = C
1
e
−3t
+ 4C
2
e
2t
, x
2
(t) = −C
1
e
−3t
+ C
2
e
2t
.
Complex Eigenvalues
Consider the ﬁrstorder homogeneous system
x
(t) = Ax(t), (1)
where A is a real matrix. Suppose λ=α+i β is an eigenvalue, i.e., det(A−λI) =0.
Since the characteristic equation has real coefﬁcients, then
¯
λ=α−i β is also an
eigenvalue.
Let the complex vector v be an eigenvector corresponding to λ, i.e.,
(A − λI)v = 0. (2)
Then x
1
(t) =e
λt
v is a solution of the homogeneous system (1).
7.4 the matrix method 329
Taking complex conjugate of equation (2), one has
(A − λI)v = (
¯
A −
¯
λ
¯
I) ¯ v = (A −
¯
λI) ¯ v = 0, A and I are real matrices.
implying that ¯ v is aneigenvector corresponding to eigenvalue
¯
λ. Thus x
2
(t) =e
¯
λt
¯ v
is a solution of the homogeneous system (1).
Corresponding to the eigenvalues α±i β, one obtains the complementary solu
tion
x(t) = C
1
e
λt
v + C
2
e
¯
λt
¯ v, v = v
R
+ i v
I
,
where C
1
and C
2
are complex constants, and v
R
and v
I
are, respectively, the real
and imaginary parts of the eigenvector v. Applying Euler’s formula
e
(α±i β)t
= e
αt
(cos βt ± i sin βt)
leads to
x(t) = C
1
e
αt
(cos βt +i sin βt)(v
R
+i v
I
) + C
2
e
αt
(cos βt −i sin βt)(v
R
−i v
I
)
= e
αt
_
(C
1
+C
2
)(v
R
cos βt −v
I
sin βt) + i (C
1
−C
2
)(v
R
sin βt +v
I
cos βt)
_
.
For the solution x(t) to be real, one must have C
1
+C
2
=A, i (C
1
−C
2
) =B, where
A and B are real constants. This can be accomplished if
¯
C
1
=C
2
. Hence
x(t) = Ae
αt
(v
R
cos βt − v
I
sin βt) + Be
αt
(v
R
sin βt + v
I
cos βt)
= ARe(e
λt
v) + B Im(e
λt
v).
Complex Eigenvalues
Suppose that matrix A of the homogeneous system x
(t) =Ax(t) is a real ma
trix. If λ=α+i β is an eigenvalue with the corresponding eigenvector v, then,
corresponding to the eigenvalues α±i β,
x
1
(t) = Re(e
λt
v) = e
αt
_
Re(v) cos βt − Im(v) sin βt
_
,
x
2
(t) = Im(e
λt
v) = e
αt
_
Re(v) sin βt + Im(v) cos βt
_
are two linearly independent realvalued solutions, or
x(t) = ARe(e
λt
v) + B Im(e
λt
v).
Example 7.15 7.15
Solve x
1
+ x
1
− 5x
2
= 0, ( · )
= d( · )/dt,
4x
1
+ x
2
+ 5x
2
= 0.
330 7 systems of linear differential equations
In the matrix form, the system of differential equations can be written as
x
(t) = Ax(t), x(t) =
_
x
1
x
2
_
, A =
_
−1 5
−4 −5
_
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
−1−λ 5
−4 −5−λ
¸
¸
¸
¸
¸
= λ
2
+ 6λ + 25 = 0 =⇒ λ= −3±i 4.
For eigenvalue λ= −3+i 4, the corresponding eigenvector is
(A−λI)v =
_
2−i 4 5
−4 −2−i 4
__
v
1
v
2
_
=
_
(2−i 4)v
1
+5v
2
−4v
1
−(2+i 4)v
2
_
=
_
0
0
_
.
Note that the two equations (2−i 4)v
1
+5v
2
=0 and 4v
1
+(2+i 4)v
2
=0 are the
same. Taking v
1
=5, then v
2
= −
1
5
(2−i 4)v
1
= −2+i 4,
∴ v =
_
v
1
v
2
_
=
_
5
−2+i 4
_
=
_
5
−2
_
+ i
_
0
4
_
.
Hence
e
λt
v = e
−3t
(cos 4t + i sin 4t)
__
5
−2
_
+ i
_
0
4
__
= e
−3t
___
5
−2
_
cos 4t −
_
0
4
_
sin 4t
_
+ i
__
5
−2
_
sin 4t +
_
0
4
_
cos 4t
__
.
Hence, the complementary solution is
x(t) = ARe(e
λt
v) + B Im(e
λt
v) = Ae
−3t
__
5
−2
_
cos 4t −
_
0
4
_
sin 4t
_
+ B e
−3t
__
5
−2
_
sin 4t +
_
0
4
_
cos 4t
_
,
∴ x
1
(t) = 5e
−3t
(Acos 4t + Bsin 4t),
x
2
(t) = 2e
−3t
_
(−A+2B) cos 4t − (2A+B) sin 4t
_
.
Multiple Eigenvalues
For an n×n matrix A with constant entries, if its n eigenvalues λ
1
, λ
2
, . . . , λ
n
,
either real or complex, are distinct, then the corresponding n eigenvectors v
1
, v
2
,
. . . , v
n
are linearly independent and form a complete basis of eigenvectors.
7.4 the matrix method 331
If matrix A has a repeated eigenvalue with algebraic multiplicity m>1 (the
number of times the eigenvalue is repeated as a root of the characteristic equation),
it is possible that the multiple eigenvalue has m linearly independent eigenvec
tors. However, it is also possible that there are fewer than m linearly independent
eigenvectors; in this case, matrix A is a defective or deﬁcient matrix.
In other words, an n×n matrix is defective if and only if it does not have n
linearly independent eigenvectors. A complete basis is formed by augmenting the
eigenvectors with generalized eigenvectors.
Suppose λ is an eigenvalue of multiplicity m, and there are only k<m linearly
independent eigenvectors corresponding to λ. A complete basis of eigenvectors is
obtained by including (m−k) generalized eigenvectors:
(A−λI)v
i
= 0 =⇒ v
i
, i =1, 2, . . . , k, linearly independent eigenvectors,
(A−λI)v
k+1
= v
k
=⇒ (A−λI)
2
v
k+1
= 0,
(A−λI)v
k+2
= v
k+1
=⇒ (A−λI)
3
v
k+2
= 0,
.
.
.
(A−λI)v
m
= v
m−1
=⇒ (A−λI)
m−k+1
v
m
= 0.
⎫
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎭
Generalized eigenvectors
Multiple Eigenvalues
Suppose matrix A of the homogeneous system x
(t) =Ax(t) has an eigenvalue
λ of algebraic multiplicity m>1, and a sequence of generalized eigenvectors cor
responding to λ is v
1
, v
2
, . . . , v
m
. Then, corresponding to the eigenvalues λ, λ,
. . . , λ (repeated m times), m linearly independent solutions of the homogeneous
system are
x
i
(t) = e
λt
v
i
, i = 1, 2, . . . , k, ∵ v
1
, v
2
, · · · , v
k
are eigenvectors
x
k+1
(t) = e
λt
_
v
k
t + v
k+1
_
,
x
k+2
(t) = e
λt
_
v
k
t
2
2!
+ v
k+1
t + v
k+2
_
,
.
.
.
x
m
(t) = e
λt
_
v
k
t
m−k
(m−k)!
+ v
k+1
t
m−k−1
(m−k−1)!
+ · · · + v
m−2
t
2
2!
+ v
m−1
t + v
m
_
.
Example 7.16 7.16
Solve x
1
− 4x
1
+ x
2
= 0, ( · )
= d( · )/dt,
3x
1
− x
2
+ x
2
− x
3
= 0,
x
1
− x
3
+ x
3
= 0.
332 7 systems of linear differential equations
In the matrix form, the system of differential equations can be written as
x
(t) = Ax(t), x(t) =
⎧
⎨
⎩
x
1
x
2
x
3
⎫
⎬
⎭
, A =
⎡
⎣
4 −1 0
3 1 −1
1 0 1
⎤
⎦
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
¸
4−λ −1 0
3 1−λ −1
1 0 1−λ
¸
¸
¸
¸
¸
¸
= −(λ
3
−6λ
2
+12λ−8) = −(λ−2)
3
= 0.
Hence, λ=2 is an eigenvector of multiplicity 3. The eigenvector equation is
(A−λI)v
1
=
⎡
⎣
2 −1 0
3 −1 −1
1 0 −1
⎤
⎦
⎧
⎨
⎩
v
11
v
21
v
31
⎫
⎬
⎭
=
⎧
⎨
⎩
2v
11
−v
21
3v
11
−v
21
−v
31
v
11
−v
31
⎫
⎬
⎭
=
⎧
⎨
⎩
0
0
0
⎫
⎬
⎭
.
Taking v
11
=1, then v
21
=2v
11
=2, v
31
=v
11
=1,
∴ v
1
=
⎧
⎨
⎩
v
11
v
21
v
31
⎫
⎬
⎭
=
⎧
⎨
⎩
1
2
1
⎫
⎬
⎭
.
It is not possible to ﬁnd two more linearly independent eigenvectors. Hence, matrix
A is defective and a complete basis of eigenvectors is obtained by including two
generalized eigenvectors:
(A−λI)v
2
= v
1
=⇒
⎡
⎢
⎣
2 −1 0
3 −1 −1
1 0 −1
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
v
12
v
22
v
32
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
2v
12
−v
22
3v
12
−v
22
−v
32
v
12
−v
32
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
1
2
1
⎫
⎪
⎬
⎪
⎭
.
Taking v
12
=2, then v
22
=2v
12
−1=3, v
32
=v
12
−1=1,
∴ v
2
=
⎧
⎪
⎨
⎪
⎩
v
12
v
22
v
32
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
2
3
1
⎫
⎪
⎬
⎪
⎭
.
(A−λI)v
3
= v
2
=⇒
⎡
⎢
⎣
2 −1 0
3 −1 −1
1 0 −1
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
v
13
v
23
v
33
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
2v
13
−v
23
3v
13
−v
23
−v
33
v
13
−v
33
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
2
3
1
⎫
⎪
⎬
⎪
⎭
.
Taking v
13
=1, then v
23
=2v
13
−2=0, v
33
=v
13
−1=0,
∴ v
3
=
⎧
⎪
⎨
⎪
⎩
v
13
v
23
v
33
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
1
0
0
⎫
⎪
⎬
⎪
⎭
.
7.4 the matrix method 333
Three linearly independent solutions are
x
1
(t) = e
λt
v
1
= e
2t
⎧
⎪
⎨
⎪
⎩
1
2
1
⎫
⎪
⎬
⎪
⎭
, x
2
(t) = e
λt
_
v
1
t +v
2
_
= e
2t
⎛
⎜
⎝
⎧
⎪
⎨
⎪
⎩
1
2
1
⎫
⎪
⎬
⎪
⎭
t +
⎧
⎪
⎨
⎪
⎩
2
3
1
⎫
⎪
⎬
⎪
⎭
⎞
⎟
⎠
,
x
3
(t) = e
λt
_
v
1
t
2
2
+v
2
t +v
3
_
= e
2t
⎛
⎜
⎝
⎧
⎪
⎨
⎪
⎩
1
2
1
⎫
⎪
⎬
⎪
⎭
t
2
2
+
⎧
⎪
⎨
⎪
⎩
2
3
1
⎫
⎪
⎬
⎪
⎭
t +
⎧
⎪
⎨
⎪
⎩
1
0
0
⎫
⎪
⎬
⎪
⎭
⎞
⎟
⎠
.
The complementary solution is
x(t) = C
1
x
1
(t) + C
2
x
2
(t) + 2C
3
x
3
(t)
= C
1
e
2t
⎧
⎪
⎨
⎪
⎩
1
2
1
⎫
⎪
⎬
⎪
⎭
+C
2
e
2t
⎛
⎜
⎝
⎧
⎪
⎨
⎪
⎩
1
2
1
⎫
⎪
⎬
⎪
⎭
t +
⎧
⎪
⎨
⎪
⎩
2
3
1
⎫
⎪
⎬
⎪
⎭
⎞
⎟
⎠
+2C
3
e
2t
⎛
⎜
⎝
⎧
⎪
⎨
⎪
⎩
1
2
1
⎫
⎪
⎬
⎪
⎭
t
2
2
+
⎧
⎪
⎨
⎪
⎩
2
3
1
⎫
⎪
⎬
⎪
⎭
t +
⎧
⎪
⎨
⎪
⎩
1
0
0
⎫
⎪
⎬
⎪
⎭
⎞
⎟
⎠
,
∴ x
1
(t) = e
2t
_
C
3
t
2
+(C
2
+4C
3
)t +(C
1
+2C
2
+2C
3
)
_
,
x
2
(t) = e
2t
_
2C
3
t
2
+2(C
2
+3C
3
)t +(2C
1
+3C
2
)
_
,
x
3
(t) = e
2t
_
C
3
t
2
+(C
2
+2C
3
)t +(C
1
+C
2
)
_
.
Example 7.17 7.17
Solve
x
(t) = Ax(t), x(t) =
⎧
⎨
⎩
x
1
x
2
x
3
⎫
⎬
⎭
, A =
⎡
⎣
−2 1 −2
1 −2 2
3 −3 5
⎤
⎦
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
¸
−2−λ 1 −2
1 −2−λ 2
3 −3 5−λ
¸
¸
¸
¸
¸
¸
= −(λ
3
− λ
2
− 5λ − 3)
= −(λ+1)
2
(λ−3) = 0.
Hence, λ
1
=λ
2
= −1 is an eigenvalue of multiplicity 2, and λ
3
=3. The eigenvec
tors are determined as follows.
(1) λ=λ
1
=λ
2
= −1:
(A−λI)v = 0 =⇒
⎡
⎣
−1 1 −2
1 −1 2
3 −3 6
⎤
⎦
⎧
⎨
⎩
v
1
v
2
v
3
⎫
⎬
⎭
=
⎧
⎨
⎩
−v
1
+v
2
−2v
3
−(−v
1
+v
2
−2v
3
)
−3(−v
1
+v
2
−2v
3
)
⎫
⎬
⎭
=
⎧
⎨
⎩
0
0
0
⎫
⎬
⎭
,
which leads to v
1
=v
2
−2v
3
.
334 7 systems of linear differential equations
Taking v
2
=1, v
3
=0 =⇒ v
1
=1; taking v
2
=0, v
3
=1 =⇒ v
1
= −2;
∴ v
1
=
⎧
⎨
⎩
1
1
0
⎫
⎬
⎭
, v
2
=
⎧
⎨
⎩
−2
0
1
⎫
⎬
⎭
.
(2) λ=λ
3
=3:
(A−λI)v
3
= 0 =⇒
⎡
⎣
−5 1 −2
1 −5 2
3 −3 2
⎤
⎦
⎧
⎨
⎩
v
13
v
23
v
33
⎫
⎬
⎭
=
⎧
⎨
⎩
−5v
13
+v
23
−2v
33
v
13
−5v
23
+2v
33
3v
13
−3v
23
+2v
33
⎫
⎬
⎭
=
⎧
⎨
⎩
0
0
0
⎫
⎬
⎭
.
Taking v
33
=3, then v
13
= −1, v
23
=1,
∴ v
3
=
⎧
⎨
⎩
v
13
v
23
v
33
⎫
⎬
⎭
=
⎧
⎨
⎩
−1
1
3
⎫
⎬
⎭
.
The complementary solution is
x(t) =
3
k=1
C
k
e
λ
k
t
v
k
= C
1
e
−t
⎧
⎨
⎩
1
1
0
⎫
⎬
⎭
+ C
2
e
−t
⎧
⎨
⎩
−2
0
1
⎫
⎬
⎭
+ C
3
e
3t
⎧
⎨
⎩
−1
1
3
⎫
⎬
⎭
,
∴ x
1
(t) =(C
1
−2C
2
)e
−t
−C
3
e
3t
, x
2
(t) =C
1
e
−t
+C
3
e
3t
, x
3
(t) =C
2
e
−t
+3C
3
e
3t
.
Remarks: Although λ= −1 is an eigenvalue of multiplicity 2, two linearly
independent eigenvectors do exist.
7.4.2 Particular Solutions
The method of variation of parameters is applied to ﬁnd a particular solution of the
nonhomogeneous system
x
(t) = Ax(t) + f (t).
The complementary solution of the homogeneous system x
(t) =Ax(t) has
been obtained as x(t) =X(t)C, where X(t) is a fundamental matrix, whose
columns are linearly independent andeachis a solutionof the homogeneous system,
i.e., X
(t) =AX(t), and C is an ndimensional constant vector.
Applying the method of variation of parameters, vary the constant vector C in
the complementary solution x(t) =X(t)C to make it a vector of functions of t, i.e.,
C⇒c(t). Thus a particular solution is assumed to be of the form
x(t) = X(t)c(t).
7.4 the matrix method 335
Differentiating with respect to t yields
x
(t) = X
(t)c(t) + X(t)c
(t) = Ax(t) + f (t).
Substituting X
(t) =AX(t) and x(t) =X(t)c(t) yields
AX(t) c(t) + X(t)c
(t) = AX(t)c(t) + f (t),
∴ X(t)c
(t) = f (t) =⇒ c
(t) = X
−1
(t)f (t).
Integrating with respect to t gives
c(t) = C +
_
X
−1
(t)f (t)dt.
Hence, the general solution is given by
x(t) = X(t)c(t) = X(t)
_
C +
_
X
−1
(t) f (t)dt
_
.
