Unit 2 Constrained Motion and the D’Alembert’s Principle

UNIT 2
CONSTRAINED MOTION
AND THE D'ALEMBERT’S
PRINCIPLE
Structure

2.1 Introduction 2.3 D'Alembert’s Principle

Expected Learning Outcomes Virtual Work
2.2 Constraints D'Alembert’s Principle
Examples of Constrained Motion 2.4 Generalized Coordinates
Classification of Constraints 2.5 Summary
2.6 Terminal Questions
2.7 Solutions and Answers

2.1 INTRODUCTION
In Unit 1, we revised the basic concepts of Newtonian Mechanics as applied to
the motion of a single particle, as well as to a many-particle system. Often the
study of mechanics is classified into three broad areas, the Newtonian
formulation, the Lagrangian formulation and the Hamiltonian Formulation. In
this Unit we study the foundations of the Lagrangian formulation of classical
mechanics, which is a whole new way of solving problems in mechanics.
You saw in Unit 1, that central to the application of Newton’s second law in
writing down the differential equation governing the motion of the system, is a
knowledge of the forces acting upon the system. The position of a particle is
completely defined by its position vector, which translates to three position
coordinates in the Cartesian coordinate system. For an n-particle system, you
get 3n second order differential equations and on solving these differential
equation you know everything that can be known about the motion. However
this method could become tedious in certain situations, for example when the
motion of the particle or system of particles in constrained in some way. We
discuss the notion of constraints in motion in Sec. 2.2. Constrained motion
forces us to revisit the notion of degrees of freedom and the number of
coordinates required to study the motion of a system. In Sec. 2.3 we first
introduce the idea of virtual work and the principle of virtual work to define
the condition of equilibrium of a system. The Euler-Lagrange equations of
motion, which we shall discuss in Unit 3, are derived from a fundamental result
of classical mechanics, the D'Alembert ’s principle, sometimes called the
Lagrange-D'Alembert principle, developed by D'Alembert in 1743 to solve 41
Block 1 The Lagrangian Formulation of Mechanics
dynamical problems. D'Alembert’s principle is a statement of the fundamental
laws of motion and we discuss this next in Sec. 2.3. We introduce the
concept of generalized coordinates in Sec. 2.4.
Expected Learning Outcomes
After studying this unit, you should be able to:
 identify and classify the constraints in a dynamical system;
 determine the number of degrees of freedom in a dynamical system;
 state and derive the principle of virtual work;
 apply the principle of virtual work to obtain the condition of static
equilibrium and acceleration of a constrained dynamical system;
 state and derive the D'Alembert’s principle;
 determine a set of generalized coordinates for a constrained dynamical
system; and
 apply D’Alembert’s principle to constrained dynamical systems.

2.2 CONSTRAINTS
We discussed the idea of degrees of freedom in Unit 1. Remember that for
the unconstrained motion of a particle, which essentially means that the
particles can move anywhere in (three-dimensional) space, the number of
(dynamical) degrees of freedom are 3 for a single particle and 3N for an N-
particle system respectively. So the position of the particle is described by 3
coordinates and for an N-particle system the configuration is completely
specified by 3N coordinates.
But what if the motion of the system is constrained in some way? Suppose
the particle or system of particles is restricted to move in a particular way,
such that the coordinates of the particle are required to satisfy certain given
conditions? The prescribed path or surface along which motion takes place is
called a constraint and the equations that describe the path or surface along
which motion is constrained to take place are called the constraint equations.
That would imply that the coordinates for each particle may not vary
independently of each other. There are several dynamical systems in which
this is true. There are offcourse certain forces that are responsible for creating
these constraints on the motion of the particle/particles in the system. These
forces are called the forces of constraint. For example:
 A box sliding down an inclined plane is constrained to slide along the
surface of the plane, and for that the surface of the plane exerts a contact
force(the normal force) on the box at all times.
 When a gas is confined in a container, the walls of the container exert a
force on the molecules to keep them confined within the container.
To use Newtonian mechanics we have to incorporate the forces of constraint
along with the other forces into our equations of motion and then find a
solution to the dynamical problem. In general, the equation of motion for a
single particle undergoing constrained motion can be written as:
42
Unit 2 Constrained Motion and the D’Alembert’s Principle
  
