Faculty of Civil Engineering and Geosciences

Section of Structural Mechanics

Dynamics, Slender Structures and an

Introduction to Continuum Mechanics
CT 4145
Module

Dynamics of Mechanical Systems and Slender Structures

A.V. Metrikine

Delft University of Technology

ii

iii

When one has a good understanding of the problem,

one must rid it of all the superfluous concepts
and reduce it to the simplest elements.
Rne Descartes

iv

CONTENTS
CHAPTER 1. SIGNIFICANCE OF DYNAMICS FOR CIVIL ENGINEERING.1
1.1. Earthquake-Induced Vibrations. 1
1.2. Wind-Induced Vibrations...... 2
1.3. Traffic-Induced Vibrations 4
1.4. Construction- and Production-Induced Vibrations... 5
1.5. Wave- and Flow-Induced Vibrations 6
CHAPTER 2. DYNAMICS OF MASS-SPRING-DASHPOT SYSTEMS.. 8
2.1. Fundamental Assumptions Leading to Models with Finite Number
of Degrees of Freedom (Mass-Spring-Dashpot Models... 8
2.2. One Degree of Freedom Systems without Damping. 9
2.2.1. Free Vibrations..10
2.2.2. Forced Vibrations under Harmonic Force 14
2.2.3. Forced Vibrations under General Disturbing Force......19
2.3. One Degree of Freedom Systems with Viscous Damping...29
2.3.1. Free Vibrations..30
2.3.2. Forced Vibrations under Harmonic Force.... 35
2.3.3. Forced Vibrations under General Disturbing Force..43
2.4. Two Degrees of Freedom Systems without Damping
2.4.1. Free Vibrations..47
2.4.2. Forced Vibrations under Harmonic Force.53
CHAPTER 3. ONE-DIMENSIONAL (SLENDER) STRUCTURES... 57
3.1. Fundamental Assumptions Leading to One-Dimensional
Continuous Models....... 57
3.2. Formulation of Equations of Motion for One-dimensional Models 59
3.2.1. Transverse Motion of a String...59
3.2.2. Longitudinal Motion of a Rod...60
3.2.3. Torsion of a Rod62
3.2.4. Pure Shear of a Beam64
3.2.5. Bending of a Beam66
3.2.6. Viscous-Elastic Kelvin Foundation...68
3.2.7. Combination of Models.....69
3.3. Boundary and Interface Conditions for One-dimensional Models.................. 70
3.3.1. Transverse Motion of a String ..... 71
3.3.2. Longitudinal Motion of a Rod...................72
3.3.3. Torsion of a Rod73
3.3.4. Pure Shear of a Beam74
3.3.5. Bending of a Beam75
3.4. Introduction to Dynamics of One-Dimensional Structures......................78
3.4.1. Longitudinal Vibrations of a Rod..79
3.4.2. Bending Vibrations of a Beam..85
3.5. Elastostatics of One-Dimensional Structures...................88
3.5.1. Static Response of a Rod to Axial Loading.. 89
3.5.2. Static Response of a Rod on Elastic Foundation to Axial Loading.. 99
3.5.3. Static Response of a Beam..104
3.5.4. Static Response of a Beam on Elastic Foundation..118

vi

vii

NOTATIONS

f0
f
G
g
G (t )

ross-sectional area, displacement

amplitude
radius, semi-major axis
displacement
semi-minor axis
constants of integration
damping coefficient, wave speed
damping coefficient per unit length
total mechanical energy
Youngs modulus
force
amplide of harmonic force, constant force
amplitude of harmonic force per unit mass
frequency
shear modulus
gravity acceleration
original function

H (...)
H
I
i
J
Jt
K
k
K
Kd
kd
L
M
m
Mt
mt
n
P
P
Q
q
q1
s
T

Heaviside step-function
horizontal force
moment of inertia of a cross-section
1
polar moment of inertia
torque constant
kinetic energy
spring constant
spring constant, dimensionless stiffness
dimensionless stiffness
stiffness per unit length
Lagrange function
mass, moment of force
mass
torque
linear density of torque
amount of viscous damping, integer
force
potential energy
generalized force
body force
force per unit length
characteristic exponent, Laplace variable
tension

A
A0
a
B
b
A, B, C
c
cd
E
E
F
F0

g ( s ) image function

viii

Tn
t, t
t
U ( x)
u
V
v0
v
w
X
Xc, Xs
x0
x
x

natural period
time
time interval
function of coordinate
axial displacement
transverse force, shear force
initial velocity
velocity increment
transverse displacement
complex amplitude
amplitudes
initial displacement
coordinate, displacement from the equilibrium position
velocity
angle
wavenumber
dimensionless amount of viscous damping, shear strain
axial strain
initial phase
phase lag

( t ) function of time

a , b
n , n

angle of twist
mass ratio
slope
mass density
normal stress
shear stress
radial frequency, radial frequency of excitation
partial frequencies
natural frequencies
x , F vector quantities
K , M matrix quantities

CHAPTER 1
SIGNIFICANCE OF DYNAMICS FOR CIVIL ENGINEERING
The analysis and design of structures subject to time dependent forces constitute the
field of structural dynamics. For example, the dynamic forces that are transmitted to a
structure that supports oscillating machinery must be considered in its analysis and
design. This is also true for a bridge facilitating car or train traffic, for a structure that
is subjected to blast, wind or water waves, or for a building whose foundation is
disturbed by an earthquake. In this Section, these and other examples of civil
engineering systems and structures are discussed to underline the significance of
dynamics for modern civil engineering. All examples are subdivided into 5 categories
in accordance with different types of causes of vibrations. These causes are:
earthquake, wind, traffic, construction and production, and water waves.

1.1.

Earthquake-Induced Vibrations

An earthquake is the shaking of the earth caused by pieces of the crust of the Earth
that suddenly shift. The crust, the thin outer layer of the Earth, is mostly cold and
brittle rock compared to the hot rock deeper inside. This crust is full of large and
small cracks called faults. Although these faults can be hundreds of miles long,
usually you cannot see the cracks because they are buried deep underground and
because the pieces of crust are compressed together very tightly. The powerful forces
that compress these crust pieces also cause them to move very slowly. When two
pieces that are next to each other get pushed in different directions, they will stick
together for a long time (many years), but eventually the forces pushing on them will
force them to break apart and move. This sudden shift in the rock shakes all of the
rock around it. The resulting vibrations propagate in the form of seismic waves and
can cause significant damage of buildings. Figure 1.1 shows a high-rise building
damaged by the Taiwan Earthquake in 1999.

Figure 1.1. High-rise building damaged by the Taiwan Earthquake in 1999

To protect buildings in seismically active areas, seismic isolators, such as friction

pendulums, loose links, kinematic bases, etc. are used. If installation of seismic
isolators is not possible, every building should be designed to withstand an incidental
earthquake. To ensure this, a dynamic analysis should be performed considering the
building vibrations that might be caused by earthquakes. Simplistic models used in

F(t)
Figure 1.2. Earthquake-induced excitation and simple structural model for a building

such an analysis utilize beams, which describe both the bearing walls and the floors,
see Figure 1.2. The force exerted by an earthquake is often modelled as a shear force
applied to the building footing.

1.2.

Wind-Induced Vibrations

Skyscrapers, Chimneys, Wind-Energy Generators, Cable-Stayed Bridges

As modern structures move towards taller and more flexible designs, the problems of
wind effect have become increasingly apparent. The most well known examples of
structures vulnerable to wind effects are suspension bridges (see Figure 1.3 of the
famous Tacoma Narrows Bridge), skyscrapers (Figure 1.4), chimneys (Figure 1.5)
and wind-energy generators (Figure 1.6).

Figure 1.3. Vibrating Tacoma Narrows Bridge

Figure 1.5. Chimney

Figure 1.4. Skyscrapers

Figure 1.6. Wind-farm

Performing structural analysis of wind-induced vibrations, one has to know

characteristics of the wind velocity and the mechanisms of excitation of structural
vibrations.
The most common description of the wind velocity superimposes a mean wind
component, described by a mean velocity profile, with a fluctuating velocity
component. The mean wind velocity is time-independent but depends on the height
referenced to the terrain. The fluctuating wind field is both time and height dependent
and is characterized by temporal averages, variances of velocity components, energy
spectra, associated length scales, and space-time correlations.
The oncoming wind imposes loads on the structure that vary both spatially and
temporally. The fluctuating wind velocity translates directly into fluctuating positive
pressures distributed across the buildings windward face. Corresponding negative
pressure acts on the leeward face of the structure. The wind incident on the windward
face is deflected around the structure and accelerated such that it cannot negotiate the
sharp corners and thus separates from the building, leaving a region of high negative
pressure. This separated flow forms a shear layer on each side, and subsequent
interaction between the layers results in the formation of discrete vortices, which are
shed alternately. This region is generally known as the wake region. The three
dimensional simultaneous loading of the structure due to its interaction with the wind
results in three structural response components. The first, termed the alongwind
component, primarily results from pressure fluctuations in the incident flow, leading
to a sway of the structure in the direction of the wind. The acrosswind component
constitutes a sway motion perpendicular to the direction of the wind and is introduced
by side-face pressure fluctuations primarily induced by the fluctuations in the
separated shear layers, vortex shedding and wake flow fields. The final torsional
component results from imbalances in the instantaneous pressure distribution on the
building surfaces. These wind load effects are further amplified on asymmetric
buildings as a consequence of inertial coupling of the structural components of the
building.
Wind-induced structural vibrations can alter the aerodynamic forces acting on the
structure, thus setting up an interaction between the elastic response and aerodynamic
forces commonly referred to as aeroelasticity. Aeroelastic contributions to the overall
aerodynamic loading are distinguished from other unsteady loads by recognizing that
aeroelastic loads vanish when there is no structural motion. Different types of
aeroelastic effects are commonly distinguished from each other. They include vortexinduced vibration, galloping and flutter. The vortex-induced vibrations (VIV) occur as
the frequency of the acrosswind component of the force, which is determined by
vortex shedding, approaches one of the natural frequencies of a structure. The VIVphenomenon is more complex than a mere resonant forcing problem. Nonlinear
interaction between the structural motion and the wake behind the structure results in
the locking in of the wake to the oscillation frequency of the structure over a much
larger velocity range than would be predicted by the resonance theory. VIV is a
common issue for tall slender structures like chimneys and wind-energy generators.
Galloping occurs for structures of certain cross sections at frequencies below those of
vortex-induced vibration. One widely known example of galloping is the large acrosswind amplitudes exhibited by power lines when freezing rain has resulted in a change
of their cross section. Galloping is normally considered a quasi-steady phenomenon
because knowledge of the average aerodynamic coefficients of a given structure (i.e.,
mean lift and drag forces on a stationary model) allows quite reliable prediction of
galloping behavior. Flutter is a common design issue for long span bridges. It can

lead to catastrophic failure as in the case of the Tacoma Narrows Bridge see Figure
1.3. Flutter is associated with so-called self-excited vibrations, during which the
structural velocity is in phase with the aerodynamic force. The term self-excited
originates from the fact that the structure itself chooses to move in phase with the
aerodynamic excitation. Such a motion is associated with negative damping
(because positive damping implies that the force is in anti-phase with structural
velocity), which leads to structural instability. In an unstable scenario, the structure
extracts energy from the wind amplifying its own motion, which can lead to
catastrophic failure. Flutter is of significant danger for structures and therefore must
be avoided for the wind velocity range of interest by means of structural optimization
and aerodynamic tailoring.
There exist a wide variety of prediction models for wind-induced vibrations. They
range from simplistic one- and two-degree of freedom systems to sophisticated threedimensional elasticity models. In the majority of these models, however, the
aeroelastic effects are described simplistically, especially at high Reynolds numbers
associated with strong wind and large structural dimensions. This is because it is
extremely difficult to model the shear layers around the structure. The modeling is
normally done by introducing added mass, negative damping and additional dynamic
stiffness of the structure. All these added properties are associated with the
aerodynamic force, induced by structural motion.

1.3.

Traffic-Induced Vibrations
Bridges, Railway Lines

When a vehicle such as a car or a train passes through an urban area (see Figure 1.7 of
a bullet train in Kyoto City), building occupants in the neighbourhood may experience
significant discomfort associated with vibrations induced by a vehicle. There can be
various reasons for these vibrations to occur: uneven or bumpy roads, curves of the
traffic lines, imperfection of vehicle suspension system, unevenness of train wheels,
corrugation of rails, etc. To predict the level of building vibrations induced by traffic,
dynamic model for a vehicle and for a building have to be integrated into a module
that predicts propagation of elastic waves through the soil from the vehicle to the
building.

Figure 1.7. Bullet train in Kyoto City

Figure 1.8. Bullet train passing a bridge

Traffic-induced vibrations are important for civil engineering not only because of
the discomfort they can cause to people but also because of their significant influence
on durability and reliability of structures such as bridges, see Figure 1.8. Bridges must
be designed such that no perceptible traffic-induced vibrations occur during
exploitation. This is especially important for bridges facilitating high-speed train

traffic, since such a train can cause considerable dynamic amplification of bridge
vibrations. Sometimes, train-induced vibrations of rails can lead to catastrophic
consequences. This can happen with continuously welded railway tracks during a hot
summer. Due to the temperature extension, a considerable axial compression can
occur in the rails. This compression makes the rails vulnerable both to static and
dynamic buckling see Figure 1.9 and 1.10. The static buckling occurs in poorly
ballasted tracks (Figure 1.9), while the dynamic buckling can be initiated even in a
well-ballasted track under the action of a passing train (Figure 1.10).

Figure 1.9. Buckled railway line

1.4.

Figure 1.10. Rail vulnerable to dynamic buckling

Construction- and Production-Induced Vibrations

Pile Driving, Machine Foundations

Construction sites are well known as a source of noise and ground vibrations. The
latter are mainly associated with pile driving, see Figure 1.11. Every nock of the
hammer on the pile is accompanied by powerful vibrations of the surrounding ground.
In some cases these vibrations can be damaging to surrounding buildings let alone
their annoyance to people living around. Thus, before opening a new construction site,
the level of vibrations must be assessed that can be expected in the surrounding
buildings.

Figure 1.11. Pile driving

Figure 1.12. Construction of machine foundation

As well as construction sites, production facilities, which use heavy machinery,

may serve as a source of perceptible vibrations. To avoid this, massive machine
foundations (see Figure 1.12) should be installed to isolate machines from other
structural elements of the facility.
Prediction models for building vibrations generated by pile driving should account
for dynamic coupling between the soil, the pile and the buildings surrounding the
construction site. Equally important is the soil-structure interaction for assessing
vibrations induced by heavy machinery.

1.5.

Wave- and Flow-Induced Vibrations

Offshore Vessels and Platforms, Floating Airports, Oil/Gas Offshore Riser

Waves and currents in the ocean cause vibrations of floating vessels and submerged
structural elements. The wave-induced vibrations play an increasingly important role
in dynamics of offshore platforms, especially in that of the semi-submersible
platforms, see Figure 1.13. These platforms facilitate the deep-water oil and gas
exploitation, which is becoming more and more important once easily accessible
oil/gas reservoirs are getting exhausted. These platforms float in the Sea being moored
to the seabed by flexible cables, which keep the platform at place but almost do not
resist to relatively small vibrational motions. These motions, however, can be
damaging to the pipes, through which oil/gas is transported up to the platform.
Therefore, the floating platforms should be designed such that their response to waves
is minimized. Normally, this is achieved by building platforms with Frequency
Response Functions, which do not intersect with the Wave Spectrum.

Figure 1.13. Semi-submersible offshore platform

Figure 1.14. Mega-Float in Tokyo Bay

Recently, another fascinating engineering development has emerged, which is very

susceptible to the wave loading. This development is known as Mega-Floats, or
simply floating airports, one of which is shown in Figure 1.14. As well as the semisubmersible platforms, these airports should be almost immovable in waves.
Not only waves generate vibrations of floating structures but also currents, which
are always present in the ocean. Most vulnerable to the currents are flexible pipelines,
which facilitate transportation of oil/gas up to the offshore vessels. These pipelines are
conventionally referred to as risers.

Figure 1.15. Floating offshore facility

Figure 1.16. Installation of a riser

In Figure 1.15, an artist impression of a system of risers is shown, which connect a

floating vessel to the seabed. Installation of a riser is shown in Figure 1.16. Depending
on the site, the risers can be of different length. In the Golf of Mexico, for example,
risers of 2000m and longer are required.
As well as high-rise structures such as skyscrapers and chimneys, offshore risers
can (and usually do) experience vortex-induced vibrations caused by currents. This
phenomenon is of the hydro-elastic nature and exhibits itself in vibrations of a riser in
the direction perpendicular to that of a current. It is caused by alternatively shed
vortices, which are generated by currents as they are deflected by a riser. Relative to
VIV of high-rise structures in the air, VIV in water is complicated by large added
mass, which is involved in the structural motion. Additionally, because of an immense
length of currently operated risers, a so-called multi-mode VIV may arise. The term
multi-mode implies that a few structural modes are simultaneously excited by the
vortices.
Not only currents can destabilize the risers but also a flow of oil or gas through the
riser. If the flow has a relatively high speed, it can cause substantial centrifugal and
Coriolis forces, which destabilise the lateral motion of the riser. This phenomenon
gives the upper limit of the speed, with which oil and gas may be pumped through the
riser.

CHAPTER 2
DYNAMICS OF MASS-SPRING-DASHPOT SYSTEMS
Many of the vibratory phenomena associated with complicated systems may be
understood by studying the behaviour of simple systems. The most basic vibrating
system consists of a single lumped mass and a spring. This is said to have one degree
of freedom, implying that there is only one possible direction of movement for the
lumped mass. In reality, structures may move in a few different directions, rotate, etc.
Therefore, real structures are systems with many degrees of freedom. In many
practical situations, however, just a few degrees of freedom are activated by the load.
In these situations, models accounting for one or two degrees of freedom can provide
a good estimate for dynamic structural processes. In this Chapter, attention is focused
on one- and two-degrees of freedom systems, which are most facilitating for
understanding of the fundamentals of structural dynamics.

2.1. Fundamental Assumptions Leading to Models with Finite

Number of Degrees of Freedom (Mass-Spring-Dashpot Models)
All models with finite number of degrees of freedom are composed of lumped
elements, whose displacements (velocities, accelerations, etc.) depend on time and are
independent of spatial coordinates. These models are governed by ordinary
differential equations.
Strictly speaking, lumped models should be applied to mechanical systems, the
material points of which do not move relative to each other, i.e. to rigid bodies. In
practice, however, lumped models are widely used to describe flexible structures as
well. Although such a description is approximate, it often provides good practical
predictions. The reason for this is that in many practical situations only one or two
modes (spatial forms) of structural vibrations are excited by the load and, therefore,
all remaining modes may be disregarded.
Two examples of structures, whose dynamic behaviour may be approximated by
one-degree-of-freedom systems, are shown in Figures 2.1 and 2.2.

x(t)
m

Figure 2.1. Idealization of horizontal motion of an offshore platform in waves

F(t)

Unloaded
static
position

x(t)
F(t)
Figure 2.2. Idealization of vertical vibration of a lorry on a bridge

In Figure 2.1 the relatively heavy deck of the offshore platform is replaced by a
lumped mass while the braced steel jacket structure is replaced by an elastic spring,
which models the lateral stiffness of the structure. The wave load is assumed to be
applied to the deck. In Figure 2.2the bridge girder is idealized as a leaf spring with its
mass concentrated at midspan being dynamically loaded by a heavy lorry. Both these
idealizations disregard all but one possible motion and implicitly assume that the
loading of the structure is such that only one mode of structural vibrations is activated.
Although, the idealizations may be thought to be apparently crude, a skilled engineer
can make surprisingly accurate predictions of the behavior of real structures by
intelligent choice of the parameters of simple systems.

2.2.

One Degree of Freedom Systems without Damping

Before going on to mathematical studies of equations of motion for one-degree of

freedom systems, it is an essential preliminary to understand the derivation of these
equations.
Consider the system shown in Figure 2.3, a body of mass m , which will be
referred to as a mass, rests on a frictionless horizontal plane, is attached to a massless
spring of stiffness k , and is subject to an external horizontal force F ( t ) .
x(t)
m

F(t)
x

Figure 2.3. Mass-spring system

For a constant mass system, Newtons second law states that

Force = mass acceleration

(2.1)

Both force and acceleration are vector quantities, mass being scalar, therefore a force
exerted on mass causes acceleration in the direction of the force.

10

In the case under consideration, there are two forces acting on the mass: the
external force F ( t ) and the force exerted by the spring FSPRING ( t ) . Therefore, by
Newtons second law
F ( t ) + FSPRING ( t ) = m
x

(2.2)

where x ( t ) is the displacement of the mass from its equilibrium position, in which
the spring is undisturbed. Let us underline that in Eq.(2.2) all terms are vectors. In
accordance with Hookes law,
FSPRING ( t ) = kx

(2.3)

Substituting this expression into Eq.(2.2) and projecting the result onto the x axis,
which is chosen to be directed rightward, the following scalar equation is obtained
mx+ kx = F ( t )

(2.4)

This is the equation of motion, which describes vibrations of the mass-spring system
around its equilibrium position under the action of the external force F ( t ) .
2.2.1. Free Vibrations. If the external force F ( t ) is absent, the equation of motion
(it is a second order ordinary linear differential equation) for the mass-spring system
reduces to
mx+ kx = 0

(2.5)

In this case, the system can move only if given an initial displacement or/and an initial
velocity:
x ( 0 ) = x0
x ( 0 ) = v0

(2.6)

To find x ( t ) that satisfies both the equation of motion Eq.(2.5) and the initial
conditions Eq.(2.6), two steps are to be made. First, the general solution of Eq.(2.5)
should be found, which will contain two unknown constants. The number of these
unknown constants is equal to the order of the differential equation. Secondly, these
two constants should be determined by employing the initial conditions.
To find the general solution of Eq.(2.5), a theorem can be used that states that the
general solution of an ordinary linear differential equation of N th order with
constant coefficients can be written as ( x ( t ) is the unknown function in this equation)
N

x ( t ) = X n exp ( sn t )
n =1

(2.7)

11

where X n and sn are complex constants and are referred to as the complex amplitude
and the characteristic exponent, respectively.
For the second order differential equation Eq.(2.5), the general form of solution
Eq.(2.7) reduces to
2

x ( t ) = X n exp ( sn t )

(2.8)

n =1

Substituting Eq.(2.8) into the equation of motion Eq.(2.5), one obtains

2

n =1

n =1

m X n sn2 exp ( sn t ) + k X n exp ( sn t ) = 0

X ( ms
n

2
n

n =1

(2.9)

+ k ) exp ( sn t ) = 0

If a non-trivial solution X n 0 is sought for, then Eq.(2.9) can be satisfied if and only
if the following characteristic equation is satisfied:
msn2 + k = 0

(2.10)

The characteristic exponents sn are the roots of this equation. In the case under
consideration s1,2 read
s1 = i k m ,

s2 = i k m

(2.11)

where i = 1 is the imaginary unit.

Thus, the general solution of Eq.(2.5) can be written as

x ( t ) = X 1 exp it k m + X 2 exp it k m

(2.12)

where the exponents with the imaginary arguments are understood through the Euler
formula, which reads
exp ( i ) = cos ( ) + i sin ( )

(2.13)

Applying this formula to Eq.(2.12), this equation can be rewritten as

( (

)) + X ( cos (t k m ) i sin (t
) cos ( t k m ) + i ( X X ) sin ( t k m )
)

x ( t ) = X 1 cos t k m + i sin t k m
= ( X1 + X 2

= A cos (n t ) + B sin (nt )

k m

))
(2.14)

12

where A and B are unknown constants (which must be real) and

n = k m

(2.15)

is the natural frequency of the mass-spring system. As can be seen from Eq.(2.14),
this is the frequency, with which the mass-spring system would vibrate given arbitrary
initial conditions.
To determine the unknown constants A and B , Eq.(2.14) is to be substituted into
the initial conditions Eq.(2.6). This yields
A = x0

(2.16)

Bn = v0

Thus, the deflection of the mass-spring system from its equilibrium position reads
x ( t ) = x0 cos (n t ) +

v0

sin (n t )

(2.17)

Introducing the amplitude A0 and initial phase 0 in accordance with the following
relationships:
x0 = A0 cos (0 )

(2.18)

v0 n = A0 sin (0 )

and combining the right-hand side of Eq.(2.17) trigonometrically, we can rewrite this
equation in the following, widely used form
x ( t ) = A0 cos (n t 0 )

(2.19)

where
A0 = x02 + ( v0 n )
v
0 = arctan 0
x0n

(2.20)

The displacement x ( t ) of the mass-spring system can be plotted versus time as shown
in Figure 2.4

13
2/n
v0
A0

x0

Figure 2.4. Free vibrations of mass-spring system

Figure 2.4 shows that vibrations of the mass-spring system are perfectly sinusoidal
and last forever. This is because of an assumption of zero damping, which is not
accurate. In reality, there almost always exists some damping (energy loss) in
vibratory systems. This damping causes the oscillatory motion induced by the initial
disturbance be reduced to zero in the course of time.
For a better understanding of the dynamic behaviour of the mass-spring system it is
useful to consider variation of the mechanical energy in the system with time. The
energy E of the system is composed of the kinetic energy K, which is associated with
the mass motion, and the potential energy P, which is associated with contractions
and expansions of the spring. This energy is given as
E =K+P =

1
1
1
2
2
m ( x ) + kx 2 = m ( x ) + n2 x 2
2
2
2

(2.21)

Substituting into this equation the expression for the mass-spring motion Eq.(2.17),
one obtains
2

2
v0
1
2
E = m ( x0n sin ( n t ) + v0 cos (n t ) ) + n x0 cos (n t ) + sin (n t )
2
n

Performing standard algebraic evaluations, this expression can be reduced to

E=

1
1
1
m ( x02n2 + v02 ) = mv02 + kx02
2
2
2

(2.22)

Eq.(2.22) shows that the mechanical energy of the mass-spring system is time
invariant and equals the sum of the kinetic and potential energies, which the system is
supplied with at the initial time moment.