For the nonhomogeneous system x
(t) =Ax(t)+f (t) with the initial condition
x(t
0
) =x
0
, the general solution can be written as
x(t) = X(t)
_
C +
_
t
t
0
X
−1
(t) f (t)dt
_
,
with
x(t
0
) = X(t
0
)C =⇒ C = X
−1
(t
0
)x(t
0
),
which yields
x(t) = X(t)
_
X
−1
(t
0
)x(t
0
) +
_
t
t
0
X
−1
(t) f (t)dt
_
.
To ﬁnd a particular solution using the method of variation of parameters, one
must evaluate the inverse X
−1
(t) of a fundamental matrix X(t). In the following,
the GaussJordan method is brieﬂy reviewed.
GaussJordan Method for Finding the Inverse of a Matrix
To ﬁnd the inverse of an n×n matrix A, det(A) =0, augment matrix A with the
n×n identity matrix I as
_
A
¸
¸
I
_
=
⎡
⎢
⎢
⎢
⎣
a
11
a
12
· · · a
1n
a
21
a
22
· · · a
2n
.
.
.
.
.
.
· · ·
.
.
.
a
n1
a
n2
· · · a
nn
¸
¸
¸
¸
¸
¸
¸
¸
¸
1 0 · · · 0
0 1 · · · 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 · · · 1
⎤
⎥
⎥
⎥
⎦
.
336 7 systems of linear differential equations
Apply a series of elementary row operations, such as
❧ exchange row k
and row l
,
❧ multiply row k
by α =0,
❧ multiply row l
by β =0 and add to row k
,
to convert the left half of the augmented matrix to the identity matrix. Then the
right half of the augmented matrix becomes A
−1
:
⎡
⎢
⎢
⎢
⎣
1 0 · · · 0
0 1 · · · 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 · · · 1
¸
¸
¸
¸
¸
¸
¸
¸
¸
b
11
b
12
· · · b
1n
b
21
b
22
· · · b
2n
.
.
.
.
.
.
· · ·
.
.
.
b
n1
b
n2
· · · b
nn
⎤
⎥
⎥
⎥
⎦
=
_
I
¸
¸
A
−1
_
.
Use k
×α+ l
×β to denote the elementary row operations:
❧ multiply row k
by α =0;
❧ multiply row l
by β =0 and add to row k
.
In particular, for a 2×2 matrix,
A=
_
a
11
a
12
a
21
a
22
_
=⇒ A
−1
=
1
det(A)
_
a
22
−a
12
−a
21
a
11
_
, det(A) =a
11
a
22
−a
12
a
21
.
Example 7.18 7.18
Solve x
1
+ 3x
1
+ 4x
2
= 2e
−t
,
x
1
− x
2
+ x
2
= 0.
In the matrix form, the system of differential equations can be written as
x
(t) = Ax(t) + f (t), x(t) =
_
x
1
x
2
_
, A =
_
−3 −4
1 1
_
, f (t) =
_
2e
−t
0
_
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
−3−λ −4
1 1−λ
¸
¸
¸
¸
= λ
2
+ 2λ + 1 = 0 =⇒ λ= −1, −1.
Hence, λ= −1 is an eigenvalue of multiplicity 2. The eigenvector equation is
(A−λI)v
1
=
_
−2 −4
1 2
_ _
v
11
v
21
_
=
_
0
0
_
=⇒ v
11
+ 2v
21
= 0.
Taking v
21
= −1, then v
11
= −2v
21
=2,
∴ v
1
=
_
v
11
v
21
_
=
_
2
−1
_
.
7.4 the matrix method 337
A second linearly independent eigenvector does not exist. Hence, matrix A is
defective anda complete basis of eigenvectors is obtainedby including a generalized
eigenvector:
(A−λI)v
2
= v
1
=⇒
_
−2 −4
1 2
_ _
v
12
v
22
_
=
_
2
−1
_
=⇒ v
12
+ 2v
22
= −1.
Taking v
22
= −1, then v
12
= −1−2v
22
=1,
∴ v
2
=
_
v
12
v
22
_
=
_
1
−1
_
.
Two linearly independent solutions are
x
1
(t) = e
λt
v
1
= e
−t
_
2
−1
_
, x
2
(t) = e
λt
_
v
1
t +v
2
_
= e
−t
__
2
−1
_
t +
_
1
−1
__
.
A fundamental matrix is
X(t) =
_
x
1
(t) x
2
(t)
_
=
_
2e
−t
(2t +1)e
−t
−e
−t
−(t +1)e
−t
_
, det(X) = −e
−2t
,
and its inverse is obtained as
X
−1
(t) =
1
−e
−2t
_
−(t +1)e
−t
−(2t +1)e
−t
e
−t
2e
−t
_
=
_
(t +1)e
t
(2t +1)e
t
−e
t
−2e
t
_
.
It is easy to evaluate
_
X
−1
(t) f (t)dt =
_
_
(t +1)e
t
(2t +1)e
t
−e
t
−2e
t
__
2e
−t
0
_
dt
=
_
_
2(t +1)
−2
_
dt =
_
t
2
+2t
−2t
_
.
The general solution is
x(t) = X(t)
_
C +
_
X
−1
(t) f (t)dt
_
=
_
2e
−t
(2t +1)e
−t
−e
−t
−(t +1)e
−t
__
C
1
+t
2
+2t
C
2
−2t
_
,
x
1
(t) =e
−t
_
−2t
2
+2(C
2
+1)t +(2C
1
+C
2
)
_
, x
2
(t) =e
−t
_
t
2
−C
2
t −(C
1
+C
2
)
_
.
Example 7.19 7.19
Solve x
1
− x
1
+ x
2
= sec t,
2x
1
− x
2
− x
2
= 0.
338 7 systems of linear differential equations
In the matrix form, the system of differential equations can be written as
x
(t) = Ax(t), x(t) =
_
x
1
x
2
_
, A =
_
−1 5
−4 −5
_
, f (t) =
_
sec t
0
_
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
1−λ −1
2 −1−λ
¸
¸
¸
¸
¸
= λ
2
+ 1 = 0 =⇒ λ = ±i .
For eigenvalue λ=i , the corresponding eigenvector is
(A−λI)v =
_
1−i −1
4 −1−i
__
v
1
v
2
_
=
_
(1−i )v
1
−v
2
2v
1
−(1+i )v
2
_
=
_
0
0
_
.
Taking v
1
=1, then v
2
=(1−i )v
1
=1−i ,
∴ v =
_
v
1
v
2
_
=
_
1
1−i
_
=
_
1
1
_
+ i
_
0
−1
_
.
Hence, using Euler’s formula e
i θ
= cos θ +i sin θ,
e
λt
v = (cos t + i sin t)
__
1
1
_
+ i
_
0
−1
__
=
_
1
1
_
cos t +
_
0
1
_
sin t + i
__
0
−1
_
cos t +
_
1
1
_
sin t
_
.
Two linearly independent realvalued solutions are
x
1
(t) = Re(e
λt
v) =
_
cos t
sin t +cos t
_
, x
2
(t) = Im(e
λt
v) =
_
sin t
sin t −cos t
_
.
A fundamental matrix is
X(t) =
_
x
1
(t) x
2
(t)
_
=
_
cos t sin t
sin t +cos t sin t −cos t
_
, det(X) = −1,
and its inverse is obtained as
X
−1
(t) =
_
cos t −sin t sin t
cos t +sin t −cos t
_
.
Evaluate the integral
_
X
−1
(t) f (t)dt =
_
_
1−tan t
1+tan t
_
dt =
_
t +ln
¸
¸
cos t
¸
¸
t −ln
¸
¸
cos t
¸
¸
_
.
7.4 the matrix method 339
The general solution is
x(t) = X(t)
_
C +
_
X
−1
(t) f (t)dt
_
=
_
cos t sin t
sin t +cos t sin t −cos t
__
C
1
+t +ln
¸
¸
cos t
¸
¸
C
2
+t −ln
¸
¸
cos t
¸
¸
_
,
∴ x
1
(t) = (t +C
1
) cos t + (t +C
2
) sin t + (cos t −sin t) ln
¸
¸
cos t
¸
¸
,
x
2
(t) = (C
1
−C
2
) cos t + (2t +C
1
+C
2
) sin t + 2 cos t ln
¸
¸
cos t
¸
¸
.
Example 7.20 7.20
Solve
x
(t) =Ax(t)+f (t), x(t) =
⎧
⎨
⎩
x
1
x
2
x
3
⎫
⎬
⎭
, A=
⎡
⎣
2 −1 −1
2 −1 −2
−1 1 2
⎤
⎦
, f (t) =
⎧
⎨
⎩
2e
t
4e
−t
0
⎫
⎬
⎭
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
¸
2−λ −1 −1
2 −1−λ −2
−1 1 2−λ
¸
¸
¸
¸
¸
¸
= −(λ
3
−3λ
2
+3λ−1) = −(λ−1)
3
= 0.
Hence, λ=1 is an eigenvector of multiplicity 3. The eigenvector equation is
(A−λI)v =
⎡
⎢
⎣
1 −1 −1
2 −2 −2
−1 1 1
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
v
1
v
2
v
3
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
v
1
−v
2
−v
3
2(v
1
−v
2
−v
3
)
−(v
1
−v
2
−v
3
)
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
0
0
0
⎫
⎪
⎬
⎪
⎭
,
which leads to v
1
=v
2
+v
3
. As a result, there are two linearly independent eigen
vectors. Taking v
21
=1 and v
31
= −1, then v
11
=v
21
+v
31
=0,
∴ v
1
=
⎧
⎪
⎨
⎪
⎩
v
11
v
21
v
31
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
0
1
−1
⎫
⎪
⎬
⎪
⎭
.
However, v
2
cannot be chosen arbitrarily; it has to satisfy a condition imposed by
v
3
, which will be clear in a moment.
A third linearly independent eigenvector does not exist. Hence, matrix A is de
fective anda complete basis of eigenvectors is obtainedby including one generalized
eigenvector:
(A−λI)v
3
=v
2
=⇒
⎡
⎢
⎣
1 −1 −1
2 −2 −2
−1 1 1
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
v
13
v
23
v
33
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
v
13
−v
23
−v
33
2(v
13
−v
23
−v
33
)
−(v
13
−v
23
−v
33
)
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
v
12
v
22
v
32
⎫
⎪
⎬
⎪
⎭
.
340 7 systems of linear differential equations
If v
13
−v
23
−v
33
=v
12
=a, one must have v
22
=2a, v
32
= −a. Taking a=1, then
v
12
=1, v
22
=2, v
32
= −1. Taking v
13
=v
23
=0, then v
33
=v
13
−v
23
−a= −1.
Hence
v
2
=
⎧
⎪
⎨
⎪
⎩
v
12
v
22
v
32
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
1
2
−1
⎫
⎪
⎬
⎪
⎭
, v
3
=
⎧
⎪
⎨
⎪
⎩
v
13
v
23
v
33
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
0
0
−1
⎫
⎪
⎬
⎪
⎭
.
Three linearly independent solutions are
x
1
(t) = e
λt
v
1
= e
t
⎧
⎪
⎨
⎪
⎩
0
1
−1
⎫
⎪
⎬
⎪
⎭
, x
2
(t) = e
λt
v
2
= e
t
⎧
⎪
⎨
⎪
⎩
1
2
−1
⎫
⎪
⎬
⎪
⎭
,
x
3
(t) = e
λt
_
v
2
t +v
3
_
= e
t
⎛
⎜
⎝
⎧
⎪
⎨
⎪
⎩
1
2
−1
⎫
⎪
⎬
⎪
⎭
t +
⎧
⎪
⎨
⎪
⎩
0
0
−1
⎫
⎪
⎬
⎪
⎭
⎞
⎟
⎠
.
A fundamental matrix is
X(t) =
_
x
1
(t) x
2
(t) x
3
(t)
_
= e
t
⎡
⎢
⎣
0 1 t
1 2 2t
−1 −1 −(t +1)
⎤
⎥
⎦
.
Apply the GaussJordan method to ﬁnd the inverse of fundamental matrix X:
⎡
⎢
⎣
0 1 t
1 2 2t
−1 −1 −(t +1)
¸
¸
¸
¸
¸
¸
¸
1 0 0
0 1 0
0 0 1
⎤
⎥
⎦
Exchange 1
and 2
==⇒
3
×(−1)
⎡
⎢
⎣
1 2 2t
0 1 t
1 1 t +1
¸
¸
¸
¸
¸
¸
¸
0 1 0
1 0 0
0 0 −1
⎤
⎥
⎦
3
− 1
==⇒
⎡
⎢
⎣
1 2 2t
0 1 t
0 −1 1−t
¸
¸
¸
¸
¸
¸
¸
0 1 0
1 0 0
0 −1 −1
⎤
⎥
⎦
1
− 2
×2
==⇒
3
+ 2
⎡
⎢
⎣
1 0 0
0 1 t
0 0 1
¸
¸
¸
¸
¸
¸
¸
−2 1 0
1 0 0
1 −1 −1
⎤
⎥
⎦
2
− 3
×t
==⇒
⎡
⎢
⎣
1 0 0
0 1 0
0 0 1
¸
¸
¸
¸
¸
¸
¸
−2 1 0
1−t t t
1 −1 −1
⎤
⎥
⎦
,
X
−1
(t) = e
−t
⎡
⎢
⎣
−2 1 0
1−t t t
1 −1 −1
⎤
⎥
⎦
.
Evaluate the integral
_
X
−1
(t) f (t) dt =
_
⎡
⎢
⎣
−2e
−t
e
−t
0
(1−t)e
−t
t e
−t
t e
−t
e
−t
−e
−t
−e
−t
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
2e
t
4e
−t
0
⎫
⎪
⎬
⎪
⎭
dt
7.4 the matrix method 341
=
_
⎧
⎪
⎨
⎪
⎩
−4+4e
−2t
2−2t + 4t e
−2t
2−4e
−2t
⎫
⎪
⎬
⎪
⎭
dt =
⎧
⎪
⎨
⎪
⎩
−4t −2e
−2t
2t −t
2
−(2t +1)e
−2t
2t +2e
−2t
⎫
⎪
⎬
⎪
⎭
.
The general solution is
x(t) = X(t)
_
C +
_
X
−1
(t) f (t)dt
_
=
⎡
⎢
⎣
0 e
t
t e
t
e
t
2e
t
2t e
t
−e
t
−e
t
−(t +1)e
t
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
C
1
−4t −2e
−2t
C
2
−t
2
+2t −(2t +1)e
−2t
C
3
+2t +2e
−2t
⎫
⎪
⎬
⎪
⎭
,
∴ x
1
(t) =
_
C
2
+C
3
t +t
2
+2t
_
e
t
− e
−t
,
x
2
(t) =
_
C
1
+2C
2
+2C
3
t +2t
2
_
e
t
− 4e
−t
,
x
3
(t) = −
_
C
1
+C
2
+C
3
(t +1)+t
2
_
e
t
+ e
−t
.
Example 7.21 7.21
Solve x
1
= 3x
1
− 3x
2
+ x
3
+ 2e
t
,
x
2
= 3x
1
− 2x
2
+ 2x
3
,
x
3
= −x
1
+ 2x
2
, x
1
(0) = 3, x
2
(0) = 2, x
3
(0) = 1.
In the matrix form, the system of differential equations can be written as
x
(t) = Ax(t) + f (t), x(t) =
⎧
⎪
⎨
⎪
⎩
x
1
x
2
x
3
⎫
⎪
⎬
⎪
⎭
, A =
⎡
⎢
⎣
3 −3 1
3 −2 2
−1 2 0
⎤
⎥
⎦
, f (t) =
⎧
⎪
⎨
⎪
⎩
2e
t
0
0
⎫
⎪
⎬
⎪
⎭
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
¸
¸
3−λ −3 1
2 −2−λ 2
−1 2 −λ
¸
¸
¸
¸
¸
¸
¸
= −λ
3
+ λ
2
− 2
= −(λ+1)(λ
2
−2λ+2) = 0,
which gives the eigenvalues λ= −1, 1±i .
(1) λ
1
= −1:
(A−λ
1
I)v
1
=
⎡
⎢
⎣
4 −3 1
3 −1 2
−1 2 1
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
v
11
v
21
v
31
⎫
⎪
⎬
⎪
⎭
=⇒
_
4v
11
−3v
21
+v
31
= 0, (1)
−v
11
+2v
21
+v
31
= 0. (2)
Eqn (1)−Eqn (2): 5v
11
−5v
21
=0. Taking v
21
=1 =⇒ v
11
=v
21
=1.
342 7 systems of linear differential equations
From Eqn (1): v
31
=3v
21
−4v
11
= −1. Hence
v
1
=
⎧
⎪
⎨
⎪
⎩
v
11
v
21
v
31
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
1
1
−1
⎫
⎪
⎬
⎪
⎭
.
(2) λ=1+i :
⎡
⎢
⎣
2−i −3 1
3 −3−i 2
−1 2 −1−i
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
v
1
v
2
v
3
⎫
⎪
⎬
⎪
⎭
=⇒
_
3v
1
−(3+i)v
2
+2v
3
= 0, (3)
−v
1
+2v
2
−(1+i)v
3
= 0. (4)
Eqn (3) +3×Eqn (4): (3−i)v
2
−(1+3i )v
3
=0.
Taking v
3
=1 =⇒ v
2
=
1+3i
3−i
=
(1+3i )(3+i)
(3−i)(3+i)
= i.