mr  F E  F C (2.1)
C
where F is the force of constraint. Some typical constraint forces that you are
familiar with are the contact forces, tension in an ideal string and the force of
static friction ( when the object is not slipping). Typically, unlike the external
forces, the constraint forces are not explicilty known to us and do not usually
follow a force law. Therefore Eq. (2.1) is a set of three scalar differential
equations for x(t), y(t), z(t) (in the Cartesian coordinate system though you
could also choose any other set of three coordinates) and six unknowns,
which are the position coordinates and the three components of the constraint
force.
Clearly you cannot arrive at a unique solution for the motion of the particle
from Eq. (2.1) unless you know the forces of constraint explicitly.
Let us consider the following examples of constrained motion, which you all
familiar with.
2.2.1 Examples of Constrained Motion
a) Motion of a Simple Pendulum
The position coordinates of the bob at any instant of time (Fig. 2.1) are
( x, y , z ) . Since the simple pendulum moves in a plane, we know that the
z coordinate of the bob is always zero:
z0 (2.2a)
Further the motion of the bob is also restricted by the condition that the
bob is tied to the support by string of constant length L0, hence:
z
2 2 2
x  y  L0 (2.2b) O
x
Eq. (2.2a and b) are the equations defining the constraints in the system.
 L0
x and y are no longer independent variables because: y

x  L02  y 2 (2.3a)
x
And any variation of x, implies a corresponding variation in y. To locate the
the position of the pendulum in the plane of its motion we need just one y
dynamical variable, which could be either x or y or even some other
Fig. 2.1: Simple
variable. For example, you can see that the x and y coordinates are
Pendulum.
related to the angle  as follows:
x  L0 sin  ; y  L0 cos  (2.3b)
So the position of the pendulum can also be uniquely described by just
the one angular variable . What is important is that, because of the two
constraints in the system (Eq. 2.2a and b ), the number of independent
dynamical variables in the system is reduced from three to one and we
say that the pendulum has just one dynamical degree of freedom.
In general constraints in the system reduce the number of dynamical
degrees of freedom of the system.
The force of constraint in the system is the tension in the string. You may
remember that whenever you have studied the motion of the simple
43
 Block 1 The Lagrangian Formulation of Mechanics
pendulum in mechanics, you would have derived the expression for the
tension in the string to solve the problem.
Let us now look at another simple example of constrained motion.
b) Atwood Machine
z
The Atwood machine, consists of two blocks of wood (Fig. 2.2) connected
O by an inextensible string. The coordinates of the two blocks are x1, y1, z1 
x and x2 , y 2 , z2  respectively. However since the motion of both blocks is
effectively one dimensional ( x1  x 2  z1  z 2  0 ) let us say that we
y2
describe the motion using two dynamical variables, y1 and y2. So the
y1
system has two degrees of freedom. However because both the blocks are
m2 connected by a fixed length of string, any increase in y1, means an equal
y decrease in y2. Effectively, assuming that the pulley has zero radius, we
m1 can write:

y1  y 2  l (2.4)
Fig. 2.2: Atwood’s
Machine. where l is the length of the string joining the two blocks. This is the
constraint equation. So now, we can describe the motion of the system
z using just one variable either y1 or y 2 (Fig. 2.2) and the system has just
one degree of freedom. This is another example of a constraint, which
reduces the number of degrees of freedom.
 
 ri  r j
ri Let us look at another example where the effect of constraints is striking.
 c) Rigid Body
rj
A rigid body (Fig 2.3) is a system of n particles in which each particle is at
y a fixed distance from every other particle. For every pair of particles, we
can write:
x  
Fig. 2.3: Rigid
ri  r j  c ij , i , j  1,2,...., n (2.5)
Body.
where c ij  c ji  0 . However an unconstrained rigid body, irrespective of
the number of particles in it, can have at most 6 degrees of freedom, and
z
not 3n, which is a result of the constraint contained Eq. (2.5). You will
understand how that is so when you study rigid body dynamics in your
O x
A
next semester course Classical Mechanics-II.
LA
yA There are several other dynamical systems in which the motion is constrained
yB in some way.
xA
A Example 2.1
B LB
Consider a system consisting of two simple pendulums connected in series
B (Fig. 2.4). A system of this kind is also called a double pendulum. Write
xB down the equations of constraint and determine the number of degrees of
freedom of the system.
y
Solution : To begin with, we have two bobs A and B, with coordinates
Fig. 2.4: Double x A, y A, zA  and xB , y B , zB  respectively, choosing the point of suspension O
Pendulum. to be the origin of the coordinate system. Since the pendulums move in a
plane:
44
Unit 2 Constrained Motion and the D’Alembert’s Principle
z A  zB  0 (i)