14

2.2.2. Forced Vibrations under Harmonic Force. Consider the dynamic response of
the mass-spring system to a harmonic (sinusoidal) force. The importance of
considering such a force descends from the fact that any time-dependent force can be
represented as a superposition of harmonic forces with different amplitudes and
phases (see Eq.(2.19) for definition of the amplitude and the phase of the harmonic
motion). Thus, the dynamic behaviour of the mass-spring system under a harmonic
force may be considered as a fundamental knowledge, which facilitates understanding
of the behaviour of the system under a force of arbitrary time signature.
Assume that the external force acting on the mass-spring system is given as
F ( t ) = F0 cos (t ) . In this case the equation of motion Eq.(2.4) takes the form

 kx = F0 cos (t )
mx+

(2.23)

The dynamic behaviour of the mass-spring system subject to an external load is

driven by two mechanisms. The first mechanism is the motion due to initial
conditions, i.e. the free vibrations. The second mechanism is the motion due to the
external force, i.e. the forced vibrations. In the mathematical terms, the free vibrations
correspond to a solution of the homogeneous equation Eq.(2.5), whereas the forced
vibrations are linked to one of the particular solutions of the complete equation
Eq.(2.23). This is a manifestation of a general mathematical property of ordinary,
linear, inhomogeneous differential equations exemplified by Eq.(2.23). The general
solution of this kind of equations is the sum of the general solution of the
homogeneous equation and a particular solution of the complete equation.
The general solution of Eq.(2.5) is given by expression Eq.(2.14). A particular
solution of Eq.(2.23) can be found by employing a simple experience-backed idea that
the forced response of the mass-spring system should be of the same frequency as the
force. Thus, it is reasonable to search for a particular solution in the following form:
x part = X cos (t )

(2.24)

Substituting Eq.(2.24) into the equation of motion Eq.(2.23), the following expression
can be found for the constant X in the particular solution:
X=

F0
F
1
= 0
2
k m
k 1 2 n2

(2.25)

Thus, the general solution of Eq.(2.23) can be written as

x ( t ) = A cos (nt ) + B sin (n t ) +

F0
1
cos (t )
k 1 2 n2

(2.26)

The unknown constants A and B depend on the initial conditions. Employing the
initial conditions Eq.(2.6), the following system of two algebraic equations can be
formulated for determining A and B:

15

A+

F0
1
= x0
k 1 2 n2

(2.27)

Bn = v0

from which A and B can be readily found to give

A = x0
B=

F0
1
k 1 2 n2

(2.28)

v0

Substituting Eq.(2.28) into Eq.(2.26), the following final expression is obtained that
describes the dynamic behavior of the mass-spring system:
x ( t ) = x0 cos (n t ) +

v0

sin (n t ) +

F0
1
( cos (t ) cos (nt ) )
k 1 2 n2

(2.29)

This expression clearly shows that the mass-spring system vibrates at two frequencies,
one being the natural frequency n and the other being the frequency of the
external force. This is also true if the initial displacement x0 and the initial velocity v0
are zero.
If the natural frequency and the frequency of the force are close, an interesting
phenomenon occurs, which is called beating. This is shown in Figure 2.5, where the
ratio x ( t ) ( F0 k ) (note that F0 k is the static displacement of the mass spring
system under the force F0 ) is plotted versus time in accordance with Eq.(2.29). To
plot this figure, the initial conditions were assumed as zero,
= 2 rad s , n = 2.1rad s.
30
20

x(t)k/F 0

10
0
-10
-20
-30
0

50

100
150
time t [s]

200

250

Figure 2.5. Vibration of the mass-spring system in the case that the natural frequency is close to the
forcing frequency

16

Figure 2.5 shows that the vibrations beat with a long period, which is inverse
proportional to the difference between n and . The beating can emerge only in the
systems without (which is unrealistic) of with a small damping, which suppresses it in
the course of time. As will be discussed in Section 2.3, after a sufficiently long time,
the dynamic behavior of the mass spring system subject to a small damping is
approximately described by the particular solution Eq.(2.24). In this case this solution
is called the steady-state solution. It is important to underline that the steady-state
solution is independent of the initial conditions.
Let us study the steady-state solution, which, in accordance with Eqs.(2.24) and
(2.25) is given as
xsteady = X cos (t ) =

F0
1
cos (t )
k 1 2 n2

(2.30)

The dependence X xstatic , xstatic = F0 k being the static deflection, on the ratio n
of the forcing and the natural frequencies is shown in Figure 2.6.
5
4
3

X/xstatic

2
1
0
-1
-2
-3
-4
-5
0

/n

Figure 2.6. Resonance diagram

Figure 2.6, as well as Eq.(2.30), show that for n < 1 the ratio X xstatic is positive ,
whereas for n > 1 it is negative. In order to understand the meaning of this sign
change, we return to Eq.(2.23) and the assumption that the steady-state solution has
the form given by Eq.(2.24). It appears that in the region n > 1 the value of X is
negative. But we can write
X cos (t ) = + X cos (t + )

(2.31)

which shows that a negative amplitude X is equivalent to a positive amplitude of

vibration, which is merely 180 deg. out of phase with (in opposition to) the original
vibration. Physically this means that, while for n < 1 force and motion are in
phase, they are in opposition for n > 1 . This implies that whereas for n < 1
the mass is located to the right from the equilibrium position when the force pushes

17

rightward, for n > 1 the mass is located to the left from the equilibrium position
when the force pushes rightward.
Thus, one may distinguish the amplitude-frequency characteristic and the phasefrequency characteristic in vibrations of the mass spring system (the former is often
referred to as the frequency-response function). To find expressions for these
dependencies, we rewrite Eq.(2.30) as

xsteady

F0
1
cos (t ) ,
k 1 2 n2

n < 1

F0
1
cos (t + ) ,
k 1 2 n2

n > 1

(2.32)

This expression shows that the amplitude of vibrations of the mass-spring system is
given by

X =

F0
1
k 1 2 n2

(2.33)

for all frequencies of the force, whereas the phase lag between the force and the mass
changes abruptly from 0 to as the forcing frequency becomes larger than the
natural frequency:
0,

=
,

n < 1

(2.34)

n > 1

The amplitude- and phase-frequency dependencies are shown in Figures 2.7a and
2.7b.
5

X/xstatic

0
0

/n

Figure 2.7a. Amplitude-frequency characteristic (magnification factor)

18

180

90

0
0

/n

Figure 2.7b. Phase-frequency characteristic

Analyzing the amplitude-frequency characteristic depicted in Figure 2.7a, a few

remarks can be made. First, consider the part of the characteristic, which is close to
= 0 . The forcing frequency here is extremely slow, and the mass will be deflected
by the force to its static deflection only. This is the reason for the curve to be nearly
equal to unity. On the other hand, for very high frequencies n >> 1 , the force
changes its direction so fast that the mass simply has no time to follow and the
amplitude of the response is very small.
The most interesting thing happens at n = 1 , where the amplitude of the
response becomes infinitely large. This can be explained in the following manner. At
n = 1 , the frequency of the force coincides exactly with the natural frequency.
Therefore, the force can push the mass always in the direction, in which the mass
moves itself, thereby increasing the amplitude of vibrations indefinitely. This
phenomenon is known as resonance.
It is clear that any force needs some time (actually, an infinite time) to amplify the
mass response to infinity. To find how the process of the amplitude increase takes
place in time, we return to Eq.(2.29). Assuming for simplicity that the initial
conditions are zero (the time signature of the resonant growth of the mass amplitude
cannot be affected by initial conditions), this equation can be reduced to
x (t ) =

F0
1
( cos (t ) cos (nt ) )
k 1 2 n2

(2.35)

Obviously, at n = 1 both the numerator and the denominator of this expression are
zero. To get rid of this uncertainty, one may differentiate the numerator and the
denominator with respect to (according to the LHospitals rule). This yields
x (t ) =

F0 t sin (t )
F
= 0 t sin (t )
2
k 2 n = 2k
n

(2.36)

19

Eq.(2.36) shows that the amplitude of resonant vibrations grows in time linearly
and the displacement of the mass-spring system increases as shown in Figure 2.8 for
= 2 rad s .
20

x(t)k/F0

10

-10

-20
0

10

15

20

25

t [s]

Figure 2.8. Development of resonance in time

As it was shown in the previous section, the energy of free vibrations of the massspring system does not change in time. The presence of an external force, naturally,
changes the situation leading to variation of the energy of the mass-spring system in
time. This energy remains limited unless resonance occurs.
2.2.3. Forced Vibrations under General Disturbing Force. In most practical
applications the dynamic loading F ( t ) is irregular and non-periodic. In this section
two ways of analysis are presented of the response of the mass-spring system to such
loading. The first way is based on the representation of a time-dependent force by a
sequence of short impulses. Superposition of the system responses to these impulses
is found in the form of a convolution integral, which is often called Duhamels
integral after the French mathematician J.M.C. Duhamel. The second way is based on
application of the Laplace transform (named after the French scientist P.S. Laplace).
In many cases, application of the Laplace transform implies physically that the
loading F ( t ) is first represented as a continuous superposition (integral) of its
harmonic components so that the response to each of these harmonics can be found
separately. These responses are then integrated by means of the inverse Laplace
transform to give the dynamic response to the original non-harmonic load.
Consider Eq.(2.4) with the initial conditions Eq.(2.6):

mx+ kx = F ( t )
x ( 0 ) = x0

(2.37)

x ( 0 ) = v0

assuming that the external force F ( t ) is an arbitrary function of time, an example of

which is shown in Figure 2.9.

20

t
t'

dt'

Figure 2.9. Time dependent force of general type

For any instant of time we may consider one elementary impulse F ( t ) dt as

represented by the shaded strip in Figure 2.9. In accordance with Newtons second
law, this one impulse imparts to the mass an instantaneous increase in velocity
dx = F ( t ) dt m regardless of what other forces, such as the spring force, may be
acting upon the mass and regardless of its displacement and velocity at the instant t .
Treating this increment of velocity as if it were an initial velocity at the instant t and
using Eq.(2.17), we conclude that the corresponding displacement of the mass-spring
system at any later time will be
F ( t ) dt
(2.38)
x (t ) =
sin (n ( t t ) )
mn
Since each impulse F ( t ) dt between t = 0 and t = t has a like effect, we obtain, as

a result of the continuous action of the external force F ( t ) , the following

displacement of the mass
x (t ) =

1
F ( t ) sin (n ( t t ) ) dt
mn 0

(2.39)

This expression still does not include the effect of any initial displacement x0 or any
initial velocity v0 when t = 0 . These effects, however, are exactly those represented
by Eq.(2.17). Hence, for a complete solution of Eq.(2.37), we may write
v0

1
x ( t ) = x0 cos (n t ) + sin (n t ) +
F ( t ) sin (n ( t t ) ) dt
mn 0
n

(2.40)

The integral in Eq.(2.40) is called Duhamels integral.

At first glance it might appear that the first two terms in this expression represented
free vibrations and the last term forced vibrations as distinguished in the previous
Section. This, however is not quite the case. Actually, the first two terms represent the
effect of initial displacement and initial velocity only, whereas the last term represents
the complete effect of the external force F ( t ) . This force produces on its own

21

account both free and forced vibrations and all this together is accounted for by the
integral Eq.(2.39).
The solution Eq.(2.40) was obtained using some physical considerations. It is
necessary, therefore, to check whether this solution indeed satisfies the original initialvalue problem Eq.(2.37). To perform this check, one has to substitute Eq.(2.40) into
both the equation of motion and the initial conditions. Such a substitution requires
taking the time derivative of the integral in Eq.(2.40), which has the upper limit
dependent on the time t . The following general formula should be used in this case,
which can be found in any book on advanced mathematical analysis:
b( t )

b(t )

db
da
f ( , t ) d =
f ( , t ) d + f ( b ( t ) , t ) f ( a ( t ) , t )

dt a ( t )
dt
dt
t
a(t )

(2.41)

Using this formula the first and the second time-derivatives of Eq.(2.40) can be
written as
x ( t ) = n x0 sin (n t ) + v0 cos (nt ) +

1
F ( t ) cos (n ( t t ) ) dt
m 0


x ( t ) = n2 x0 cos (nt ) n v0 sin (n t )

n
m

F ( t) sin ( ( t t)) dt +
n

F (t )

(2.42)

Substitution of these expressions into the initial-value problem Eq.(2.37) proves that
Eq.(2.40) is indeed the solution to this problem.
For some applications, it is convenient to rewrite the integral in Eq.(2.40) using the
following trigonometric relation:
sin (nt nt ) = sin (n t ) cos (nt ) cos (n t ) sin (n t )
Substituting this relation in Eq.(2.40), we obtain
x ( t ) = x0 cos (n t ) +

v0

sin (n t ) + A ( t ) cos (n t ) + B ( t ) sin (n t )

(2.43)

where
A (t ) =

1
F ( t ) sin (n t )dt ,
mn 0

B (t ) =

1
F ( t ) cos (n t )dt
mn 0

(2.44)

As a first example, let us consider the case of harmonic excitation

F ( t ) = F0 cos (t ) , which was studied in the previous section. In this case, the
integral in Eq.(2.40) becomes

22
t

F
1
F ( t ) sin (n ( t t ) )dt = 0 cos (t ) sin (n ( t t ) )dt

mn 0
mn 0

(2.45)

F0
F
1
cos (t ) cos (nt ) ) = 0
=
( cos (t ) cos (nt ) )
2
2 (
k 1 2 n2
m ( n )

Substituting this into Eq.(2.40), we obtain the following expression, which coincides
with Eq.(2.29)
x ( t ) = x0 cos (n t ) +

v0

sin (n t ) +

F0
1
( cos (t ) cos (nt ) )
k 1 2 n2

An advantage of using the general formula Eq.(2.40) relative to the approach

discussed in the previous section is that applying Eq.(2.40) one does not need to make
any physical guess about the form of a particular solution but just calculate the
Duhamels integral. On the other hand, excluding the physical guess from the
procedure introduces a risk of loosing the link with underlying physics.
The general formula Eq.(2.40) can be used to advantage in studying the motion of
the mass-spring system under the action of a series of discontinuous impulses, see
Figure 2.10 for two examples. If, owing to such impulses, the mass obtains increments
of velocity v0 , v1 , v2 ,... at the moments t0 , t1 , t2 ,... , we have (in the case of zero
initial conditions)
x=

( v sin ( ( t t ) ) + v sin ( ( t t ) ) + v sin ( ( t t ) ) + ...)

0

(2.46)

(b)

(a)
2

t2

t1

t0

/n

-1

1
-2

/n

Figure 2.10. Discontinuous impulses

If the pulses follow each other with a constant time delay equal to 2 n as show in
Figure 2.10(a), this delay being equal to the period of natural oscillations of the massspring system, then Eq.(2.46) takes the form (which is obtained by substitution
tn = t0 + 2 n n )

23

x=

( v sin ( ( t t )) + v sin ( (t t ) ) + v sin ( (t t )) + ...)

0

=
=
=

( v sin ( (t t )) + v sin ( ( t t 2 )) + v sin ( (t t 4 ) ) + ...)

( v sin ( (t t )) + v sin ( ( t t ) 2 ) + v sin ( (t t ) 4 ) + ...)

sin (n ( t t0 ) ) vk

k =0

This expression clearly shows that the amplitude of the response will be built up
without limit in the course of time. The same result will follow from a succession of
positive and negative impulses, which are shown in Figure 2.10(b). In this case the
time delay equals n so that tn = t0 + n n should be substituted in Eq.(2.46) to
give
x=

( v sin ( (t t ) ) + v sin ( ( t t ) ) + v sin ( ( t t ) ) + v sin ( ( t t ) ) + ...)

0

( v sin ( ( t t ) ) + v sin ( ( t t
0

n ) )

+v2 sin (n ( t t0 2 n ) ) + v3 sin (n ( t t0 3 n ) ) + ...

=

( v sin ( ( t t ) ) + v sin ( ( t t ) )
0

+v2 sin (n ( t t0 ) 2 ) + v3 sin (n ( t t0 ) 3 ) + ...

( v sin (
0

( t t0 ) ) v1 sin (n ( t t0 ) )

+v2 sin (n ( t t0 ) ) v3 sin (n ( t t0 ) ) + ...

=

sin (n ( t t0 ) ) vk + vk
n
k = 0,2,4
k =1,3,5

These observations lead to the important practical conclusion that any periodic
external force, the period of which equals the period of natural vibrations causes
resonance, i.e. a large-amplitude response.
As a third example consider the external force to be suddenly applied to the mass
and then remain constant, i.e.

0,
F (t ) =
F0 ,

t<0
t0

Substituting this expression in the general formula Eq.(2.40), we obtain

(2.47)

24

x ( t ) = x0 cos (n t ) +

v0

sin (n t ) +

F0
(1 cos (nt ) )
mn2

(2.48)

Noting that
F0
F
= 0 = xstatic
2
mn
k

representing the static displacement corresponding to the constant force F0 , we

rewrite Eq.(2.48) in the form
x ( t ) = xstatic + ( x0 xstatic ) cos (n t ) +

v0

sin (n t )

(2.49)

From this expression we see that a suddenly applied constant force produces free
vibrations of amplitude
A=

( x0 xstatic )

v
+ 0
n

(2.50)

superimposed upon the static displacement. In the particular case where the initial
conditions are zero, Eq.(2.50) for the amplitude, reduces to A = xstatic and we have
xmax = 2 xstatic and xmin = 0 . Thus, a suddenly applied constant force produces a
maximum deflection twice as great as the static effect of the same force.
Eq.(2.47) corresponds to the constant force, once applied, to act indefinitely. If it
acts only for an interval of time t and then is suddenly removed, i.e.
t < 0 and t > t

0,
F (t ) =
F0 ,

(2.51)

0 t t

then the resulting displacement of the mass can be found as

F0
t t
m 2 (1 cos (nt ) ) ,
v0
n
x ( t ) = x0 cos (nt ) + sin (nt ) +
(2.52)
F
n
0

cos (nt ) cos (n ( t t ) ) , t t

mn2
Note that the response x ( t ) is expressed differently at the time intervals t t and

t t . This is because of different upper limits, which should be used in the

Duhamels integral when calculating the response at these intervals. At t t the
upper limit should be t , while at t t it should be t .
The expression for x ( t ) at t t (after the external force has stopped acting) can
be rewritten as

x ( t ) = cos (nt ) x0 + xstatic ( cos (n t ) 1) + sin (nt ) 0 + xstatic sin (n t )

n

(2.53)

25

This shows that after removal of the force, the mass-spring system oscillates at its
natural frequency. The amplitude of these vibrations in the case of zero initial
conditions reads
A = xstatic

( cos ( t ) 1)
n

t
2
t
+ sin (n t ) = 2 xstatic sin n = 2 xstatic sin
(2.54)
2
Tn

where Tn = 2 n is the natural period of oscillations. Eq.(2.54) shows that this

amplitude depends on the ratio t Tn , i.e. on the ratio of the duration of the constant
force to the natural period of the mass-spring system. By taking t Tn = 1 2 , we
obtain A = 2 xstatic . By taking t Tn = 1 , we obtain A = 0 , i.e. there will be no
vibrations after removal of the constant force F0 . In the first case, F0 acts through the
displacement from 0 to A and does positive work on the system. After removal of the
force in the extreme position, the system retains its energy and we have free vibrations
equivalent to those excited due to initial displacement x0 = 2 xstatic . In the second case,
the constant force does positive work from 0 to A and negative work from A back to
0, thus the nett work becomes zero and the system acquires no increase of energy. If
the mass was initially at rest in its equilibrium position, it must remain at rest in its
equilibrium when the force is taken away.
The general formula Eq.(2.40) allows to calculate the dynamic behaviour of the
mass-spring system under arbitrary external force and initial conditions. Therefore, it
might seem unnecessary to discuss other methods of analysis. This is true but only for
simple dynamical systems, for which an analogue of Eq.(2.40) is known or can be
found straightforwardly. For systems with many degrees of freedom finding an
analogue to Dunamels integral proves to be laborious and the other method becomes
more advantageous, namely the method based on Laplace transform. This method is
briefly discussed below.
The Laplace transform of a function G ( t ) is defined as follows:

g ( s ) = G ( t ) exp ( st ) dt

(2.55)

where G ( t ) is a function of the real variable t and s is a complex variable. G ( t ) is

called the original function and g ( s ) is called the image function.

The operations, which are valid for the Laplace transform are shown in Table 2.1.
In the inversion formula in this table, is a real value, which should be greater than
the real parts of all singularities of the image function g ( s ) . In the translation
property in Table 2.1 (the last row), a generalized Unit Step Function is introduced.
This function is often called the Heaviside function and is defined as

0,
H ( t t0 ) =
1,

t < t0
t > t0

(2.56)

26

Original Function G ( t )

G (t )

Definition

Image Function g ( s )

G ( t ) exp ( st ) dt
0

g ( s)

+ i

Inversion
Formula

1
g ( s ) exp ( st ) dt
2 i i

Linearity
Property
Differentiation

AG1 ( t ) + BG2 ( t )
G ( t )

sg ( s ) G ( 0 )

G ( t )
G(

n)

Ag1 ( s ) + Bg 2 ( s )
s 2 g ( s ) sG ( 0 ) G ( 0 )

(t )

s n g ( s ) s n 1G ( 0 ) ...G (

Integration

Convolution
Theorem

G ( t ) G ( ) d
1

(0)

1
g (s)
s

G ( ) d

n 1)

g1 ( s ) g 2 ( s )

Translation

G (t b) H (t b) , b > 0

exp ( bs ) g ( s )

Table 2.1. Operations for Laplace transform

To demonstrate the way, in which Laplace transform should be applied, let us

solve the initial-value problem Eq.(2.37). Applying Laplace transform to this problem
x ( t ) , we obtain the following
and using the differentiation property for transforming 
algebraic equation from Eq. (2.37):
m ( s 2 x ( s ) sx0 v0 ) + kx ( s ) = f ( s )

(2.57)

The image function can be promptly found from Eq.(2.57) to give

x (s) =

f ( s ) + msx0 + mv0

(2.58)

ms 2 + k

The remaining task is to apply the inversion formula to determine the original
function x ( t ) . This can be done by applying the contour integration method
supplemented by the residue theorem. In this development, however, there is no room
for studying this method. Therefore, a table of Laplace transforms will be used, a
short version of which is presented below. For finding other pairs of Laplace
transforms, one may use MAPLE.

g ( s)

G (t )

1s
1 s2

1
t

1 sn

t n 1 ( n 1)!

27

g ( s)

G (t )

1 t

( n +1 2 )

, n = 1, 2,3,...

2n t n 1 2

1 (s + a)

exp ( at )

1 (s + a)

t exp ( at )

1 (s + a)

t n 1 exp ( at ) ( n 1) !

(a b)

( s + a )( s + b )
s

(a b)

( s + a )( s + b )

exp ( at ) exp ( bt )
ba
a exp ( at ) b exp ( bt )
a b

1
s + a2
s ( s2 + a2 )

1
sin ( at )
a
cos ( at )

1 ( s2 a2 )

sinh ( at ) a
cosh ( at )

s
s a2
1
2
s ( s + a2 )
2

1
(1 cos ( at ) )
a2

1
s ( s + a2 )

1
( at sin ( at ) )
a3

1
( sin ( at ) at cos ( at ) )
2a 3

(s

+a

2 2

(s

+a

(s

t
sin ( at )
2a

2 2

1
( sin ( at ) + at cos ( at ) )
2a

+ a2 )

t cos ( at )

s2 a2

(s
(s

+a

+a

)( s

+b

2 2

(( 2n 1)! )

(a

1
( s + a 2 ) + b2
2

s+a
( s + a 2 ) + b2
2

b2 )

cos ( at ) cos ( bt )
b2 a2
1
exp ( at ) sin ( bt )
b

exp ( at ) cos ( bt )

Table 2.2. Table of Laplace transforms

28

Let us now transform Eq.(2.58) to the time domain. To this end we rewrite this
equation as follows:
x (s) =

1
1
s
1
+ x0 2
+ v0 2
f (s) 2
2
2
m
s + n
s + n
s + n2

(2.59)

In accordance with the linearity property (see Table 2.1), every of the three terms in
Eq.(2.59) may be inverted separately. To invert the first term, we use the convolution
theorem (Table 2.1), which states that a multiplication of images is inverted into
convolution of the originals. Since, by definition, the original to f ( s ) is F ( t ) ,
whereas the original to 1 ( s 2 + n2 ) is 1 n sin (n t ) (Table 2.2), the first term is
inverted into the Duhamels integral:
t

1
1
1

f (s) 2
F ( t ) sin (n ( t t ) )dt
2
m
s + n mn 0

(2.60)

In Eq.(2.60) and in what follows the sign indicates a pair of Laplace-original and
Laplace-image.
The second and third terms in Eq.(2.60), according to Table 2.2, have the following
originals:
x0

s
x0 cos (nt )
s + n2
2

(2.61)

v
1
0 sin (n t )
v0 2
2
s + n n

Thus the complete result of inversion coincides with Eq.(2.40):

v0

1
x ( t ) = x0 cos (n t ) + sin (n t ) +
F ( t ) sin (n ( t t ) ) dt
n
mn 0

(2.62)

Let us consider the case of a suddenly applied constant load given by Eq.(2.47). The
Laplace image f ( s ) of this load, according to the first row of Table 2.2 reads
f (s) =

F0
s

Thus, the first term in the general image-expression Eq.(2.59) takes the form
F0
1
2
m s ( s + n2 )

In accordance with Table 2.2 this corresponds to the following Laplace-original

29

F0
F
1
0 2 (1 cos (n t ) )
2
2
m s ( s + n ) mn

(2.63)

Combining Eq.(2.63) and Eq.(2.61), we obtain the solution given by Eq.(2.48).

Concluding this section, we can say that both the method of convolution integral
and the method of Laplace transform may be applied to find the behaviour of a
dynamic system. The latter method, however, is formalised much better and can be
used for analysis of complex dynamic systems.

2.3.

One Degree of Freedom Systems with Viscous Damping

In the previous section, we considered the mass-spring system subject to no damping.