From Eqn (4): v
1
=2v
2
−(1+i)v
3
=2i −(1+i) = −1+i. Hence
v =
⎧
⎪
⎨
⎪
⎩
v
1
v
2
v
3
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
−1+i
i
1
⎫
⎪
⎬
⎪
⎭
,
e
λt
v = e
t
(cos t + i sin t)
⎛
⎜
⎝
⎧
⎪
⎨
⎪
⎩
−1
0
1
⎫
⎪
⎬
⎪
⎭
+ i
⎧
⎪
⎨
⎪
⎩
1
1
0
⎫
⎪
⎬
⎪
⎭
⎞
⎟
⎠
= e
t
⎡
⎢
⎣
⎧
⎪
⎨
⎪
⎩
−1
0
1
⎫
⎪
⎬
⎪
⎭
cos t −
⎧
⎪
⎨
⎪
⎩
1
1
0
⎫
⎪
⎬
⎪
⎭
sin t + i
⎛
⎜
⎝
⎧
⎪
⎨
⎪
⎩
1
1
0
⎫
⎪
⎬
⎪
⎭
cos t +
⎧
⎪
⎨
⎪
⎩
−1
0
1
⎫
⎪
⎬
⎪
⎭
sin t
⎞
⎟
⎠
⎤
⎥
⎦
.
A fundamental matrix is
X(t) =
_
e
λ
1
t
v
1
(t) Re(e
λt
v) Im(e
λt
v)
_
=
⎡
⎢
⎣
e
−t
−e
t
(cos t +sin t) e
t
(cos t −sin t)
e
−t
−e
t
sin t e
t
cos t
−e
−t
e
t
cos t e
t
sin t
⎤
⎥
⎦
.
Apply the GaussJordan method to ﬁnd the inverse of fundamental matrix, X
−1
(t):
⎡
⎢
⎣
e
−t
−e
t
(cos t +sin t) e
t
(cos t −sin t)
e
−t
−e
t
sin t e
t
cos t
−e
−t
e
t
cos t e
t
sin t
¸
¸
¸
¸
¸
¸
¸
1 0 0
0 1 0
0 0 1
⎤
⎥
⎦
2
− 1
==⇒
3
+ 1
⎡
⎢
⎣
e
−t
−e
t
(cos t +sin t) e
t
(cos t −sin t)
0 e
t
cos t e
t
sin t
0 −e
t
sin t e
t
cos t
¸
¸
¸
¸
¸
¸
¸
1 0 0
−1 1 0
1 0 1
⎤
⎥
⎦
7.4 the matrix method 343
1
+ 2
− 3
==⇒
3
×cos t + 2
×sin t
⎡
⎢
⎣
e
−t
0 0
0 e
t
cos t e
t
sin t
0 0 e
t
¸
¸
¸
¸
¸
¸
¸
−1 1 −1
−1 1 0
cos t −sin t sin t cos t
⎤
⎥
⎦
2
− 3
×sin t
==⇒
⎡
⎢
⎣
e
−t
0 0
0 e
t
cos t 0
0 0 e
t
¸
¸
¸
¸
¸
¸
¸
−1 1 −1
−1−cos t sin t +sin
2
t 1−sin
2
t −sin t cos t
cos t −sin t sin t cos t
⎤
⎥
⎦
1
× e
t
2
× e
−t
/cos t
==⇒
3
× e
−t
⎡
⎢
⎣
1 0 0
0 1 0
0 0 1
¸
¸
¸
¸
¸
¸
¸
−e
t
e
t
−e
t
−e
−t
(cos t +sin t) e
−t
cos t e
−t
sin t
−e
−t
(cos t −sin t) e
−t
sin t e
−t
cos t
⎤
⎥
⎦
.
Evaluate the integral
_
X
−1
(t) f (t) dt =
_
⎡
⎢
⎣
−e
t
e
t
−e
t
−e
−t
(cos t +sin t) e
−t
cos t e
−t
sin t
−e
−t
(cos t −sin t) e
−t
sin t e
−t
cos t
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
2e
t
0
0
⎫
⎪
⎬
⎪
⎭
dt
=
_
⎧
⎪
⎨
⎪
⎩
−2e
2t
−2(cos t +sin t)
2(cos t −sin t)
⎫
⎪
⎬
⎪
⎭
dt =
⎧
⎪
⎨
⎪
⎩
−e
2t
−2(sin t −cos t)
2(sin t +cos t)
⎫
⎪
⎬
⎪
⎭
.
Given the initial condition x(0), the vector C is given by
C = X
−1
(0) x(0) −
_ _
X
−1
(t) f (t)dt
_
t=0
=
⎡
⎢
⎣
−1 1 −1
−1 1 0
1 0 1
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
3
2
1
⎫
⎪
⎬
⎪
⎭
−
⎧
⎪
⎨
⎪
⎩
−1
2
2
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
−1
−3
2
⎫
⎪
⎬
⎪
⎭
.
The solution satisfying the initial condition is
x(t) = X(t)
_
C +
_
X
−1
(t) f (t)dt
_
=
⎡
⎢
⎣
e
−t
−e
t
(cos t +sin t) e
t
(cos t −sin t)
e
−t
−e
t
sin t e
t
cos t
−e
−t
e
t
cos t e
t
sin t
⎤
⎥
⎦
⎧
⎪
⎨
⎪
⎩
−1−e
2t
−3−2(sin t −cos t)
2+2(sin t +cos t)
⎫
⎪
⎬
⎪
⎭
=
⎧
⎪
⎨
⎪
⎩
−e
−t
+ e
t
(5 cos t +sin t −1)
−e
−t
+ e
t
(2 cos t +3 sin t +1)
e
−t
+ e
t
(−3 cos t +2 sin t +3)
⎫
⎪
⎬
⎪
⎭
.
344 7 systems of linear differential equations
7.4.3 Response of Multiple DegreesofFreedom Systems
As shown in Section 7.1, the equations of motion of an n degreesoffreedom
system occur naturally as a system of n coupled secondorder linear differential
equations of the form
M¨ x(t) + C ˙ x(t) + Kx(t) = F(t), (1)
where M, C, K are the mass, damping, and stiffness matrices of dimension n×n,
respectively, and F(t) is the load vector of dimension n. Matrices M and K are
symmetric, i.e., M
T
=M, K
T
=K, and positive deﬁnite.
Instead of converting system (1) to a system of 2n ﬁrstorder differential equa
tions, it is more convenient and physically meaningful to study system (1) directly.
Undamped Free Vibration
The equations of motion of undamped free vibration, or complementary solution,
are given by
M¨ x(t) + Kx(t) = 0. (2)
Seeking a solution of the form x(t) = ˆ x sin(ωt +θ) and substituting into equation
(2) yield
(K−ω
2
M) ˆ x sin(ωt +θ) = 0.
Since sin(ωt +θ) is not identically zero, one must have
(K−ω
2
M) ˆ x = 0. (3)
Equation (3) is a system of n homogeneous linear algebraic equations. To have
nonzero solutions for ˆ x, the determinant of the coefﬁcient matrix must be zero:
det(K−ω
2
M) = 0, (4)
which leads to the characteristic equation, a polynomial equation in ω
2
of degree
n. Equation (4) is also called the frequency equation. Since the mass and stiffness
matrices M and K are symmetric and positive deﬁnite, it can be shown that all
roots ω
2
of the frequency equation are real and positive. The ith root ω
i
, which is
called the ith eigenvalue (ω
1
<ω
2
< · · · <ω
n
), is the natural circular frequency of
the ith mode of the system or the ith modal frequency.
Corresponding to the ith eigenvalue ω
i
, a nonzero solution ˆ x
i
of system (3),
(K−ω
2
i
M) ˆ x
i
= 0, i =1, 2, . . . , n, (5)
is the ith eigenvector or the ith mode shape.
The response of the undamped free vibration is then given by
x(t) = a
1
x
1
sin(ω
1
t +θ
1
) + a
2
x
2
sin(ω
2
t +θ
2
) + · · · + a
n
x
n
sin(ω
n
t +θ
n
),
7.4 the matrix method 345
where the 2n constants a
1
, a
2
, . . . , a
n
, θ
1
, θ
2
, . . . , θ
n
are determined using the
initial conditions x(0) and ˙ x(0).
Orthogonality of Mode Shapes
Corresponding to the ith mode, equation (3) becomes
K ˆ x
i
= ω
2
i
Mˆ x
i
.
Multiplying this equation by ˆ x
T
j
from the left yields
ˆ x
T
j
K ˆ x
i
= ω
2
i
ˆ x
T
j
Mˆ x
i
. (6)
Similarly, corresponding to the jth mode, equation (3) becomes
K ˆ x
j
= ω
2
j
Mˆ x
j
.
Multiplying this equation by ˆ x
T
i
from the left gives
ˆ x
T
i
K ˆ x
j
= ω
2
j
ˆ x
T
i
Mˆ x
j
. (7)
Taking transpose of both sides of equation (7) leads to
ˆ x
T
j
K
T
(ˆ x
T
i
)
T
= ω
2
j
ˆ x
T
j
M
T
(ˆ x
T
i
)
T
=⇒ ˆ x
T
j
K ˆ x
i
= ω
2
j
ˆ x
T
j
Mˆ x
i
. (7
)
Subtracting equation (7
) from equation (6) results in
(ω
2
i
−ω
2
j
) ˆ x
T
j
Mˆ x
i
= 0.
Since ω
i
=ω
j
for different modes i =j, one has
ˆ x
T
j
Mˆ x
i
= 0, i =j.
From equation (6), one has
ˆ x
T
j
K ˆ x
i
= 0, i =j.
The orthogonality conditions can be written as
ˆ x
T
j
Mˆ x
i
=
_
0, j =i,
m
i
, j =i,
ˆ x
T
j
K ˆ x
i
=
_
0, j =i,
ω
2
i
m
i
, j =i.
(8)
Construct the modal matrix fromthe eigenvectors (mode shapes) ˆ x
1
, ˆ x
2
, . . . , ˆ x
n
as
=
_
ˆ x
1
, ˆ x
2
, . . . , ˆ x
n
_
.
The orthogonality conditions can then be written as
T
M =
⎡
⎣
m
1
m
2
.
.
.
m
n
⎤
⎦
,
T
K =
⎡
⎢
⎣
m
1
ω
2
1
m
2
ω
2
2
.
.
.
m
n
ω
2
n
⎤
⎥
⎦
. (9)
346 7 systems of linear differential equations
Undamped Forced Vibration
The equation of motion of the undamped forced vibration is
M¨ x(t) + Kx(t) = F(t). (10)
Letting x(t) =q(t), substituting into equation (10) and multiplying
T
from
the left yields
T
M¨ q(t) +
T
Kq(t) =
T
F(t).
Using the orthogonality conditions (9), one obtains a set of n uncoupled single
degreeoffreedom systems
m
i
¨ q
i
(t) + m
i
ω
2
i
q
i
(t) = f
i
(t), i =1, 2, . . . , n, (11)
where f
i
(t) = ˆ x
T
i
F(t). Each of equations (11) can be solved using the methods
presented in Chapters 5 and 6.
Having obtained q
i
(t), the response of the undamped forced vibration is
x(t) = q(t) = q
1
(t) ˆ x
1
+ q
2
(t) ˆ x
2
+ · · · + q
n
(t) ˆ x
n
=
n
i=1
q
i
(t) ˆ x
i
. (12)
Multiplying equation (12) by ˆ x
T
j
M from the left yields
ˆ x
T
j
Mx(t) =
n
i=1
q
i
(t) ˆ x
T
j
Mˆ x
i
= m
j
q
j
(t), j =1, 2, . . . , n.
Hence, the initial conditions q
i
(0) and ˙ q
i
(0) are then given by
q
i
(0) =
ˆ x
T
i
Mx(0)
m
i
, ˙ q
i
(0) =
ˆ x
T
i
M˙ x(0)
m
i
. (13)
Damped Forced Vibration
The same approach for undamped system can be applied for damped system.
Letting x(t) =q(t), substituting into the equation of motion (1) of damped
system and multiplying
T
from the left yields
T
M¨ q(t) +
T
C˙ q(t) +
T
Kq(t) =
T
F(t).
Assuming that the orthogonality condition applies to the damping matrix
ˆ x
T
j
C ˆ x
i
=
_
0, j =i,
c
i
, j =i,
the equations of motion are then decoupled
m
i
¨ q
i
(t) + c
i
˙ q
i
(t) + m
i
ω
2
i
q
i
(t) = f
i
(t), i =1, 2, . . . , n,
7.5 summary 347
or, in the standard form,
¨ q
i
(t) + 2ζ
i
ω
i
˙ q
i
(t) + ω
2
i
q
i
(t) =
f
i
(t)
m
i
, i =1, 2, . . . , n, (14)
where ζ
i
=c
i
/(2ω
i
m
i
) is the ith modal damping coefﬁcient (ratio).
Remarks: Inpractice, it is often not practical to set upthe damping matrix C by
evaluating its elements. It is generally more convenient and physically reasonable
to define the damping of a multiple degreesoffreedom system by specifying the
modal damping coefficients ζ
i
, i =1, 2, . . . , n, because the modal damping coef
ficient ζ
i
can be determined experimentally or estimatedwith adequate precision
in many engineering applications.
Amore detailed discussion of damping in a multiple degreesoffreedomsystem
is beyond the scope of this book, and can be found in standard textbooks on
structural dynamics.
Except for the damping terms, the procedure of analysis for damped forced vibra
tion is the same as that for undamped forced vibration.
An example of vibration of a twostory shear building is present in Section 8.4.
7.5 Summary
In this chapter, three methods, i.e., the method of operator, the method of Laplace
transform, and the matrix method, are introduced for solving systems of linear
ordinary differential equations.
7.5.1 The Method of Operator
The method of operator is an extension of the approach presented in Chapter 4 for
nthorder linear ordinary differential equations with constant coefﬁcients.
Consider a system of linear ordinary differential equations with independent
variable t and n dependent variables x
1
, x
2
, . . . , x
n
,
φ
11
(D) x
1
+ φ
12
(D) x
2
+ · · · + φ
1n
(D) x
n
= f
1
(t), D(·) ≡ d(·)/dt,
φ
21
(D) x
1
+ φ
22
(D) x
2
+ · · · + φ
2n
(D) x
n
= f
2
(t),
· · · · · ·
φ
n1
(D) x
1
+ φ
n2
(D) x
2
+ · · · + φ
nn
(D) x
n
= f
n
(t).
The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
φ
ij
(D)
¸
¸
.
Suppose the order of operator φ(D) is N.
348 7 systems of linear differential equations
Complementary Solutions
The characteristic equation is φ(λ) =0, which is a polynomial equation of degree
N and gives N characteristic numbers λ
1
, λ
2
, . . . , λ
N
. The complementary
solutions for x
1
, x
2
, . . . , x
n
all have the same formand can easily be obtained from
the characteristic numbers, denoted as x
iC
(t; C
i1
, C
i2
, . . . , C
iN
), i =1, 2, . . . , n.
Particular Solutions
When the righthand side functions f
i
(t), i =1, 2, . . . , N, are of the form
e
αt
_
(a
0
+ a
1
t + · · · + a
k
t
k
) cos βt + (b
0
+ b
1
t + · · · + b
k
t
k
) sin βt
_
,
it is advantageous to obtain particular solutions using the method of Doperator
x
iP
(t) =
i
(t)
φ(D)
, i = 1, 2, . . . , n,
where
i
(t) is obtained by replacing the ith column of φ(D) with the righthand
side vector.
Otherwise, particular solutions have to be obtained using the method of varia
tion of parameters.
General Solutions
The general solutions are
x
i
(t) = x
iC
(t; C
i1
, C
i2
, . . . , C
iN
) + x
iP
(t), i = 1, 2, . . . , n.
Since the order of φ(D) is N, there should only be N arbitrary constants in the
general solutions. The extra constants can be eliminated by substituting the general
solutions into one of the original differential equations.
7.5.2 The Method of Laplace Transform
The procedure of the method of Laplace transform for solving systems of linear
ordinary differential equations is the most straightforward.
Applying the Laplace transform to a system of linear differential equations con
verts it to a system of linear algebraic equations for the Laplace transforms, which
can easily be solved using Gaussian elimination or Cramer’s Rule. The solutions of
the systemof linear ordinary differential equations can then be obtained by ﬁnding
the inverse Laplace transforms.
The method of Laplace transform is advantageous in solving linear differential
equations with the righthand side functions involving the Heaviside step function
and the Dirac delta function. The restriction of this method is that the Laplace
transform of the righthand sides should be easily obtained.
7.5 summary 349
7.5.3 The Matrix Method
The matrix method is the most general and systematic approach, especially in
dealing with systems of higher dimensions. However, this method is the most difﬁ
cult to master because of the challenging concepts in eigenvalues and eigenvectors,
particularly when multiple eigenvalues are involved.
Rewrite a system of linear ordinary differential equations in the standard form
of a system of n ﬁrstorder linear ordinary differential equations
x
(t) =Ax(t)+f (t), x(t) =
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
x
1
x
2
.
.
.
x
n
⎫
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎭
, A=
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
a
11
a
12
· · · a
1n
a
21
a
22
· · · a
2n
.
.
.
.
.
.
· · ·
.
.
.
a
n1
a
n2
· · · a
nn
⎫
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎭
, f (t) =
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
f
1
(t)
f
2
(t)
.
.
.
f
n
(t)
⎫
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎭
.