The motion of the particles is further constrained by the (inextensible) strings

of lengths L A and LB , connecting A to O and B to A, respectively, as shown
in the figure. The equations of constraint are:

x A 2  y A 2  LA 2 (ii)

And
x B  x A 2  y B  y A 2  LB 2 (iii)

We have four constraints in the system (Eqs. i, ii and iii). So we need to use
just two coordinates to locate A and B in the xy-plane. These could be either
the  x A , x B  or y A , y B  . As in the case of the simple pendulum, we could
also choose the angles  A and B , which are the angular displacement of the
masses relative to the vertical, to describe the motion(as shown in Fig. 2.4).

SAQ 1
Write down the equation(s) of constraint for the following systems:
a) A tiny particle of mass m sliding without friction down the surface of a
sphere of radius R under the action of gravity.
b) A simple pendulum of mass m, and length L0 , suspended from a point of
support of mass M which is sliding horizontally.
c) A bead sliding due to gravity on a fixed elliptical wire.
d) A spherical pendulum consisting of a mass m fixed to the end of light
inextensible string of length L0 where the mass is free to move in any
direction (as long as the string remains taut).

Sometimes, inspite of constraints, the dynamics of the system can easily be

determind using the Newtonian formulation, but not always. We now move
towards a more formal description of constraints within a dynamical system as
a first step towards a different formulation of mechanics in which this problem
can be addressed.
2.2.2 Classification of Constraints
The conditions defining the constraints in a system are typically described in
terms of relationships between the dynamical variables of the system and
time. Based on the nature of these relationships, the constraints may be
classified in the following different ways:
i) Holonomic and Nonholonomic Constraints
Consider an N-particle system described by the set of position vectors
  
r1, r2 ,......, rN . 3N coordinates are required to describe the unconstrained
motion of this system. Suppose now that there are constraints in the
system. If a constraint condition (there may be more than one, say k ) is
defined by an equation connecting the coordinates of the particles (and
possibly time), of the following form:
   45
f ( r1, r2 ,......, rN , t )  0,   1, 2, 3...., k (2.6)
Block 1 The Lagrangian Formulation of Mechanics
the constraint is said to be holonomic constraint.

If not, the constraint is said to be nonholonomic. For example, a

constraint expressed as an equation containing the position as well as the
velocity coordinates, of the kind:
     
f (r1, r2 ,......, rn , r1, r2 ,......, rN , t )  0 (2.7)

Or a constraint is expressed as an inequality:

  
f (r1, r2 ,......, rN , t )  0 (2.8)

Is usually classified as nonholonomic.

So the holonomic constraint:

 is expressed as an equality

 may contain time explicitly

 does not depend on the velocity of the particles.

Typically, holonomic constraints are used to reduce the number of dynamical

degrees of freedom in a system, as you will see when you work out problems.

Going back to the example of the simple pendulum, the constraint equations
Eq. (2.2b) clearly defines a holonomic constraint since it can be written as an
equality: x 2  y 2  L0 2  0 Similarly Eq. (2.5) for the rigid body is also an
example of a holonomic constraint .

On the other hand the constraint of SAQ 1(a) cannot be written as equality
connecting the coordinates and/or time, so it is a nonholonomic constraint.

Example 2.2

Write down the equations of constraint for the following systems and classify
them as holonomic/non-holonomic:

a) a dumbbell moving in space

b) the motion of a billiard ball of radius a on a billiard table , whose length and
breadth are l and b respectively.

c) a circular disc of radius a rolling without slipping on a flat plane.