This model is very helpful for understanding of fundamentals of the vibration theory
but is unrealistic. Every engineering system experiences damping that dissipates the
energy of vibrations. In this section, the mass-spring system is considered to be
subject to viscous damping, which is proportional to the velocity of the mass.
The system under consideration is shown in Figure 2.11, which differs from Figure
2.3 by a dashpot with damping coefficient c that connects the mass and the wall in
parallel with the spring.
x(t)
m

F(t)
x

Figure 2.11. Mass-spring-dashpot system

The dashpot introduces an additional force FDASHPOT ( t ) on the mass relative to the
mass-spring system depicted in Figure 2.3. Therefore, the balance of forces as dictated
by Newtons second law, has one additional term as compared to Eq.(2.2). This
balance reads

F ( t ) + FSPRING ( t ) + FDASHPOT ( t ) = m

x

(2.64)

By definition, the viscous resistance associated with the dashpot is given by

FDASHPOT ( t ) = cx

(2.65)

Substituting this expression into Eq.(2.64) and projecting the result onto the x axis,
the following scalar equation is obtained
 kx = F ( t )
mx+ cx+

(2.66)

30

which describes small vibrations of the mass-spring-dashpot system near its

equilibrium position under the action of the external force F ( t ) .
2.3.1. Free Vibrations. If the external force F ( t ) is absent, the equation of motion
for the mass-spring-dashpot system reduces to
mx+ cx + kx = 0

(2.67)

Dividing through by m and introducing the notations

k m = n ,

c m = 2n

(2.68)

where n is the natural frequency of the undamped (corresponding to c = 0 ) system,

and n is a measure for the viscous damping in the system, we obtain the following
linear homogeneous differential equation of the second order:

x + 2nx + n2 x = 0

(2.69)

This equation governs small free vibrations of the mass-spring-dashpot system.

The general solution of the second order differential equation Eq.(2.69) can be
written as
2

x ( t ) = X n exp ( sn t )

(2.70)

n =1

Substituting this expression in the equation of motion Eq.(2.69), the following

characteristic equation is obtained
sn2 + 2ns + n2 = 0

(2.71)

Note that the characteristic equation Eq.(2.71) could be obtained by applying the
Laplace transform with zero initial conditions to Eq.(2.69).
The characteristic exponents sn (the roots of Eq.(2.71)) can be easily found for the
second-order polynomial Eq.(2.71) to give

s1 = n + n 2 n2 ,

s2 = n n 2 n2

(2.72)

Hence, the general solution of Eq.(2.69) can be written as

x ( t ) = X 1 exp ( s1t ) + X 2 exp ( s2t )

(2.73)

where X 1 and X 2 are arbitrary constants.

To attach any significance to this solution, we must distinguish between two cases,
depending upon whether the radical

n 2 n2 is real or imaginary, i.e., on whether

31

n > n or n < n . Going back to notations introduced by Eq.(2.68), we see that this
rests on the relative magnitude of the damping coefficient c and the spring constant
k . Generally speaking, a large coefficient of damping and a small spring constant will
result in real values of s1 and s2 while, for the reverse of these conditions, s1 and s2
will be complex numbers.
Case 1, n > n . In this case, s1 and s2 have real values. To evaluate the constants
X 1 and X 2 in the general solution Eq.(2.73), we must impose the initial conditions of
the motion. Assuming as a particular case, that
x ( 0 ) = x0 ,

x ( 0 ) = 0

(2.74)

we find by substituting Eq.(2.73) into these conditions that

X1 =

s2 x0
,
s1 s2

X2 =

s1 x0
s1 s2

(2.75)

With these values, Eg.(2.73) becomes

x (t ) =

x0
( s1 exp ( s2t ) s2 exp ( s1t ) )
s1 s2

(2.76)

In connection with this solution, it should be noted that both s1 and s2 are negative,
s2 having a bigger absolute value. Thus, the displacement x of the mass has the same
sign as x0 and approaches zero as a limit when the time tends to infinity. The
displacement-time diagram plotted in accordance with Eq.(2.76) is shown in Figure
2.12.

x0

Figure 2.12. Aperiodic free motion of mass-spring-dashpot system; n > n

We see that the motion is not a vibration at all but simply one in which the suspended
mass, after its initial displacement, gradually creeps back toward the equilibrium
position but takes theoretically infinite time to get there. This is called aperiodic
motion. It results from the fact that the damping coefficient is too large relative to the
spring constant.
In the special case n = n , we also obtain aperiodic motion, and the corresponding
value of the damping coefficient

32

c = 2nm = 2 km

(2.77)

is called critical damping.

Case 2, n < n . In mechanics, more often we encounter the case in which n < n
so that the roots s1 and s2 are complex. In discussing this case, it is convenient to
change the form of Eq.(2.72) in order to see more clearly its physical significance.
Introducing the following real positive value

1 = n2 n2

(2.78)

which has the dimension of radiant frequency, we rewrite Eq.(2.72) as

s1 = n + i1 , s2 = n i1

(2.79)

Using these expressions for the characteristic exponents, and applying the Euler
formula in the same way, as it was done in Eq.(2.14), Eq.(2.73) may be rewritten as
x ( t ) = exp ( nt ) ( A cos (1t ) + B sin (1t ) )

(2.80)

where A and B are new arbitrary constants. To evaluate these constants, we assume
the following general initial conditions:
x ( 0 ) = x0

(2.81)

x ( 0 ) = v0

Substituting Eq.(2.80) into the initial conditions Eq.(2.81) and resolving the soobtained system of two algebraic equations, we find
A = x0 ,

B=

v0

nx0

(2.82)

and our solution becomes

v nx
x ( t ) = exp ( nt ) x0 cos (1t ) + 0 + 0 sin (1t )
1 1

(2.83)

Similar to Eq.(2.17), Eq.(2.83) can be written as

x ( t ) = A0 exp ( nt ) cos (1t 0 )
where the amplitude A0 and initial phase 0 are given as

(2.84)

33

v nx
A0 = x + 0 + 0
1 1

2
0

(2.85)

v + nx0
0 = arctan 0

x01

A displacement-time curve, plotted in accordance with Eq.(2.85), is shown in

Fugure 2.13, and we see that the motion now is vibratory in nature and represents
damped free vibrations. The main difference between the damped free vibrations of
the mass-spring-dashpot system (Figure 2.13) and those of the mass-spring system
(Figure 2.4), which are undamped, is the attenuation of vibrations in the former case.
Thus, the main effect of damping on free vibrations is that it extracts energy from the
system leading to a decay of the amplitude of vibrations.

A0exp(-nt)

A0
T 1=

2
1

x0
0
1

t
-A0 exp(-nt)

Figure 2.13. Free vibrations of mass-spring-dashpot system; n < n

Figure 2.13 shows that each time that cos (1t 0 ) becomes equal to 1 , the

time-displacement curve is tangent to one of the envelopes x = A0 exp ( nt ) . This

quantity is called the amplitude of vibration. We see that, owing to damping, the
amplitude gradually diminishes with time and that the rate of decay depends on the
damping factor n . When n = 0 , (no damping), the amplitude of vibrations remains
constant as shown in Figure 2.4. The time T1 = 2 1 required to complete one cycle
of the motion is called the period of vibration, and its reciprocal f1 = 1 2 is called
the frequency of vibration. Eq.(2.78) shows that these quantities also depend on the
damping parameter n . The initial phase 0 , as defined by Eq.(2.85), is often also
called the phase angle. Figure 2.13 shows that it determines the lapse of time between
the initial instant t = 0 and the first point of tangency with the envelopes
x = A0 exp ( nt ) .
Let us consider the amplitude, period and phase angle in more detail. We begin
with a consideration of the amplitude of vibration as represented graphically by the
exponential curves x = A0 exp ( nt ) in Figure 2.13. This figure shows that the
displacement-time curve is tangent to these curves at the ordinates t = 0 1 ,
t = (0 + ) 1 , t = (0 + 2 ) 1 , etc. It can be noted that these points do not

34

coincide exactly with the points of the curve representing extreme positions of the
vibrating mass and that, owing to damping, the time interval needed for the mass to
move from an extreme position to a subsequent middle position is greater than that
needed to move from a middle position to the next extreme position. The rate at which
the amplitude diminishes depends upon the damping factor n and can be calculated in
the following manner. Let x1 be the displacement at the first point of tangency with
the exponential curve and xs the displacement after ( s 1) complete cycles. Then it is
evident from Eq.(2.84) that
x1
= exp ( ( s 1) nT1 ) or
xs

x
ln 1 = ( s 1) nT1
xs

(2.86)

These expressions show that the amplitudes at the ends of successive cycles diminish
as a geometric progression. The quantity
nT1 =

2 n

(2.87)

n2 n 2

on which the rate of decay depends, is called the logarithmic decrement of the
amplitude. It can be determined experimentally by observing in what proportion the
amplitude is diminished after an arbitrary number of cycles and then using Eq.(2.86).
Using notation (2.78), we can write the expression for the period of damped free
vibrations as
T1 =

(2.88)

n2 n 2

As the value of n varies from 0 (no damping) to n (critical damping), the period T1
varies from the natural period of undamped vibrations 2 n to infinity, while the
frequency of vibration

n2 n 2
1
f1 = =
2
T1

(2.89)

varies from the frequency of the undamped vibrations n 2 to 0. For small values
of n n , which we usually have in practice, Eq.(2.89), taking the first two terms in
the Taylors expansion of the radical, can be simplified to

n2
f1 = f n 1 2 ,
2n

where

fn =

n
2

(2.90)

This expression shows that the frequency of the damped vibrations differs from that of
the undamped vibrations only by a small quantity of second order. Consequently, in

35

most practical cases, the period of the damped vibrations is considered to be

approximately the same as the natural period of the undamped vibrations:
2
T1 Tn =
(2.91)

We note that the period of vibrations is completely independent of the initial

conditions and depends only on physical constants of the system.
The initial phase (phase angle) 0 is of less practical interest than either the
amplitude or the period. By a proper choice of the initial velocity v0 and initial
displacement x0 , the phase angle 0 can be always set to zero.
2.3.2. Forced Vibrations under Harmonic Force. Consider the dynamic response of
the mass-spring system to a harmonic (sinusoidal) force, which is given as
F ( t ) = F0 cos (t ) . In this case the equation of motion of the mass-spring-dashpot
system under an external force Eq. (2.66) takes the form:
 kx = F0 cos (t )
mx+ cx+

(2.92)

Dividing through by m and introducing the notation f 0 = F0 m , we obtain


x + 2nx + n2 x = f 0 cos (t )

(2.93)

where n2 and n are defined by Eq.(2.68).

The general solution of Eq.(2.92) is the sum of the general solution of the
homogeneous equation and a particular solution of the complete equation. The general
solution of Eq.(2.92) is given by expression Eq.(2.80). A particular solution of
Eq.(2.92) cannot be found in the form
x part = X cos (t )

(2.94)

like it was done for the undamped mass-spring system. The reason is that substituting
Eq.(2.94) into Eq.(2.93) we would get the following equation

+ n2 ) X cos (t ) 2n X sin (t ) = f 0 cos (t )

(2.95)

in which the time-dependent terms can not be cancelled.

One form of a particular solution, which can be used is given as
x part = X c cos (t ) + X s sin (t )

(2.96)

where X c and X s are constants, the subscripts of which show that they are the
multipliers of the cosinus and sinus, respectively. To determine these constants we
substitute Eq.(2.96) into Eq.(2.93) and obtain

36

( X
2

+ 2n X s + n2 X c f0 ) cos (t ) + ( 2 X s 2n X c + n2 X s ) sin (t ) = 0 (2.97)

This equation can be satisfied for all values of t only if the expressions in parentheses
vanish. Thus, for calculating X c and X s , we have the following system of linear
algebraic equations
2 X c + 2n X s + n2 X c = f0

(2.98)

2 X s 2n X c + n2 X s = 0

from which we find

X c = f0
X s = f0

2
n

2
n

2
n

2 )

2 ) + 4n2 2
2

(2.99)

2n

2 ) + 4n2 2
2

Substituting these constants into Eq.(2.96) we obtain the particular solution of

Eq.(2.93).
The particular solution Eq.(2.96) can be also found in a slightly different manner
by searching for it in the following form (Re implies the real part )

x part = Re X exp ( it )

Substituting
Eq.(2.100)
into
Eq.(2.93)
cos (t ) = Re ( exp ( it ) ) , we obtain

(2.100)
and

using

Re ( 2 + 2in + n2 ) X exp ( it ) f 0 exp ( it ) = 0

the

fact

that

(2.101)

This equation can be satisfied for all values of t only if the following equation with
complex coefficients is satisfied:

+ 2in + n2 X = f 0

(2.102)

from which the complex amplitude X of vibrations can be found as

X =

f0
+ 2in + n2
2

Thus, the particular solution is given by the following expression

(2.103)

37

f0
exp ( it )
x part = Re
2
2
+ 2in + n

f0
cos (t ) + i sin (t ) )
= Re
2
2 (
+ 2in + n

( 2 2 ) 2in

= f 0 Re
( cos (t ) + i sin (t ) )
( 2 2 )2 + 4n 2 2
n

f0
=
(n2 2 ) cos (t ) + 2n sin (t )
2
2
2
2 2
( n ) + 4 n

(2.104)

which coincides with Eq.(2.96) supplemented by Eq. (2.99). The latter way of
obtaining the particular solution might seem to be more laborious but it is the way that
is mainly used in advanced research on dynamics of systems. The reason for that is
that the complex form of the particular solution
x part = X exp ( it )

(2.105)

is equally suitable for finding this solution in the case of the cosinusoidal external
force F ( t ) = F0 cos (t ) and the sinusoidal one F ( t ) = F0 sin (t ) . In the later case
the particular solution is simply given by

x part = Im X exp ( it )

(2.106)

where the complex amplitude X is defined by Eq.(2.103). A particular implication of

this is that the complex amplitude carries information concerning not only the
dynamic response to the sinusoidal or cosinusoidal force but to a harmonic force of
arbitrary phase.
Let us write now the general solution of Eq.(2.93), which is the sum of Eq.(2.80)
and Eq.(2.96):
x ( t ) = exp ( nt ) ( A cos (1t ) + B sin (1t ) ) + X c cos (t ) + X s sin (t )

(2.107)

This solution represents the superposition of damped free vibrations as represented by

the first two terms, and damped forced vibrations, as represented by the last two
terms. The free vibrations have the period
T1 =

as discussed in the preceding section, and the forced vibrations have the period
T=

38

identical with the period of the external force that produces them.
Because of the factor exp ( nt ) , which is due to the viscous damping, the free
vibrations gradually subside, leaving only the steady forced vibrations represented by
the last two terms, i.e., by the particular solution Eq. (2.96). These forced vibrations
are maintained indefinitely by the action of the disturbing force and therefore are of
great practical importance.
Let us consider the forced vibrations in detail. To facilitate this, we rewrite the
steady-state solution Eq.(2.96) in a new form, introducing the amplitude and phase lag
of vibrations, similar to Eq.(2.31). This new form reads
x part = X cos (t )

(2.108)

where
X = X c2 + X s2 =

Xc
Xs

= arctan

f0

(2.109)

(n2 2 ) + 4n2 2
2

2n

arctan
=

( n 2 )

(2.110)

Note that the amplitude X and the phase lag can be obtained directly from the
complex amplitude X of the steady-state vibrations, which is defined by Eq.(2.103).
Indeed, as any complex value, this complex amplitude can be rewritten as

( ( ))

X = X exp i arg X

(2.111)

where

( )

X = Re X

( )

+ Im X

( )
( )

f0

2
n

2 ) + 4n 2 2
2

Im X

= arctan 2n =
arg X = arctan
2
2

Re X
( n )

( )

(2.112)

It is easy to see that the amplitude and phase lag of the steady-state vibrations are
equal to the modulus and the argument (with minus) of the complex amplitude X .
From Eq.(2.108) we see that steady-state forced vibration with viscous damping is
a simple harmonic motion having constant amplitude X and phase lag with the
force. As mentioned before, the period of the forced vibration T = 2 is the same
as that of the external force regardless of the natural period of the system and
regardless of the amount of damping.

39

Let us study the amplitude-frequency characteristic X ( ) . Extracting explicitly

the static deflection xstatic = F0 k as a multiplier in Eq.(2.109), we obtain
F
1
1
X = 0
= xstatic
2
2
2
k
2
2
2 2n 2
2
1
1

2
2
n2
n n n
n

(2.113)

where the new notation is introduced

2n

(2.114)

The ratio of the amplitude of the steady-state vibration X and the corresponding
static deflection xstatic = F0 k is referred to as magnification factor or dynamic
amplification factor. In accordance with Eq.(2.113), it reads

X
xstatic

(2.115)

2
1 2 + 2
n
n
2

One can see that this magnification factor depends on the amount of damping, as
represented by , and on the ratio n of the frequency of the external force and
the natural frequency of the undamped vibrations.
In Figure 2.14, the values of the magnification factor X xstatic are plotted for
various values of the amount of damping against the ratio n .
5.0

=0

4.5
4.0

= 0.2

X /xstatic

3.5
3.0

| 2.5
| 2.0
1.5
1.0
0.5

= 0.3
= 0.5

=1
=2

0.0
0.0

0.5

1.0

1.5

2.0

2.5

/n

Figure 2.14. Magnification factor

From this figure we see that when the frequency of the force is small compared
with the natural frequency n , the value of the magnification factor is not greatly

40

different from unity, regardless of the amount of damping. This means that the
amplitude of the steady-state vibration at the frequency band << n is
approximately equal to the static displacement xstatic . Owing to this property, the
vibrations at this frequency band are often called quasi-static vibrations.
When >> n , i.e., when the frequency of the force is much greater that the
natural frequency, the value of the magnification factor tends to zero, regardless of the
amount of damping. This means that a high-frequency external force produces
practically no forced vibrations of a system that has a low natural frequency.
Since in both extreme cases << n and >> n , the damping has only a
marginal effect on the magnification factor, it is justifiable in discussing forced
vibrations to neglect the effect of damping entirely, in which case Eq.(2.115) takes the
much simpler form given by Eq.(2.33):
X =

F0
1
k 1 2 n2

As the value of approaches that of n , i.e. as the frequency of the force gets
close to the natural frequency, the magnification factor grows rapidly and, as we see
from Figure 2.14, its value is very sensitive to the amount of damping as represented
by . It can also be noted that the maximum value of the magnification factor occurs
for a value of n slightly smaller than unity, i.e., slightly below resonance. Setting
the derivative of the magnification factor with respect to n to zero, we find that
the maximum occurs when

2
= 1
n
2

(2.116)

Substituting this relationship into Eq.(2.115), it can be found that the maximum
magnification depends on the amount of damping according to the following formula:
X
2

=
2
xstatic max 4

(2.117)

From this formula it follows that if 2 , the magnification factor as a function of the
frequency of external force has no maximum. This means that in this case, the largest
value of the magnification factor is equal to unity and is achieved at the zero
frequency of the external force. In yet other words, this implies that no dynamic
amplification is possible if 2 . Note that the damping that corresponds to = 2 is
precisely the critical damping ccrit = 2 km as defined by Eq.(2.77).
In most practical situations, << 1 and it is justifiable to consider that the
maximum value of the magnification factor occurs when n and the maximum of
the magnification factor can be approximated by
X
1

xstatic max

(2.118)

41

Both Eq.(2.117) and Eq.(2.118) show that for small damping, the amplitude of forced
vibration can become extremely large under the condition of resonance. Since large
amplitudes mean large stresses in the spring, the condition of resonance is usually to
be regarded as a dangerous one and is to be avoided whenever possible.
Let us turn now to the phase relationship between the forced vibrations and the
external force producing them. This is represented by the phase angle in Eq.(2.108)
, the expression for which can be rewritten making use of , Eq.(2.114), as

2
2
n 1 n

= arctan

(2.119)

We recall that since the external force varies according to cos (t ) and the forced

vibrations according to cos (t ) , the phase angle represents the phase lag of

the vibrations behind the external force. That is, when the external force F ( t ) in
Figure 2.11 is directed rightward, the mass on which it acts is not yet in its utmost
right position but arrives there seconds later.
Eq.(2.119) shows that the value of the phase lag, like that of the magnification
factor, depends both upon the relative amount of damping and upon the ratio
n . The curves in Figure 2.15 show the variation of the phase lag with the ratio

n for several values of the damping factor . Note that arctan ( x ) is a multivalued function, therefore, to obtain the continuous graphs in Figure 2.15, one has to
plot the principal value Arctan ( x ) if the argument x is positive and the principal
value plus (if the graph is in radians) or plus 180o (if the graph is in degrees) if the
argument is negative.
180.0

=0
= 0.25
[deg]

= 0.5
90.0

=2

=1

0.0
0.0

0.5

1.0

1.5

2.0

2.5

/n

Figure 2.15. Phase lag

We see that when = 0 , the phase lag is discontinuous (this was also shown in
Figure 2.8). We recall that in this case the forced vibrations are exactly in phase with
the disturbing force for all values of n smaller than unity and out of phase for all
values of n greater than unity. When damping is present, we can see a continual
change in as the ratio n increases. Also, regardless of damping, = 90 0 = 2

42

at resonance. That is, at resonance, the forced vibrations lag behind the external force
by one quarter cycle.
For values of n either well below resonance or well above it, we note that a
moderate amount of damping ( << 1) has only a secondary effect on the phase lag.
That is, well below resonance, it is practically zero, while well above resonance, it is
practically 180 0 = . This means, again, that in discussing forced vibrations well
away from the condition of resonance, the effect of damping can be ignored.
The frequency dependence of the phase lag presented in Figure 2.14 allows us to
identify the frequency bands, at which particular elements of the mass-spring-dashpot
system play a major role. To carry out such identification, we can use Eq.(2.101),
which, after multiplying through by m , reads

Re ( m 2 + ic + k ) X exp ( it ) F0 exp ( it ) = 0

(2.120)

Employing the representation for the complex amplitude X given by Eq. (2.111) and
taking into account that X = X and arg X = , we rewrite Eq.(2.120) as

( )

Re ( m 2 + ic + k ) X exp ( i ) exp ( it ) F0 exp ( it ) = 0

(2.121)

To satisfy this equation, the expression in the figure brackets must vanish, hence
dividing by exp ( it i ) we obtain

( m

+ ic + k X = F0 exp ( i )

This equation, using the following equalities substantiated by the Euler formula
Eq.(2.13),
1 = exp ( i ) , i = exp ( i 2 )

(2.122)

can be rewritten as
m 2 X exp ( i ) + c X exp ( i 2 ) + k X = F0 exp ( i )

(2.123)

The three terms on the left-hand side represent the forces produced by the mass,
dashpot and spring respectively, whereas the term on the right-hand side is the
external force. We can clearly see now that while the external force leads the
displacement by the phase angle , the phases of the other forces are as follows:
the spring-force is in phase with the displacement,
the dashpot-force leads the displacement by 2 ,
the mass-force leads the displacement by .
Using this information, we can take a fresh look at the frequency dependence of
the phase lag of vibrations presented by Fig. 2.15. This figure shows that when
<< n , the phase lag between the force and the displacement is almost zero, which
coincides with the phase of the spring-force. This implies that the major role in low-

43

frequency vibrations is played by the spring. Alternatively, when >> n ,

approaches , which implies that the major role is taken over by the mass-force. Near
resonance, varies very fast near 2 so that the dashpot-force plays an important
role, which corresponds to high sensitivity of the magnification factor to the amount
of damping, which is shown in Figure 2.13.
2.3.3. Forced Vibrations under General Disturbing Force. The solution of
Eq.(2.66) in the case of an arbitrary time dependence of the external force can be
obtained in exactly the same manner as it was done in Section 2.2.3 for the undamped
mass-spring system. Let us briefly repeat this procedure.
Consider Eq.(2.66) divided through by m with the initial conditions Eq.(2.6):

x + 2nx + n2 x = f ( t ) ,
2n = c m , n2 = k m, f ( t ) = F ( t ) m
x ( 0 ) = x0

(2.124)

x ( 0 ) = v0

assuming that the external force per unit mass f ( t ) is an arbitrary function of time.

For any instant of time we may consider one elementary impulse f ( t ) dt as

represented by the shaded strip in Figure 2.9. In accordance with Newtons second
law, this one impulse imparts to each unit of mass an instantaneous increase in
velocity dx = f ( t ) dt . Treating this increment of velocity as if it were an initial
velocity at the instant t and using Eq.(2.83), we conclude that the corresponding
displacement of the mass-spring system at any later time will be
x ( t ) = exp ( n ( t t ) )

f ( t ) dt

sin (1 ( t t ) )

(2.125)

where 1 = n2 n2 . Note that Eq.(2.125) is valid both for n2 > n 2 and n2 < n 2 . In
the latter case, 1 has an imaginary value.
Since each impulse f ( t ) dt between t = 0 and t = t has a like effect, we obtain,
as a result of the continuous action of the external force, the following displacement
of the mass
x (t ) =

1 0

f ( t ) exp ( n ( t t ) ) sin (1 ( t t ) ) dt

(2.126)

The complete solution to the initial-value problem Eq.(2.124) is obtained by summing

up Eq.(2.126) and Eq.(2.83), the latter representing the system response to the initial
conditions:

44

v nx
x ( t ) = exp ( nt ) x0 cos (1t ) + 0 + 0 sin (1t )
1 1

1 0

(2.127)

f ( t ) exp ( n ( t t ) ) sin (1 ( t t ) ) dt

As an example, let us consider the external force to be suddenly applied to the

mass and then remain constant, i.e.,

0,
F (t ) =
F0 ,

t<0

(2.128)

t0

Substituting this expression in the general formula Eq.(2.40), and assuming that the
initial displacement x0 and initial velocity v0 are zero, we obtain
x=

F0 1 exp ( nt ) (1 cos (1t ) + n sin (1t ) )

m1
n 2 + 12

(2.129)

Dividing through by the static deflection xstatic = F0 k , using defined by Eq.(2.114)

, and introducing the new dimensionless time

= n t ,

(2.130)

we rewrite Eq.(2.129) as
x
xstatic

4 2
4 2

= 1 exp cos
+
sin

2
2
2
4 2

(2.131)

The curves in Figure 2.16 show the variation of the relative displacement x xstatic with
the dimensionless time for three values of the damping as represented by .
2.0

=0
= 0.5

x/xstatic

1.5

1.0

0.5

= 1.9
0.0
0

10

15

20

25

Figure 2.16. Response to suddenly applied constant force

45

This figure shows that when = 0 the mass oscillates indefinitely long with the
amplitude equal to the static deflection around the static equilibrium x xstatic = 1 . This
result coincides with that of Section 2.2.3. When 0 , i.e., in the presence of
damping, the mass sooner or later comes to its static equilibrium. The larger the
damping, the faster this happens. Two scenarios are possible here. If damping is
moderate as exemplified by = 0.5 some oscillations around the static equilibrium
take place before the mass gets in rest in its static position. If damping is large, as
represented by = 1.9 , the mass just creeps to the static position demonstrating no
oscillations whatsoever. Thus, under the action of a suddenly applied constant force,
any system with damping (read: any engineering system) will sooner or later get to its
static equilibrium, regardless of the initial conditions. If the constant force is removed
at a time instant the system, owing to damping, will get to its undisturbed equilibrium
x =0.