Complementary solutions are the solutions of the homogeneous differential
equations with f (t) =0 being set to zero: x
(t) =Ax(t). The characteristic equa
tion of matrix A is
det(A − λI) = 0,
which is a polynomial equation in λ of degree n. A solution of the characteristic
equation λ
k
is called an eigenvalue and the corresponding eigenvector v
k
is given
by
(A − λ
k
I)v
k
= 0.
Case 1. Distinct Eigenvalues
When the n solutions λ
1
, λ
2
, . . . , λ
n
of the characteristic equations are distinct,
the n corresponding eigenvectors v
1
, v
2
, . . . , v
n
will be linearly independent. n
linearly independent solutions for the homogeneous system are
x
1
(t) = e
λ
1
t
v
1
, x
2
(t) = e
λ
2
t
v
2
, . . . , x
n
(t) = e
λ
n
t
v
n
.
Case 2. Complex Eigenvalues
Suppose λ=α+i β is an eigenvalue with the corresponding eigenvector v. Then
λ=α−i β is also an eigenvalue. Corresponding to the eigenvalues α±i β, two
linearly independent realvalued solutions of the homogeneous system are
x
1
(t) = Re(e
λt
v) = e
αt
_
Re(v) cos βt − Im(v) sin βt
_
,
x
2
(t) = Im(e
λt
v) = e
αt
_
Re(v) sin βt + Im(v) cos βt
_
.
Case 3. Multiple Eigenvalues
Suppose λ is an eigenvalue of multiplicity m, and there are k linearly independent
eigenvectors corresponding to λ. If k<m, then matrix A is defective and a com
plete basis of eigenvectors is obtained by including m−k generalized eigenvectors
350 7 systems of linear differential equations
obtained as follows:
(A−λI)v
i
= 0 =⇒ v
i
, i =1, 2, . . . , k, linearly independent eigenvectors,
(A−λI)v
k+1
= v
k
=⇒ (A−λI)
2
v
k+1
= 0,
(A−λI)v
k+2
= v
k+1
=⇒ (A−λI)
3
v
k+2
= 0,
.
.
.
(A−λI)v
m
= v
m−1
=⇒ (A−λI)
m−k+1
v
m
= 0.
⎫
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎭
Generalized eigenvectors
m linearly independent solutions of the homogeneous system are
x
i
(t) = e
λt
v
i
, i = 1, 2, . . . , k,
x
k+1
(t) = e
λt
_
v
k
t + v
k+1
_
,
x
k+2
(t) = e
λt
_
v
k
t
2
2!
+ v
k+1
t + v
k+2
_
,
.
.
.
x
m
(t) = e
λt
_
v
k
t
m−k
(m−k)!
+ v
k+1
t
m−k−1
(m−k−1)!
+ · · · + v
m−2
t
2
2!
+ v
m−1
t + v
m
_
.
Complementary Solutions
In general, if x
1
(t), x
2
(t), . . . , x
n
(t) are n linearly independent solutions of the
homogeneous system, a fundamental matrix is
X(t) =
_
x
1
(t) x
2
(t) · · · x
n
(t)
_
.
The complementary solution is given by
x
C
(t) = X(t)C = C
1
x
1
(t) + C
2
x
2
(t) + · · · + C
n
x
n
(t).
General Solutions
A particular solution is x
P
(t) =X(t)
_
X
−1
(t)f (t)dt.
The general solution of the nonhomogeneous system is given by
x(t) = x
C
(t) + x
P
(t) = X(t)
_
C +
_
X
−1
(t) f (t)dt
_
.
If the nonhomogeneous system satisﬁes the initial condition x(t
0
) =x
0
, then
x(t) = X(t)
_
X
−1
(t
0
) x
0
+
_
t
t
0
X
−1
(t) f (t)dt
_
.
problems 351
Problems
The Method of Operator
Solve the following systems of differential equations using the method of operator.
7.1 (D+2)x − y = 0, x + (D−2) y = 0, D(·) ≡d(·)/dt
A
NS x = C
1
e
√
3t
+ C
2
e
−
√
3t
, y = (2+
√
3)C
1
e
√
3t
+ (2−
√
3)C
2
e
−
√
3t
7.2 (2D+1)x − (5D+4) y = 0, (3D−2)x − (4D−1) y = 0, D(·) ≡d(·)/dt
A
NS x = 3C
1
e
t
+ C
2
e
−t
, y = C
1
e
t
+ C
2
e
−t
7.3 (D−1)x + 3y = 0, 3x − (D−1) y = 0, D(·) ≡d(·)/dt
A
NS x = e
t
(Acos 3t + Bsin 3t), y = e
t
(Asin 3t − Bcos 3t)
7.4 (D
2
+D)x + (D−2) y = 0, (D+1)x − Dy = 0, D(·) ≡d(·)/dt
A
NS x = C
1
e
t
+ C
2
e
−t
+ C
3
e
−2t
, y = 2C
1
e
t
+
1
2
C
3
e
−2t
7.5 (D
2
−3)x − 4y = 0, x + (D
2
+1) y = 0, D(·) ≡d(·)/dt
A
NS x = −2e
t
(C
1
+ C
2
+ C
2
t) − 2e
−t
(C
3
− C
4
+ C
4
t)
y = e
t
(C
1
+ C
2
t) + e
−t
(C
3
+ C
4
t)
7.6
dy
1
dx
− y
2
=0, 4y
1
+
dy
2
dx
− 4y
2
− 2y
3
=0, −2y
1
+ y
2
+
dy
3
dx
+ y
3
=0
A
NS y
1
= C
1
+ C
2
e
x
+ C
3
e
2x
, y
2
= C
2
e
x
+ 2C
3
e
2x
, y
3
= 2C
1
+
1
2
C
2
e
x
7.7 (D−2) y
1
+ 3y
2
− 3y
3
= 0, −4y
1
+ (D+5) y
2
− 3y
3
= 0,
−4y
1
+ 4y
2
+ (D−2) y
3
= 0, D(·) ≡d(·)/dx
A
NS y
1
= C
1
e
−x
+ C
3
e
2x
, y
2
= C
1
e
−x
+ C
2
e
−2x
+ C
3
e
2x
y
3
= C
2
e
−2x
+ C
3
e
2x
7.8 (D+1)x + 2y = 8, 2x + (D−2) y = 2e
−t
− 8
A
NS x = C
1
e
−2t
+ C
2
e
3t
+ e
−t
, y =
1
2
C
1
e
−2t
− 2C
2
e
3t
+ 4
7.9
dx
dt
= 2x − 3y + t e
−t
,
dy
dt
= 2x − 3y + e
−t
A
NS x = C
1
+ C
2
e
−t
− t
2
e
−t
, y =
2
3
C
1
+ C
2
e
−t
− (t
2
− t)e
−t
7.10 (D−1)x − 2y = e
t
, −4x + (D−3) y = 1
A
NS x = C
1
e
−t
+ C
2
e
5t
+
1
4
e
t
−
2
5
, y = −C
1
e
−t
+ 2C
2
e
5t
−
1
2
e
t
+
1
5
352 7 systems of linear differential equations
7.11 (D−4)x + 3y = sin t, −2x + (D+1) y = −2 cos t
A
NS x = C
1
e
t
+ C
2
e
2t
+ cos t − 2 sin t, y = C
1
e
t
+
2
3
C
2
e
2t
+ 2 cos t − 2 sin t
7.12
dx
dt
− y = 0, −x +
dy
dt
= e
t
+ e
−t
A
NS x = C
1
e
t
+ C
2
e
−t
+
1
2
te
t
−
1
2
te
−t
y =
_
C
1
+
1
2
_
e
t
+
_
−C
2
−
1
2
_
e
−t
+
1
2
te
t
+
1
2
te
−t
7.13 (D+2)x + 5y = 0, −x + (D−2) y = sin 2t
A
NS x = Acos t + Bsin t +
5
3
sin 2t
y = −
1
5
(2A + B) cos t +
1
5
(A − 2B) sin t −
2
3
(cos 2t + sin 2t)
7.14 (D−2)x + 2Dy = −4e
2t
, (2D−3)x + (3D−1) y = 0
A
NS x = C
1
e
−2t
+ C
2
e
t
+ 5e
2t
, y = −C
1
e
−2t
+
1
2
C
2
e
t
− e
2t
7.15 (3D+2)x + (D−6) y = 5e
t
, (4D+2)x + (D−8) y = 5e
t
+ 2t − 3
A
NS x = Acos 2t + Bsin 2t + 2e
t
− 3t + 5, y = Bcos 2t − Asin 2t + e
t
− t
7.16 (D−5)x + 3y = 2e
3t
, −x + (D−1) y = 5e
−t
A
NS x = C
1
e
2t
+ 3C
2
e
4t
− e
−t
− 4e
3t
, y = C
1
e
2t
+ C
2
e
4t
− 2e
−t
− 2e
3t
7.17 (D−2)x + y = 0, x + (D−2) y = −5e
t
sin t
A
NS x =C
1
e
t
+C
2
e
3t
+e
t
(2 cos t −sin t), y =C
1
e
t
−C
2
e
3t
+e
t
(3 cos t +sin t)
7.18 (D+4)x + 2y =
2
e
t
−1
, 6x − (D−3) y =
3
e
t
−1
A
NS x = C
1
+ 2C
2
e
−t
+ 2e
−t
ln
¸
¸
e
t
−1
¸
¸
, y = −2C
1
− 3C
2
e
−t
− 3e
−t
ln
¸
¸
e
t
−1
¸
¸
7.19 (D−1)x + y = sec t, −2x + (D+1) y = 0
A
NS x = C
1
cos t + C
2
sin t + t (cos t +sin t) + (cos t −sin t) ln
¸
¸
cos t
¸
¸
y = (C
1
−C
2
) cos t + (C
1
+C
2
) sin t + 2t sin t + 2 cos t ln
¸
¸
cos t
¸
¸
The Method of Laplace Transform
Solve the following differential equations using the method of Laplace transform.
7.20
dx
dt
− x − 2y = 16t e
t
, 2x −
dy
dt
− 2y = 0, x(0) = 4, y(0) = 0
A
NS x = −e
t
(12t +13) + e
−3t
+ 16e
2t
, y = −2e
t
(4t +3) − 2e
−3t
+ 8e
2t
7.21
dx
dt
− 2x + y = 5e
t
cos t, x +
dy
dt
− 2y = 10e
t
sin t, x(0) = y(0) = 0
A
NS x = 5e
t
(1 − cos t + sin t), y = 5e
t
(1 − cos t)
Problems 353
7.22
dx
dt
− 4x + 3y = sin t, 2x −
dy
dt
− y = 2 cos t, x(0) = x
0
, y(0) = y
0
A
NS x = (−2x
0
+3y
0
−4)e
t
+ 3(x
0
−y
0
+1)e
2t
+ cos t − 2 sin t
y = (−2x
0
+3y
0
−4)e
t
+ 2(x
0
−y
0
+1)e
2t
+ 2 cos t − 2 sin t
7.23
dx
dt
− 2x − y = 2e
t
, x −
dy
dt
+ 2y = 3e
4t
, x(0) = x
0
, y(0) = y
0
A
NS x =
1
2
(x
0
+y
0
+4)e
3t
+
1
2
(x
0
−y
0
+2t −2)e
t
− e
4t
y =
1
2
(x
0
+y
0
+4)e
3t
−
1
2
(x
0
−y
0
+2t)e
t
− 2e
4t
7.24
d
2
x
dt
2
+
dx
dt
+
dy
dt
− 2y = 40e
3t
,
dx
dt
+ x −
dy
dt
= 36e
t
x(0) = 1, y(0) = 3, x
(0) = 1
A
NS x =22e
−2t
−33e
−t
−3e
t
(2t −3)+3e
3t
, y =11e
−2t
−12e
t
(t +1)+4e
3t
7.25
dx
dt
− 2x − y = 2e
t
,
dy
dt
− 2y − 4z = 4e
2t
, x −
dz
dt
− z = 0
x(0) = 9, y(0) = 3, z(0) = 1
A
NS x = 3t + 2 + 2e
t
− 3e
2t
+ 8e
3t
, y = −6t − 1 − 4e
t
+ 8e
3t
z = 3t − 1 + e
t
− e
2t
+ 2e
3t
7.26
d
2
x
dt
2
+ 2x − 2
dy
dt
= 0, 3
dx
dt
+
d
2
y
dt
2
− 8y = 240e
t
x(0) = y(0) = x
(0) = y
(0) = 0
A
NS x = 12 cos 2t − 24 sin 2t − 10e
−2t
+ 30e
2t
− 32e
t
y = −12 cos 2t − 6 sin 2t + 15e
−2t
+ 45e
2t
− 48e
t
7.27
dx
dt
− x − 2y = 0, x −
dy
dt
= 15 cos t H(t −π), x(0) = x
0
, y(0) = y
0
A
NS x =
2
3
(x
0
+y
0
)e
2t
+
1
3
(x
0
−2y
0
)e
−t
+
_
4e
2(t−π)
+ 5e
−(t−π)
+ 9 cos t + 3 sin t
_
H(t −π)
y =
1
3
(x
0
+y
0
)e
2t
−
1
3
(x
0
−2y
0
)e
−t
+
_
2e
2(t−π)
− 5e
−(t−π)
− 3 cos t − 6 sin t
_
H(t −π)
7.28
dx
dt
−x+y =2 sin t
_
1−H(t −π)
_
, 2x−
dy
dt
−y =0, x(0) = y(0) =0
A
NS x = (t +1) sin t − t cos t +
_
−(t −π +1) sin t + (t −π) cos t
_
H(t −π)
y = 2(sin t − t cos t) + 2
_
−sin t + (t −π) cos t
_
H(t −π)
354 7 systems of linear differential equations
7.29 2
dx
dt
+ x − 5
dy
dt
− 4y = 28e
t
H(t −2), 3
dx
dt
− 2x − 4
dy
dt
+ y = 0
x(0) = 2, y(0) = 0
A
NS x = −e
−t
+ 3e
t
+
_
5e
4−t
− (6t −7)e
t
_
H(t −2)
y = −e
−t
+ e
t
+
_
5e
4−t
− (2t +1)e
t
_
H(t −2)
The Matrix Method
Solve the following differential equations using the matrix method, in which
( · )
=d( · )/dt.