( x1, y1, z1) Solution : a) Let the coordinates of the two masses be ( x1, y1, z1) and
m1 ( x2, y 2, z2 ) and the distance between them be d (Fig. 2.5). The constraint
d on the motion of the two masses can be written as

m2 d 2  ( x 22  x12 )  ( y 2  y 1 ) 2  ( z 2  z1 ) 2 (i)
( x 2 , y 2 , z2 )
This is a holonomic constraint since it is expressed as an equality and
Fig. 2.5: Dumb bell.
depends only on the coordinates.
b) Let the coordinate of the ball be ( x, y , z ) and the origin of the coordinate
system be at one corner of the table. Since the ball has to move on the
table, the equations of constraint are:
46
Unit 2 Constrained Motion and the D’Alembert’s Principle
z0 (ii)

a  x l a (iii)

a y ba (iv)

Eq. (ii) defines a holonomic constraint, but Eqs. (iii) and (iv) are
z
nonholonomic constraints.
c) Let the position coordinates of the centre (P) of the disc be ( x, y , z )
(Fig. 2.6). Since the disc is always vertical, the z coordinate of the centre is O
always at a fixed height which is the radius of the disc, so y
P
za (v) 
x 
v
This is a holonomic constraint. The point of contact between the disc and 
the plane has the coordinates (x,y). The angle of rotation of the disc is 
and the orientation of the disc is given by the angle  .  is the angular Fig. 2.6: Vertical disc
rolling on a flat plane.
speed of the disc about its axis. To specify the motion of the disc we need
the coordinates x, y ,  and  . The condition for rolling without sliding
implies that the point of contact of the disc with the plane must be
stationary. Therefore, we must have

v  a  ; x  v cos  ; y  v sin  (vi)

Where Eq. (vi) defines the constraints on the motion which depend on the
velocity, so these are non-holonomic constraints. From Eq. (vi) we can
also write the constraints in the following form:
dx  a cos  d ; dy  a sin  d (vii)

You may also have noticed that these constraints do not depend on time,
however that may not always be the case. This brings us to another important
classification of constraints.
ii) Scleronomic and Rheonomic Constraints
Constraints defined by conditions that do not depend explicitly on time are
called scleronomic constraints. Constraints defined by conditions that
depend explicitly on time are rheonomic constraints.
An example of an equation for a scleronomic costraint would be:
     
f (r1, r2 ,......, rn , r1, r2 ,......, rN )  0 (2.9)

A typical rheonomic costraint would be:

     
f (r1, r2 ,......, rn , r1, r2 ,......, rN , t )  0 (2.10)

A scleronomic constraint can be either holonomic or nonholonomic. A

scleronomic constraint which is also holonomic would satisfy an equation
of the kind:
  
f (r1, r2 ,......, rN )  0 (2.11)
47
Block 1 The Lagrangian Formulation of Mechanics
Let us consider an example on the classification of constraints as
holonomic/nonholonomic/scleronomic/rheonomic.

Example 2.3
O
x a) Consider a simple pendulum whose length is varying with time t as
L(t)
 L  L(t )  L0 t 2 (Fig. 2.7). Write down the equation of constraint and
y
classify the constraint.

x b) Two particles are connected by a massless rigid rod of length a. The

system is moving in a plane such that the centre of the rod in the direction
y of the rod (Fig. 2.8).

Fig. 2.7: Simple Solution : a) Here the constraint equation is x 2  y 2  L0 2 t 4 which is

Pendulum. holonomic and rheonomic.
b) The coordinates of the two masses (Fig. 2.8) are ( x1, y1, z1) and
( x2, y 2, z2 ) . Clearly, since the masses move in the xy-plane, one constraint
is that
z y z1  z2  0 (i)

V Another constraint is that the distance between the two masses is fixed
a ( x1, y1, z1) and equal to a, so
P
( x1  x 2 )2  ( y1  y 2 )2  a 2 (ii)
( x 2 , y 2 , z2 )
x  x  x 2 y1  y 2 
O The coordinates of the centre of the rod (P) are  1 , ,0  . The
 2 2 

Fig. 2.8: Example 2.3(b) velocity of the centre of the rod, V  VVˆ is along the direction of the rod,
where Vˆ is the unit vector along the direction of the rod:

1

Vˆ  ( x1  x 2 )iˆ  ( y 1  y 2 ) ˆj
a
 (iii)

The velocity is

  x  x   y  y 
V  1 2 iˆ 
  1  1 2 
2 ˆj  V ( x  x )iˆ  ( y  y ) ˆj
1 2  (iv)
 2   2  a

From Eq. (iv) we can write:

x1  x 2 V y1  y 2 V
 ( x1  x 2 ) and  ( y1  y 2 ) (v)
2 a 2 a

This gives us the following constraint on the motion:

x1  x 2 y1  y 2
 (vi)
x1  x 2 y1  y 2

The constraints defined in Eqs. (i) and (ii) are holonomic and scleronomic,
but the constraint defined in Eq. (vi) is nonholomonic and scleronomic.