2.4.

Two Degrees of Freedom Systems without Damping

In the preceding sections we discussed the theory of the vibrations of a system with a
single degree of freedom. Though the exact idealized system with which the theory
dealt occurs rarely, a number of actual cases are sufficiently close to the ideal to
permit conclusions of practical importance. The theory of the single-degree-offreedom system enables us to explain the resonance phenomenon in many machines,
to calculate natural frequencies of a number of structures, and to discuss the main
principles of vibration isolation.
This exhausts the possibilities of application pretty thoroughly, and in order to
explain additional phenomena it is necessary to develop the theory of more
complicated systems. As a first step, in this section we consider a system with two
degrees of freedom, which will yield the explanation of most vibration dampers
applied in practice.
The most general undamped two-degree-of-freedom system can be reduced to that
shown in Figure 2.17 and consists of two masses, which rest on a frictionless
horizontal plane, and three springs. The springs k1 and k2 connect the respective
masses to the walls, while the spring k3 ties the masses together, thereby coupling
their motion. The external loading is represented by the forces F1 ( t ) and F2 ( t ) ,
which are applied to the masses m1 and m2 , respectively.

m1
k1

F1(t)

F2(t)

x1(t)
k3

x2(t)
m2
k2
x

Figure 2.17. Undamped two-degrees-of-freedom system with spring coupling

To write down the equations of motion for this system, we use the second Newtons
law. Each mass is subject to three forces, namely the external force and two springforces. Therefore, the vector-form of the equations of motion reads

46

m1 x1 = F1 ( t ) + FSPRING 1-1 ( t ) + FSPRING 3-1 ( t )

m2 x2 = F2 ( t ) + FSPRING 2-2 ( t ) + FSPRING 3-2 ( t )

(2.132)

where x1 ( t ) and x2 ( t ) are the displacements of the masses from their equilibrium
positions, in which the springs are undisturbed. The spring-forces imply the
following: FSPRING 1-1 is the vector-force imposed on m1 by spring k1 ; FSPRING 2 2 is
the vector-force imposed on m2 by spring k2 ; FSPRING 3-1 is the vector-force imposed
on m1 by spring k3 ; FSPRING 3-2 is the vector-force imposed on m2 by spring k3 .
According to the Hookes law, the vector-spring-forces can be expressed as
FSPRING 1-1 = k1 x1 ,

FSPRING 2-2 = k2 x2 ,

FSPRING 3-1 ( t ) = k3 ( x1 x2 ) , FSPRING 3-2 ( t ) = k3 ( x2 x1 )

(2.133)

Substituting these expressions into Eqs.(2.132) and projecting the result onto the
x axis, the following scalar equations are obtained
m1 
x1 + k1 x1 + k3 ( x1 x2 ) = F1 ( t )

m2 
x2 + k2 x2 + k3 ( x2 x1 ) = F2 ( t )

(2.134)

These are the equations of motion, which describe small vibrations of the undamped
two-degrees-of-freedom system near its equilibrium position under the action of the
external forces F1 ( t ) and F2 ( t ) .
Retrieving the equations of motion of a dynamic system, one has to project the
balances of forces written in the vector-form on the chosen coordinate axes. If the
system can move in one direction only, the operation of projecting the forces is easy.
If, however, a planar or spatial motion is considered, this operation can prove to be
quite laborious and vulnerable to men-introduced errors. To avoid such errors, the socalled Lagrangian formalism can be used, which is shortly described below.
The Lagrangian formalism is based on the Lagrangian equations, which read
d L L

= Qs
dt qs qs

(2.135)

where
L =KP

(2.136)

is the Lagrange function, which is defined as the difference between the kinetic
energy and the potential energy of the system under consideration; qs and qs ,
s = 1, 2,3,... are the generalized coordinates and velocities of the system; Qs are the
generalized forces acting on the system. The term generalized reflects the fact that
qs can represent not only a displacement but an angle as well, or, for example, the
voltage in an electric machine. The system presented in Figure 2.16 has two degrees

47

of freedom, which are specified by the two displacements. This implies that for this
system s = 1, 2 , q1 = x1 , q2 = x2 , Q1 = F1 , Q2 = F2 .
The Lagrangian equations are based on a very general principle that any system is
smart enough and if it had to move from one point to another one, being
unconstraint, it would choose the way, which requires the minimum amount of energy
to be spent. The mathematical developments, which lead to the Lagrangian equations,
can be found in any text book on theoretical mechanics.
As an example, let us apply the Lagrangian equations to obtain the equations of
motion for the system shown in Figure 2.17. To this end, we have to write the kinetic
and the potential energies of this system. The kinetic energy is the sum of the kinetic
energies of the two masses, whereas the potential energy is the sum of the potential
energies of the three springs. Therefore, the following expressions hold

(
(

1
2
2
K
= m1 ( x1 ) + m2 ( x2 )
2
1
2
2
2
P = k1 ( x1 ) + k2 ( x2 ) + k3 ( x1 x2 )
2

(2.137)

Now we can calculate L qs = L xs and L qs = L xs , s = 1, 2. This yields

L L K
=
=
= m1 x1 ,
q1 x1 x1

L L K
=
=
= m2 x2 ,
q2 x2 x2

L L
P
=
=
= ( k1 x1 + k3 ( x1 x2 ) ) ,
q1 x1
x1

(2.138)

L L
P
=
=
= ( k2 x2 k3 ( x1 x2 ) )
q2 x2
x2

Substitution of these expressions into the Lagrangian equations Eq.(2.135) gives the
equations of motion, which coincide with Eq.(2.134).
The equations of notion, Eq.(2.134) can be written in the matrix form, which is
widely used for numerical analyses. The matrix form of Eq. (2.134) reads
M 
x+K x=F

(2.139)

where x = { x1 , x2 }
vector, and

m 0
M = 1
,
0 m2

and F = { F1 , F2 }

k + k
K = 1 3
k3

are the displacement vector and the force

k3
k2 + k3

(2.140)

are the mass matrix and the stiffness matrix, respectively.

2.4.1. Free Vibrations. If the external forces F1 ( t ) and F2 ( t ) are absent, the
equations of motion Eq.(2.134) reduce to the following system of two linear, coupled
homogeneous differential equations of the second order with constant coefficients:

48

m1 
x1 + k1 x1 + k3 ( x1 x2 ) = 0

(2.141)

m2 
x2 + k2 x2 + k3 ( x2 x1 ) = 0

The general solution of Eq.(2.141) can be written as

4

x1 ( t ) = X n( ) exp ( snt ),

x2 ( t ) = X n( ) exp ( sn t )

n =1

(2.142)

n =1

so that each displacement is a superposition of four terms in correspondence with the

order of the system of equations. Substituting Eq.(2.142) into the equation of motion
Eq.(2.141), we obtain the following system
4
4
4
4 1

1
1
2
m1 X n( ) sn2 exp ( snt ) + k1 X n( ) exp ( snt ) + k3 X n( ) exp ( snt ) X n( ) exp ( snt ) = 0
n =1
n =1
n =1
n =1

4
4
4
4

2
2
2
1
m2 X n( ) sn2 exp ( snt ) + k2 X n( ) exp ( snt ) + k3 X n( ) exp ( snt ) X n( ) exp ( snt ) = 0
n =1
n =1
n =1
n =1

Which can be reduced to

{ X ( ) ( m s

+ k1 + k3 ) + X n(

{ X ( m s

+ k2 + k3 ) + X n ( k3 ) exp ( snt ) = 0

1
n

2
1 n

n =1
4

( 2)

n =1

2
2 n

2)

( k3 )} exp ( snt ) = 0

(1)

(2.143)

In order these equations be satisfied at any instant of time, the expressions in the
figure brackets must vanish, that is the following system of algebraic equations must
be satisfied:
X n( ) ( m1sn2 + k1 + k3 ) + X n(
1

X n(

2)

(m s

2
2 n

2)

( k3 ) = 0

+ k2 + k3 ) + X n( ) ( k3 ) = 0
1

(2.144)

This system of equations (2.144) has a non-trivial solution if the determinant of its
coefficient matrix vanishes. The coefficient matrix of Eq.(2.144) is given as

m1sn2 + k1 + k3

k3
2

= Msn + K
2

k
m
s
+
k
+
k
3
2 n
2
3

(2.145)

Thus, the characteristic exponents sn must satisfy the following characteristic

equation:

(m s
1

+ k1 + k3 )( m2 s 2 + k2 + k3 ) k32 = 0

which can be rewritten as

49

k +k k +k k k +k k +k k
s4 + s2 1 3 + 2 3 + 1 2 1 3 2 3 = 0
m2
m1m2
m1

(2.146)

Eq.(2.146) is a biquadratic equation, which defines two values of s 2 . These values

can be shown to be real and negative, so that the corresponding four values of s are
all imaginary, each two of them being complex conjugate. This implies that the
motion of the two-degrees-of-freedom system shown in Figure 2.16 is vibratory and
takes place at two distinct frequencies. To find and analyse these frequencies, we
replace s in Eq.(2.146) by i and introduce the following notations

a2 =

k1 + k3
k +k
k3
, b2 = 2 3 , ab2 =
m1
m2
m1m2

(2.147)

The quantities a and b are the frequencies of the system in which one of the
masses is held clamped (the so-called partial frequencies), whereas ab represents the
strength of the coupling between the masses. With these notations and s = i ,
Eq.(2.146) can be written as

4 2 (a2 + b2 ) + (a2b2 ab4 ) = 0

(2.148)

The roots of this equation are the natural frequencies of the two-degrees-of-freedom
system under consideration.
It is interesting to note that Eq.(2.148) can be analysed geometrically with the help
of Mohrs circle diagram, which may be familiar to the reader from two-dimensional
elasticity. The Mohrs circle diagram for Eq.(2.147) is shown in Fig.2.18.

C
O

Figure 2.18. Mohrs circle for determining natural frequencies

In this diagram
OA = a2

OB = b2

BC = ab2

The circle is drawn through C about the mid-point between A and B as centre. The
new points D and E thus found, determine the natural frequencies of the system:

n21 = OD

n22 = OE

50

which can be verified from the equation. Figure 2.17 shows, in particular, that the
interval between the natural frequencies of n1 and n 2 is always greater than or
equal to that between the partial frequencies a and b .
The analytical expressions for the natural frequencies can be found from
Eq.(2.148) to give
1
a2 + b2
2

1
n 2 =
a2 + b2 +
2

n1 =

2
a

b2 ) + 4ab4
2

(2.149)
2
a

2 2
b

+ 4

4
ab

It is obvious that the expression under the inner square root in Eq.(2.149) is positive.
It can also be easily checked that a2 + b2 >

2
a

b2 ) + 4ab4 , which implies that

2

both natural frequencies are real. Thus, the general solution Eq.(2.142) of Eq.(2.141),
following similar rearrangements as presented in Eq.(2.14), can be rewritten as
x1 ( t ) = A1 cos (n1t ) + B1 sin (n1t ) + C1 cos (n 2t ) + D1 sin (n 2t )

x2 ( t ) = A2 cos (n1t ) + B2 sin (n1t ) + C2 cos (n 2t ) + D2 sin (n 2t )

(2.150)

Eq.(2.150) shows that the free vibrations of the system take place at two distinct
frequencies. The amplitudes of vibrations at these frequencies depend on specific
initial conditions. Correspondingly, the complete vibration pattern can be quite
different, depending on the natural frequencies and initial conditions. We may expect,
however, that if the natural frequencies are close to each other, the beats should occur
leading to a time-displacement diagram similar to that depicted in Figure 2.5.
The free vibrations of the one-degree-of freedom system considered in the
preceding sections are fully characterised by its natural frequencies and the initial
conditions. This is not the case, however, with the two-degree-of freedom system. The
latter requires knowledge of the relationship between the amplitudes of vibrations of
the two degrees-of-freedom, that is A2 A1 , B2 B1 , C2 C1 , D2 D1 . It should be
underlined that these ratios are independent of the initial conditions. To find these
ratios, Eq.(2.150) can be substituted into the equations of motion Eq.(2.141), which,
employing the notations Eq.(2.147), can be rearranged to

x1 + a2 x1 ab2 x2 = 0

x2 + b2 x2 ab2 x1 = 0

(2.151)

where = m2 m1 . Substituting Eq.(2.150) into this system of equations, we obtain

51

cos (n1t ) A1 (a2 n21 ) A2ab2 + sin (n1t ) B1 (a2 n21 ) B2ab2

+ cos (n 2t ) C1 (a2 n22 ) C2ab2 + sin (n 2t ) D1 (a2 n22 ) D2ab2 = 0

cos (n1t ) A2 (b2 n21 ) A1ab2 + sin (n1t ) B2 (b2 n21 ) B1ab2

+ cos (n 2t ) C2 (b2 n22 ) C1ab2 + sin (n 2t ) D2 (b2 n22 ) D1ab2 = 0

To have these equations satisfied at any time instant, all expression in the figure
brackets must vanish. Thus, the following equations must be satisfied:
A2
1
=
(a2 n21 ) ,
A1 ab2

A2
ab2
=
A1 (b2 n21 )

B2
1
=
(a2 n21 ) ,
B1 ab2

ab2
B2
=
B1 (b2 n21 )

C2
1
=
2 n22 ) ,
2 ( a
C1 ab

C2

=
C1 (b2 n22 )

D2
1
=
2 n22 ) ,
2 ( a
D1 ab

ab2
D2
=
D1 (b2 n22 )

(2.152)

2
ab

We see that there are two expressions for every ratio of the amplitudes, which should
be satisfied simultaneously. It is easy to check that they are satisfied simultaneously
indeed, since n21 and n22 are the roots of Eq.(2.148). This, of course, is to be
expected since Eq.(2.148) is nothing else but the condition of linear dependence of
the equations in the system Eq.(2.141), which implies that if one of the two equation is
satisfied then the other one is satisfied automatically.
Using the ratios of amplitudes given by Eq.(2.152), the general solution Eq.(2.150)
can be given the following more specific form:
x1 ( t ) = A1 cos (n1t ) + B1 sin (n1t ) + C1 cos (n 2t ) + D1 sin (n 2t )
x2 ( t ) = ( A1 cos (n1t ) + B1 sin (n1t ) )

+ ( C1 cos (n 2t ) + D1 sin (n 2t ) )

2
ab

2
ab

2
a

n21 )

2
a

(2.153)

n22 )

This form of the general solution contains four unknown constants, which can be
uniquely determined employing the initial conditions (the initial displacement and
initial velocity for x1 and those for x2 ).
The general solution Eq.(2.153) shows that the amplitudes of the free vibrations of
the masses m1 and m2 are intrinsically related to each other. Thus, the free vibrations
of the two-degrees-of-freedom system are characterised not only by the natural
frequencies but also by the shapes of vibrations at these frequencies. These shapes are

52

normally referred to as normal modes and imply a specific motion of the masses with
respect to each other.
To gain insight in what the normal modes mean physically, let us simplify the
system somewhat by making it symmetrical. To this end we assume that k1 = k2 = k
and m1 = m2 = m . In this case, a2 = b2 , = 1 and the expressions Eq.(2.149) for the
natural frequencies reduce to

n1 = a2 ab2 = k m
n 2 = a2 + ab2 =

( k + 2k3 )

Consequently, the ratios of the amplitudes, Eq.(2.152) become

A2 B2
=
=1
A1 B1
C2 D2
=
= 1
C1 D1

(2.154)

The physical significance of these results is obvious. The fact that A2 A1 = B2 B1 = 1

implies that the two masses move in the same direction through the same distance.
The coupling spring k3 is not stretched or compressed in this process. Naturally, the
frequency of this motion is n1 = k m , since the system reduces to two independent
single-degree-of-freedom systems. The fact that C2 C1 = D2 D1 = 1 means that the
two masses move through the same distance but in opposition to each other. This
motion is completely symmetrical, so that the mid-point of the coupling spring does
not move. If this mid-point were held clamped, no change in the motion would take
place. Thus the system is again split into two independent single-degree-of-freedom
systems. This time, however, each mass is subject to two springs, one of stiffness k
and another of stiffness 2k3 , so that the frequency is n 2 = ( k + 2k3 ) m .
Thus, there are two natural modes of motion of the two-degrees-of-freedom
system, each with its corresponding natural frequency. In the case of the symmetrical
system, both modes are naturally symmetrical. In the case that no symmetry exists,
these modes are somewhat more complex. In any case, however, the general solution
Eq.(2.153) shows that any motion of the system is a superposition of the two natural
modes of vibrations.
Concluding the discussion of the free vibrations of the two-degrees of freedom
system, it is important to note that the operations of finding the natural frequencies
and general solution can be significantly shortened by employing the matrix notation,
Eq.(2.139). Using this notation, the equations describing the free vibrations read
M 
x+K x=0

(2.155)

To find the natural frequencies of the system, the following determinant should be
set to zero:

53

det 2 M + K = 0

which is equivalent to finding the eigenvalues of the matrix KM 1 . The natural

frequencies n21 and n22 can be obtained in both ways. The normal modes of the free
vibrations can be found by searching for the eigenvectors of the matrix KM 1 .
2.4.2. Forced Vibrations under Harmonic Force. Consider the dynamic response of
the mass-spring system to harmonic (sinusoidal) forces which are given as
1
2
F1 ( t ) = F0( ) cos (t ) , F2 ( t ) = F0( ) cos (t + 0 ) . In this case the equations of motion
for the two-degrees-of-freedom system depicted in Figure 2.16, Eq.(2.134) take the
form:

m1 
x1 + k1 x1 + k3 ( x1 x2 ) = F0( ) cos (t )
1

(2.156)

m2 
x2 + k2 x2 + k3 ( x2 x1 ) = F0( ) cos (t + 0 )
2

We will study only the forced steady-state vibrations of the system, since these
vibrations are commonly of most practical importance.
We already know that the steady-state vibrations take place at the frequency of the
force. Therefore, these vibrations may be sought in the form, which is a supermosition
of cosinusoidal vibrations dictated by the forces on the right-hand side of Eq. (2.156),
x1 = A1 cos (t ) + B1 cos (t + 0 )

(2.157)

x2 = A2 cos (t ) + B2 cos (t + 0 )

Substituting Eq.(2.157) into Eq.(2.156), we obtain

{ m A + k A + k A k A F ( )} cos (t )
1

1 1

1 1

+ { 2 m1 B1 + k1 B1 + k3 B1 k3 B2 } (t + 0 ) = 0

{ m A + k A
2

+ k3 A2 k3 A1} cos (t )

(2.158)

+ 2 m2 B2 + k2 B2 + k3 B2 k3 B1 F0( 2) (t + 0 ) = 0
To satisfy these equations at any time instant the expressions in the figure brackets
must vanish. This leads to the following two systems of algebraic equations, from
which the amplitudes of the steady-state vibrations can be found:

2 m1 A1 + k1 A1 + k3 A1 k3 A2 = F0(1)
2
m2 A2 + k2 A2 + k3 A2 k3 A1 = 0

(2.159)

2 m1 B1 + k1 B1 + k3 B1 k3 B2 = 0
2
(2)
m2 B2 + k2 B2 + k3 B2 k3 B1 = F0

(2.160)

54

Let us consider from hereon a simplified but a very practically important example,
which is depicted in Figure 2.19. In this example, we consider a two-degrees-offreedom system, which consists of a main system K , M subject to a harmonic force
F0 cos (t ) and an attached auxiliary system k , m . This model can schematise, for
example a bridge with an auxiliary mass elastically attached in the middle, Fig.
2.20(a), or a vibrating machine with an auxiliary mass-spring system, Fig. 2.20(b).
This auxiliary mass-spring system is normally referred to as an undamped dynamic
vibration absorber. It will be shown that the parameters of this absorber can be
chosen such that the main system (the bridge or the machine) does not vibrate
although it is subject to the external harmonic force.
F0 cos (t )
x1(t)
k

x2(t)
m

K
x
Figure 2.19. Vibrations absorber k-m attached to the main system K-M (bridge, machine, etc.)

(a)

Bridge beam

(b)

k
k

m
K/2

K/2

Auxiliary mass
Figure 2.20. Examples of application of the vibration absorber: (a) bridge with an elastically
suspended mass; (b) machine with an elastically mounted mass

The amplitudes of the steady-state vibrations of the system, which is depicted in

Figure 2.18 can be retrieved from Eqs.(2.159) and (2.160) by setting
k1 = K , m1 = M , F0(1) = F0 , k3 = k ,

(2.161)

k2 = 0, m2 = m, F0( ) = 0
2

As follows from Eq.(2.160), since F0( ) = 0 , the amplitudes B1 and B2 are trivially
zero. The system of equations for A1 and A2 , Eq.(2.159), is simplified to
2

2 MA1 + KA1 + kA1 kA2 = F0

2
mA2 + kA2 kA1 = 0

(2.162)

55

Introducing in this system of equations the partial frequencies a and b , the

coupling frequency ab , and the mass ratio in accordance with the following
definitions (the same as for the free vibrations of the mass)
K +k
k
k
m
(2.163)
a2 =
, b2 = , ab2 =
, =
M
m
M
mM
we can write Eq.(2.163), as

A1 (a2 2 ) ab2 A2 = F0 M

2
2
2
A2 (b ) ab A2 = 0

(2.164)

Solving this system, we obtain

A1 = xstatic
A2 = xstatic

2n (b2 2 )

4 2 (a2 + b2 ) + (a2b2 ab4 )

2nab2

4 2 (a2 + b2 ) + (a2b2 ab4 )

(2.165)

where n = K M is the natural frequency of the main system and xstatic = F0 K is

the static deflection under the force F0 .
We can see from Eq.(2.165) that the denominator in the expressions for both
amplitudes resembles the frequency equation, Eq.(2.148), which determines the
natural frequencies of the system. This means that the masses vibrate with infinite
amplitudes when the frequency of the force becomes equal to one of the natural
frequencies of the system. Furthermore, if the frequency of the force becomes
equal to the partial frequency b , i.e., to the natural frequency k m of the auxiliary
system, the amplitude of the main system becomes zero. This means that if the
auxiliary system is chosen such that its natural frequency equals the frequency of the
external force, the main system will not vibrate under the external force, effectively
passing all the energy introduced into the system to the auxiliary system. This is why
the latter is referred to as dynamic absorber. Coming back to the practical example
depicted in Figure 2.20, we can now make some practical conclusions. Considering
the bridge, Fig. 2.20(a), it becomes clear that to dampen the vibrations of the bridge,
the auxiliary system has to have the natural frequency equal the first natural frequency
of the bridge, since that is the first natural frequency of the bridge, which captures the
main energy of the vibrations induced by traffic on the bridge. Considering the
machine, Fig. 2.20(a), we see that the auxiliary system has to be chosen such that it is
equal to the main frequency of the machine vibrations.
Let us analyse the amplitudes A1 and A2 in more detail. Making use of the natural
frequencies of the system defined by Eq.(2.149), dividing by xstatic , and noting that

ab2 = a2 , we write Eq.(2.165) as

56

A1
xstatic
A2
xstatic

=
=

n2 (b2 2 )
2

n21 )( 2 n22 )

(
2

2
n
2
n1

2
ab

)(

(2.166)

n22 )

A/x static

In Figure 2.21 the displacement-frequency curves A1 xstatic and A2 xstatic are plotted
qualitatively versus the frequency of the external force.

n1

n2

A1
A2

Figure 2.21. Displacement-frequency curves for the two-degrees-of-freedom system

Figure 2.21 shows clearly two resonances at the natural frequencies of the system.
Furthermore, as noticed earlier, we see that the main system does not move at the
partial frequency b of the auxiliary system. Concerning the phases of vibrations,
Figure 2.21 allows to conclude that (i) for < n1 the masses vibrate in phase with
the load and the phase lag between vibrations of the masses is zero, although the
amplitudes of the vibrations are different; (ii) for n1 < < b the masses are out of
phase with the load but the phase lag between vibrations of the masses is still zero;
(iii) for b < < n 2 the masses start to vibrate out of phase with each other, the main
mass keeping in phase with the load; (iii) for n 2 < the masses are out of phase
with each other, but the main mass vibrates out of phase with the load.
Concluding this section, it is important to note that the displacement-frequency
curves in Figure 2.21 are plotted for the undamped two-degrees-of-freedom system. If
the viscous damping were included, these curves would be different, especially in the
vicinity of resonances and around b .

57

CHAPTER 3
ONE-DIMENSIONAL STRUCTURES
Most civil engineering systems and structures are continuous and have distributed
material properties. A bridge, a chimney, and a railway track, for example, can be
considered as beams with uniform mass per unit length and uniform bending stiffness.
Landing strips of an airport have properties distributed over their surface. The soil
properties are distributed all over the volume. The fact that the material properties are
distributed continuously suggests that to describe the mechanical behaviour of these
systems and structures, we need to quantify their mechanical behaviour with
characteristics (displacements, forces, etc.) which depend both on time and position.
Analyses of models, which take into account both these dependencies, form the
subject of continuum mechanics.
The mechanical behaviour of many structures such as long bridges, chimneys,
railway tracks, skyscrapers, etc., which are much longer in one direction than in the
two other ones, depends mainly on the time and position along the structure.
Consequently, modelling these structures, one may assume a priory that the
displacements, forces and moments in these structures depend on time and the alongthe-structure co-ordinate only. Such models are referred to as one-dimensional models
and the structures, which these models represent, are often called one-dimensional
structures.
Though the mechanical characteristics of the one-dimensional structures depend on
one co-ordinate only, this dependence is continuous. This makes a principal
theoretical difference between the lumped models considered in the previous chapter
and the continuous models, which will be treated in this chapter. To describe the
mechanical behaviour of a continuous system, theoretically speaking, we have to track
the displacement time diagrams, for example, of infinitely many points. Therefore,
continuous models are often referred to as models with infinite number of degrees of
freedom. In practice, however, mechanical properties of structures vary in space
relatively slowly. Therefore, it is always possible to characterise the mechanical
behaviour of such structures using a finite-degree-of-freedom system, increasing the
accuracy of characterization by increasing the number of the degrees of freedom to be
considered.

3.1.