7.30 x
1
= x
1
− x
2
, x
2
= −4x
1
+ x
2
A
NS x
1
= C
1
e
−t
+ C
2
e
3t
, x
2
= 2C
1
e
−t
− 2C
2
e
3t
7.31 x
1
= x
1
− 3x
2
, x
2
= 3x
1
+ x
2
A
NS x
1
= e
t
(Acos 3t + Bsin 3t), x
2
= e
t
(Asin 3t − Bcos 3t)
7.32 x
1
= 5x
1
+ 3x
2
, x
2
= −3x
1
− x
2
, x
1
(0) = 1, x
2
(0) = −2
A
NS x
1
= (−3t +1)e
2t
, x
2
= (3t −2)e
2t
7.33 x
1
= 2x
1
− x
2
+ x
3
, x
2
= x
1
+ 2x
2
− x
3
, x
3
= x
1
− x
2
+ 2x
3
A
NS x
1
= C
2
e
2t
+ C
3
e
3t
, x
2
= C
1
e
t
+ C
2
e
2t
, x
3
= C
1
e
t
+ C
2
e
2t
+ C
3
e
3t
7.34 x
1
= 3x
1
− x
2
+ x
3
, x
2
= x
1
+ x
2
+ x
3
, x
3
= 4x
1
− x
2
+ 4x
3
A
NS x
1
= C
1
e
t
+ C
2
e
2t
+ C
3
e
5t
, x
2
= C
1
e
t
− 2C
2
e
2t
+ C
3
e
5t
x
3
= −C
1
e
t
− 3C
2
e
2t
+ 3C
3
e
5t
7.35 x
1
= 2x
1
+ x
2
, x
2
= x
1
+ 3x
2
− x
3
, x
3
= −x
1
+ 2x
2
+ 3x
3
A
NS x
1
= Ce
2t
+ e
3t
(Acos t +Bsin t), x
2
= e
3t
_
(A+B) cos t +(B−A) sin t
_
x
3
= Ce
2t
+ e
3t
_
(2A−B) cos t + (2B+A) sin t
_
7.36 x
1
= 3x
1
− 2x
2
− x
3
, x
2
= 3x
1
− 4x
2
− 3x
3
, x
3
= 2x
1
− 4x
2
A
NS x
1
=(C
1
+2C
2
)e
2t
+C
3
e
−5t
, x
2
=C
2
e
2t
+3C
3
e
−5t
, x
3
=C
1
e
2t
+2C
3
e
−5t
7.37 x
1
= x
1
− x
2
+ x
3
, x
2
= x
1
+ x
2
− x
3
, x
3
= −x
2
+ 2x
3
x
1
(0) = 1, x
2
(0) = −2, x
3
(0) = 0
A
NS x
1
= t e
t
+ e
2t
, x
2
= (t −2)e
t
, x
3
= (t −1)e
t
+ e
2t
7.38 x
1
= −x
1
+ x
2
− 2x
3
, x
2
= 4x
1
+ x
2
, x
3
= 2x
1
+ x
2
− x
3
A
NS x
1
=
_
C
1
+C
2
(t −1)
_
e
−t
, x
2
= −
_
2C
1
+C
2
(2t −1)
_
e
−t
+ 2C
3
e
t
x
3
= −(C
1
+C
2
t)e
−t
+ C
3
e
t
Problems 355
7.39 x
1
= 2x
1
+ x
2
+ 26 sin t, x
2
= 3x
1
+ 4x
2
A
NS x
1
=C
1
e
t
+C
2
e
5t
−10 cos t −11 sin t, x
2
=−C
1
e
t
+3C
2
e
5t
+9 cos t +6 sin t
7.40 x
1
= −x
1
+ 8x
2
+ 9t, x
2
= x
1
+ x
2
+ 3e
−t
A
NS x
1
= 2C
1
e
3t
+ 4C
2
e
−3t
− 3e
−t
+ t − 1, x
2
= C
1
e
3t
− C
2
e
−3t
− t
7.41 x
1
= −x
1
+ 2x
2
, x
2
= −3x
1
+ 4x
2
+
e
3t
e
2t
+1
A
NS x
1
= C
1
e
t
+ 2C
2
e
2t
− e
t
ln(e
2t
+1) + 2e
2t
tan
−1
e
t
x
2
= C
1
e
t
+ 3C
2
e
2t
− e
t
ln(e
2t
+1) + 3e
2t
tan
−1
e
t
7.42 x
1
= −4x
1
− 2x
2
+
2
e
t
−1
, x
2
= 6x
1
+ 3x
2
−
3
e
t
−1
A
NS x
1
=2
_
C
1
+ln
¸
¸
e
t
−1
¸
¸
_
e
−t
+C
2
, x
2
= −3
_
C
1
+ln
¸
¸
e
t
−1
¸
¸
_
e
−t
−2C
2
7.43 x
1
= x
1
+ x
2
+ e
2t
, x
2
= −2x
1
+ 3x
2
A
NS x
1
=e
2t
(C
1
cos t +C
2
sin t −1), x
2
=e
2t
_
(C
1
+C
2
)cos t +(C
2
−C
1
)sin t −2
_
7.44 x
1
= −x
1
− 5x
2
, x
2
= x
1
+ x
2
+
4
sin 2t
A
NS x
1
= (C
1
−2C
2
+10t) cos 2t + (2C
1
+C
2
) sin 2t −5 sin 2t ln
¸
¸
sin 2t
¸
¸
x
2
= −(C
1
+2t) cos 2t − (C
2
−4t) sin 2t + (2 cos 2t +sin 2t) ln
¸
¸
sin 2t
¸
¸
7.45 x
1
= 2x
1
+ x
2
+ 27t, x
2
= −x
1
+ 4x
2
A
NS x
1
= (C
1
+C
2
t)e
3t
− 12t − 5, x
2
=
_
C
1
+C
2
(t +1)
_
e
3t
− 3t − 2
7.46 x
1
= 3x
1
− x
2
+ e
t
, x
2
= 4x
1
− x
2
A
NS x
1
= e
t
_
C
1
+ C
2
(t +1) + t
2
+ t
_
, x
2
= e
t
_
2C
1
+ C
2
(2t +1) + 2t
2
_
7.47 x
1
= 3x
1
− 2x
2
, x
2
= 2x
1
− x
2
+ 35e
t
t
3/2
A
NS x
1
= e
t
_
C
1
+ C
2
(2t +1) − 8t
7/2
_
, x
2
= e
t
(C
1
+ 2C
2
t − 8t
7/2
+ 14t
5/2
)
7.48 x
1
= x
1
− x
2
+ x
3
, x
2
= x
1
+ x
2
− x
3
+ 6e
−t
, x
3
= 2x
1
− x
2
A
NS x
1
= −(C
1
+1)e
−t
+ C
2
e
t
+ C
3
e
2t
, x
2
= 3(C
1
−1)e
−t
+ C
2
e
t
x
3
= (5C
1
−1)e
−t
+ C
2
e
t
+ C
3
e
2t
7.49 x
1
= x
1
− 2x
2
− x
3
, x
2
= −x
1
+ x
2
+ x
3
+ 12t, x
3
= x
1
− x
3
A
NS x
1
= 3C
1
e
2t
+ 6t
2
+ 6t + (C
3
+3), x
2
= −2C
1
e
2t
+ C
2
e
−t
− 6
x
3
= C
1
e
2t
− 2C
2
e
−t
+ 6t
2
− 6t + (C
3
+9)
356 7 systems of linear differential equations
7.50 x
1
= −3x
1
+ 4x
2
− 2x
3
+ e
t
, x
2
= x
1
+ x
3
, x
3
= 6x
1
− 6x
2
+ 5x
3
A
NS x
1
= C
1
e
−t
+ (−t +C
2
+1)e
t
, x
2
= (−t +C
2
−1)e
t
+ C
3
e
2t
x
3
= −C
1
e
−t
−3e
t
+ 2C
3
e
2t
7.51 x
1
= x
1
− x
2
− x
3
+ 4e
t
, x
2
= x
1
+ x
2
, x
3
= 3x
1
+ x
3
A
NS x
1
= 2e
t
(−C
2
sin 2t + C
3
cos 2t), x
2
= e
t
(C
1
+1 + C
2
cos 2t + C
3
sin 2t)
x
3
= e
t
(−C
1
+ 3 + 3C
2
cos 2t + 3C
3
sin 2t)
7.52 x
1
= 2x
1
− x
2
+ 2x
3
, x
2
= x
1
+ 2x
3
, x
3
= −2x
1
+ x
2
− x
3
+ 4 sin t
A
NS x
1
= −2(C
2
+2t) cos t − 2(C
3
−1) sin t
x
2
= 2C
1
e
t
− 2(C
2
+2t) cos t − 2(C
3
−1) sin t
x
3
= C
1
e
t
+ (C
2
−C
3
+2t −1) cos t + (C
2
+C
3
+2t −1) sin t
7.53 x
1
= 4x
1
− x
2
− x
3
+ e
3t
, x
2
= x
1
+ 2x
2
− x
3
, x
3
= x
1
+ x
2
+ 2x
3
x
1
(0) = 1, x
2
(0) = 2, x
3
(0) = 3
A
NS x
1
= 5e
2t
+ 2(t −2)e
3t
, x
2
= 5e
2t
+ (t −3)e
3t
, x
3
= 5e
2t
+ (t −2)e
3t
7.54 x
1
= 2x
1
− x
2
− x
3
+ 2e
2t
, x
2
= 3x
1
− 2x
2
− 3x
3
, x
3
= −x
1
+ x
2
+ 2x
3
A
NS x
1
= C
1
+ (C
2
+C
3
)e
t
+ 3e
2t
, x
2
= 3C
1
+ C
2
e
t
+ 3e
2t
x
3
= −C
1
+ C
3
e
t
− e
2t
7.55 x
1
= 2x
1
− x
3
+ 24t, x
2
= x
1
− x
2
, x
3
= 3x
1
− x
2
− x
3
A
NS x
1
= t
4
+ 8t
3
+ (C
3
+12)t
2
+ (C
2
+2C
3
)t + (C
1
+C
2
+2C
3
)
x
2
= t
4
+ 4t
3
+ C
3
t
2
+ C
2
t + (C
1
+2C
3
)
x
3
= 2t
4
+ 12t
3
+ 2C
3
t
2
+ 2(C
2
+C
3
)t + (2C
1
+C
2
+2C
3
)
8
C H A P T E R
Applications of Systems of
Linear Differential Equations
In this chapter, examples are presented to illustrate engineering applications of
systems of linear differential equations.
8.1 Mathematical Modeling of Mechanical Vibrations
In many engineering applications, such as vibration of mechanical systems, the sys
tems are usually complex and have to be modeled as multiple degreesoffreedom
systems, resulting in systems of linear ordinary differential equations. Modeling a
complex engineering system as an appropriate, mathematically tractable problem
and establishing the governing differential equations are often the ﬁrst challenging
step. In this section, a free vibration problem of a simple two degreesoffreedom
system is ﬁrst considered to illustrated the basic procedure. The equations of mo
tion of a more complex problem, i.e., the vibration of an automobile, which is
modeled as a four degreesoffreedom system, are then established.
Example 8.1 — Two DegreesofFreedom System 8.1
Two uniform rods AB and CD of mass density per unit length ρ are hinged at A
and C. Rotational springs of stiffnesses κ
1
and κ
2
provide resistance to rotations
of end A and end C, respectively, as shown. The lengths of AB and CD are L
1
and L
2
, respectively. Rod AB carries a concentrated mass M at end B. The rods
are connected by a spring of stiffness k. When the rods are hanging freely, they are
vertical and there is no force in the spring.
357
358 8 applications of systems of linear differential equations
The motion of the system is described by two angles θ
1
and θ
2
. Consider only
small oscillations, i.e.,
¸
¸
θ
1
¸
¸
1 and
¸
¸
θ
2
¸
¸
1, and neglect the effect of gravity.
1. Set up the differential equations governing the angles of rotation θ
1
and θ
2
.
2. For the special case when L
1
=L, L
2
=2L, M=ρL, κ
1
=κ
2
=0, and ini
tial conditions θ
1
(0) =θ
10
, θ
2
(0) =θ
20
,
˙
θ
1
(0) =
˙
θ
2
(0) =0, determine the re
sponses of the free vibration.
3. For the special case when L
1
=L
2
=L, M=0, κ
1
=κ
2
=kL
2
, discuss the re
sponses of the free vibration.
A
A
B
D
B
D
C
C
ρ, L
1
k(x
2
−
x
1
) k(x
1
−
x
2
)
ρ, L
2
k
M
θ
1
θ
2
R
Ax
R
Ay
R
Cx
R
Cy
x
1
x
2
Mx
1
J
1
θ
1
κ
1
θ
1
κ
1
J
2
θ
2
κ
2
θ
2
κ
2
1. Whenthe rotationof rod AB about point A is θ
1
, the horizontal displacement of
point B is x
1
=L
1
sin θ
1
≈L
1
θ
1
. The mass of rod AB is m
1
=ρL
1
and its moment
of inertia about point A is
J
1
=
1
3
m
1
L
2
1
=
1
3
ρL
3
1
.
When the rotation of rod CD about point C is θ
2
, the horizontal displacement of
point D is x
2
=L
2
sin θ
2
≈L
2
θ
2
. The mass of rod CD is m
2
=ρL
2
and its moment
of inertia about point C is
J
2
=
1
3
m
2
L
2
2
=
1
3
ρL
3
2
.
To establish the differential equations governing the angles θ
1
and θ
2
, consider
the freebody diagrams of rods AB and CD, respectively. D’Alembert’s Principle
is applied to set up the equation of motion.
❧ Rod AB. To ﬁnd the spring force applied at end B, observe the spring at end D:
end B moves toward the right by x
1
so that the spring is compressed and the spring
force pushes end B toward the left. Since end D also moves toward the right by x
2
,
the net compression of the spring is x
1
−x
2
; hence, the spring force is k(x
1
−x
2
).
8.1 mathematical modeling of mechanical vibrations 359
Since the acceleration of point B is ¨ x
1
=L
1
¨
θ
1
toward the right, the inertia force
of mass M is ML
1
¨
θ
1
toward the left.
Since the angular acceleration of rod AB about point A is
¨
θ
1
counterclockwise,
the inertia moment of rod AB is J
1
¨
θ
1
clockwise.
Replace the hinge at A by two reaction force components R
Ax
and R
Ay
. The
rotational spring at A provides a clockwise resisting moment κ
1
θ
1
.
Applying D’Alembert’s Principle and summing up moments about point A yield
M
A
=0: J
1
¨
θ
1
+ ML
1
¨
θ
1
· L
1
+ k(x
1
−x
2
) · L
1
+ κ
1
θ
1
= 0,
∴
_
1
3
ρL
2
1
+ML
1
_
¨
θ
1
+ k(L
1
θ
1
−L
2
θ
2
) +
κ
1
L
1
θ
1
= 0.
❧ Rod CD. To ﬁnd the spring force applied at end D, observe the spring at end B:
end D moves toward the right by x
2
so that the spring is extended and the spring
force pulls end D toward the left. Since end B also moves toward the right by x
1
,
the net extension of the spring is x
2
−x
1
; hence, the spring force is k(x
2
−x
1
).
Since the angular acceleration of rod CD about point C is
¨
θ
2
counterclockwise,
the inertia moment of rod CD is J
2
¨
θ
2
clockwise.
Replace the hinge at C by two reaction force components R
Cx
and R
Cy
. The
rotational spring at C provides a clockwise resisting moment κ
2
θ
2
.
Applying D’Alembert’s Principle and summing up moments about point C give
M
C
=0: J
2
¨
θ
2
+ k(x
2
−x
1
) · L
2
+ κ
2
θ
2
= 0,
∴
1
3
ρL
2
2
¨
θ
2
+ k(L
2
θ
2
−L
1
θ
1
) +
κ
2
L
2
θ
2
= 0.
2. When L
1
=L, L
2
=2L, M=ρL, κ
1
=κ
2
=0, the equations of motion become
_
1
3
ρL
2
+(ρL)L
_
¨
θ
1
+ k
_
Lθ
1
−(2L)θ
2
_
= 0,
1
3
ρ(2L)
2
¨
θ
2
+ k
_
(2L)θ
2
−Lθ
1
_
= 0,
∴
¨
θ
1
+
ω
2
0
3
(θ
1
−2θ
2
) = 0,
¨
θ
2
+
ω
2
0
3
(2θ
2
−θ
1
) = 0,
ω
2
0
=
9k
4ρL
.
Using the Doperator, D≡d/dt, the equations of motion can be written as
_
D
2
+
ω
2
0
3
_
θ
1
−
2ω
2
0
3
θ
2
= 0,
−
ω
2
0
3
θ
1
+
_
D
2
+
2ω
2
0
3
_
θ
2
= 0.
360 8 applications of systems of linear differential equations
The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
¸
¸
¸
¸
¸
¸
D
2
+
ω
2
0
3
−
2ω
2
0
3
−
ω
2
0
3
D
2
+
2ω
2
0
3
¸
¸
¸
¸
¸
¸
¸
¸
= D
4
+ ω
2
0
D
2
.
The characteristic equation is then given by φ(λ) =0, which gives
λ
4
+ ω
2
0
λ
2
= 0 =⇒ λ
2
(λ
2
+ ω
2
0
) = 0 =⇒ λ =0, 0, ±i ω
0
.
Hence, the complementary solution for θ
2
is given by
θ
2
(t) = C
0
+ C
1
t + Acos ω
0
t + Bsin ω
0
t.
The complementary solution of θ
1
can be obtained as
θ
1
(t) =
3
ω
2
0
_
D
2
+
2ω
2
0
3
_
θ
2
(t)
=
3
ω
2
0
(−Aω
2
0
cos ω
0
t −Bω
2
0
sin ω
0
t) + 2(C
0
+C
1
t +Acos ω
0
t +Bsin ω
0
t)
= 2(C
0
+ C
1
t) − Acos ω
0
t − Bsin ω
0
t.
Hence, the complementary solutions or the responses of free vibrationof the system
are given by
θ
1
(t) = 2(C
0
+ C
1
t) − (Acos ω
0
t + Bsin ω
0
t),
θ
2
(t) = (C
0
+ C
1
t) + (Acos ω
0
t + Bsin ω
0
t),
. ,, . . ,, .
Nonoscillatory motion Harmonic oscillation at ω
0
in which the ﬁrst two terms correspond to nonoscillatory motion with θ
1
=2θ
2
(the length of the spring is not changed); whereas the last two terms correspond to
harmonic oscillation with θ
1
= −θ
2
at circular frequency ω
0
. These two motions
are shown schematically in the following ﬁgure.
A
Nonoscillatory motion Harmonic oscillation at frequency ω
0
C
A
C
A
C
θ
1
θ
1
θ
2
θ
2
θ
1
θ
2
8.1 mathematical modeling of mechanical vibrations 361
For the initial conditions θ
1
(0) =θ
10
, θ
2
(0) =θ
20
,
˙
θ
1
(0) =
˙
θ
2
(0) =0, and noting
˙
θ
1
(t) =2C
1
+Aω
0
sin ω
0
t −Bω
0
cos ω
0
t,
˙
θ
2
(t) =C
1
−Aω
0
sin ω
0
t +Bω
0
cos ω
0
t,
one has
θ
1
(0) = 2C
0
− A = θ
10
, θ
2
(0) = C
0
+ A = θ
20
,
˙
θ
1
(0) = 2C
1
− Bω
0
= 0,
˙
θ
2
(0) = C
1
+ Bω
0
= 0,
which gives
C
0
=
1
3
(θ
10
+θ
20
), C
1
= 0, A =
1
3
(2θ
20
−θ
10
), B = 0.
The responses of the free vibration are
θ
1
(t) =
2
3
(θ
10
+θ
20
) −
1
3
(2θ
20
−θ
10
) cos ω
0
t,
θ
2
(t) =
1
3
(θ
10
+θ
20
) +
1
3
(2θ
20
−θ
10
) cos ω
0
t.
3. When L
1
=L, L
2
=L, M=0, κ
1
=κ
2
=kL
2
, the equations of motion become
1
3
ρL
2
¨
θ
1
+ k(Lθ
1
−Lθ
2
) +
kL
2
L
θ
1
= 0,
1
3
ρL
2
¨
θ
2
+ k(Lθ
2
−Lθ
1
) +
kL
2
L
θ
2
= 0,
which yield
¨
θ
1
+ ω
2
(2θ
1
−θ
2
) = 0,
¨
θ
2
+ ω
2
(2θ
2
−θ
1
) = 0,
ω
2
=
3k
ρL
.
Using the Doperator, D≡d/dt, the equations of motion can be written as
(D
2
+2ω
2
)θ
1
− ω
2
θ
2
= 0,
−ω
2
θ
1
+ (D
2
+2ω
2
)θ
2
= 0.