48
Unit 2 Constrained Motion and the D’Alembert’s Principle
SAQ 2
Write down the equations for the constraints in the following systems and
classify them:
a) A rigid rod of length L is constrained to move inside a thin walled hollow
sphere of radius A, such that both its end touch the walls of the sphere at
all times.
b) Three point masses connected by rigid rods of length l.
c) An Atwood machine with a variable length of string.

The study of constrained motion in Newtonian mechanics is difficult for the

following two reasons:
 The coordinates specifying the motion of the particles can no longer all
vary independently of each other, because they are connected by the
equations of constraint.
 Knowing or determining the constaint forces becomes an integral part of
solving the dynamical problem using the Newtonian formulation. We may
not always know what exactly these constraint forces are and they may
not be described by a known force law. As we have explained at the
beginning of this Unit, there are more unknowns than equations available
to us, even assuming that you have the equation of constraint at your
disposal. You may offcourse ask how you have solved so many problems
of constrained motion using Newtonian mechanics. When you look back
you will see that you have used equations signifying additional conditions
like the torque equations or the conservation of energy, etc in addition to
the second law. So Newton’s second law by itself would not be sufficient to
arrive at a complete solution to the dynamical problem of constrained
motion.
For example in the problem of the pendulum, while we are interested in
knowing how the pendulum moves, yet in the Newtonian formulation we have
to calculate the constraint force, which is the tension in the string before we
get to determine the motion of the pendulum.
Most often, however, we are interested only in the description of
unconstrained motion of the system. The question now is, can we now
formulate a general framework for the description of motion where:
i) we do not need to, apriori, determine the forces of constraint in the system
as we have to in the Newtonian framework, and,
ii) we work with fewer variables, which describe the unconstrained motion of
the system
In the next section we see how the these two objectives may be achieved by
using the concept of virtual work, from which we can arrive at the
D'Alembert’s Principle.

2.3 D'ALEMBERT’S PRINCIPLE

You are familiar with the concept of the work done by force acting on a particle
as it moves along a path under the action of the force. In this section we
describe the concept of virtual work.
49
Block 1 The Lagrangian Formulation of Mechanics
2.3.1 Virtual Work
We first introduce the idea of virtual displacement. A virtual diplacement, is an
infinitesimal change in the configuration of a system resulting from any
The concept of virtual 
arbitrary infinitesimal change in the coordinates of the system denoted by r ,
displacement was first
introduced by Bernoulli to at any given instant of time. This displacement is consistent with all the
explain static equilibrium forces and constraints present in the system at that instant of time. The
in 1703. displacement is called virtual because it is purely an imaginary displacement.
It has the following properties:

i) the virtual displacement r takes place at a fixed instant of time, as

opposed to a real displacement dr which take place in a time interval dt.
Therefore the applied forces as well the forces of constraint do not
change during a virtual displacement. You can imagine the system is
frozen in its motion at an instant of time, say t, and then moved without
violating any of the constraints .
ii) it is infinitesimal.

iii) the time derivatives r do not change during a virtual displacement.
 
iv) there can be a virtual displacement ri corresponding to each variable ri
being used to describe the motion.

v) the virtual displacement obeys the constraints of motion.

Suppose now, that the system we are studying is in equilibrium. Then at  any

instant of time, the total force acting on each particle must be zero or Fi  0

where Fi is the force acting on the ith particle. Therefore we can write:
 
Fi .r i  0 (2.12)

We now define the virtual work done by force Fi for the virtual displacement

ri as the quantity:
 
W i  Fi .r i (2.13)

Note
 that the virtual work is defined in exactly the same way as the real work
( Fi .dri ) except that now we have a virtual displacement in place of a real
displacement. Given the condition expressed in Eq. (2.12), we can sum over
all i to write:
 

Wi  
Fi .ri  0 (2.14)
i i
 
Let us now write the force Fi as a sum of the applied force Fi A and the force of

constraint Fi C . Then Eq. (2.14) can be rewritten as:
   
 
Fi A .r i  Fi C .r i  0 (2.15)
i i

If we now restrict ourselves to systems where the net virtual work done
by the forces of constraint in the system is zero, that is:
 
Fi C .r i  0 (2.16)
50 i
Unit 2 Constrained Motion and the D’Alembert’s Principle
The condition for the equlibrium of the system can be rewritten as: 
  N
 Fi A.ri  0 (2.17)
i

That is to say that, for the static equilibrium of the system, the virtual work s
done by the applied forces is zero. This (Eq. 2.17) is called the Principle of
Virtual Work. It is stated as follows:
The necessary condition for the static equilibrium of a dynamical (a)
system, is that the virtual work done by all the applied forces on the
system is zero, provided that the virtual work done by all the constraint O
x
forces is zero.
  