Fundamental Assumptions Leading to One-Dimensional

Continuous Models

One-dimensional continuous structural models are characterized by their

displacements and velocities, which depend both on time and the position along the
system. These models are governed by partial differential equations. The onedimensional models are applicable only for description of spatial scales, which are
much greater than the lateral dimensions of the modelled structure.
Let us consider the same two structures as in Section 2.1 and show how the
representation of their dynamic behaviour can be improved by using the onedimensional continuous models instead of the one-degree-of-freedom models.

58

F(t)

u (t, z )

Figure 3.1. Idealization of motion of an offshore platform on steel jacket structure

by a beam with a lumped mass on the top

w (t, x )
x

F(t)
Figure 3.2. Idealization of vertical vibration of a lorry on a bridge by a simply supported beam
with a lumped mass in the middle

In Figure 3.1. the relatively heavy deck of the offshore platform is replaced by a
lumped mass while the braced steel jacket structure is replaced by a beam, which may
perform the bending motion in the plane of the figure. The model in Figure 3.1 is
more advanced than that in Figure 2.1 because the distributed mass and distributed
stiffness of the supporting structure is now taken into account. Basically, accounting
for the distributed properties of the supporting structure implies that we take account
of the dependence of the support reaction on both the amplitude and frequency of the
platform vibrations. This was not the case with the model in Figure 2.1, where the
reaction was considered as proportional to the displacement of the platform only.
In Figure 3.2. the bridge girder is idealized as a simply supported beam with its
mass and stiffness uniformly distributed over the bridge length. The advantage of the
model in Figure 3.2 relative to that in Figure 2.2 is that now the reaction of the bridge
against the lorry depends not only on the amplitude but also on the frequency of the
lorry vibrations.
Finalizing this section, it is worth noting that the larger the mass of the platform
and the lorry in Figures 3.1 and 3.2, the more accurate predictions can be obtained of
vibrations of the platform and lorry using the one-degrees-of-freedom systems
depicted in Figures 2.1 and 2.2. Thus, the one-degrees-of-freedom models are often
capable of predicting vibrations of heavy bodies attached to elastic structures. What
they are incapable of, however, is the prediction of the shape of elastic structures such
as the braced steel jacket structure and the bridge in Figures 3.1 and 3.2. Thus, to
study this shape and, accordingly, the deformations and stresses in these structures,
the continuous models should be applied.
In the sections to follow five one-dimensional continuous models will be first
formulated and then analyzed. These models represent the following basic
deformations of one-dimensional structural elements: (1) transverse deflection of a
cable; (2) longitudinal compression and extension of a rod; (3) torsion of a rod; (4)
pure shear of a beam; (5) bending of a beam.

59

3.2.

Formulation of Equations of Motion for One-Dimensional

Models

3.2.1. Transverse Motion of a String. The string is an idealization of a long flexible

structure, which is tensioned by external forces so that the restoring force associated
with this tension is much greater than that associated with the bending stiffness. The
string can serve as a model for cables of electric transmission lines, overhead power
lines of electric transportation means, mooring systems of floating vessels, and deepwater catenary offshore risers.
Let us obtain the governing equation for the taut string subject to a horizontal force
H , which stretches it, and a distributed vertical force with the density per unit length
q1 ( x, t ) . Once displaced from its rest horizontal position, every segment of the string
is subject to the inertia force, the external vertical force with the density per unit
length q1 ( x, t ) and the tensile axial forces T acting on the string both from the left
and from the right as shown in Figure 3.1 (a).

q1 ( x, t )
T

x
V

T
T

q1 ( x, t )

w
s

w
(a)

V + V

H
T + T

x
(b)

Figure 3.1. Taut string: (a) taut string subject to a distributed force; (b) differential element of a
displaced from the rest position (horizontal line) string

Consider a differential element x of the string. The forces acting on this element
once it is displaced from its rest position are shown in Figure 3.1(b). Assuming that
the horizontal motion of the string is negligible so that every element of the string
moves just vertically, we may write the following equation, which is the projection of
the vector form of second Newtons law on the vertical (directed downward) axis:

Ax

2w
= V + V V + q1x = V + q1x
t 2

(3.1)

where w ( x, t ) is the vertical deflection of the string, is the mass density of the
string material and A is the cross-sectional area of the string. Note that on the lefthand side of Eq.(3.1) we introduced the partial derivative with respect to time, which
is not necessarily equal to the ordinary derivative as it is applied to a function of both
time and space.
Dividing Eq.(3.1) by x and taking into account that the vertical component V of
the tensile force T is related to its horizontal component H as

60

V = H tan ( )

(3.2)

we write Eq.(3.1) in the following form

2 w ( H tan ( ) )
=
+ q1
x
t 2

(3.3)

Assume that the vibrations of the string are small so that the following relation
holds

FG w IJ
H x K

<< 1.

(3.4)

which implies that the slope of the string should be small. This condition is usually
referred to as the condition of small vibrations. Under this condition, the following
approximation holds
tan ( )

w
x

(3.5)

Substituting Eq.(3.5) into Eq.(3.3), and taking the limit x 0 we obtain the
following equation

2 w ( x, t )
t 2

w ( x, t )
H
+ q1 ( x, t )
x
x

(3.6)

Within the approximation of small vibrations,

H = T cos ( ) T
Eq.(3.6) can be written as

2 w ( x, t )
t 2

w ( x, t )
T
+ q1 ( x, t )
x
x

(3.7)

Equations (3.6) and (3.7) are partial differential equations that describe small
vibrations of the string about its equilibrium. These equations can predict the string
response both to initial conditions and the external loading q1 ( x, t ) provided that the
predicted vibrations do not violate the assumption Eq.(3.4). Note that tension T can
be a function of the co-ordinate x in Eq.(3.7), which is the case, for example, of long
catenary risers is considered.
3.2.2. Longitudinal Motion of a Rod. Modelling the longitudinal motion of a thin
rod it is assumed that its particles move along the main axis only. Such motion is
important for piles driven into soil. In statics, the rod model can be applied for
studying stresses in such structures as chimneys and skyscrapers to roughly analyze

61

stresses induced by the dead weight. Another important application of the model is
found when stresses are analysed in constraint structural elements undergoing thermal
expansion.
Consider a straight, prismatic rod as shown in Figure 3.2(a). The co-ordinate x
refers to a cross-section of the rod, while the longitudinal displacement of this section
is given by u x , t . We presume the rod to be under a dynamically varying stress field
x , t , so that adjacent sections are subjected to varying stresses. A body force
q x , t is also assumed present.

b g
b g

b g

(a)

bg

x + x

(b)
Figure 3.2. A thin rod: (a) with co-ordinate x and displacement
on element x of the rod

u of a section; (b) the stresses acting

With these assumptions the equation of motion of the rod in the x direction reads

Ax

2u
= ( x ) A + ( x + x ) A + qAx
t 2

(3.8)

where is the mass density of the rod material and A is the cross-sectional area of
the rod. The latter is a constant in this development, since we are considering a
prismatic rod. We note that tensile stress is assumed positive.
By taking into account the following Taylor expansion of x + x

g bg

x + x x +

x
x

(3.9)

Eq.(3.8) reduces to

2u
=
+q
t 2 x

(3.10)

Material effects have not been introduced yet, so equation (3.10) is applicable to nonelastic as well as elastic problems. We now presume that the material behaves
elastically and follows the simple Hookes law

62

= E

(3.11)

where E is the Youngs modulus and is the axial strain that for the case at hand is
defined by

= u x

(3.12)

Using (3.11) and (3.12) in the equation of motion, we obtain

2u u
=
E
+q
t 2 x x

(3.13)

One can easily recognise this equation to be identical in form to Eq.(3.6) derived for
the taut string.
If the rod is homogeneous so that E does not vary with x , equation (3.13) may be
written as
2u
2u
2 = E 2 +q
t
x

(3.14)

On the other hand, if the rod is not prismatic, so that its cross-sectional area A is a
function of x , Eq.(3.13) can be generalized to
2u
u
A 2 = EA + qA
t
x
x

(3.15)

These equations can predict the longitudinal motion of the thin rod both due to initial
conditions and the external loading q ( x, t ) provided that the strain-state remains
elastic and approximately uniaxial.
3.2.3. Torsion of a Rod. The torque may occur in many structures. For example, in
bridges loaded by heavy vehicles, in skyscrapers under wind loading, in rails loaded
by fright trains, etc. In reality, any rotational motion of a structure is accompanying by
warping, that is by deviation of every cross section from its original plane shape. The
less similar the structural cross-section to the circle, the greater the warping.
Performing a rough structural analysis, however, the warping can be neglected, which
will be followed in this development.
Let us obtain the equation of motion governing small torsion of a generalized rod.
To this end, we consider a straight, cylindrical rod, a differential element of which is
shown in Figure 3.3 (the circular cross-section is shown as an example). The coordinate x refers to a cross-section of the rod, while the angle of twist is given by
( x, t ) .

63

( x + x )

M t ( x + x )
Mt ( x)
x

m1

x + x

Figure 3.3. Differential element of a rod subject to end torques and distributed body torque m1 (x,t)

We assume that the rod is subject to a dynamically varying torque M t ( x, t ) and a

body torque with the density per unit length m1 ( x, t ) .
The equation of motion of the element in Figure 3.3 reads

J x

2
= M t ( x ) + M t ( x + x ) + m1 x
t 2

(3.16)

where is the mass density of the rod material and J is the polar moment of inertia.
By employing the following Taylor expansion of M t ( x + x )
M t ( x + x ) M t ( x ) +

M t
x
x

(3.17)

Eq.(3.8) can be reduced to

2 M t
=
+ m1
x
t 2

(3.18)

The torque M t is related to the angle of twist by

M t = GJ t

(3.19)

where G is the shear modulus of the rod material and J t is the torque constant. For
one-dimensional structural elements with circular cross-section this constant is
numerically equal to the polar moment of inertia J . For all other shapes of the crosssection, J and J t differ. As established from elasticity analysis, the torque constants
for the following four cross-sections are:

64

circle of radius a
Jt = J =

a 3b 3
a2 + b2

triangle: equilateral, of altitude a

Jt =

ellipse: semi-major and semi-minor axis of a and b ,

Jt =

a4

a4
15 3

narrow rectangle: width 2a , depth 2b , b >> a

Jt =

a 3b
3

Substituting the constitutive equation Eq.(3.19) into Eq.(3.18), we obtain

= GJ t
2
x
x
t

+ m1

(3.20)

One can easily recognise this equation to be identical in form to Eq.(3.6) derived for
the taut string and Eq.(3.13) derived for the rod in longitudinal motion.
Equation can predict a small rotational motion of one-dimensional structures both
due to initial conditions and the external torque m1 ( x, t ) in the case that the structural
shape does not cause considerable wrapping.
3.2.4. Pure Shear of a Beam. Some structures such as multi-storey buildings with
weak main walls may experience deformations which are very close to pure shear
deformations. For example, each storey of a high-rise building shaken by an
earthquake would move nearly horizontally. This deflection can be roughly predicted
with a model, which can be referred to as a shear beam, which is introduced in this
section.
Consider a straight, prismatic beam as shown in Figure 3.4. The co-ordinate x
refers to a cross-section of the beam, while the vertical displacement of this section is
given by w ( x, t ) . We presume the beam to be under a dynamically varying shear

stress field ( x, t ) , so that adjacent sections are subjected to varying stresses. A

b g

vertical body force q x , t is also assumed present.

65

x
x

x
w

( x + x )

( x)
q
x

Figure 3.4 A beam in pure shear

With these assumptions the equation of motion of the beam in the vertical direction
reads

Ax

2w
= ( x ) A + ( x + x ) A + qAx
t 2

(3.21)

where is the mass density of the beam material and A is the cross-sectional area of
the beam, the latter being a constant in this development.
Expanding ( x + x ) into the Taylor series and dividing by Ax we obtain from
Eq.(3.21)

2 w
=
+q
t 2 x

(3.22)

We now assume that the beam material behaves elastically and follows the simple
Hookes law while experiencing shear deformation:

= G

(3.23)

where G is the shear modulus and is the shear strain that for the case at hand is
defined by

= w x

(3.24)

Substituting the constitutive relation Eq.(3.23) into the equation of motion Eq.(3.22),
and using the kinematic relation Eq.(3.24), we obtain

2 w w
=
G
+q
t 2 x x

(3.25)

66

This equation is once again identical in form to the equations of motion obtained in
the preceding sections.
If the beam is homogeneous so that G does not vary with x , equation (3.13) may
be written as

2w
2w
=
+q
G
t 2
x 2

(3.26)

If the beam is not prismatic, so that its cross-sectional area A is a function of x ,

Eq.(3.25) can be generalized to

2w
w
= GA + qA
2
t
x
x

(3.27)

These equations can predict the shear motion of the thin beam both due to initial
conditions and the external loading q ( x, t ) provided that the shear strain plays the
major role in the deformation process.
3.2.5. Bending of a Beam. Bending of beams is the most common phenomenon in
structural engineering. Nearly every structure contains beams as structural elements
and a beam-in-bending model can be extremely instrumental both for dynamic and
static analyses of many structures.
Consider a straight beam undergoing transverse motion, as shown in Figure 3.5(a)
and consider a differential element of the beam as isolated in Figure 3.5(b).

q1 ( x, t )

x
M (x )

q1 ( x, t )
M ( x + x )

V (x )

V ( x + x )

(a)

(b)

Figure 3.5. (a) Beam undergoing transverse motion and (b) differential element of the beam subject
to the shear force, bending moment and external load

Bending moment M , shear force V and a distributed external force with density
per unit length q1 ( x, t ) act on the beam element. We invoke the basic hypothesis of
the Euler-Bernoulli theory of beams: namely that the plane cross-sections initially
perpendicular to the axis of the beam remain plane and perpendicular to the neutral
axis during bending. This assumption implies that the longitudinal strains vary
linearly across the beam and that, for elastic behaviour, the neutral axis of the beam
passes through the centroid of the cross-section. According to this theory the bending
moment and curvature are related as

67

M = EI

2w
x 2

(3.28)

where w( x, t ) is the transverse deflection of the neutral axis of the beam, E is the
Youngs modulus, and I is the moment of inertia of the cross-section. The result
Eq.(3.28) carries the assumption that the slope of the beam is small, see Eq.(3.4).
In accordance with Figure 3.5, the equation of motion of the beam element in the
vertical direction (if the slopes are small, the shear forces V ( x ) and V ( x + x ) can be
considered as if they were directed vertically) can be written as

Ax

2 w
= V ( x ) + V ( x + x ) + q1 x
t 2

(3.29)

Applying the Taylor expansion and dividing through by x , this equation can be
reduced to

2 w V
=
+ q1
x
t 2

(3.30)

In addition to the equation of motion, the moment equilibrium should be employed to

describe the beam motion. If we neglect the effect of rotational inertia, the moment
equilibrium is, effectively, that of statics and reads
V=

M
x

(3.31)

where the higher order contributions of the external load q to the moment are
neglected.
Substitution of Eq.(3.31) into the equation of motion Eq.(3.30), yields
2 w 2 M
A 2 = 2 + q1
t
x

(3.32)

Employing the constitutive relation Eq.(3.28), this equation can be written in terms of
the transverse deflection w of the beam neutral line:

2 w 2
+
t 2 x 2

2 w
EI 2 = q1
x

(3.33)

This equation governs the transverse motion of the beam as it responds to the external
loading q ( x, t ) and initial conditions. The application of this equation is constraint by
the requirements that the slopes of the beam are small, the rotational inertia effects are
neglected and the Euler-Bernoulli assumptions are not violated significantly.
It is important to note that Eq.(3.33) is qualitatively different from all equations of
motion obtained in the preceding sections. In contrast to all previously obtained
equations of motion, Eq.(3.33) is of the order four with respect to the spatial coordinate, whereas the other equations are of the second order. The implications of this
difference will become clear later in this chapter.

68

3.2.6. Viscous-Elastic Kelvin Foundation. In many cases structural elements are

subject to some reactions of the surrounding media all along their surface. For
example, a railway track is subject to reaction of the soil on which it is laid, a pile
being driven into the ground experiences resistance along its sides, an offshore riser
connecting a well at the seabed to a floating vessel is all surrounded by water, etc.
The simplest way to account for such reactions is by means of so called Kelvin
foundation, which is assumed to have certain stiffness and viscosity per unit length.
Therefore, it opposes both structural deflection and structural velocity.
As an example of incorporation of the Kelvin foundation into the equations of
motion, let us consider the vertical vibrations of a railway track on soil. The situation
is schematized in Figure 3.6. The railway track is approximated by the beam, whereas
the soil reaction is accounted for by the Kelvin foundation.

x
Figure 3.6. Beam on Kelvin foundation as a model for a railway track

The vertical motion of the beam is described by Eq.(3.33), which, assuming that
the railway track is homogeneous along its length, reduces to

2 w
4 w
+
= q1
EI
t 2
x 4

(3.34)

There are two external vertical forces, which act on the railway track (the beam) in the
absence of a train. One force is the gravity force, whereas the other one is the soil
reaction. The density per unit length of the gravity force is Ag , where g is the
gravity acceleration. The soil reaction we model by the Kelvin foundation, which
implies that the density per unit length of this reaction is given by
q1,SOIL = kd w cd

w
t

(3.35)

where k d is the soil stiffness per unit length and cd is the soil damping coefficient per
unit length. Substituting this expression into Eq.(3.34) and accounting for the gravity
force, we obtain

2w
4w
w
+
+ kd w + cd
= Ag
EI
2
4
t
x
t

(3.36)

This equation allows to analyse vibrations of the railway track in the gravity field due
to imposed initial conditions.

69

3.2.7. Combination of Models. The models formulated in this Section can be

superimposed upon each other. For example, the viscous-elastic foundation can be
used not only to oppose the transverse motion of a beam but also that of the string.
Equivalently, it can be used to oppose the shear motion of a beam, the rotational
motion of a rod and its longitudinal motion. In the latter case, the model can be
schematized as shown in Figure 3.7. In this case, we may speak about the longitudinal
cd

b g

u x, t

x
kd

Figure 3.7. Rod on visco-elastic foundation

motion of the rod on viscous-elastic foundation. The equation of motion for this
model can be deduced from the equation of motion for the rod. Assuming for
simplicity that the rod is prismatic and homogeneous, we employ Eq.(3.14) and
replace in this equation the body force q by the reaction of the foundation, that is by

q1 =

1
u
kd u cd
A
t

(3.37)

The right-hand side in Eq.(3.37) is divided by A to reformulate the surface reaction

of the foundation into the body force. Substitution of Eq.(3.37) into Eq.(3.14) gives
the following equation of longitudinal motion of the rod on elastic foundation subject
to no external force:

2u
2u 1
u
E

+ kd u + cd
=0
2
2
A
t
x
t

(3.38)

Another example of combination of the above-formulated models can be a

tensioned or axially compressed beam. Suppose that the beam is tensioned by an axial
force T as shown in Figure 3.8, and we can neglect neither bending nor tension in
calculation of the restoring force.
x
q1 ( x, t )
T

Figure 3.8. Tensioned beam

70

To obtain the equation of motion for this beam, we could repeat the regular procedure
of equilibrating the forces and moments, like in Section 3.2.4, accounting for the
additional tensile force. There is, however, a more economical approach of just
combining the equation of motion for the beam, Eq.(3.33), and that for the string,
Eq.(3.7). Indeed, these two equations can be considered as two particular cases of the
equation of motion for the tensioned beam. In this perspective, Eq.(3.7) for the string
can be considered as that for the tensioned beam with the restoring force associated
with the bending stiffness neglected. On the other hand, Eq.(3.33) for the beam can be
considered as that for the tensioned beam with the restoring force associated with the
tension neglected. Thus, to account for these restoring forces simultaneously, we may
add the term responsible for the bending force from Eq.(3.33) into Eq.(3.7), or,
alternatively, we may add the term responsible for the tensile force from Eq.(3.7) into
Eq.(3.33). In both cases, the result reads

2 w 2
+
t 2 x 2

2 w w ( x, t )
+ q1
EI 2 = T
x
x x

(3.39)

This is the equation for the transverse motion of the tensioned beam.
If the beam were not tensioned but axially compressed, the form of Eq.(3.39)
would stay intact, but the tensile force T would have to be replaced by a compressive
force P .

3.3.

Boundary and Interface Conditions for One-dimensional

Models

Every structure has a finite length and is fixed or attached to another structure at its
ends. Thus, to predict the mechanical behaviour of a structure, we have to describe
mathematically the conditions at the ends of the structure. This is done with the help
of the boundary conditions.
There are two types of the boundary conditions, namely: kinematic and dynamic.
Often, these are also referred to as essential and natural. A boundary condition is
called kinematic or essential if the kinematic characteristics of the structure are
prescribed at its ends such as displacements, angles or slopes. A boundary condition is
called dynamic or natural if forces or moments are prescribed at the ends. The number
of the boundary conditions, which have to be formulated at an end of the structure,
depends on the equation of motion, which is adopted for prediction of structural
mechanics. This number is always twice as small as the order of this equation with
respect to the spatial co-ordinate.
Sometimes, the mechanical characteristics of a structure, such as the crosssectional area or an elasticity modulus, change abruptly along the structure. The crosssection at which this happens is called an interface. At the interface, the equations of
motion are not valid and the interface conditions have to be formulated. These
conditions can also be subdivided into the kinematic and dynamic. However, at any
interface, both kinematic and dynamic conditions must be applied simultaneously,
which is not a necessity in the case of the boundary conditions. The number of the
interface conditions is always equal to the order of the equation of motion with
respect to the spatial co-ordinate.
In what follows, the boundary and interface conditions are formulated for the onedimensional models, which were introduced in the preceding section.

71

3.3.1. Transverse Motion of a String. The equation of motion of a string, Eq.(3.7) is

of the second order with respect to the spatial co-ordinate. Therefore, at every end of
the string, one boundary condition has to be formulated. The kinematic condition may
prescribe only the transverse deflection. The slope is prescribed by a dynamic
boundary condition, if this is formulated. The dynamic boundary conditions represent
the balance of forces acting on the ends of the string. The string participates in this
boundary conditions with the force, which it imposes on the end-cross-section. The
expression for this transverse force as it acts on the left-end of the string is given by
Eq.(3.2), which under the assumption of small slopes, Eq.(3.4), can be written as
V =T

w
x

(3.40)

In correspondence with Fig.3.1(b) this expression has to be taken with the minus sign
if applied to the right-end of the string.
Besides the force imposed by the string, the dynamic boundary conditions depend
on the external element, to which the boundary is attached. This can be a spring, a
dashpot or a mass. The forces imposed by these elements are known to us from the
courses on basic mechanics and Chapter 2 of this development.
Table 3.1 shows all basic boundary conditions (BC), which the string can be
subject to. In this table, A ( t ) and B ( t ) are the vertical displacements of the ends, k ,
c, and m are the stiffness, dashpot coefficient and mass of the boundary element,
respectively.

Type of BC
Kinematic
(displacement)

Left-end
w = A (t )

Dynamic
(external force)

w
= P
x

Dynamic
(elastic)

w
= kw
x

Dynamic
(viscous)

w
w
=c
x
t

Dynamic
(inertial)

w
2w
=m 2
x
t

Right-end
w = B (t )

w
=P
x

w
= kw
x

w
w
= c
x
t

w
2w
= m 2
x
t

Table 3.1. Boundary conditions for a string

Two particular, practically important, boundary conditions can be deduced from this
table, namely: the fixed end and the free end. For the fixed and not displaced end, the
displacements A ( t ) or B ( t ) are zero and the boundary condition reads w = 0 . For
the free end (subject to no external reaction), k = c = m = 0, so that the boundary
condition reads w x = 0 . Let us note that if we assumed that the string were
supported by the viscous-elastic Kelvin foundation, the boundary conditions would

72

not have changed. The reason for this is that the Kelvin foundation is not able to
provide a reaction at a cross-section (at an end) but only at an element of onedimensional system of a finite extension.
The interface conditions at a cross-section of the string, where the tension changes
abruptly should ensure that the string does not break and the balance of vertical forces
is not violated. This means that at interface x = c the following conditions must hold
simultaneously:
w ( x, t )
= w ( x, t ) x = c 0
x =c + 0

w
w
=T
T
x x =c 0
x x =c + 0

(3.41)

Naturally, these conditions must also hold at any other cross-section of the string.
3.3.2. Longitudinal Motion of a Rod. The equation of motion for the longitudinal
motion of a rod, Eq.(3.15) is of the second order with respect to the spatial coordinate, as well as the string-equation. Therefore, the boundary conditions for the rod
are mathematically similar to those for the string.
The kinematic boundary condition for the rod prescribes the longitudinal
displacement, whereas the dynamic boundary condition represents the balance of
longitudinal forces. The expression for the force imposed by the rod on its left-end can
be deduced from Eqs.(3.11) and (3.12) to give

f L = A = EA u x

(3.42)

In correspondence with Fig.3.2(b) this expression has to be taken with the minus sign
if applied to the right-end of the rod.
The basic boundary conditions for the rod in longitudinal motion are summarized
in Table 3.2. As in the previous sub-section, A ( t ) and B ( t ) are the displacements of
the ends, now longitudinal.
Type of BC
Kinematic
(displacement)

Left-end
A(t )

u = A(t )

Dynamic
(external force)

EA

u
= P
x

Dynamic
(elastic)

EA

u
= ku
x

Dynamic
(viscous)

EA

u
u
=c
x
t

Dynamic
(inertial)

EA

u
2u
=m 2
x
t

u = B (t )

EA

u
=P
x

EA

u
= ku
x

EA

u
u
= c
x
t

EA

u
2u
= m 2
x
t

Right-end

Table 3.2. Boundary conditions for a rod in longitudinal motion

B (t )

73

The interface conditions should be imposed at a cross-section of the rod, where

either the cross-sectional area or the Youngs modulus change abruptly. The interface
conditions must ensure continuity of the displacements and forces (stresses may be
discontinuous!). Therefore, at interface x = c the following conditions must hold
simultaneously:
u ( x, t )
= u ( x, t ) x = c 0
x =c + 0

u
u
= EA
EA
x x =c 0
x x =c + 0

(3.43)

3.3.3. Torsion of a Rod. The equation of motion for the rotational motion of a rod,
Eq.(3.20), is of exactly the same form as that for the string and the rod in longitudinal
motion. Therefore, the boundary and interface conditions must have the same form as
those in the preceding two sub-sections.
The kinematic boundary condition for the rod in torsion prescribes the angle of
rotation of a cross-section, whereas the dynamic boundary condition represents the
balance of torques. The expression for the torque imposed by the rod on its left-end is
given by Eq.(3.19) and reads
M t = GJ t

(3.44)

In correspondence with Fig.3.3 this expression has to be taken with the minus sign if
applied to the right-end of the rod.
The basic boundary conditions for the rod in torsion are summarized in Table 3.3.
In this table, A ( t ) and B ( t ) are the angles of rotation of the end-cross-sections,
kt [ N m ] , ct [ N s m] , and I t kg m 2 are the stiffness, dashpot coefficient, and
moment of inertia of the boundary element opposing the torque.