The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
¸
¸
¸
D
2
+2ω
2
−ω
2
−ω
2
D
2
+2ω
2
¸
¸
¸
¸
¸
= D
4
+ 4ω
2
D
2
+ 3ω
4
.
The characteristic equation is then given by φ(λ) =0, which gives
λ
4
+ 4ω
2
λ
2
+ 3ω
4
= 0 =⇒ (λ
2
+ω
2
)(λ
2
+3ω
2
) = 0 =⇒ λ = ±i ω, ±i
√
3ω.
Hence, the complementary solution for θ
1
is given by
θ
1
(t) = A
1
cos ωt + B
1
sin ωt + A
2
cos
√
3ωt + B
2
sin
√
3ωt.
362 8 applications of systems of linear differential equations
The complementary solution of θ
2
can be obtained as
θ
2
(t) =
1
ω
2
(D
2
+2ω
2
)θ
1
(t)
=
1
ω
2
_
(−Aω
2
cos ωt −Bω
2
sin ωt − A
2
· 3ω
2
cos
√
3ωt − B
2
· 3ω
2
sin
√
3ωt)
+ 2ω
2
(A
1
cos ωt + B
1
sin ωt + A
2
cos
√
3ωt + B
2
sin
√
3ωt)
_
= A
1
cos ωt + B
1
sin ωt − A
2
cos
√
3ωt − B
2
sin
√
3ωt.
The complementary solutions or the responses of free vibration of the system are
θ
1
(t) = (A
1
cos ωt + B
1
sin ωt) + (A
2
cos
√
3ωt + B
2
sin
√
3ωt),
θ
2
(t) = (A
1
cos ωt + B
1
sin ωt) − (A
2
cos
√
3ωt + B
2
sin
√
3ωt).
. ,, . . ,, .
First Mode Oscillation at ω Second Mode Oscillation at
√
3ω
The four constants A
1
, B
1
, A
2
, and B
2
are determined from the initial conditions
θ
1
(0), θ
2
(0),
˙
θ
1
(0), and
˙
θ
1
(0). The ﬁrst two terms in the solutions correspond to
the ﬁrst mode of harmonic oscillation at circular frequency ω
1
=ω with θ
1
=θ
2
;
whereas the last two terms correspond to the second mode of oscillation at circular
frequency ω
2
=
√
3ω with θ
1
= −θ
2
. Note that natural frequencies are always or
dered in ascending order, i.e., ω
1
<ω
2
. These two motions are shown schematically
in the following ﬁgure.
A C
First Mode
Oscillation at frequency ω, θ
1
=θ
2
Second Mode
Oscillation at frequency 3ω, θ
1
=−θ
2
θ
1
θ
2
A C
θ
1
θ
2
A C
θ
1
θ
2
A C
θ
1
θ
2
Example 8.2 — Vibration of an Automobile 8.2
To study the dynamic response of an automobile on a wavy road, a simpliﬁed
mechanical model shown in Figures 8.1(a) and (b) is used.
❧ The body of the vehicle is modeled as a rigid body of mass m and moment of
inertia J about its center of gravity C.
❧ The front shock absorbers are modeled as a spring of stiffness k
f
and a damper
of damping coefﬁcient c
f
. The rear shock absorbers are modeled as a spring of
stiffness k
r
and a damper of damping coefﬁcient c
r
.
❧ The front axle is modeled as a point mass m
f
. The rear axle is modeled as a point
mass m
r
.
8.1 mathematical modeling of mechanical vibrations 363
A B
L
f
L
r
k
r
k
rt
m
r
c
r
k
f
k
ft
m
f
c
f
C
k
r
(y
A
−y
r
)
k
r
(y
r
−y
A
)
k
rt
(y
r
−y
r0
)
c
r
(y
A
−y
r
)
c
r
(y
r
−y
A
)
my
m, J
m
r
y
r
k
r
k
rt
m
r
c
r
k
f
k
ft
m
f
c
f
y
r
A
B
y
A
y
B
C
y
θ
y
r0
y
f0
y
f
O
Reference
Positions
Body
Shock absorbers
Axle
Tires
(a)
(b)
(c)
x
y
0
m
r
A
B
L
f
L
r
C
k
f
(y
B
−y
f
)
k
f
(y
f
−y
B
)
k
ft
(y
f
−y
f0
)
c
f
(y
B
−y
f
)
c
f
(y
f
−y
B
)
m
f
y
f
m
f
J θ
U
Figure 8.1 An automobile moving on a wavy road.
364 8 applications of systems of linear differential equations
❧ The front tires are modeled as a spring of stiffness k
ft
. The rear tires are modeled
as a spring of stiffness k
rt
.
❧ The front axle, shock absorbers, and tires support the vehicle body at point B
with BC=L
f
. The rear axle, shock absorbers, and tires support the vehicle body
at point A with AC=L
r
.
❧ The wavy road is described as y
0
(x). The vehicle travels at a constant speed U.
At time t =0, the front tires of the vehicle is at the origin of the road, i.e., x =0.
❧ The vibration of the vehicle is described by four variables: the vertical displace
ment y(t) andthe rotation θ(t) about the center of gravity C of the vehicle body,
the vertical displacement y
f
(t) and y
r
(t) of the front and rear axles. Vertical
displacements are positive upward, and rotations are positive counterclockwise.
Hence, the vehicle is modeled as a four degreesoffreedom system.
Assuming small angle of rotation θ, derive the equations of motion of the vehicle.
Since the front tires are at x =0 whentime t =0, andthe vehicle travels at a constant
speed U, the front tires are at x =Ut at time t. The vertical displacement of the
front tires is y
f 0
(t) = y
0
(Ut). At time t, the rear tires are at x =Ut −AB=Ut −L,
L=L
f
+L
r
, and the vertical displacement of the rear tires is y
r0
(t) = y
0
(Ut −L).
Relative to the reference positions, impose vertical displacement y(t) to the
center of gravity C and a rotation θ(t) about C to the vehicle body, vertical
displacement y
f
(t) to the front axle (mass m
f
), and vertical displacement y
r
(t) to
the rear axle (mass m
r
) as shown in Figure 8.1(b).
❧ The Body of the Vehicle. The vertical upward displacement of point B is
y
B
= y+L
f
sin θ ≈ y+L
f
θ for small angle of rotation. To ﬁnd the spring force
applied at B by the front shock absorbers, observe spring k
f
at the front axle: point
B moves upward by y
B
so that the spring k
f
is extended and the spring force pulls
point B downward. Since the front axle also moves upward by y
f
, the net extension
of the spring is y
B
−y
f
; hence the spring force is k
f
( y
B
−y
f
) downward.
Since the damping force is proportional to the relative velocity of the two ends
of the damper, the damping force applied at point B by the front shock absorber is
c
f
( ˙ y
B
− ˙ y
f
) downward.
Similarly, the vertical upward displacement of point A is y
A
= y−L
r
θ. The
spring force and damping force applied at point A by the rear shock absorbers are
k
r
( y
A
−y
r
) and c
r
( ˙ y
A
− ˙ y
r
), respectively, both downward.
Since the acceleration of the center of gravity C is ¨ y upward, the inertia force is
m¨ y downward.
Since the angular accelerationof the vehicle body about C is
¨
θ counterclockwise,
the inertia moment is J
¨
θ clockwise.
8.1 mathematical modeling of mechanical vibrations 365
Applying D’Alembert’s Principle, the freebody of the vehicle body as shown in
Figure 8.1(c) is in dynamic equilibrium. Hence
↓
F
y
=0: m¨ y + c
f
( ˙ y
B
− ˙ y
f
) + k
f
( y
B
−y
f
) + c
r
( ˙ y
A
− ˙ y
r
) + k
r
( y
A
−y
r
) =0,
M
C
=0: J
¨
θ +
_
c
f
( ˙ y
B
− ˙ y
f
) + k
f
( y
B
−y
f
)
_
L
f
−
_
c
r
( ˙ y
A
− ˙ y
r
) + k
r
( y
A
−y
r
)
_
L
r
=0.
❧ The Front Axle. To determine the spring force applied on the front axle m
f
by the front shock absorbers, observe spring k
f
at point B: the front axle m
f
moves upward by y
f
so that the spring k
f
is compressed and the spring force
pushes the front axle m
f
downward. Since point B also moves upward by y
B
,
the net compression of the spring is y
f
−y
B
; hence the spring force is k
f
( y
f
−y
B
)
downward. Similarly, the damping force applied on the front axle m
f
by the front
shock absorber is c
f
( ˙ y
f
− ˙ y
B
) downward.
To determine the spring force applied on the front axle m
f
by the front tires,
observe spring k
ft
at the bottomof the front tires: the front axle m
f
moves upward
by y
f
so that the spring k
ft
is extended and the spring force pulls the front axle m
f
downward. Since the bottom of the front tires also moves upward by y
f 0
, the net
extension of the spring is y
f
−y
f 0
; hence the spring force is k
ft
( y
f
−y
f 0
) downward.
Since the acceleration of the front axle m
f
is ¨ y
f
upward, the inertia force is m
f
¨ y
f
downward.
Applying D’Alembert’s Principle, the freebody of the front axle as shown in
Figure 8.1(c) is in dynamic equilibrium. Hence
↓
F
y
=0: m
f
¨ y
f
+ c
f
( ˙ y
f
− ˙ y
B
) + k
f
( y
f
−y
B
) + k
ft
( y
f
−y
f 0
) =0.
❧ The Rear Axle. Similarly, the equation of motion of the rear axle is
↓
F
y
=0: m
r
¨ y
r
+ c
r
( ˙ y
r
− ˙ y
A
) + k
r
( y
r
−y
A
) + k
rt
( y
r
−y
r0
) =0.
Noting y
B
=y+L
f
θ, y
A
=y−L
r
θ, y
f 0
(t) =y
0
(Ut), y
r0
(t) =y
0
(Ut −L), the equa
tions of motion of the automobile are four secondorder differential equations
m¨ y + (c
f
+c
r
) ˙ y + (c
f
L
f
−c
r
L
r
)
˙
θ − c
f
˙ y
f
− c
r
˙ y
r
+(k
f
+k
r
) y + (k
f
L
f
−k
r
L
r
)θ − k
f
y
f
− k
r
y
r
= 0,
J
¨
θ + (c
f
L
f
−c
r
L
r
) ˙ y + (c
f
L
2
f
+c
r
L
2
r
)
˙
θ − c
f
L
f
˙ y
f
+ c
r
L
r
˙ y
r
+(k
f
L
f
−k
r
L
r
) y + (k
f
L
2
f
+k
r
L
2
r
)θ − k
f
L
f
y
f
+ k
r
L
r
y
r
= 0,
m
f
¨ y
f
− c
f
˙ y − c
f
L
f
˙
θ + c
f
˙ y
f
− k
f
y − k
f
L
f
θ + (k
f
+k
ft
) y
f
= k
ft
y
0
(Ut),
m
r
¨ y
r
− c
r
˙ y + c
r
L
r
˙
θ + c
r
˙ y
r
− k
r
y + k
r
L
r
θ + (k
r
+k
rt
) y
r
= k
rt
y
0
(Ut −L).
366 8 applications of systems of linear differential equations
8.2 Vibration Absorbers or Tuned Mass Dampers
In engineering applications, many systems can be modeled as single degreeof
freedom systems. For example, a machine mounted on a structure can be modeled
using a massspringdamper system, in which the machine is considered to be rigid
with mass m and the supporting structure is equivalent to a spring k and a damper
c, as shown in Figure 8.2. The machine is subjected to a sinusoidal force F
0
sin t,
which can be an externally applied load or due to imbalance in the machine.
x(t)
Supporting
Structure
F
0
sint
c
m
k
Machine
Supporting Structure
Mathematical
Modeling
Figure 8.2 A machine mounted on a structure.
From Chapter 5 on the response of a single degreeoffreedom system, it is well
known that when the excitation frequency is close to the natural frequency of
the system ω
0
=
_
k/m, vibration of large amplitude occurs. In particular, when
the system is undamped, i.e., c =0, resonance occurs when =ω
0
, in which the
amplitude of the response grows linearly with time.
To reduce the vibration of the system, a vibration absorber or a tuned mass
damper (TMD), which is an auxiliary massspringdamper system, is mounted
on the main system as shown in Figure 8.3(a). The mass, spring stiffness, and
damping coefﬁcient of the viscous damper are m
a
, k
a
, and c
a
, respectively, where
the subscript “a” stands for “auxiliary.”
To derive the equation of motion of the main mass m, consider its freebody
diagram as shown in Figure 8.3(b). Since mass m moves upward, spring k is
extended and spring k
a
is compressed.
❧ Because of the displacement x of mass m, the extension of spring k is x. Hence
the spring k exerts a downward force kx and the damper c exerts a downward
force c ˙ x on mass m.
❧ Because the mass m
a
also moves upward a distance x
a
, the net compression in
spring k
a
is x−x
a
. Hence the spring k
a
and damper c
a
exert downward forces
k
a
(x−x
a
) and c
a
(˙ x−˙ x
a
), respectively, on mass m.
Newton’s Second Law requires
↑ m¨ x =
F : m¨ x = −kx − c ˙ x − k
a
(x−x
a
) − c
a
(˙ x−˙ x
a
) + F
0
sin t,
8.2 vibration absorbers or tuned mass dampers 367
x(t)
(a) (b)
Vibration
Absorber
x
a
(t)
k
a
(x
a
−x)
k
a
(x−x
a
)
kx cx
F
0
sint
c
m
k
c
a
m
a
k
a
x(t)
x
a
(t)
F
0
sint
c
m
k
c
a
m
a
k
a
c
a
(x
a
−x)
c
a
(x−x
a
)
Figure 8.3 A vibration absorber mounted on the main system.
or
m¨ x + (c +c
a
) ˙ x + (k+k
a
)x − c
a
˙ x
a
− k
a
x
a
= F
0
sin t.
Similarly, consider the freebody diagram of mass m
a
. Since mass m
a
moves
upward a distance x
a
(t), spring k
a
is extended. The net extension of spring k
a
is
x
a
−x. Hence, the spring k
a
and damper c
a
exert downward forces k
a
(x
a
−x) and
c
a
(˙ x
a
−˙ x), respectively. Applying Newton’s Second Law gives
↑ m
a
¨ x
a
=
F : m
a
¨ x
a
= −k
a
(x
a
−x) − c
a
(˙ x
a
−˙ x),
∴ m
a
¨ x
a
+ c
a
˙ x
a
+ k
a
x
a
− c
a
˙ x − k
a
x = 0.
The equations of motion can be written using the Doperator as
_
mD
2
+ (c +c
a
)D + (k+k
a
)
_
x − (c
a
D+k
a
)x
a
= F
0
sin t,
−(c
a
D+k
a
)x + (m
a
D
2
+c
a
D+k
a
)x
a
= 0.
Because of the existence of damping, the responses of free vibration (com
plementary solutions) decay exponentially and approach zero as time increases.
Hence, it is practically more important and useful to study responses of forced
vibration (particular solutions). The determinant of the coefﬁcient matrix is
φ(D) =
¸
¸
¸
¸
¸
mD
2
+ (c +c
a
)D + (k+k
a
) −(c
a
D+k
a
)
−(c
a
D+k
a
) m
a
D
2
+c
a
D+k
a
¸
¸
¸
¸
¸
=
_
mD
2
+ (c +c
a
)D + (k+k
a
)
_
(m
a
D
2
+c
a
D+k
a
) − (c
a
D+k
a
)
2
=
_
(mD
2
+k)(m
a
D
2
+k
a
) + k
a
m
a
D
2
+ c
a
c D
2
_
+
_
c
a
(mD
2
+k) + c(m
a
D
2
+k
a
) + c
a
m
a
D
2
_
D,
368 8 applications of systems of linear differential equations
and
1
=
¸
¸
¸
¸
¸
F
0
sin t −(c
a
D+k
a
)
0 m
a
D
2
+c
a
D+k
a
¸
¸
¸
¸
¸
= (m
a
D
2
+c
a
D+k
a
)F
0
sin t.
Applying Cramer’s Rule yields the particular solution x
P
(t) or response of forced
vibration due to the excitation
x
P
(t) =
1
φ(D)
= F
0
(m
a
D
2
+k
a
) + c
a
D
_
_
(mD
2
+k)(m
a
D
2
+k
a
) + k
a
m
a
D
2
+ c
a
c D
2
_
+
_
c
a
(mD
2
+k) + c(m
a
D
2
+k
a
) + c
a
m
a
D
2
_
D
_
sin t
= F
0
(k
a
−m
a
2
) + c
a
D
_
_
(k−m
2
)(k
a
−m
a
2
) − k
a
m
a
2
− c
a
c
2
_
+
_
c
a
(k−m
2
) + c(k
a
−m
a
2
) − c
a
m
a
2
_
D
_
sin t
Theorem 3 of Chapter 4: replace D
2
by −
2
.
= F
0
(k
a
−m
a
2
) + c
a
D
A + BD
sin t,
where
A = (k−m
2
)(k
a
−m
a
2
) − k
a
m
a
2
− c
a
c
2
,
B = c
a
(k−m
2
) + c(k
a
−m
a
2
) − c
a
m
a
2
.
Hence
x
P
(t) = F
0
_
(k
a
−m
a
2
) + c
a
D
_
(A − BD)
(A + BD)(A − BD)
sin t
= F
0
(k
a
−m
a
2
)A − c
a
BD
2
+
_
−(k
a
−m
a
2
)B + c
a
A
_
D
A
2
− B
2
D
2
sin t
= F
0
_
(k
a
−m
a
2
)A + c
a
B
2
_
sin t +
_
−(k
a
−m
a
2
)B + c
a
A
_
cos t
A
2
+ B
2
2
Replace D
2
by −
2
and then evaluate using Dsin t =cos t.