In general the applied force Fi A  0 and neither is Fi A .r i . 
T 
r
Let us discuss in some detail about the condition defined in Eq. (2.17). In
some constrained systems that you study, the constraint force on each particle
is a normal force and hence perpendicular to the displacement. In that case y
the virtual work due to the constraint force for every individual particle would (b)
 
vanish separately( Fi C .ri  0 ). Let us see some examples:

i) For a block sliding down a frictionless inclined plane(Fig. 2.9a) the force of O
 x
constraint is the normal force N and the virtual displacement is along the
   
plane. Hence F C .s  N.s  0 .
 
ii) For a simple pendulum the force of constraint is the tension in the string T T
(Fig. 2.9b). But as you can see the virtual displacement is perpendicular to
     
T and so: F C .r  T .r  0 . T y 
y 2
In that case the virtual work due to the constraint force for every individual
 
particle would vanish separately( Fi C .r i  0 ). In some systems however the
constraint force may be along the direction of displacement and in that case 
y1
the net virtual work only may be zero. This is the case for the Atwood’s (c)
       
machine (Fig. 2.9c) where F C .r  T .y 1  T .y 2  0 because y 1  y 2 Fig. 2.9: a) Bock on an
inclined plane; b) simple
Let us now work out the condition of static equilibrium for two different pendulum; and c) Atwood’s
systems using the principle of virtual work. Remember this is only for systems machine.
where Eq. (2.16) is valid.
Example 2.4 O
x

a) A mass M attached to a string of length l suspended from the ceiling of a a
 
carriage in a train which has an acceleration a , deflects by an angle  
from the vertical. Using the principle of virtual work determine the  r
 T 
magnitude of a .

b) Determine the condition for static equilibrium in an Atwood’s machine. y


Mg
Solution : a) The motion is in the xy-plane, so let us write the virtual
displacement as (Fig. 2.10): Fig. 2.10


r  xiˆ  y ˆj (i)
51
 Block 1 The Lagrangian Formulation of Mechanics

The workdone by the force of constraint, which is the tension T in the

string is T .r  0 Suppose that under this virtual displacement the angle
 increases to    .Then
x  l cos   ; y  l sin   (ii)
O
x The virtual work done by the applied forces is


W   Fi A . r  Ma iˆ. l cos    iˆ  (Mg ˆj ).  l sin    ˆj 

T  Ma l cos   M g l sin  (iii)

 m2 By the principle of virtual work (Eq. 2.17), for static equilibrium W  0

y 
T y 2  a  g tan  (iv)
m1  b) Let us assume that in the virtual displacement the mass m1 moves down
 m2 g
y1 by y1 and the mass m2 moves up by y 2 , so (Fig. 2.11),
 
  y 1  y 1 ˆj ; y 2  y 1 ˆj (v)
m1g
The applied forces are m1g ˆj on the mass m1 and m2g jˆ on the mass m 2 .
Fig. 2.11: Static
equilibrium of the So the total virtual work done by the applied forces is:
Atwood’s machine.
W  m1g ˆj . y 1 ˆj  m 2 g ˆj . ( y 1 ˆj )

 m1g  y1  m2g y1  (m1  m 2 ) g y 1 (vi)

For static equilibrium, W  0, and hence we must have

m1  m2  0  m1  m2 (vii)
 
N This is the condition for static equilibrium of the Atwood’s machine.
dr2

dr
 The condition on the virtual work done by the constraint forces (Eq. 2.16) is
dr1
true for most constrained systems you study, in one form or the other.
However you may have noticed that in these systems the real work done by
the forces of constraint is also zero. So you may wonder why we need a
Fig. 2.12: Block sliding principle of virtual work, using a virtual displacement.
down a moving inclined
plane. It is because there are some situations where the real work done by the
constraint forces is not zero. Let us look at some such systems. This is usually
O
in systems where we have time varying constraints of the kind shown in
x
Figs. 2.12 and 2.13:
i) In Fig. 2.12 you see a block sliding down a moving frictionless plane. Now
  because the inclined plane is moving, in the time interval t to t + dt, the
T dr   
 actual displacement of the particle dr  dr1  dr 2 has a component
r 
normal to the surface, that is, in the direction of the force of constraint ( N ).
y In this situation, the real work done by the force of constraint is not zero.
But, for a virtual displacement, dt =0. Therefore, the virtual displacement
Fig. 2.13: Simple
of the particle is still along the plane and the virtual work done by the
pendulum with varying
length. force of constraint will be zero.