Type of BC
Kinematic
(displacement)
Dynamic
(external torque)
Dynamic
(elastic)

Left-end
= A (t )
GJt

Right-end
= B (t )

= M text
x

GJt

= M text
x

= kt
x

GJt

= kt
x

GJt

Dynamic
(viscous)

GJ t

= ct
x
t

GJ t

= ct
x
t

Dynamic
(inertial)

GJ t

2
= It 2
x
t

GJ t

2
= It
x
t2

Table 3.3. Boundary conditions for a rod in torsion

The interface conditions should be imposed at a cross-section of the rod, where

either the torque constant or the shear modulus change abruptly. The interface

74

conditions must ensure continuity of the rotation angle and torque. Therefore, at
interface x = c the following conditions must hold:
( x, t )
= ( x, t ) x = c 0
x=c+ 0

= GJ t
GJ t
x x =c + 0
x x = c 0

(3.45)

3.3.4. Pure Shear of a Beam. The shear motion of a beam is governed by Eq.(3.27),
which is identical in form to the equations of motion for the string and rod in both
longitudinal and rotational motion.
The kinematic boundary condition for the beam in pure shear prescribes the
transverse displacement. The dynamic boundary condition represents the balance of
shear forces. The expression for the shear force imposed by the beam on its left-end
can be deduced from Eqs.(3.23) and (3.24) to give
f S = A = GA

w
x

(3.46)

In correspondence with Fig.3.4 this expression has to be taken with the minus sign if
applied to the right-end of the beam.
The basic boundary conditions for the beam in pure shear are summarized in Table
3.4, using the same notations as for the string (sub-section 3.3.1).
Type of BC
Kinematic
(displacement)

Left-end
w = A (t )

Dynamic
(external force)

GA

Dynamic
(elastic)

GA

w
= kw
x

Dynamic
(viscous)

GA

w
w
=c
x
t

Dynamic
(inertial)

GA

w
2w
=m 2
x
t

Right-end
w = B (t )

w
= P
x

GA
P

w
=P
x

GA

w
=kw
x

w
w
GA =c
x
t

w
2 w
GA =m 2
x
t

Table 3.4. Boundary conditions for a beam in pure shear

The interface conditions should be imposed at a cross-section of the beam, where

either the cross-sectional area or the shear modulus change abruptly. The interface
conditions must ensure continuity of the transverse displacements and shear forces
(shear stresses may be discontinuous!). Therefore, at interface x = c the following
conditions must hold simultaneously:

75

w ( x, t )
= w ( x, t ) x = c 0
x =c + 0

w
w
= GA
GA
x x = c + 0
x x =c 0

(3.47)

3.3.5. Bending of a Beam. The equation of motion for the beam in bending, Eq.(3.33)
, is of the order four with respect to the spatial co-ordinate. Therefore, at each end of
the beam, two boundary conditions have to be formulated, in contrast to all previously
considered cases.
The kinematic condition may prescribe the transverse deflection and the slope of
the beam. The dynamic boundary conditions represent the balance of forces and the
balance of moments acting on the end of the beam. The beam participates in these
boundary conditions with the shear force and moment, which it imposes on the endcross-sections. The expressions for the moment and shear force as they act on the leftend of the beam are given by Eqs.(3.28) and (3.31), and read

2w
x 2
3w
V = EI 3
x
M = EI

(3.48)
(3.49)

In correspondence with Fig.3.5(b) these expressions are to be taken with the minus
signs if applied to the right-end of the beam.
In all preceding cases, at an end-cross-section, it was allowed to pose one boundary
condition (because the equations were of the second order), implying that we could
pose either kinematic or dynamic boundary condition and never both of them. In the
case of the beam in bending, two boundary conditions should be posed at each endcross-section. These can be two kinematic, two dynamic or kinematic and dynamic
boundary conditions.
In Table 3.5 all basic boundary conditions are presented, which the beam can be
subject to. In this table,
A ( t ) and B ( t ) are the vertical displacements of the ends
A ( t ) and B ( t ) are the slopes at the ends

k , c, and m are the stiffness, dashpot coefficient and mass of the boundary
element, acting in the vertical direction.
kr [ N m] , cr [ N s m ] , and I r kg m 2 are the stiffness, dashpot coefficient, and

moment of inertia of the boundary element opposing the rotation of the

beam in the plane of the figure.

76

Type of BC
Kinematic
(displacement)+
Kinematic
(slope)
Kinematic
(displacement)+
Dynamic Moment
(elastic)
Kinematic
(displacement)+
Dynamic Moment
(viscous)
Kinematic
(displacement)+
Dynamic Moment
(inertial)
Kinematic (slope)
+
Dynamic Force
(elastic)

Left-end
w = A(t )
A
w
= A ( t )
x

w
= B (t )
x

w = A(t )

2 w kr w
=
x2 EI x

kr

w = A( t )

2 w cr 2 w
=
x2 EI t x

cr

w = A( t )
2w Ir 3w
=
x2 EI t 2 x

Ir

w
= A ( t )
x
k
k
3w
w
=

EI
x3

Kinematic (slope)
+
Dynamic Force
(inertial)

w
= A ( t )
x
m
m 2w
3w
=
EI t 2
x3

EI

2w
=M
x2

EI

w
=P
x3

EI

w
=
x2

w = B (t )
k w
2w
= r
2
EI x
x
w = B(t )
cr 2w
2w
=
EI t x
x2

kr
B

cr

w = B( t )

w
= A ( t )
x
c w c
3w
=

EI t
x3

Dynamic Moment
(elastic + viscous
+inertial)
+
Dynamic Force
(elastic + viscous
+inertial)

w = B (t )

Kinematic (slope)
+
Dynamic Force
(viscous)

Dynamic Moment
(external)
+
Dynamic Force
(external)

Right-end

I r 3w
2w
=

EI t 2 x
x2
w
= B (t )
x
3w k
w
=
x3 EI
w
= B (t )
x
3 w c w
=
x3 EI t
w
= B (t )
x
3w m 2 w
=
x3 EI t 2

EI

2w
= M
x2

EI

w
= P
x3

EI

w
=
x2

Ir

w
2w
k
w
c
I
+
+
r

r
r
x
t
t 2

3w
w
2w
EI 3 = kw + c + m 2
x
t
t

w
2w
+
+
k
w
c
I
r

r
r
x
t
t 2
3w
w
2w
EI 3 = kw + c
+m 2
x
t
t

Table 3.5. Boundary conditions for a beam in bending

77

The last row in this table presents the two dynamic boundary conditions in the most
general form, not separating the elastic, viscous and inertial cases.
There are a few types of the beam fixation, which are widely used in practice,
namely:
pinned-pinned (at both ends the displacement and moment are zero);
clamped-clamped (at both ends the displacement and slope are zero);
clamped-free or cantilever (at one end the displacement and slope are zero,
whereas at the other end the moment and force are zero);
clamped-pinned (at one end the displacement and slope are zero, whereas at the
other end the displacement and moment are zero);
pinned-free (at one end the displacement and moment are zero, whereas at the other
end the moment and force are zero).
Though all these types of fixation are particular cases of the general ones presented in
Table 3.5, it is customary to collect these in a separate table. This is done in Table 3.6
presented below.
Type of fixation
Pinned-Pinned

Schematization
x=a

x=b

Boundary Conditions
x = a : w = 2 w x 2 = 0
x = b : w = 2 w x 2 = 0

Clamped-Clamped

Clamped-Free

x=a

x=b

x=a

x=b

x = a : w = w x = 0
x = b : w = w x = 0

x = a : w = w x = 0
2

x = b : 2 w x = 3 w x = 0

Clamped-Pinned

x=a

x=b

x = a : w = w x = 0
x = b : w = w2 x 2 = 0

Pinned-Free

x=a

x=b

x = a : w = 2 w x = 0
2

x = b : 2 w x = 3 w x = 0
Table 3.6 Typical beam fixations

There must be four interface conditions (in accordance with the spatial order of the
equation of motion) imposed at each cross-section of the beam, where the bending
stiffness changes abruptly. These conditions should ensure that the displacement,
slope, moment and shear force at the interface are continuous. Thus, at interface x = c
the following conditions must be satisfied simultaneously:

78

w ( x, t )
= w ( x, t ) x = c 0
x =c + 0

w
w
=
x
x x =c 0
x =c + 0

2w
2w
=
EI
EI
x 2
x 2 x =c 0
x=c+0

3w
3w
EI 3
= EI 3
x x = c 0
x x =c + 0

(3.50)

If the beam is homogeneous, the interface conditions, Eq.(3.50), reduce to the

requirement that the displacement and its derivatives up to the third order must be
continuous at each cross-section of the beam.

3.4.

Introduction to Dynamics of One-Dimensional Structures

In the preceding sections we formulated the equations of motion for a series of onedimensional structural models and discussed the boundary conditions, which
correspond to these models. This gives us the mathematical basis for studying the
mechanical behaviour of one-dimensional structures.
In this section, we will introduce the method of separation of variables, which
generally allows to study the dynamic behaviour of a large class of one-dimensional
models, as long as they are governed by linear differential equations.
Two models will be considered in this section, namely: the model for longitudinal
motion of a rod and that for the bending motion of a beam. It will be assumed that
both the rod and the beam are cylindrical and homogeneous. The results, which will
be obtained for the rod in longitudinal motion, can be easily generalized to describe
the motion of the string, rod in torsion and beam in pure shear. The possibility of such
generalization is based on the equations of motion for these four models, which have
exactly the same format. Indeed, each of these equations can be written in the
following generalized form (note that we assume all parameters to be independent of
the co-ordinate)
2
2w
2 w
=c
+Q
t 2
x 2

(3.51)

where c has the physical meaning of the wave speed, whereas Q represents a
generalized external force. To obtained from Eq.(3.51) the particular models, which
were introduced in Section 2 of this Chapter, the following substitutions are to be
made:
c 2 = T ( A) , Q = q1 ( A ) for the string, Eq.(3.7);
c 2 = E , Q = q for the rod in longitudinal motion, Eq.(3.14);
c 2 = GJ t ( J ) , Q = m1 ( J ) for the rod in torsion, Eq.(3.20);
c 2 = G , Q = q for the beam in pure shear, Eq.(3.26);

79

Thus, since in what follows we will consider dynamics of a rod in longitudinal

motion, the obtained results can be reformulated in the terms of any remaining three
models (except for beam in bending) by using the following formal replacement of
parameters:
rod in longitudinal motion to string: E T , A, q q1 ;
rod in longitudinal motion to rod in torsion: E GJ t , J , q m1 ;
rod in longitudinal motion to beam in pure shear: E G, , q q .

The bending motion of a beam will be considered separately since the equation of
motion in this case is different from the other ones.
3.4.1. Longitudinal Vibrations of a Rod. The complete mathematical statement of a
problem describing the dynamic behaviour of a continuous system includes three
components, namely: the equation(s) of motion, the boundary conditions and the
initial conditions. Depending on the loading type and the type of the boundary
conditions, different mathematical methods may be applied to achieve the most
efficient solution procedure. In this section, the method of separation of variables will
be employed, which can handle almost any loading type but, in its classical form, is
only applicable to time-independent boundary conditions.
In this sub-section we consider vibrations of a rod, which is cylindrical,
homogeneous and has the length L . The equation of motion governing the
longitudinal motion of such a rod and the most general form of the initial conditions
read

2u
2u
=
+ q ( x, t )
E
t 2
x 2

(3.52)

u ( x, 0 ) = u 0 ( x )
u
= v0 ( x )
t t =0

(3.53)

To demonstrate the method of separation of variables we will assume that the rod is
fixed at both ends as shown in Figure 3.9. Note, that as mentioned in the beginning of
this section, the method can be equally easily applied to any other time-independent
boundary conditions, which account for a constant displacement or a constant force
applied at the ends of the rod.

u ( x, t )

x=0

q ( x, t )

x=L
Figure 3.9. Fixed-fixed rod

80

The boundary conditions for the rod fixed at both ends read
u ( 0, t ) = u ( L, t ) = 0

(3.54)

The method of separation of variables is based on the assumption that the time- and
space- dependent generalized displacement of a continuous system can be represented
as a multiplication of two functions, one dependent only on the time, and the other
dependent only on the spatial coordinate. In accordance with this method, we assume
that the longitudinal displacement of the rod can be represented in the following form
u ( x, t ) = U ( x ) ( t )

(3.55)

where U ( x ) and ( t ) are unknown functions of the coordinate along the rod and
time, respectively.
Let us first assume that the external loading q is absent and the equation of
motion, Eq.(3.52), reads
2
2u
2 u
=c
t 2
x 2

(3.56)

where c = E is the speed of longitudinal waves in the rod.

Substitution of Eq.(3.55) into the equation of motion, Eq.(3.56) gives
U

d 2
2U
2
c
=

dt 2
x 2

(3.57)

Dividing this equation through by U , we obtain

2
1 d 2
2 1 U
c
=
U x 2
dt 2

(3.58)

The left-hand-side of Eq.(3.58) depends only on time, whereas its right-hand-side

depends only on the coordinate. Thus, both the left- and the right-hand-side must be
equal to a constant, otherwise it would be impossible to equalize two functions of two
different independent variables:
1 d 2
= 2

dt 2

2
c 2 1 U = 2
U x 2

Momentarily, 2 in Eq.(3.59) is a separation constant.

The first equation in Eq.(3.59) can be written as

(3.59)

81

d 2
+ 2 = 0
2
dt

(3.60)

The general solution of this equation is known to us, see Eq.(2.12). It reads
= A sin (t ) + B cos (t )

(3.61)

This form of the solution represents simple harmonic motion at the frequency ,
provided that the separation constant in Eq.(3.59) is positive, i.e., 2 > 0 . If we had
chosen a separation constant less than zero or complex in Eq.(3.60), we would than
obtain a non-harmonic solution, which is definitely incapable of representing the
periodic motion, which, as the common sense suggests, must be possible in the rod
without damping. Thus, the separation constant 2 in Eq.(3.59) must be positive.
Consequently, can be given the physical sense of a frequency of vibrations.
The second equation in Eq.(3.59) can be written as
d 2U 2
+ U =0
dx 2 c 2

(3.62)

The general solution of this equation under the condition that > 0 reads
U = C sin ( x ) + D cos ( x )

(3.63)

where

(3.64)

Now we will use the boundary conditions. Eq.(3.54). Substituting Eq.(3.55) into
these boundary conditions we obtain the following equations

U ( 0 ) ( t ) = 0

U ( L ) ( t ) = 0
which have to be satisfied at every time instant. This can be achieved if and only if the
following equations hold
U ( 0) = U ( L ) = 0

(3.65)

Substituting the general expression for U into these boundary conditions, and
imposing the condition of non-triviality that the coefficients C and D may not vanish
simultaneously, we obtain
D=0

(3.66)

sin ( L ) = 0

(3.67)

82

Equation (3.67) is called the frequency equation, since it determines the natural
frequencies of the system. For the case at hand, we have

n L = n

n = c n = n c L,

n = 1, 2,3,...

(3.68)

The frequencies n in Eq.(3.68) represent the discrete frequencies at which the rod is
capable of undergoing harmonic motion. For a given value of n , we thus have the
vibration pattern of the rod described by
U n ( x ) = Cn sin ( n x )

(3.69)

where U n are usually called the normal modes of the system.

Several of the resulting vibration patterns are shown in Figure 3.10.
u

n=1

n=2
u

x
n=3

x
n=4

Figure 3.10. Symmetric (n = 1,3) and antisymmetric (n = 2,4) modes of a fixed-fixed rod

From this figure we can note that the modes with n = 1,3,... give symmetric motion
while antisymmetric modes result from n = 2,4,. The points of zero displacement
are called the nodes of vibration, while the points of maximum vibration are called the
antinodes.
Combining the time dependence, Eq.(3.61), and the spatial dependence, Eq.(3.63)
for a given value of n, we obtain
un ( x, t ) = ( An sin (n t ) + Bn cos (n t ) ) sin ( n x )

(3.70)

where the constant Cn of Eq.(3.69) has been absorbed in the constants An and Bn .
The general solution of Eq.(3.56) is obtained by the superposition of all particular
solutions of the form (3.70). This solution reads

83

u ( x, t ) = ( An sin (nt ) + Bn cos (n t ) ) sin ( n x )

(3.71)

n =1

To determine the unknown constants An and Bn , the general solution, Eq.(3.71), is

to be substituted into the initial conditions, Eq.(3.53) . This yields

n =1

sin ( n x ) = u0 ( x )

(3.72)

A
n

n =1

sin ( n x ) = v0 ( x )

To find the constants An and Bn from Eqs.(3.72), we multiply each of these equations
by sin ( m x ) and integrate over the interval 0 x L . Noting that

sin ( x ) sin ( x ) dx = 0,
L

L 2,

m=n
mn

(3.73)

we obtain
L

2
An =
v0 ( x ) sin ( n x ) dx
Ln 0
L

(3.74)

2
Bn = u0 ( x ) sin ( n x ) dx
L0

This establishes the free motion of the rod.

Let us note that Eq.(3.71) is effectively a statement that the rod is a system with
infinite number of degrees of freedom. We recall that in the vibrations of mass-spring
systems, the problem is to describe the motion of each mass element. If there are n
elements, which can move only in one direction, the system will have n degrees of
freedom and its motion will be described by displacements u1 ( t ) , u2 ( t ) ,..., un ( t ) . In a
continuous system, n varies from one to infinity, therefore any infinite system has an
infinite number of degrees of freedom. In the normal mode representation, Eq.(3.71),
each mode corresponds to a single degree of freedom.
Let us now return to Eq.(3.52), which has the forcing term on the right-hand side.
The solution of this equation can be sought for as the superposition of vibrations on
the normal modes. Such form of the solution allows to satisfy automatically the
boundary conditions. For the fixed-fixed rod, the normal modes are given by
Eq.(3.69) so that the form, in which the solution to Eq.(3.52) can be sought for, reads

u ( x, t ) = n ( t ) sin ( n x )

(3.75)

n =1

where the coefficients n ( t ) should be functions of time to represent the time-

dependent character of q ( x, t ) .
Substitution of Eq.(3.75) into Eq.(3.52) gives

84

d 2 n

1
2

2 + n n sin ( n x ) = q ( x, t )

n =1 dt

(3.76)

Multiplying this equation by sin ( m x ) and integrating over the interval 0 x L , we

obtain
L
d 2 n
L
1
2
+
x
x
dx
=
q ( x, t ) sin ( m x ) dx

sin
sin
( n ) ( m )

2
n n
0
n =1 dt
0

which, employing Eq.(3.73), can be reduced to

d 2 n
2
+ n2 n =
q ( x, t ) sin ( n x ) dx = Qn ( t )
2
L 0
dt
L

(3.77)

This equation has exactly the same form as Eq.(2.36), therefore it must have the
solution of the form given by Eq.(2.39), i.e.,

n ( t ) = n ( 0 ) cos (nt ) +

t
1 d n ( 0 )
1
sin (nt ) + Qn ( t ) sin (n ( t t ) ) dt (3.78)
n dt
n 0

where n ( 0 ) and d n ( 0 ) dt should be found from the initial conditions. This is

done by substitution of Eq.(3.78) into Eq.(3.75) with the subsequent substitution into
the initial conditions, Eq.(3.53). This yields

( 0 ) sin ( x ) = u ( x )
n

n =1

d n ( 0 )

n =1

dt

(3.79)
sin ( n x ) = v0 ( x )

This system of equations can be resolved in exactly the same manner as Eqs.(3.72) to
give
d n ( 0 )
dt

2
= v0 ( x ) sin ( n x ) dx
L0
L

(3.80)

2
n ( 0 ) = u0 ( x ) sin ( n x ) dx
L0

This completes the solution of Eq. (3.52).

Let us now pose the question: what is the relation between the general dynamic
solution, Eq.(3.70), of the homogeneous equation, Eq.(3.56), and the static solution of
this equation. If the statics is considered, all time derivatives in the equations of
motion should be set to zero. Thus, the longitudinal displacement of the rod in statics,
in accordance with Eq.(3.56) is governed by the following equation:

85

2u
=0
x 2

(3.81)

the general solution of which is

u = C1 + C2 x

(3.82)

From the first glance, it seems that there can not be any relationship between the static
solution, Eq.(3.82), and the dynamic solution, Eq.(3.70). Note, however, that the
dynamic solution, Eq.(3.70), was obtained under the requirement that the frequency
is not greater than zero. But in statics we have = 0 . Therefore, the general
solution of Eq.(3.62), has to be rewritten for the case = 0 to correspond to statics.
This gives the following expression for U ( x ) (instead of Eq.(3.63))
U = C1 + C 2 x

(3.83)

which, taking into account that there is only one natural frequency, which is equal to
zero, represents the complete solution to the problem.
Thus, the dynamic solution can be reduced to the static one by taking the limit of
all natural frequencies going to zero. This confirms mathematically the obvious fact
that statics is a particular case of dynamics.
3.4.2. Bending Vibrations of a Beam. The method of separation of variables can be
applied to the problem of bending motion of a beam in the same manner as described
in the previous sub-section.
Consider the bending motion of a beam, which is cylindrical, homogeneous and
has the length L . In this sub-section we will consider only the free vibrations of the
beam, i.e., the external force q1 will be neglected. The equation of motion governing
the beam motion in this case and the general form of the initial conditions read
4
2w
2 w
a
+
= 0,
t 2
x 4

a2 =

EI
A

(3.84)

u ( x, 0 ) = u 0 ( x )
u
= v0 ( x )
t t =0

(3.85)

Let us assume that the beam is pinned at both ends so that the boundary conditions at
x = 0 (the left-end) and x = L (the right-end) are given as
2w
w ( 0, t ) = 2 ( 0, t ) = 0
x

(3.86)

2w
( L, t ) = 0
x 2

(3.87)

w ( L, t ) =

86

In accordance with the method of separation of variables, we search for the

solution of the problem in the following form
w ( x, t ) = W ( x ) ( t )

(3.88)

where W ( x ) and ( t ) are unknown functions of the coordinate along the rod and
time, respectively.
Substituting Eq.(3.88) into the equation of motion, Eq.(3.84) and dividing the
obtained result by W , we obtain
4
1 d 2
2 1 W
a
+
=0
W x 4
dt 2

(3.89)

Since the first term in Eq.(3.89) depends only on time, whereas the second term
depends only on the coordinate, this equation can be satisfied if and only if these
terms are equal to a separation constant. This constant has to be introduced in the
following manner, to ensure that the beam is capable of performing harmonic
vibrations:
1 d 2
= 2

2
dt

4
a 2 1 W = 2
W x 4

(3.90)

where is a real value representing a frequency of vibrations.

The general solution of the first equation in Eqs.(3.90) reads
= A sin (t ) + B cos (t )

(3.91)

and represents simple harmonic motion at the frequency .

The second equation in Eq.(3.59) can be written as
d 4W 2
W =0
dx 4 a 2

(3.92)

The general solution of this equation, in accordance with the general representation,
Eq.(2.7), can be written as
4

W ( x ) = C n exp ( sn x )

(3.93)

n =1

Substituting Eq.(3.93) into Eq.(3.92), we obtain the following characteristic equation

sn4

2
a2

=0

(3.94)

87

Solving this characteristic equation, we find

s1 = , s2 = , s3 = i , s4 = i

(3.95)

where

=
4

(3.96)

a2

Substituting the characteristic exponents, Eq.(3.95), into Eq.(3.93), we obtain

W ( x ) = C1 exp ( x ) + C 2 exp ( x ) + C3 exp ( i x ) + C 4 exp ( i x )
= C1 sin ( x ) + C2 cos ( x ) + C3 sinh ( x ) + C4 cosh ( x )

where

C1 = i C3 C 4 , C2 = C3 + C 4 , C3 = C1 C 2 , C4 = C1 + C 2

Thus, the general solution of Eq.(3.92) can be written as
W = C1 sin ( x ) + C2 cos ( x ) + C3 sinh ( x ) + C4 cosh ( x )

(3.97)

The boundary conditions, Eqs.(3.86) and (3.87) can be rewritten in the terms of
W ( x ) as (since these should be valid at any time instant)
W (0) =

2W
(0) = 0
x 2

(3.98)

W ( L) =

2W
( L) = 0
x 2

(3.99)

Substituting the general solution, Eq.(3.97), into these boundary conditions, and
imposing the non-triviality condition that the coefficients C1 C4 may not vanish
simultaneously, we obtain
C2 = C3 = C4 = 0
sin ( L ) = 0

(3.100)

n L = n ,

n = 1, 2,3,...

(3.101)

The frequency equation, Eq.(3.101), determines the natural frequencies of the beam
vibrations, which, taking into account Eq.(3.96), can be found as
n
n = a
L

(3.102)

88

Correspondingly, the normal mode of the beam vibrations, for a given value of n ,
reads
Wn ( x ) = Cn sin ( n x )

(3.103)

which is precisely the same in the form as Eq.(3.69) derived for the rod. Therefore,
the shape of the normal modes repeats that shown in Figure 3.10.
Combining the time dependence, Eq.(3.91), and the spatial dependence, Eq.(3.103)
we obtain the following general expression describing the beam vibrations caused by
the initial conditions:

w ( x, t ) = ( An sin (n t ) + Bn cos (n t ) ) sin ( n x )

(3.104)

n =1

The constants An and Bn can be determined from the initial conditions to give the
expressions presented by Eq.(3.74), in which the natural frequencies are defined by
Eq.(3.102).
Note, that if the boundary conditions at the beam ends were chosen differently, the
frequency equation would be more complicated (transcendental) as well as the normal
modes and, consequently, the overall solution.

3.5.