= F
0
_
_
(k
a
−m
a
2
)A + c
a
B
2
_
2
+
_
−(k
a
−m
a
2
)B + c
a
A
_
2
2
A
2
+ B
2
2
sin(t +ϕ)
Use a sin θ +b cos θ =
_
a
2
+b
2
sin(θ +ϕ), ϕ = tan
−1
(b/a).
= F
0
_
_
(k
a
−m
a
2
)
2
+ c
2
a
2
_
(A
2
+ B
2
2
)
A
2
+ B
2
2
sin(t +ϕ)
= F
0
_
(k
a
−m
a
2
)
2
+ c
2
a
2
A
2
+ B
2
2
sin(t +ϕ).
8.2 vibration absorbers or tuned mass dampers 369
The Dynamic Magniﬁcation Factor (DMF) for mass mis
DMF =
¸
¸
x
P
(t)
¸
¸
max
x
static
=
F
0
_
(k
a
−m
a
2
)
2
+ c
2
a
2
A
2
+ B
2
2
F
0
k
= k
_
(k
a
−m
a
2
)
2
+ c
2
a
2
A
2
+ B
2
2
.
Adopting the following notations
ω
2
0
=
k
m
, c =2mζ ω
0
, r =
ω
0
, μ=
m
a
m
, ω
2
a
=
k
a
m
a
, c
a
=2m
a
ζ
a
ω
0
, r
a
=
ω
a
ω
0
,
one has
(k
a
−m
a
2
)
2
+ c
2
a
2
= m
a
_
k
a
m
a
−
2
_
2
+ (2m
a
ζ
a
ω
0
)
2
2
= m
2
a
ω
4
0
_
_
ω
2
a
−
2
ω
2
0
_
2
+
_
2ζ
a
ω
0
ω
2
0
_
2
_
= m
2
a
ω
4
0
_
(r
2
a
−r
2
)
2
+ (2ζ
a
r)
2
_
,
A
2
+B
2
2
=
_
(k−m
2
)(k
a
−m
a
2
) − k
a
m
a
2
− c
a
c
2
_
2
+
_
c
a
(k−m
2
) + c(k
a
−m
a
2
) − c
a
m
a
2
_
2
2
=
_
m
a
m
_
k
m
−
2
__
k
a
m
a
−
2
_
−m
2
a
k
a
m
a
2
−(2m
a
ζ
a
ω
0
)(2mζ ω
0
)
2
_
2
+
_
(2m
a
ζ
a
ω
0
)m
_
k
m
−
2
_
+(2mζ ω
0
)m
a
_
k
a
m
a
−
2
_
−(2m
a
ζ
a
ω
0
)m
a
2
_
2
2
=
_
m
a
mω
4
0
_
ω
2
0
−
2
ω
2
0
__
ω
2
a
−
2
ω
2
0
_
−m
2
a
ω
4
0
ω
2
a
ω
2
0
2
ω
2
0
−m
a
mω
4
0
· 4ζ
a
ζ
2
ω
2
0
_
2
+
_
m
a
mω
3
0
· 2ζ
a
_
ω
2
0
−
2
ω
2
0
_
+m
a
mω
3
0
· 2ζ
_
ω
2
a
−
2
ω
2
0
_
−m
2
a
ω
3
0
· 2ζ
a
2
ω
2
0
_
2
2
= m
2
a
m
2
ω
8
0
_
_
(1−r
2
)(r
2
a
−r
2
) − μr
2
a
r
2
− 4ζ
a
ζ r
2
_
2
+4r
2
_
ζ
a
(1−r
2
) + ζ(r
2
a
−r
2
) − μζ
a
r
2
_
2
_
.
The Dynamic Magniﬁcation Factor becomes
DMF=
_
(r
2
a
−r
2
)
2
+ (2ζ
a
r)
2
_
(1−r
2
)(r
2
a
−r
2
)−μr
2
a
r
2
−4ζ
a
ζ r
2
_
2
+4r
2
_
ζ
a
(1−r
2
−μr
2
)+ζ(r
2
a
−r
2
)
_
2
_1
2
.
For the special case when μ=0, r
a
=0, ζ
a
=0, the Dynamic Magniﬁcation
Factor reduces to
DMF =
1
_
(1−r
2
)
2
+ (2ζ r)
2
,
370 8 applications of systems of linear differential equations
which recovers the DMF of a single degreeoffreedomsystem, i.e., the main system
without the auxiliary vibration absorber or TMD.
0
2
4
6
8
10
12
14
16
0.6 0.7 0.8 0.9 1
Without
vibration
absorber
With
vibration
absorber
D
y
n
a
m
i
c
M
a
g
n
i
f
i
c
a
t
i
o
n
F
a
c
t
o
r
1.1 1.2 1.3
1
2
ζ
a
=0
μ=0.05
ζ=0
r
a
=1
ζ
a
=0.1
ω
0
Figure 8.4 Dynamic Magniﬁcation Factor for ζ =0.
0
2
4
6
8
10
12
14
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
Without
vibration
absorber
With
vibration
absorber
D
y
n
a
m
i
c
M
a
g
n
i
f
i
c
a
t
i
o
n
F
a
c
t
o
r
ζ
a
=0
μ=0.05
ζ=0.04
r
a
=1
ζ
a
=0.1
ω
0
Figure 8.5 Dynamic Magniﬁcation Factor for ζ =0.04.
The Dynamic Magniﬁcation Factors for an undamped main system, i.e., ζ =0,
are shown in Figure 8.4. Without the vibration absorber or TMD, the single degree
offreedom system is in resonance when r =1 or =ω
0
, where the amplitude of
the response grows linearly with time or DMF approaches inﬁnite.
In order to reduce the vibration of the main system at resonance, a vibration
absorber or TMDis attached to the main mass m. The vibration absorber is usually
tuned so that ω
a
=ω
0
or r
a
=1, hence the name tuned mass damper. In practice,
8.2 vibration absorbers or tuned mass dampers 371
the mass of the vibration absorber or TMD is normally much smaller than that of
the main mass, i.e., m
a
m or μ
1; in Figure 8.4, μ is taken as 1/20=0.05.
If the vibration absorber or TMDis undamped, i.e., ζ
a
=0, then DMF=0 when
=ω
0
, meaning that the vibration absorber eliminates vibration of the main mass
m at the resonant frequency =ω
0
. However, it is seen that the vibration absorber
or TMD introduces two resonant frequencies
1
and
2
, at which the amplitude
of vibration of the main mass m is inﬁnite. In practice, the excitation frequency
must be kept away from the frequencies
1
and
2
.
In order not to introduce extra resonant frequencies, vibration absorbers or
TMD are usually damped. A typical result of DMF is shown in Figure 8.4 for
ζ
a
=0.1. It is seen that the vibration of the main mass m is effectively suppressed
for all excitation frequencies. By varying the value of ζ
a
, an optimal vibration
absorber can be designed.
When the main system is also damped, typical results of DMF are shown in
Figure 8.5. Similar conclusions can also be drawn.
As an application, Figure 8.6 shows a schematic illustration of the tuned mass
damper (TMD) in Taipei 101. Taipei 101 is a landmark skyscraper located in Taipei,
Taiwan. The building has 101 ﬂoors above ground and 5 ﬂoors underground; the
height of the top of its spire or Pinnacle is 509.2 m. It has been extolled as one of
the Seven NewWonders of the World and Seven Wonders of Engineering.
Taiwan lies in one of the most earthquakeprone regions of the world. A catas
trophic earthquake on September 21, 1999, measured 7.6 on the Richter scale,
Cable
Bumper
system
Mass
block
Snubber
hydraulic
viscous
damper
Primary
hydraulic
viscous
damper
Figure 8.6 Tuned mass damper (TMD) in Taipei 101.
372 8 applications of systems of linear differential equations
killed over 2000 people, and seriously injured over 11,000 people in northern and
central Taiwan. From summer to fall, Taiwan is also affected by typhoons.
Taipei 101 is designed to withstand the strongest earthquakes likely to occur in a
2500year cycle and gale winds of 60 m/sec.
A tuned mass damper (TMD), the largest and heaviest of its type in the world,
is installed between the 87th and 91st ﬂoors. It consists of a sphere of 5.5 m in
diameter constructed from 41 steel plates with a total weight of 660 metric tons
(equivalent to 0.26 percent of building weight). The sphere is suspended from four
groups of steel cables (each group has four 9cmdiameter cables) and supported by
eight primary hydraulic viscous dampers as shown in Figure 8.6. A bumper system
of eight snubber hydraulic viscous dampers placed at the bottom of the sphere
absorbs vibration impacts, particularly in major typhoons or earthquakes where
the movement of the TMDexceeds 1.5 m. The period of the TMDis tuned to be the
same as that of the building, approximately 7 sec. The TMD helps to stabilize the
tower to withstand earthquakes measuring above the magnitude of 7.0 and reduces
the building’s vibration by as much as 45% in strong winds.
Another two tuned mass dampers, each weighing 6 metric tons, are installed at
the Pinnacle to reduce windinduced fatigue of its steel structure, which vibrates
approximately 180,000 cycles every year. The two PinnacleTMD’s reduce the
vibration of the Pinnacle by 40%.
8.3 An Electric Circuit
Consider the electric circuit shown in the following ﬁgure. The differential equa
tions governing i
C
(t) and i
L
(t) for t >0 are derived as follows.
V
1
(t)H(t) V
2
(t)H(t)
L
R
C
v
R
v
L
v
C
i
C
i
R
i
L
1
❧ For t 0
−
, V
1
(t)H(t) =0 and V
2
(t)H(t) =0. The circuit is source free and
v
C
(0
−
) =0, i
L
(0
−
) =0.
❧ For t 0
+
, V
1
(t)H(t) =V
1
(t) and V
2
(t)H(t) =V
2
(t). Since the current in an
inductor cannot change abruptly, i
L
(0
+
) =i
L
(0
−
) =0. Since the voltage across a
capacitor cannot change abruptly, v
C
(0
+
) =v
C
(0
−
) =0. Hence,
v
R
(0
+
) = v
1
(0
+
) =⇒ i
C
(0
+
) = i
R
(0
+
) =
V
1
(0
+
)
R
.
8.3 an electric circuit 373
Applying Kirchhoff ’s Current Law at node 1 yields i
R
=i
C
−i
L
.
Applying Kirchhoff ’s Voltage Law on the left mesh gives
−V
1
(t) + v
C
+ v
R
= 0 =⇒
1
C
_
t
−∞
i
C
dt + Ri
R
= V
1
(t).
Differentiating with respect to t leads to
1
C
i
C
+ R
d(i
C
−i
L
)
dt
=
dV
1
(t)
dt
.
Applying Kirchhoff ’s Voltage Law on the right mesh results in
−v
R
+ v
L
+ V
2
(t) =0 =⇒ R(i
C
−i
L
) − L
di
L
dt
= V
2
(t).
Hence, the system of differential equations governing i
C
(t) and i
L
(t) are
R
di
C
dt
+
1
C
i
C
− R
di
L
dt
=
dV
1
(t)
dt
,
Ri
C
− L
di
L
dt
− Ri
L
= V
2
(t), i
C
(0
+
) =
V
1
(0
+
)
R
, i
L
(0
+
) =0.
Suppose that R=2 , C=
5
8
F, L=1 H, V
1
(t) =6V, and V
2
(t) is as shown in
the following ﬁgure, which can be expressed as V
2
(t) =24t −24(t −1)H(t −1).
V
2
(t) (V)
t (sec)
1 0
24
The system of differential equations becomes
5
di
C
dt
+ 4i
C
− 5
di
L
dt
= 0,
2i
C
−
di
L
dt
− 2i
L
= 24t −24(t −1)H(t −1), i
C
(0
+
) =3, i
L
(0
+
) =0.
The method of Laplace transform and the matrix method will be applied to solve
the differential equations for i
C
(t) and i
L
(t).
Method of Laplace Transform
Let I
C
(s) =L
_
i
C
(t)
_
and I
L
(s) =L
_
i
L
(t)
_
. Taking the Laplace transform of
both sides of the differential equations gives
5
_
s I
C
(s)−i
C
(0
+
)
_
+ 4I
C
(s) − 5
_
s I
L
(s)−i
L
(0
+
)
_
= 0,
374 8 applications of systems of linear differential equations
2I
C
(s) −
_
s I
L
(s)−i
L
(0
+
)
_
− 2I
L
(s) =
24(1−e
−s
)
s
2
.
These equations lead to two algebraic equations for I
C
(s) and I
L
(s)
(5s +4)I
C
(s) − 5s I
L
(s) = 15, 2I
C
(s) − (s +2)I
L
(s) =
24(1−e
−s
)
s
2
,
which can be solved using Cramer’s Rule
=
¸
¸
¸
¸
¸
5s +4 −5s
2 −(s +2)
¸
¸
¸
¸
¸
= −(5s
2
+4s +8),
L
=
¸
¸
¸
¸
¸
¸
5s +4 15
2
24(1−e
−s
)
s
2
¸
¸
¸
¸
¸
¸
=
120s +96
s
2
(1−e
−s
) − 30,
∴ I
L
(s) =
L
=
−120s −96
s
2
(5s
2
+4s +8)
(1−e
−s
) +
30
5s
2
+4s +8
.
Using partial fractions, one has
−120s −96
s
2
(5s
2
+4s +8)
=
A
s
2
+
B
s
+
Cs +D
5s
2
+4s +8
.
Comparing the coefﬁcients of the numerators leads to
1: 8A = −96 =⇒ A = −12,
s : 4A + 8B = −120 =⇒ B = −15 −
1
2
A = −9,
s
2
: 5A + 4B + D = 0 =⇒ D = −5A − 4B = 96,
s
3
: 5B + C = 0 =⇒ C = −5B = 45.
Taking the inverse Laplace transform, one has
i
L
(t) = L
−1
__
−12
s
2
+
−9
s
+
45s +96
5s
2
+4s +8
_
(1−e
−s
) +
30
5s
2
+4s +8
_
= L
−1
_
_
−
12
s
2
−
9
s
_
(1−e
−s
) +
9
_
s +
2
5
_
+ 13
_
6
5
_
_
s +
2
5
_
2
+
_
6
5
_
2
_
1 − e
2
5
· e
−(s +
2
5
)
_
+
5
_
6
5
_
_
s +
2
5
_
2
+
_
6
5
_
2
_
L
−1
_
F(s −a)
_
=e
at
L
_
F(s)
_
= −12t −9−
_
−12(t −1)H(t −1)−9H(t −1)
_
L
−1
_
e
−as
F(s)
_
= f (t −a)H(t −a)
+ e
−
2t
5
L
−1
_
9s + 13
_
6
5
_
s
2
+
_
6
5
_
2
_
1 − e
2
5
· e
−s
_
+
5
_
6
5
_
s
2
+
_
6
5
_
2
_
8.3 an electric circuit 375
= −12t − 9 + (12t −3)H(t −1) + e
−
2t
5
_
_
9 cos
6t
5
+ 13 sin
6t
5
_
− e
2
5
_
9 cos
6(t −1)
5
+ 13 sin
6(t −1)
5
_
H(t −1) + 5 sin
6t
5
_
= −12t − 9 + (12t −3)H(t −1) + e
−
2t
5
_
9 cos
6t
5
+ 18 sin
6t
5
_
− e
−
2(t−1)
5
_
9 cos
6(t −1)
5
+ 13 sin
6(t −1)
5
_
H(t −1).
The current i
C
(t) can also be obtained in the same way as i
L
. However, since i
C
(t)
can be expressed in terms of i
L
(t) from the second differential equation of the
system, one can obtain i
C
using
i
C
(t) = 12t − 12(t −1)H(t −1) + i
L
(t) +
1
2
di
L
(t)
dt
.
Notingthat
dH(t −a)
dt
=δ(t −a), f (t)δ(t −a) = f (a)δ(t −a), andusingthe prod
uct rule, one has
di
L
dt
= −12 + 12H(t −1) + (12t −3)δ(t −1)
−
2
5
e
−
2t
5
_
9 cos
6t
5
+ 18 sin
6t
5
_
+ e
−
2t
5
_
−
54
5
sin
6t
5
+
108
5
cos
6t
5
_
+
2
5
e
−
2(t−1)
5
_
9 cos
6(t −1)
5
+ 13 sin
6(t −1)
5
_
H(t −1)
− e
−
2(t−1)
5
_
−
54
5
sin
6(t −1)
5
+
78
5
cos
6(t −1)
5
_
H(t −1)
− e
−
2(t−1)
5
_
9 cos
6(t −1)
5
+ 13 sin
6(t −1)
5
_
δ(t −1)
= −12 + 12H(t −1) + 18e
−
2t
5
_
cos
6t
5
− sin
6t
5
_
+ e
−
2(t−1)
5
_
−12 cos
6(t −1)
5
+ 16 sin
6(t −1)
5
_
H(t −1),
∴ i
C
(t) = −15 + 15H(t −1) + 9e
−
2t
5
_
2 cos
6t
5
+ sin
6t
5
_
− 5e
−
2(t−1)
5
_
3 cos
6(t −1)
5
+ sin
6(t −1)
5
_
H(t −1).
Matrix Method
From the second differential equation, one has
di
L
dt
= 2i
C
− 2i
L
− 24t + 24(t −1)H(t −1).