Elastostatics of One-Dimensional Structures

In a great deal of practical situations, the major contribution to the stress-state of civil
engineering structures is associated with static loads. These loads may be associated
with the direct gravity loading, the dead weight of various loads, the mean component
of the drag force in a flow, the temperature extension of materials, etc. Therefore, it is
of immense importance for civil engineering to study the statics response of structures
to time-invariant loads.
The static response of an elastic structure is governed by the balance of forces and
moments acting on this structure. This balance can be straightforwardly deduced from
the corresponding equation of motion by setting the time derivatives to zero and
neglecting the time dependence of the loading term. For example, consider the
equation of longitudinal motion of the rod, Eq.(3.15):

2u
u
= EA + q ( x, t ) A
2
t
x
x

(3.105)

Assume that the external force q ( x ) acting on the rod is time-independent. In this
case, after a sufficiently long time after the application of the load, the structure will
assume the static equilibrium, in which its displacement u is time independent. In this
case the time derivatives of u vanish and Eq.(3.105) takes the form

u
EA = q ( x ) A
x
x

89

which represents the balance of axial forces acting on a differential element of the rod.
The same idea holds for the boundary conditions. To study the static response of
structures, we may use the boundary conditions formulated in Section 3.3 setting to
zero those terms in these conditions, which include the time derivatives.
In this section we will consider the static response of the rod to longitudinal forces
and that of the bending beam to transverse forces and moments. Various forces and
boundary conditions will be treated. Additionally, both the rod and beam will be
considered subject to the elastic foundation and the effect will be studied of this
foundation on the static response and the method of analysis. The static response of
the other three above-introduced models (string, rod in torsion and beam in shear) will
not be studied, because of the mathematical similarity of these models to that of the
rod.
3.5.1. Static Response of a Rod to Axial Loading. Let us shortly repeat the
equations, which govern the axial deflection of the rod. These are:
the equilibrium of axial forces, as follows from Eq.(3.8):
d
( A ) = qA
dx

(3.106)

Note that this form of the equilibrium is valid not only for the cylindrical rods but
also for the rods with variable cross-sections.
the constitutive equation, Eq.(3.11):

= E

(3.107)

the kinematic relation, Eq.(3.12)

= u x

(3.108)

Studying statics of one-dimensional structures, it is customary to work with forces

rather than with stresses. The axial deflection of the rod takes place under the action
of axial forces, which are also called normal forces, underlying that they are normal to
the cross-section, on which they act. For the rod with a homogeneous cross-section,
the normal force is defined as
N = A

(3.109)

which, using the kinematic relation and constitutive equation, results in

N = EA

du
dx

(3.110)

Making use of the normal force, the equilibrium equation can be written as
dN
= qA
dx

(3.111)

90

or, substituting Eq.(3.110), as

d
du
AE
= qA
dx
dx

(3.112)

which is a particular case of Eq.(3.15).

The general solutions to Eqs. (3.111) and (3.110) can be easily found to give
x

N ( x ) = q ( x ) A ( x ) dx + N ( 0 )

(3.113)

u ( x) =
0

N ( x )
dx + u ( 0 )
E ( x ) A ( x )

(3.114)

If the cross-sectional area A , the Youngs modulus E , and the body force q are
invariant along the rod, the general solutions, Eqs.(3.113) and (3.114) reduce to
N ( x ) = q1 x + C1
u ( x) =

(3.115)

1 1

2
q1 x + C1 x + C2
EA 2

(3.116)

where q1 = qA is the density per unit length of the external force.

The boundary conditions, which are applicable to the rod in statics, can be deduced
from Table 3.2 to form the following table:
Type of BC
Kinematic
(displacement)

Left-end
u=A

Dynamic
(external force)

N = P

Dynamic
(elastic)

N = ku

Right-end
u=B
N=P
N = ku

Table 3.7. Boundary conditions for static axial deformation of a rod

Note that the dynamic boundary condition in Table 3.7 is given in terms of the normal
force, which is more suitable for static analyses.
The interface conditions at a cross-section x = c of the rod, where an abrupt
change of the Youngs modulus or the cross-sectional area takes place, are given by
Eq.(3.43). These conditions can also be written making use of the normal force to give
u ( x )
= u ( x ) x =c 0
x =c + 0

N ( x ) x =c + 0 = N ( x ) x = c 0

(3.117)

91

Equations (3.110) and (3.111) supplemented by Table 3.7 and the interface
conditions, Eq.(3.117), allow us to analyse the axial deflections and normal forces in
the rod induced by any axial force. In what follows we consider a number of possible
loads and boundary conditions.
Example A. Consider a fixed-free cylindrical and homogeneous rod of the length L as
shown in Figure 3.11(a). Assume that this rod is subject to a distributed axial force
with the constant density per unit length q1 .
The boundary conditions in the case at hand are given as
u ( 0) = 0

(3.118)

N ( L) = 0

(3.119)

Since the rod is cylindrical and homogeneous, and the external force is constant, the
simplified expressions can be applied for the normal force and the axial displacement,
Eqs.(3.115) and (3.116).
Substituting the simplified expression for the normal force, Eq. (3.115), into the
boundary condition at the right end, Eq.(3.119), gives C1 = q1 L , which implies that
the normal force in the rod is given as:
N = q1 ( L x )

(3.120)

Thus, the axial force varies linearly along the rod as shown in Figure 3.11(b). Because
the rod is cylindrical, the variation of the axial stress along the rod is also linear, as
well as that of the axial strain, in correspondence with the constitutive equation,
Eq.(3.107).
(a)

q1

u=0

N =0
L

(b)

N
(c)

u
Figure 3.11. (a) Fixed-free rod under uniform axial loading; (b) normal force; (c) axial deflection

To find the axial displacement of the rod, we substitute the simplified expression for
this displacement, Eq.(3.116), into the boundary condition at the right end of the rod.
This yields

92

0=

1
( 0 + 0 + C2 )
EA

C2 = 0

(3.121)

Thus, the variation of the axial deflection along the rod is parabolic and given as
u=

1
q1 x ( 2 L x )
2 EA

(3.122)

which is shown in Figure 3.11(c).

Example B. Consider a fixed-fixed cylindrical and homogeneous rod of the length L
as shown in Figure 3.12(a). Assume that this rod is subject to a distributed axial force
with the constant density per unit length q1 .
The boundary conditions in this case are given as
u ( 0) = 0

(3.123)

u ( L) = 0

(3.124)

To solve this problem we apply the simplified expression for axial displacement,
Eq.(3.116). Substituting this expression into the boundary conditions, we obtain:
(a)

q1

u=0

u=0
L

(b)

(c)

Figure 3.12. (a) Fixed-fixed rod under uniform axial loading; (b) normal force; (c) axial deflection

93

0=

1
( 0 + 0 + C2 )
EA

C2 = 0

0=

1 1 2

q1 L + C1 L + C2
EA 2

C1 =

(3.125)
1
q1 L
2

Thus, the normal force and axial displacement, in accordance with Eqs.(3.115) and
(3.116) become
1

N = q1 L x ,
2

u=

1
q1 x ( L x )
2 EA

(3.126)

These are shown in Figure 3.12(b,c).

Example C. Consider a fixed-fixed cylindrical and homogeneous rod of the length L
as shown in Figure 3.13(a). Assume that this rod is subject to a distributed axial force
with the variable density per unit length q1 = q0 cos ( x L ) shown in Figure 3.13(b).
The boundary conditions in this case are the same as in the previous one and are
given by Eqs.(3.123) and (3.124).
Since the external force is not constant along the rod, we should apply the general
expressions for the normal force and axial displacement, Eqs.(3.113) and (3.114).
Substituting q1 = q0 cos ( x L ) into Eq.(3.113) and evaluating the integral, we obtain
(a)

q1

u=0

u=0
L

(b)

q1
(c)

N
(d)

u
Figure3.13. (a) Fixed-fixed rod under cosinusoidal axial loading; (b) external axial force; (c)
normal force in the rod; (d) axial deflection

94

N ( x) =

x
sin
+ N ( 0)

q0 L

(3.127)

where N ( 0 ) is unknown. Substituting the obtained expression for the normal force,
Eq.(3.127), into the general expression for the axial displacement, Eq.(3.114), we
obtain
u ( x) =

q0 L2
x N (0)
cos
1 +
x + u (0)

2
EA L EA

(3.128)

To determine the two unknowns, namely N ( 0 ) and u ( 0 ) , we substitute Eq.(3.128)

into the boundary conditions, Eqs.(3.123) and (3.124). This yields
0 = u ( 0)

u (0) = 0

0 = 2

q0 L2 N ( 0 )
+
L + u ( 0)
2 EA EA

N (0) = 2

q0 L

Thus, the normal force and axial displacement, in accordance with Eqs.(3.127) and
(3.128) become
N=

q0 L
x
2 sin

2

L

(3.129)

qL

x
u = 20 L cos
L + 2x
EA
L

These are shown in Figure 3.13(c,d).

Example D. Consider a homogeneous vertically positioned rod (a column) of the

length L , with an unknown cross-sectional area, which may depend on the height as
shown in Figure 3.14. Assume that this rod is subject to the gravity force and a
concentrated vertical load P applied at the top of the column. The bottom side of the
rod is fixed. The task is to find such dependence of the cross-sectional area on the
height that the axial stress is constant along the rod and equal to 0 .
P

x, u

q1 ( x )
Figure 3.14. A vertical column with a variable cross-section loaded by its deadweight

95

The body force, which the gravity imposes in the column is q = g , where is the
mass density of the column material and g is the gravity acceleration. Consequently,
the equilibrium of vertical forces in the column (expressed through the normal stress),
Eq.(3.106), takes the form
0

dA ( x )
dx

= gA ( x )

(3.130)

This is an ordinary differential equation of the first order, the solution to which can be
sought in the form
A ( x ) = A0 exp ( sx )
Substituting this into Eq.(3.130), we obtain

0s = g

s = g 0

(3.131)

Thus the cross-section should have the following exponentially depending of the
height form:
g
A ( x ) = A0 exp
0

(3.132)

The unknown constant A0 should be found from the equilibrium condition at the top
of the column, which reads
N ( 0) = P

(3.133)

By definition, the normal force equals the multiplication of the normal stress and the
cross-sectional area, see Eq.(3.109). Thus, Eq.(3.133) can be written as
0 A ( 0 ) = P ,
which, after substitution of Eq.(3.132), yields the following expression for A0 and the
final expression for the cross-sectional area:
A0 =

A( x) =

g
exp
0
0
P

(3.134)

Example E. Consider a cylindrical and homogeneous rod of the length L , the left end
of which is fixed, whereas the right end is attached to a spring with the stiffness k as
shown in Figure 3.15(a). Assume that this rod is subject to a distributed axial force
with the constant density per unit length q1 .

96

(a)

q1

u=0

3L 5
(b)

N = ku
2
q1 L
5

N
1
q1 L2
10

(c)

u
Figure 3.15. (a) Fixed-elastically supported rod under uniform axial loading; (b) normal force; (c)
axial deflection. The results are presented for K = 4

The boundary conditions in this case, in correspondence with Table 3.7, read
u ( 0) = 0

(3.135)

N ( L ) = ku ( L )

(3.136)

Since the rod is cylindrical and homogeneous, and the external force is constant, the
simplified expressions can be applied for the normal force and the axial displacement,
Eqs.(3.115) and (3.116).
Substituting the simplified expression for the displacement, Eq.(3.116), into the
boundary condition at the left end of the rod, Eq.(3.135), we obtain
0=

1
( 0 + 0 + C2 )
EA

C2 = 0

With this constant defined, substitution of the simplified expression for the normal
force, Eq. (3.115), into the boundary condition at the right end, Eq.(3.136), gives
q1 L + C1 =

k 1 2

q1 L + C1 L
EA 2

(3.137)

Solving Eq.(3.137), we find

C1 = q1 L

1+ K 2
1+ K

(3.138)

97

where K = kL ( EA ) represents the ratio of the spring stiffness k to the overall axial
stiffness of the rod EA L .
Thus, the normal force in and the axial displacement of the rod are given as
1+ K 2

N = q1 L
x
1+ K

q 1+ K 2 1 2
u = 1 xL
x
1+ K
2
EA

These are shown in Figure 3.15(b,c) for K = 4 .

Example F. Consider a homogeneous rod of the length L , which consists of two
cylindrical parts, each of the length L 2 as shown in Figure 3.16(a). Both parts are
cylindrical but the cross-sectional area of the left-hand part is AL , whereas the crosssectional area of the right-hand part is AR . Assume that this rod is subject to a
distributed axial force with the constant density per unit length q1 and fixed at x = 0
and x = L .
The boundary conditions in the case at hand read
u ( 0) = 0

(3.139)

u ( L) = 0

(3.140)

(a)

q1
u=0

L 2
7 L 12

(b)

L 2

u=0

5
q1 L
12

N
(c)

Figure 3.16. (a) Fixed-fixed combined rod under uniform axial loading; (b) normal force; (c) axial
deflection

98

Since the cross-section experiences the abrupt change at x = L 2 , we have to

formulate the interface conditions at this point. In accordance with Eq.(3.117), these
conditions read
u ( L 2 + 0 ) = u ( L 2 0 )

N ( L 2 + 0 ) = N ( L 2 0 )

(3.141)

Since both parts of the rod are cylindrical and homogeneous, the simplified
expressions for the normal force and axial displacement, Eqs.(3.115) and (3.116),
hold for each part. Thus, introducing the subscripts L and R for the left- and righthand-part related quantities, respectively, we may write
N L ( x ) = q1 x + C1

(3.142)

2
q1 x + C1 x + C2
2

N R ( x ) = q1 x + C3

uL ( x ) =

uR ( x ) =

1
EAL

1
EAR

(3.143)
(3.144)

2
q1 x + C3 x + C4
2

(3.145)

Substituting these expressions into the boundary and interface conditions, we

obtain the following system of four linear algebraic equations with respect to the
unknown constants C1 C4 :

from

u ( 0 ) = uL ( 0 ) = 0 :

from

u ( R ) = uR ( L ) = 0 :

from
from

C2 = 0

1
q1 L2 + C3 L + C4 = 0
2
1 1 2 1
1 1 2 1

u L ( L 2 ) = uR ( L 2 ) :
q1 L + C1 L + C2 =
q1 L + C3 L + C4

AL 8
2
2
AR 8

1
1
N L ( L 2) = N R ( L 2 ) : q1 L + C1 = q1 L + C3
2
2

Solving this system of equations we find

C1 =

1
1 + 3 A
q1 L
,
4
1 + A

C2 = 0,

C3 = C1 ,

C4 =

1 2 1 A
q1 L
4
1 + A

(3.146)

where A = AL AR .
Thus, the problem has been solved. For graphical representation of the results, let
us assume that the left-hand part of the rod has twice greater cross-section than the
right-hand part, i.e., A = 2 . In this case the expressions for the constants, Eq.(3.146),
reduce to
C1 =

7
q1 L,
12

C2 = 0,

C3 = C1 ,

C4 =

1
q1 L2
12

99

and the expressions for the normal forces and axial displacements, Eqs.(3.142)(3.145) become
7

N L ( x ) = q1 L x ,
12

q x 7

uL ( x ) = 1 L x ,
2 EAL 6

N R ( x ) = q1 L x
12

q 7
1

u R ( x ) = 1 Lx L2 x 2
2 EAR 6
6

(3.147)

The distribution of the normal force and that of the axial displacement along the rod
are plotted in Figure 3.16(b,c).
3.5.2. Static Response of a Rod on Elastic Foundation to Axial Loading. In this
sub-section we consider the rod as it is subject to a distributed reaction of an elastic
foundation as shown in Figure 3.17.

kd
x, u
Figure 3.17. Rod on elastic foundation

The equilibrium equation, governing the balance of axial forces in this rod reads
d
du
EA = qA + kd u
dx
dx

(3.148)

which is the static reduction of Eq.(3.38).

The constitutive equation, kinematic relation, expression for the normal force N ,
the boundary and interface conditions are the same for this rod as for the rod without
foundation.
If the cross-sectional area A and the Youngs modulus E are invariant along the
rod, Eq.(3.148) reduces to
EA

d 2u
kd u = qA
dx 2

(3.149)

If, additionally, the body force q and the stiffness per unit length kd of the elastic
foundation are also coordinate-invariant, then the general solution of this equation can
be easily found to give
kd
kd qA
u = C1 cosh
x + C2 sinh
x +
EA
EA k
d

(3.150)

100

Consequently, employing Eq.(3.110), which relates the axial displacement to the

normal force, we find
N = EA

kd
kd
du
= EAkd C1 sinh
x + C2 cosh
x

dx
EA
EA

(3.151)

Equations (3.150) can be also written in the explicitly exponential form:

kd

k qA
u = C1 exp
x + C2 exp d x +
EA

EA kd

kd

k
N = EAkd C1 exp
x C2 exp d x

EA
EA

(3.152)

(3.153)

Equations (3.150) and (3.151) show that both the axial displacement of and the
normal force in the rod on elastic foundation vary along the rod exponentially, even in
the case of a constant external force. To grasp other peculiarities of the static response
of the rod on elastic foundation to an axial force, we consider two examples, which
are discussed below.
Example A. Consider a fixed-free cylindrical and homogeneous rod of the length L
on elastic foundation as shown in Figure 3.18(a). Assume that this rod is subject to a
distributed axial force with the constant density per unit length q1 .
x

(a)

u=0

q1

N =0

(b)

N
(c)

Figure 3.18. (a) Fixed-free rod on elastic foundation under uniform axial loading; (b) normal force; (c)
axial deflection

101

The boundary conditions in the case at hand are given as

u ( 0) = 0

(3.154)

N ( L) = 0

(3.155)

Since the rod is cylindrical and homogeneous, the elastic foundation is uniform and
the external force is constant, the simplified expressions can be applied for the normal
force and axial displacement, Eqs.(3.150) and (3.151).
Substituting these expressions into the boundary conditions, we obtain the
following system of algebraic equations:
0 = C1 +

k
kd
0 = C1 sinh d L + C2 cosh
L
EA
EA

q1
,
kd

which yields the unknown constants C1 and C2 :

C1 =

q1
,
kd

C2 =

kd
q1
tanh
L
EA
kd

Consequently, the normal force and the axial displacement are given as

kd
k
kd
x + tanh d L sinh
x
1 cosh
EA
EA
EA

k
kd
kd
EA
N = q1
L cosh
x
sinh d x tanh
EA
EA
EA
kd

u=

q1
kd

(3.156)

(3.157)

Figures 3.18(b,c) visualize these expressions for L2 kd EA = 4 . The latter ratio

represents the relation between the total stiffness Lkd of the elastic foundation and
that EA L of the rod. Let us note, that Figures 3.18 (b,c) are plotted in the case that
the elastic foundation is four times stiffer than the rod. In practice, the situation is
normally opposite and the rod (for example, a concrete pile) is much stiffer than the
foundation (soil). In the latter case, the graphs in Figures 3.18 (b,c) will be similar to
those shown in Figures 3.11 (b,c).
Example B. Consider a vertically positioned cylindrical homogeneous rod of the
length L , the lower part of which (of the length L 2 ) is subject to the elastic
foundation as shown in Figure 3.19(a). Assume that this rod is subject to the gravity
force, and to a weight P = Mg (a concentrated load) on its top, x = 0 . The lower end
of the rod, x = L , is free. This model can be applied to study the axial stresses and
deflections in a tall water-tower, whose foundation is in the ground.

102

N = Mg

L 2

gA

L 2

kd

N =0

Figure 3.16. Vertical column (water tower) whose lower part is in the ground

The boundary conditions in the case at hand read

NU ( 0 ) = P

(3.158)

NL ( L) = 0

(3.159)

where the subscripts U and L mark the upper- and lower-part related quantities.
Since at x = L 2 the foundation starts acting on the rod abruptly, we have to
formulate the interface conditions at this point. In accordance with Eq.(3.117), these
conditions read
uU ( L 2 ) = uL ( L 2 )

(3.160)

NU ( L 2 ) = N L ( L 2 )

(3.161)

The normal force in and the displacement of the upper-part of the rod, since it has
no foundation, are given by Eqs.(3.115) and (3.116), whereas those of the supported
lower-part are given by Eqs.(3.152) and (3.153). Thus, these quantities read
NU = q1 x + C1
uU =

(3.162)

1 1

2
q1 x + C1 x + C2
EA 2

kd

N L = EAkd C3 exp
x C4 exp

EA

kd

k
u L = C3 exp
x + C4 exp d x +
EA

EA

(3.163)
kd
x
EA

(3.164)

q1
kd

(3.165)

where q1 = gA is the force per unit length induced by gravity.

Let us first consider the normal force. Substituting Eq.(3.162) into the boundary
condition at the top, Eq.(3.158), we immediately find

103

C1 = P
and
NU = q1 x P

(3.166)

Now we will make use of the dynamic interface condition, Eq.(3.161), and the
boundary condition at the lower end, Eq.(3.159). Substitution of Eqs.(3.162) and
(3.164) into these two conditions yields the following system of two linear algebraic
equations with respect to C3 and C4 :

k L

k L
L
P = EAkd C3 exp d
C4 exp d

EA 2

2
EA 2

kd

k
C3 exp
L C4 exp d L = 0
EA

EA

q1

which can be readily solve to give

C3 =

L Q + 2P
3

exp
K d 1 exp K d
EA 2 K d
2

L Q + 2P
C4 =
exp
EA 2 K d

Kd

) (1 exp (

Kd

))

))

(3.167)

where K d = kd L2 EA is a dimensionless stiffness representing the ratio of the

stiffness per unit length kd of the foundation and the axial stiffness per unit length
EA L2 of the rod, Q = q1 L is the total weight of the rod.
The remaining unknown constant C2 should be found from the kinematic interface
condition, Eq.(3.160). Substituting in this equation, Eqs.(3.163) and (3.165), we
obtain
kd L

kd L q1
1 1 2
L

+ C4 exp
+
q1 L + C1 + C2 = C3 exp
2
EA 8
EA 2 kd

EA 2

This equation can be solved by employing Eq.(3.167) to give the following expression
for C2

1 + exp
1
C2 =
4 K d P + 8Q + K d Q 4 K d ( Q + 2 P )
8K d
1 exp

(
(

)
)

Kd

Kd

(3.168)

which completes the problem solution.

The normal force and axial displacement are visualized in Figure 3.17 using
P Q = 1 and K d = 1/ 4 . These values of the parameters imply that the weight of the

104

column is equal to the load at the top and that the elastic foundation is four times
softer than the rod.
0.0

17.0
16.8

N/Q

u=(EA/q1/L2)

-0.4
16.6
16.4

-0.8

16.2

-1.2
16.0

-1.6

15.8
0.00

0.25

0.50

0.75

1.00

0.00

0.25

u/L

0.50

0.75

1.00

u/L

Figure 3.17. Axial displacement of and normal force in the column

Figure 3.17 shows that for the chosen set of parameters the displacement varies along
the rod only slightly (less than 6%) relative to the average displacement of the rod
( 16.3 q1L ( EA) ) . The latter displacement is almost equal to the displacement, which

the rod would undergo if it were completely rigid, that is urigid = 2 ( P + q1 L ) kd L . In

the case under consideration, in which P Q = 1 and K d = 1/ 4 , this expression can be

rewritten as urigid = 16 q1 L ( EA ) . The normal force in the rod, as can be seen from
Figure 3.17, changes its behaviour abruptly at the middle cross-section of the rod,
where the elastic foundation starts to have effect.
3.5.3. Static Response of a Beam. The equations governing bending of a beam in
statics can be deduced from sub-section 3.2.5. These equations are summarized
below.
The equilibrium of transverse forces, as follows from Eq.(3.32):
V
= q1
x

(3.169)

The equilibrium of moments, as follows from Eq.(3.31):

V=

M
x

(3.170)

The relation between the bending moment and the transverse displacement,
Eq.(3.28):
M = EI

2w
x 2

The relation between the slope and transverse displacement:

(3.171)

105

w
=
x

(3.172)

The general solutions to Eqs.(3.169)-(3.171) can be found by direct integration to

give
x

V ( x ) = q1 ( x ) dx + V ( 0 )

(3.173)

M ( x ) = V ( x ) dx + M ( 0 )

(3.174)

0
x

()
() ()

M x
  ( 0 ) x + w ( 0 )
w ( x ) =
dxdx




E
x
I
x
0 0
x x

(3.175)

If the loading q1 is invariant along the beam, the general expressions for the shear
force V and the bending moment M , Eqs.(3.173) and (3.174), reduce to
V = q1 x + C1
1
M = q1 x 2 + C1 x + C2
2

(3.176)
(3.177)

If the bending stiffness is also (as well as q1 ) invariant along the beam, the general
expression for the transverse displacement, Eq.(3.175), becomes
w=

1 1
1
1

4
3
2
q1 x + C1 x + C2 x + C3 x + C4
6
2
EI 24

(3.178)

The boundary conditions, which are applicable to the bending of a beam, can be
deduced from Table 3.5 to form the following table:
Type of BC
Kinematic
(displacement)+
Kinematic
(slope)
Kinematic
(displacement)+
Dynamic Moment
(elastic)
Kinematic (slope)
+
Dynamic Force
(elastic)
Dynamic Moment
(external)
+
Dynamic Force
(external)

Left-end
w= A
w
= A
x

Right-end
w=B
w
= B
x

w= A
M = kr

w=B
A

w
x

M = kr
kr

w
= A
x
V = kw

A
k

M = M 0
V = P

w
x

w
= B
x
V = kw
M = M0

V =P
M0

kr

B
P

106

Type of BC
Dynamic Moment
(elastic)
+
Dynamic Force
(elastic)

Left-end
w
M = kr
x
V = kw

Right-end
M = kr

w
x

V = kw

Table 3.8. Boundary conditions for static bending of a beam

Table 3.8 covers all possible types of the boundary conditions, which may be imposed
on the beam in statics. If no elastic element is attached to an end of the beam, then
only four boundary conditions are applicable at this end. These conditions are
collected in the table below.
Type of BC
Pinned

Schematization

Boundary Conditions
w=M =0

Clamped

w = = 0

Free

M =V = 0

Sliding

=V = 0
Table 3.9. The simplest boundary conditions for a beam

The interface conditions at a cross-section x = c of the beam, where an abrupt

change of the bending stiffness takes place, are the same as given by Eq.(3.50), i.e.:
w ( x ) x = c + 0 = w ( x ) x =c 0 ,

w
w
,
=
x x =c + 0 x x =c 0

2 w
2 w
,
EI 2
= EI 2
x x = c + 0
x x =c 0

3 w
3 w
EI 3
= EI 3
x x = c + 0
x x =c 0

(3.179)

These conditions can also be written in the terms of the slope , the bending moment
M and the shear force V :
w ( c + 0) = w ( c 0) ,

M ( c + 0) = M ( c 0) ,

(c + 0) = ( c 0) ,
V (c + 0) = V ( c 0)

(3.180)

Equations (3.180) underline that the displacement, slope, bending moment and shear
force must be continuous at every interface of the beam provided that neither external
moment nor external transverse force are applied at this interface.
Using the above collected equations, any elastostatic problem of the bending of a
beam can be solved. In what follows we consider a number of examples to
demonstrate the methods of solution.