376 8 applications of systems of linear differential equations
Substituting into the ﬁrst differential equation yields
di
C
dt
= −
4
5
i
C
+
di
L
dt
=
6
5
i
C
− 2i
L
− 24t + 24(t −1)H(t −1).
In the matrix form, the system of differential equations can be written as
di(t)
dt
= Ai(t), i(t) =
_
i
C
i
L
_
, A =
_
6
5
−2
2 −2
_
,
f (t) = −24
_
t +(t −1)H(t −1)
_
_
1
1
_
, i(0) =
_
i
C
(0
+
)
i
L
(0
+
)
_
=
_
3
0
_
.
The characteristic equation is
det(A−λI) =
¸
¸
¸
¸
¸
6
5
−λ −2
2 −2−λ
¸
¸
¸
¸
¸
=
1
5
(5λ
2
+ 4λ + 8) = 0 =⇒ λ= −
2
5
±i
6
5
.
For eigenvalue λ= −
2
5
+i
6
5
, the corresponding eigenvector is
(A−λI)v =
⎡
⎣
8
5
−i
6
5
−2
2 −
8
5
−i
6
5
⎤
⎦
_
v
1
v
2
_
=
⎧
⎨
⎩
_
8
5
−i
6
5
_
v
1
−2v
2
2v
1
+
_
−
8
5
−i
6
5
_
v
2
⎫
⎬
⎭
=
_
0
0
_
.
Taking v
1
=5, then v
2
=
1
2
_
8
5
−i
6
5
_
v
1
=4−i 3,
∴ v =
_
v
1
v
2
_
=
_
5
4−i 3
_
=
_
5
4
_
+ i
_
0
−3
_
.
Hence, using Euler’s formula e
i θ
= cos θ +i sin θ,
e
λt
v = e
−
2t
5
_
cos
6t
5
+ i sin
6t
5
_
__
5
4
_
+ i
_
0
−3
__
= e
−
2t
5
__
5
4
_
cos
6t
5
−
_
0
−3
_
sin
6t
5
+ i
__
0
−3
_
cos
6t
5
+
_
5
4
_
sin
6t
5
__
.
A fundamental matrix is
I(t) =
_
Re(e
λt
v), Im(e
λt
v)
_
= e
−
2t
5
⎡
⎢
⎢
⎣
5 cos
6t
5
5 sin
6t
5
3 sin
6t
5
+4 cos
6t
5
4 sin
6t
5
−3 cos
6t
5
⎤
⎥
⎥
⎦
,
and its inverse is obtained as
I
−1
(t) = −
e
2t
5
15
⎡
⎢
⎣
4 sin
6t
5
−3 cos
6t
5
−5 sin
6t
5
−3 sin
6t
5
−4 cos
6t
5
5 cos
6t
5
⎤
⎥
⎦
.
8.4 vibration of a twostory shear building 377
Evaluate the quantity
I
−1
(0)i(0) +
_
t
0
I
−1
(t) f (t)dt = −
1
15
_
−3 0
−4 5
_ _
3
0
_
+
_
t
0
−
e
2t
5
15
⎡
⎢
⎣
4 sin
6t
5
−3 cos
6t
5
−5 sin
6t
5
−3 sin
6t
5
−4 cos
6t
5
5 cos
6t
5
⎤
⎥
⎦
_
1
1
_
_
−24
_
t +(t−1)H(t−1)
_
_
dt
=
⎧
⎪
⎨
⎪
⎩
3
5
4
5
⎫
⎪
⎬
⎪
⎭
−
8
5
_
t
0
⎧
⎪
⎨
⎪
⎩
e
2t
5
_
sin
6t
5
+3 cos
6t
5
_
t
e
2t
5
_
3 sin
6t
5
−cos
6t
5
_
t
⎫
⎪
⎬
⎪
⎭
dt
+
8
5
_
t
1
⎧
⎪
⎨
⎪
⎩
e
2t
5
_
sin
6t
5
+3 cos
6t
5
_
(t −1)
e
2t
5
_
3 sin
6t
5
−cos
6t
5
_
(t −1)
⎫
⎪
⎬
⎪
⎭
dt · H(t −1)
=
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
18
5
+ e
2t
5
_
−4t sin
6t
5
−3 cos
6t
5
+sin
6t
5
_
+
_
e
2t
5
_
4t sin
6t
5
+3 cos
6t
5
−5 sin
6t
5
_
− e
2
5
_
3 cos
6
5
−sin
6
5
_
_
H(t −1)
9
5
+ e
2t
5
_
4t cos
6t
5
−cos
6t
5
−3 sin
6t
5
_
−
_
e
2t
5
_
4t cos
6t
5
−5 cos
6t
5
−3 sin
6t
5
_
+ e
2
5
_
cos
6
5
+3 sin
6
5
_
_
H(t −1)
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
.
Substituting into the solution of the system of differential equations
i(t) = I(t)
_
I
−1
(0) i(0) +
_
t
0
I
−1
(t) f (t)dt
_
,
expanding, and simplifying by combining terms using trigonometric identities,
one obtains the same results as obtained using the method of Laplace transform.
Remarks: The method of Laplace transform, while applicable, is easier than the
matrix method; whereas the matrix method is the most general approach.
8.4 Vibration of a TwoStory Shear Building
Consider the vibration of an undamped twostory shear building under harmonic
excitations with m
1
=m
2
=m, k
1
=k
2
=k, F
1
(t) =F
2
(t) =ma cos t, as shown in
Figure 8.7. A twostory shear building is a two degreesoffreedom system, i.e., it
needs two variables x
1
and x
2
to describe its motion.
From Example 7.2 in Chapter 7, equations of motion are given by
m¨ x
1
+ 2kx
1
− kx
2
= ma cos t,
m¨ x
2
− kx
1
+ kx
2
= ma cos t,
378 8 applications of systems of linear differential equations
x
2
x
1
F
2
(t)
F
1
(t)
m
m
k
k
Figure 8.7 A twostory shear building.
or, using the notations ω
0
=
_
k/m,
¨ x
1
+ 2ω
2
0
x
1
− ω
2
0
x
2
= a cos t,
¨ x
2
− ω
2
0
x
1
+ ω
2
0
x
2
= a cos t.
In the matrix form, the equations of motion are
I ¨ x + Kx = F(t),
where
K =
_
2ω
2
0
−ω
2
0
−ω
2
0
ω
2
0
_
, F(t) =
_
a cos t
a cos t
_
.
The analysis of free and forced vibration follows the procedure presented in Section
7.4.3.
8.4.1 Free Vibration—Complementary Solutions
The frequency equation is given by
¸
¸
K−ω
2
I
¸
¸
=
¸
¸
¸
¸
¸
2ω
2
0
−ω
2
−ω
2
0
−ω
2
0
ω
2
0
−ω
2
¸
¸
¸
¸
¸
= 0,
ω
4
− 3ω
2
0
ω
2
+ ω
4
0
= 0 =⇒ ω
2
=
3±
√
5
2
ω
2
0
,
ω =
_
3±
√
5
2
ω
0
=
√
6±2
√
5
2
ω
0
=
√
(
√
5)
2
±2
√
5+1
2
ω
0
=
√
5±1
2
ω
0
.
Hence, the eigenvalues or modal frequencies are
ω
1
=
√
5−1
2
ω
0
, ω
2
=
√
5+1
2
ω
0
.
☞
Natural frequencies are always ordered in ascending order, i.e., ω
1
<ω
2
< . . . .
8.4 vibration of a twostory shear building 379
Corresponding to ω
1
, the ﬁrst mode shape ˆ x
1
is the eigenvector given by
_
2ω
2
0
−ω
2
1
−ω
2
0
−ω
2
0
ω
2
0
−ω
2
1
__
ˆ x
11
ˆ x
21
_
= 0.
Taking ˆ x
11
=1, then
ˆ x
21
=
2ω
2
0
−ω
2
1
ω
2
0
= 2 −
3−
√
5
2
=
1+
√
5
2
.
Corresponding to ω
2
, the second mode shape ˆ x
2
is the eigenvector given by
_
2ω
2
0
−ω
2
2
−ω
2
0
−ω
2
0
ω
2
0
−ω
2
2
__
ˆ x
12
ˆ x
22
_
= 0.
Taking ˆ x
12
=1, then
ˆ x
22
=
2ω
2
0
−ω
2
2
ω
2
0
= 2 −
3+
√
5
2
=
1−
√
5
2
.
Hence, the two mode shapes are given by
ˆ x
1
=
_
ˆ x
11
ˆ x
21
_
=
⎧
⎨
⎩
1
1+
√
5
2
⎫
⎬
⎭
, ˆ x
2
=
_
ˆ x
12
ˆ x
22
_
=
⎧
⎨
⎩
1
1−
√
5
2
⎫
⎬
⎭
.
The responses of free vibration (or complementary solution) are of the form
x
C
(t) = a
1
ˆ x
1
sin(ω
1
t +θ
1
) + a
2
ˆ x
2
sin(ω
2
t +θ
2
),
where the constants a
1
, θ
1
, a
2
, θ
2
are determined from the initial conditions x(0)
and ˙ x(0).
1
1.618
1
−0.618
(a) First mode, ω
1
(b) Second mode, ω
2
Figure 8.8 Mode shapes of a twostory shear building.
If the initial conditions are such that a
2
=0, θ
2
=0, the systemis vibrating in the
ﬁrst mode with natural circular frequency ω
1
. The responses of free vibration are
of the form
x
C
(t) =
_
x
1C
(t)
x
2C
(t)
_
= a
1
⎧
⎨
⎩
1
1+
√
5
2
⎫
⎬
⎭
sin(ω
1
t +θ
1
).
380 8 applications of systems of linear differential equations
The ratio of the amplitudes of responses of the second and ﬁrst ﬂoors is
x
2C
(t)
¸
¸
amplitude
x
1C
(t)
¸
¸
amplitude
=
1+
√
5
2
1
=
1.618
1
.
If the initial conditions are such that a
1
=0, θ
1
=0, the systemis vibrating in the
second mode with natural circular frequency ω
2
. The responses of free vibration
are of the form
x
C
(t) =
_
x
1C
(t)
x
2C
(t)
_
= a
2
⎧
⎨
⎩
1
1−
√
5
2
⎫
⎬
⎭
sin(ω
2
t +θ
2
).
The ratio of the amplitudes of responses of the second and ﬁrst ﬂoors is
x
2C
(t)
¸
¸
amplitude
x
1C
(t)
¸
¸
amplitude
=
1−
√
5
2
1
=
−0.618
1
.
The mode shapes ˆ x
1
and ˆ x
2
giving the ratios of the amplitudes of vibration of
x
2C
and x
1C
are shown in Figure 8.8. In general, the responses of the system are
linear combinations of the ﬁrst and second modes.
8.4.2 Forced Vibration—General Solutions
The modal matrix is
=
_
ˆ x
1
, ˆ x
2
_
=
⎡
⎣
1 1
1+
√
5
2
1−
√
5
2
⎤
⎦
,
with the orthogonality conditions
T
I =
⎡
⎣
5+
√
5
2
0
0
5−
√
5
2
⎤
⎦
=
_
m
1
0
0 m
2
_
,
T
K =
⎡
⎣
m
1
ω
2
1
0
0 m
2
ω
2
2
⎤
⎦
.
Letting x(t) =q(t), substituting into the equation of motion, and multiplying
T
from the left yields
T
¨ q(t) +
T
Kq(t) =
T
F(t),
i.e.,
m
1
¨ q
1
(t) + m
1
ω
2
1
q
1
(t) = f
1
(t), m
2
¨ q
2
(t) + m
2
ω
2
2
q
2
(t) = f
2
(t),
where
_
f
1
(t)
f
2
(t)
_
=
T
F(t) =
⎡
⎢
⎣
1
1+
√
5
2
1
1−
√
5
2
⎤
⎥
⎦
_
a cos t
a cos t
_
=
⎧
⎪
⎨
⎪
⎩
3+
√
5
2
3−
√
5
2
⎫
⎪
⎬
⎪
⎭
a cos t,
8.4 vibration of a twostory shear building 381
or
¨ q
1
(t) + ω
2
1
q
1
(t) =
¯
f
1
(t), ¨ q
2
(t) + ω
2
2
q
2
(t) =
¯
f
2
(t),
where
¯
f
1
(t) =
f
1
(t)
m
1
=
5+
√
5
10
a cos t,
¯
f
2
(t) =
f
2
(t)
m
2
=
5−
√
5
10
a cos t.
Using the method of operators, particular solutions, or responses of the systemdue
to forcing, are given by
q
1P
(t) =
5+
√
5
10
a
D
2
+ω
2
1
cos t, q
2P
(t) =
5−
√
5
10
a
D
2
+ω
2
2
cos t.
Hence,
x
P
(t) = q
P
(t) =
⎡
⎣
1 1
1+
√
5
2
1−
√
5
2
⎤
⎦
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
5+
√
5
10
a
D
2
+ω
2
1
cos t
5−
√
5
10
a
D
2
+ω
2
2
cos t
⎫
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎭
= a
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
5+
√
5
10
1
D
2
+ω
2
1
cos t +
5−
√
5
10
1
D
2
+ω
2
2
cos t
5+3
√
5
10
1
D
2
+ω
2
1
cos t +
5−3
√
5
10
1
D
2
+ω
2
2
cos t
⎫
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎭
.
The general solutions, or the responses of the system, are given by
x(t) = x
C
(t) + x
P
(t),
where, as obtained earlier,
x
C
(t) = a
1
ˆ x
1
sin(ω
1
t +θ
1
) + a
2
ˆ x
2
sin(ω
2
t +θ
2
),
in which the constants a
1
, θ
1
, a
2
, θ
2
are determined fromthe initial conditions x(0)
and ˙ x(0).
Depending on the values of the excitation frequency , the particular solutions
will have different forms.
1. = ω
1
or ω
2
, i.e., No Resonance
The following operators can be evaluated using Theorem 3 of Chapter 4 as
1
D
2
+ω
2
i
cos t =
1
ω
2
i
−
2
cos t, i =1, 2. Replace D
2
by −
2
382 8 applications of systems of linear differential equations
Hence,
x
P
(t) =
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
5+
√
5
10
1
ω
2
1
−
2
+
5−
√
5
10
1
ω
2
2
−
2
5+3
√
5
10
1
ω
2
1
−
2
+
5−3
√
5
10
1
ω
2
2
−
2
⎫
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎭
a cos t.
2. = ω
1
, i.e., Resonance in the First Mode
A particular solution q
1P
(t) can be evaluated using Theorem 4 of Chapter 4
1
D
2
+ω
2
1
cos ω
1
t = Re
_
1
D
2
+ω
2
1
e
i ω
1
t
_
,
φ(D) = D
2
+ω
2
1
, φ(i ω
1
) = (i ω
1
)
2
+ω
2
1
= 0,
φ
(D) = 2D, φ
(i ω
1
) = 2i ω
1
,
Re
_
1
D
2
+ω
2
1
e
i ω
1
t
_
= Re
_
1
φ
(i ω
1
)
t e
i ω
1
t
_
= Re
_
t
2i ω
1
(cos ω
1
t + i sin ω
1
t)
_
=
√
5+1
4ω
0
t sin ω
1
t.
A particular solution q
2P
(t) can be evaluated using Theorem 3 of Chapter 4
1
D
2
+ω
2
2
cos ω
1
t =
1
ω
2
2
−ω
2
1
cos ω
1
t =
√
5
5ω
2
0
cos ω
1
t. Replace D
2
by −ω
2
1
Hence,
x
P
(t) = a
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
5+3
√
5
20
·
1
ω
0
t sin ω
1
t +
√
5−1
10
·
1
ω
2
0
cos ω
1
t
. ,, .
γ
1
5+2
√
5
10
·
1
ω
0
t sin ω
1
t +
√
5−3
10
·
1
ω
2
0
cos ω
1
t
. ,, .
γ
2
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
.
Note that, when =ω
1
, the terms involving (t sin ω
1
t) have amplitudes growing
linearly in time t and are dominant in the responses of the system. The ratio of the
amplitudes of these two terms is
γ
2
γ
1
=
5+2
√
5
10
×
20
5+3
√
5
=
1+
√
5
2
=
1.618
1
,
which conforms to the mode shape of the ﬁrst mode as depicted in Figure 8.8(a),
indicating that the system is in resonance in the ﬁrst mode.
8.4 vibration of a twostory shear building 383
3. = ω
2
, i.e., Resonance in the Second Mode
A particular solution q
1P
(t) can be evaluated using Theorem 3 of Chapter 4
1
D
2
+ω
2
1
cos ω
2
t =
1
ω
2
1
−ω
2
2
cos ω
2
t = −
√
5
5ω
2
0
cos ω
2
t. Replace D
2
by −ω
2
2
A particular solution q
2P
(t) can be evaluated using Theorem 4 of Chapter 4
1
D
2
+ω
2
2
cos ω
2
t = Re
_
1
D
2
+ω
2
2
e
i ω
2
t
_
,
φ(D) = D
2
+ω
2
2
, φ(i ω
2
) = (i ω
2
)
2
+ω
2
2
= 0,
φ
(D) = 2D, φ
(i ω
2
) = 2i ω
2
,
Re
_
1
D
2
+ω
2
2
e
i ω
2
t
_
= Re
_
1
φ
(i ω
2
)
t e
i ω
2
t
_
= Re
_
t
2i ω
2
(cos ω
2
t + i sin ω
2
t)
_
=
√
5−1
4ω
0
t sin ω
2
t.
Hence,
x
P
(t) = a
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
−
√
5+1
10
·
1
ω
2
0
cos ω