107

Example A. Consider a clamped-free (cantilevered) cylindrical and homogeneous

beam of the length L as shown in Figure 3.18 (a). Assume that this beam is subject to
a distributed transverse force of the constant density per unit length q1 .
q1

(a)

w = 0, w x = 0

M = 0, V = 0

(b)

q1 L
V
1 2
q1 L
2

(c)

M
(d)

1 q1 L4
8 EI

Figure 3.18. (a) Clamped-free beam under uniform transverse loading; (b) shear force; (c) bending
moment; (d) transverse displacement

The boundary conditions in the case at hand are given as

w (0) = 0
w
(0) = 0
x
M ( L) = 0

V ( L) = 0

(3.181)
(3.182)
(3.183)
(3.184)

Since the beam is cylindrical and homogeneous and the external force is constant, the
simplified expressions can be applied for the shear force, bending moment, and
transverse displacement, Eqs.(3.176)-(3.178).
Substitution of the simplified expressions for the shear force and bending moment,
Eqs. (3.176) and (3.177), into the boundary conditions at the right end of the beam,
Eqs.(3.183) and (3.184), gives the following system of equations:
0 = q1 L + C1
1
0 = q1 L2 + C1 L + C2
2

solving which we find

108

C1 = q1 L

(3.185)

1
C2 = q1 L2
2

This results in the following expressions for the shear force and bending moment:
V = q1 ( L x )

(3.186)

1
M = q1 ( x 2 + L ( L 2 x ) )
2

To find the displacement of the beam, we substitute the simplified expression for this
displacement, Eq.(3.178), into the boundary conditions at the left end, Eqs.(3.181) and
(3.182). This results in C3 = C4 = 0 and, consequently, in the following expression
for the beam displacement:
w=

q1 x 2 2
1 1
1
1
4
3
2

q
x
+
C
x
+
C
x
=
1
1
2
24 EI ( x + 2 L ( 3L 2 x ) )
6
2
EI 24

(3.187)

This displacement is shown in Figure 3.18 (d) together with the shear force and
bending moment, which are depicted in Figures 3.18 (b) and 3.18 (c), respectively.
Example B. Consider a cylindrical and homogeneous beam of the length L as shown
in Figure 3.19. Assume that this beam is clamped at the left end, and is subject to a
concentrated transverse force P and a concentrated moment M 0 at the right end.
x
P

M0
w = 0, w x = 0

M = M0, V = P
L

Figure 3.19. Beam clamped at the left end and subject to concentrated force and moment
at the right end

The boundary conditions in this case are given as

w (0) = 0
w
(0) = 0
x
M ( L) = M0

V ( L) = P

(3.188)
(3.189)
(3.190)
(3.191)

109

As in the previous example, to solve this problem, we may apply the simplified
expressions for the shear force, bending moment, and transverse displacement,
Eqs.(3.176)-(3.178).
Since no distributed force is applied to the beam, in accordance with Eqs. (3.176)
and (3.177), the shear force remains constant along the beam, whereas the bending
moment varies linearly. The constants C1 and C2 , which correspond to this variation,
can be easily found from the boundary conditions at the right end to give
C1 = P
C2 = M 0 PL

(3.192)

Consequently, the shear force and bending moment read:

V =P
M = P ( x L) + M 0

(3.193)

To find the displacement of the beam, we substitute the simplified expression for this
displacement, Eq.(3.178), into the boundary conditions at the left end, Eqs.(3.181) and
(3.182). This results in C3 = C4 = 0 and, consequently, in the following expression
for the beam displacement:
1 1
1
x2 1

3
2
w = C1 x + C2 x =
P ( x 3L ) + M 0

EI 6
2
2 EI 3

(3.194)

This displacement is visualised by the cubic parabola, which would reduce to the
ordinary parabola if the force P were absent.
Example C. Consider a pinned-pinned (simply supported) cylindrical and homogeneous beam of the length L as shown in Figure 3.20 (a). Assume that this beam is
subject to a distributed transverse force of the constant density per unit length q1 .
The boundary conditions in this case are given as
w (0) = 0

M (0) = 0
w( L) = 0

M ( L) = 0

(3.195)
(3.196)
(3.197)
(3.198)

To solve this problem we may apply the simplified expressions for the bending
moment and displacement, Eqs.(3.177) and (3.178). Substitution of these equations
into the boundary conditions results in the following system of four algebraic
equations with respect to the unknown constants C1 C4 :
0 = C4
0 = C2

110

1
1
1
q1 L4 + C1 L3 + C2 L2 + C3 L + C4
24
6
2
1 2
0 = q1 L + C1 L + C2
2
0=

resolving which we find

C1 =

1
q1 L,
2

C2 = 0,

C3 =

1
q1 L3 ,
24

C4 = 0

q1

(a)

(3.199)
x

w = 0, M = 0

w = 0, M = 0
L

q1 L 2

(b)

q1 L 2
V

L 2

(c)

1 2
q1 L
8

M
(d)

5 q1 L4
384 EI

Figure 3.20. (a) Pinned-pinned beam under uniform transverse loading; (b) shear force; (c) bending
moment; (d) transverse displacement

Thus, the shear force, bending moment and transverse displacement in the case at
hand are given as
1
q1 ( L 2 x )
2
1
M = q1 x ( L x )
2
qx
w= 1
x3 + L ( L2 2 x 2 )
24 EI

V=

(3.200)

These are shown in Figures 3.20 (b-d).

111

Example D. Consider a pinned-pinned cylindrical and homogeneous beam of the

length L as shown in Figure 3.21 (a). Assume that this rod is subject to a distributed
transverse force with the variable density per unit length q1 = q0 cos ( x L ) shown in
Figure 3.21(b).
The boundary conditions in this case are the same as in the previous one and are
given by Eqs.(3.195)-(3.198).
q1

(a)

w = 0, M = 0

w = 0, M = 0
L

q0
(b)

q0
q1

(c)

2
V

q0 L

(d)

M
(e)

w
Figure 3.21. (a) Pinned-pinned beam under uniform transverse loading; (b) external force; (c) shear
force; (d) bending moment; (e) transverse displacement

Since the external force is not constant along the beam, we should apply the
general expressions for the shear force, bending moment and displacement of the
beam, Eqs.(3.173)-(3.175). Substituting q1 = q0 cos ( x L ) into Eq. (3.173) and
evaluating the integral, we obtain
x

V ( x ) = q1 ( x ) dx + V ( 0 ) =
0

x
sin
+ V ( 0)

q0 L

(3.201)

112

where V ( 0 ) is unknown. Further, substituting this expression, into the general

expression for the bending moment, Eq.(3.174), we find

M ( x ) = V ( x ) dx + M ( 0 ) =
0

q0 L x x

sin
+ V ( 0 ) dx + M ( 0 )

0 L

q0 L2
x
= 2 cos
1 + V (0) x + M ( 0)
L

(3.202)

where V ( 0 ) and M ( 0 ) are unknown.

To determine the displacement of the beam we substitute Eq.(3.202) into the
general expression for this displacement, Eq.(3.175). This yields:
w( x) =

1
EI

x x

  ( 0) x + w ( 0) =
M ( x ) dxdx
0 0

q0 L2 x

 + M ( 0) dxdx
  ( 0) x + w ( 0)
V
x
cos
1
0

(
)

0 0

1 q L2 L x 1
= 0 2 sin x + V ( 0) x 2 + M ( 0) x dx ( 0) x + w ( 0)
EI 0
L 2

1
=
EI

x x

2
1 q0 L2 L2 L2
x x V ( 0) x M ( 0) x
+
2 2 2 cos +
EI
6
2
L 2
3

(3.203)

( 0) x + w ( 0)

where V ( 0 ) , ( 0 ) , M ( 0 ) and w ( 0 ) are unknown. To determine these unknowns,

we should employ the boundary conditions. This gives the following system of
algebraic equations:
0 = w (0)

0 = M ( 0)
1 q0 L4 2 1 V ( 0 ) L M ( 0 ) L
0 = 2 2 +
+
EI
2
6
2
3

0 = 2

q0 L2

( 0 ) L + w ( 0 )

+ V (0) L + M ( 0)

Solving these equations, we find

w ( 0 ) = 0, ( 0 ) =

qL
1 q0 L3 2 1
, M ( 0 ) = 0, V ( 0 ) = 2 0 2
2
2
6
EI

Thus, the expressions for the shear force, bending moment and transverse
displacement can be written as

113

q0 L 2
x
sin

2

L

q L
x
M ( x ) = 0 2 L cos
L + 2x

L

V ( x) =

w( x) =

(3.204)

x
1 q0 L L2
x
2
2
L L cos
2 x + ( 2 x 3Lx + L )
2
2
EI
L
6

These are shown in Figures 3.21 (c-e).

Example E. Consider a cylindrical and homogeneous beam of the length L , the left
end of which is clamped, whereas the right end is attached to a translational spring
with the stiffness k and to a rotational spring with a stiffness kr as shown in Figure
3.22 (a). Assume that this beam is subject to a distributed transverse force with the
constant density per unit length q1 .
The boundary conditions in this case, in correspondence with Table 3.8, read
w (0) = 0

(3.205)

w
(0) = 0
x

(3.206)

M ( L ) = kr

w
x

V ( L ) = kw

(3.207)
(3.208)

To solve this problem, the simplified expressions for the bending moment and
displacement, Eqs.(3.177) and (3.178), may be applied. Substitution of these
equations into the boundary conditions results in the following system of four
algebraic equations with respect to the unknown constants C1 C4 :
0 = C4
0 = C3
k 1
1
1

q1 L2 + C1 L + C2 = r q1 L3 + C1 L2 + C2 L + C3
2
2
EI 6

q1 L + C1 =

1
1
k 1

4
3
2
q1 L + C1 L + C2 L + C3 L + C4
EI 24
6
2

Solving this system, we obtain

C1 =

K ( 5 + K r ) + 24 (1 + K r )
1
q1 L
,
2
K ( 4 + K r ) + 12 (1 + K r )

C2 =

K ( 6 + K r ) + 24 ( 3 + K r )
1
q1 L2
,
12
K ( 4 + K r ) + 12 (1 + K r )

C3 = 0, C4 = 0

(3.209)

114

where K = kL ( EI ) is the ratio of the translational spring stiffness k to the overall

stiffness of the beam EI L , whereas K r = kr L ( EI ) is the ratio of the rotational
3

stiffness to the overall rotational stiffness of the beam EI L .

Thus, we have found the solution to the problem, which is given by the simplified
expressions for the shear force, bending moment and transverse displacement,
Eqs.(3.176)-(3.178) with the constants C1 C4 given by Eqs.(3.209).
q1

(a)

w = 0, w x = 0
M = kr w x , V = kw
L

(b)

11
q1 L
14
V
2 2
q1 L
7

(c)

M
(d)

1 q1 L4
7 EI

Figure 3.22. (a) Clamped-elastically supported beam under uniform transverse loading; (b) shear
force; (c) bending moment; (d) transverse displacement. The results are presented for K = 4, K r = 0

In Figure 3.22, a particular case is visualized of K r = 0 and K = 4 . In this case the

rotational spring is absent, whereas the stiffness of the translational spring is four
times greater than the overall stiffness of the beam. With these constants, the
expressions for the shear force, bending stiffness and transverse displacement read
q1
(11L 14 x )
14
q
M = 1 (11Lx 7 x 2 4 L2 )
14
q x2
w= 1
22 Lx 7 x 2 24 L2 )
(
168 EI

V=

(3.210)

115

Figure 3.23, presents another particular case, in which the translational spring is
absent, whereas the stiffness of the rotational spring is four times greater than the
overall rotational stiffness of the beam, i.e., K r = 4 and K = 0 . In this case the
expressions for the shear force, bending stiffness and transverse displacement read
V = q1 ( L x )
q1
30 Lx 15 x 2 11L2 )
(
30
q x2
w= 1
20 Lx 5 x 2 44 L2 )
(
120 EI
M=

(3.211)

q1

(a)

w = 0, w x = 0
M = kr w x , V = kw
L

(b)

11
q1 L
14
V
11 2
q1 L
30

(c)

M
(d)

11 q1 L4
30 EI

Figure 3.23. (a) Clamped-elastically supported beam under uniform transverse loading; (b) shear
force; (c) bending moment; (d) transverse displacement

If both the translational and the rotational springs are present, the beam pattern
depends significantly of the stiffness of these springs relative to the beam stiffness.
Example F. Consider a homogeneous beam of the length L , which consists of two
cylindrical parts, each of the length L 2 as shown in Figure 3.24(a). Both parts are
cylindrical but the cross-sectional moment of inertia of the left-hand part is I L ,
whereas the cross-sectional moment of inertia of the right-hand part is I R . Assume

116

that this beam is subject to a distributed axial force with the constant density per unit
length q1 , clamped at x = 0 and free at x = L .
The boundary conditions in the case at hand read
w (0) = 0

(3.212)

w
(0) = 0
x
M ( L) = 0

(3.213)
(3.214)

V ( L) = 0

(3.215)

Since the cross-section experiences the abrupt change at x = L 2 , we have to

formulate the interface conditions at this point. In accordance with Eq.(3.180), these
conditions read
w
w
( L 2 + 0) = ( L 2 0) ,
w
w
V ( L 2 + 0) = V ( L 2 0)

w ( L 2 + 0) = w ( L 2 0) ,
M ( L 2 + 0) = M ( L 2 0) ,

q1

(3.216)

(a)

M = 0, V = 0

w = 0, w x = 0
L 2

L 2

(b)

q1 L
V
1 2
q1 L
2

(c)

M
(d)

1 q1 L4
8 EI

Figure 3.24. (a) Clamped-free combined beam under uniform transverse loading; (b) shear force; (c)
bending moment; (d) transverse displacement

Since both parts of the beam are cylindrical and homogeneous, the simplified
expressions for the shear force, bending moment and transverse displacement,
Eqs.(3.176)-(3.178), can be applied to each part of the beam. Thus, introducing the

117

subscripts L and R for the left- and right-hand-part-related quantities, respectively,

we may write
VL = q1 x + C1
1
M L = q1 x 2 + C1 x + C2
2
1 1
1
1

4
3
2
wL =
q1 x + C1 x + C2 x + C3 x + C4
6
2
EI L 24

VR = q1 x + C5
1
M R = q1 x 2 + C5 x + C6
2
1 1
1
1

4
3
2
wR =
q1 x + C5 x + C6 x + C7 x + C8
6
2
EI R 24

(3.217)
(3.218)
(3.219)
(3.220)
(3.221)
(3.222)

Substituting the expression for the displacement of the left-hand-part, Eq.(3.219),

into the boundary conditions at the left end, Eqs.(3.212) and (3.213), we find that
C3 = C4 = 0

(3.223)

Further, subsisting the expressions for the bending moment and shear force in the
right-hand-part of the beam into the boundary conditions at the right end, Eqs.(3.214)
and (3.215), we readily obtain the following expressions for C5 and C6 :
C5 = q1 L,

1
C6 = q1 L2
2

(3.224)

To find the remaining unknown constants, the interface conditions, Eqs.(3.216),

should be employed. Substituting Eqs.(3.217)-(3.222) into these conditions and
making use of the determined constants, Eqs.(3.223) and (3.224), we obtain the
following system of four algebraic equations:
1 17
1
1
1
1 1

q1 L4 + C7 L + C8 =
q1 L4 + C1 L3 + C2 L2

I R 384
2
48
8
I L 384

1 7
1
1
1 1

3
3
2
q1 L + C7 = q1 L + C1 L + C2 L
I R 48
8
2
I L 48

1
1
1
q1 L2 = q1 L2 + C1 L + C2
8
8
2
1
1
q1 L = q1 L + C1
2
2

resolving which we find

1
7
I 1
11
I 1
C1 = q1 L, C2 = q1 L2 , C7 = q1 L3
q1 L4
, C8 =
2
48
I
384
I

(3.225)

118

where I = I L I R Thus, all unknowns in the expressions Eqs.(3.217)-(3.222) have

been found, therefore the problem has been solved.
For graphical representation of the results, let us assume that the moment of inertia
of the left-hand-part cross-section is twice as great as that of the right-hand-part, i.e.,
I = 2 . In this case Eqs.(3.225), reduce to
1
7
11
C1 = q1 L, C2 = q1 L2 , C7 = q1 L3 , C8 =
q1 L4
2
96
768

and the shear force, bending moment, and transverse displacement take the form
shown in Figures 3.24 (b-d). Note that the slope of the beam is continuous at the
interface.
3.5.4. Static Response of a Beam on Elastic Foundation. In this sub-section we
consider the beam on elastic foundation as shown in Figure 3.25.

kd

x
Figure 3.25. Beam on elastic foundation

The equilibrium equation, governing the balance of transverse forces acting on a

differential element of the beam, reads
d 2 d 2w
EI 2 = q1 kd w
dx 2
dx

(3.226)

which is the static reduction of Eq.(3.36).

All other relations, including the boundary and interface conditions are the same as
for the beam without the foundation.
If the bending stiffness is invariant along the beam, Eq.(3.226) reduces to
EI

d 4w
+ kd w = q1
dx 4

(3.227)

If, additionally, the external force q1 and the stiffness per unit length kd of the elastic
foundation are also coordinate-invariant, then the general solution of this equation can
be readily found. This solution is the superposition of a particular solution of
Eq.(3.227), which may be chosen as
wpart =

q1
kd

(3.228)

119

and the general solution of the homogeneous part of this equation. The latter solution
is to be sought in the form
4

wgen,hom = C n exp ( sn x )

(3.229)

n =1

Substituting Eq.(3.229) into Eq.(3.227) with the zero right-hand side, we obtain the
following characteristic equation
k
(3.230)
sn4 + d = 0
EI
Solving this characteristic equation, we find
s1 = (1 + i ) ,

s2 = (1 i ) ,

(3.231)

s3 = ( 1 + i ) , s4 = (1 + i )

where
4 4 =

kd
EI

(3.232)

Substituting the characteristic exponents, Eq.(3.231), into Eq.(3.229), we obtain

w = C1 exp ( (1 + i ) x ) + C2 exp ( (1 i ) x ) + C3 exp ( ( 1 + i ) x ) + C4 exp ( (1 + i ) x )

= exp ( x ) C1 exp ( i x ) + C2 exp ( i x ) + exp ( x ) C3 exp ( i x ) + C4 exp ( i x )
= exp ( x ) ( C1 cos ( x ) + C2 sin ( x ) ) + exp ( x ) ( C3 cos ( x ) + C4 sin ( x ) )

where

C1 = C1 + C 2 , C2 = i C1 C 2 , C3 = C 3 + C 4 , C4 = i C 3 C 4

Thus, the general solution of Eq.(3.227) under the condition that EI , kd , and q1 are
invariant along the beam, reads
w = exp ( x ) ( C1 cos ( x ) + C2 sin ( x ) ) + exp ( x ) ( C3 cos ( x ) + C4 sin ( x ) ) +

q1
kd
(3.233)

Consequently, employing Eqs.(3.170) and (3.171), the bending moment and shear
force can be found as

M = 2EI 2 exp ( x ) ( C1 sin ( x ) C2 cos ( x ) ) + exp ( x ) ( C3 sin ( x ) + C4 cos ( x ) )

(3.234)

120

V = 2 EI 3 exp ( x ) C1 ( sin ( x ) + cos ( x ) ) + C2 ( sin ( x ) cos ( x ) )

+ exp ( x ) C3 ( sin ( x ) cos ( x ) ) C4 ( sin ( x ) + cos ( x ) )

)}

(3.235)

These expressions show that variation of the transverse displacement of, and the
bending moment and shear force in the beam on elastic foundation are represented by
the superposition of multiplications of the sinusoidal and exponential functions.
Normally, the solution of elastostatic problems concerning the beam on elastic
foundation, requires application of a program like MAPLE, which is capable of
performing symbolic calculations. This is beyond the scope of this course. In what
follows, only one example is considered, which has a practical significance and can be
analyzed employing simple evaluations.
Example A. Consider an infinitely long beam on elastic foundation, which is loaded
by a concentrated force P as shown in Figure 3.26. This model proves useful for
studying the static deformation of rails under a train wheel.
x=0

x
wL

wR

Figure 3.26. Infinitely long beam on elastic foundation subject to concentrated load

The boundary conditions in this case should be formulated not at the ends of the
beam but at the infinite distance from the load. Obviously, the larger the distance from
the load, the smaller must be the static response of the beam. Therefore, the boundary
conditions at infinity may be written as
lim w ( x ) = 0

(3.236)

At the cross section x = 0 , where the load is applied, the interface conditions
should be formulated. These are also particular in this case, since the external load P
must be accounted for. It is not difficult, however, to understand, using Table 3.8 and
the interface conditions at a load-free cross-section, Eqs.(3.179), that the following
conditions must be satisfied at x = 0 :
w ( +0 ) = w ( 0 )

w ( +0 ) w ( 0 )
=
x
x
2
2
w ( +0 ) w ( 0 )
=
x 2
x 2
3 w ( +0 ) 3 w ( 0 )
EI

=P
3
x3
x

(3.237)

121

The first three conditions represent continuity of the displacement, slope and bending
moment. The last condition is the balance of vertical forces, which act on the crosssection of the beam loaded by the external force.
The solution procedure of the problem at hand may be simplified by making use of
the symmetry of the problem. Indeed, the problem is symmetric with respect to the
loading point. Therefore, the response of the beam must be symmetric with respect to
this point as well. To take advantage of this symmetry, we introduce
wR ( x ) = w ( x ) x > 0
wL ( x ) = w ( x ) x <0

(3.238)

where the subscripts L and R stand for the parts of the beam, which lay at the left
and at the right from the load, respectively. Employing these notations, the interface
conditions at x = 0 can be rewritten as
wR ( 0 ) = wL ( 0 )

wR ( 0 ) wL ( 0 )
=
x
x
2
2
wR ( 0 ) wL ( 0 )
=
x 2
x 2
3 wR ( 0 ) 3 wL ( 0 )

EI
=P
3
x 3
x

(3.239)
(3.240)
(3.241)
(3.242)

The requirement of the symmetry of the beam displacement with respect to the
loading point can be formulated mathematically as
wL ( x ) = wR ( x )

(3.243)

With this requirement, the first and the third boundary conditions, Eqs.(3.239)
and(3.241), are satisfied automatically, whereas the second and the fourth,
Eqs.(3.240) and (3.242), can be written in the terms of wR only to give
wR ( 0 )
=0
x
3 wR ( 0 ) 1
= P
EI
x3
2

(3.244)
(3.245)

Let us now derive wR . We start with the general expression, Eq.(3.233), which in
the case of q1 = 0 , reads
wR = exp ( x ) ( C1 cos ( x ) + C2 sin ( x ) ) + exp ( x ) ( C3 cos ( x ) + C4 sin ( x ) ) (3.246)

122

Obviously, to have the beam displacement vanish at x + , the coefficients C1 and

C2 in this equation must be zero. Thus, the expression for wR simplifies to
wR = exp ( x ) ( C3 cos ( x ) + C4 sin ( x ) )

(3.247)

Substituting Eq.(3.247) into the interface conditions, Eqs.(3.244) and (3.245), we

obtain
C3 + C4 = 0
1
EI ( C3 + C4 ) = P
2
3

C3 = C4 =

P
P
=
3
4 EI
kd

Thus, the deflection of the beam at the right of the load is given as
P
exp ( x ) ( cos ( x ) + sin ( x ) )
kd

wR =

(3.248)

Correspondingly, the bending moment and shear force at the right of the load read
P
exp ( x ) ( cos ( x ) sin ( x ) )
2
VR = P exp ( x ) cos ( x )
MR =

Using the symmetry requirement, the solution of the problem can be written such that
it becomes applicable to the entire beam. This may be accomplished by replacing x
by its absolute value in all expressions. This yields

w=

P
exp ( x ) cos ( x ) + sin ( x )
kd

M=

P
exp ( x ) cos ( x ) sin ( x )
2

(3.249)

V = P exp ( x ) cos ( x )

Note that the argument of cosinus is kept intact, since cosinus is an even function.
Equations (3.249) clearly show that the transverse displacement, bending moment
and shear force decay exponentially with the distance from the load. This decay,
however, is not monotonic but oscillatory along the beam.
The deflection of the beam along with the bending moment and shear force are
visualized in Figure 3.27 considering the vertical deflection of a railway track (two
rails) under an axle load of a train. To plot this figure, parameters of the UIC 60 rails
were employed, namely EUIC 60 =2.1 1011 N m-2, I UIC 60 = 3055 10-8 m4, and an average
effective stiffness of the soil against the railway track, k e ff = 108 N m-2 . From these
parameters we may retrieve the physical parameters of our model as

123

E = EUIC 60 = 2.1 1011 N m-2, I = 2 I UIC 60 = 6110 10-8 m4, k d = k eff = 108 N

(3.250)

The axle load was taken equal to 100 kN.

Displacement

1.2
1.0

40.0

0.8

30.0

M kN m

w mm

Bending Moment

50.0

0.6
0.4

20.0
10.0

0.2

0.0

0.0

-10.0

-0.2
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00

0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00

wm

wm

Shear Force

100.0
80.0

V kN

60.0
40.0
20.0
0.0
-20.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00

wm

Figure 3.27. Vertical displacement, bending moment and shear force for the UIC 60 railway track
under an axle loading

Figure 3.27 clearly shows that the rails are perceptibly deformed only within seven
meters from the load. Further away from the load, the rails are practically not
disturbed. This is due to the soil reaction, which we modelled with the help of elastic
foundation. Note that the softer the soil, the larger the response of the railway track
and the slower the decay of the deflection with the distance from the load.

124