52
Unit 2 Constrained Motion and the D’Alembert’s Principle

ii) In Fig. 2.13 you see a simple pendulum with a length that varies with time.
  
Here the real work done by the constraint force T ( T .dr ) is not zero,
  
though the virtual work done by T ( T .r ) is zero.

SAQ 3
Calculate the virtual work for a small bead of mass m sliding under gravity on a
fixed frictionless elliptical wire.

Let us now study how the principle of virtual work provides us with an extra
set of equations, to add to the 3N equations of motion derived from Newton’s
second law for a constrained system. This happens only if the virtual
displacements are independent of each other.
2.3.2 D'Alembert’s Principle
Notice that while the principle of virtual work, solves a part of our problem in
that it does not contain the forces of constraint, but it is still only a condition of
static equilibrium. We have yet to arrive at the dynamics of the system. Let us
see how that can be done.
  
The equation of motion for the ith particle, Fi  p i , where Fi is the net force

acting on the ith particle and pi is its linear momentum, can also be written as :
 
Fi  p i  0 (2.18)

Eq. (2.18) can also be stated as follows:

Each particle of the system is in equlibrium under the action of the
 
following two forces acting on it: Fi and  p i .
  
While the force acting on each particle Fi  Fi A  Fi C , is the sum of the

applied and constraint forces,  p may be thought of an effective reverse
i
force, sometimes called the inertial force, which would bring the system to
equilibrium. From Eq. (2.18) we can also write:

 
  
Fi  p i .r i  0 (2.19)
i

We already know that the virtual work done by force Fi for the virtual
      
displacement r i is: W i  Fi .r i  Fi A .r i  Fi C .r i

As we have already studied in the previous section, the total virtual work is
     
W  
W i  
Fi A .r i  
Fi C .r i  
Fi A .r i (2.20)
i i i i
 
since  Fi C .r i   0 . Eq. (2.19) therefore reduces to:
i

 Fi A  p i . ri
  
0 (2.21)
i

This is called the D'Alembert’s principle. The physical significance of this

principle is that it tells us that the momentum of the particle is determined only
53
Block 1 The Lagrangian Formulation of Mechanics
   
by the Fi A and not by Fi C . Remember that the statement is not Fi A  p i  0 ,
which, in general, is not true for a constrained system. It is true only when
multipled by the virtual displacement. Notice also, that Eq. (2.21) is different
from the Principle of Virtual Work, because it includes the dynamics of the

system through the term p i . In particular,

i) The forces of constraint do not figure in the equation in any way, all you
need to know are the applied forces.

ii) The constraint relation enters the equation only through the term r i ,
through the relation between the coordinates.
iii) The only condition required to be satisfied by a dynamical system for the
D'Alembert’s principle to be applicable is that the virtual work done by all
the constraint forces vanishes (Eq. 2.16).
The D'Alembert’s principle is a formulation for determining the dynamics of the
system which does not require a knowledge of the constraints. So one has
solved a part of the problem. Yet Eq. (2.21) cannot really be applied to derive
the equatins of motion of a constrained system, because the virtual

displacements ri are not all independent. You know that in a constrained
system, there are equations of constraint which connect the some of the
 
variables ri . If r i were independent variables, we could have written:

 
 
Fi A  p i . r i  0 for i  1, N (2.22)

This would then be a complete description of the constrained dynamical

system, where we have 3N differential equations for 3N unknowns (unlike
Newton’s second law). However, to actually use Eq. (2.21) to determine the
motion of a system, one has to eliminate those degrees of freedom in the
system which have no dynamics associated with them. We do this using the
constraints in the system.

SAQ 4
Show that the D'Alembert’s principle is also a statement of the conservation of
energy, if the virtual displacements are replaced by real displacements.

In the next section we introduce the concept of generalized coordinates as the

next step towards arriving at a general formalism for the study of the motion of
constrained systems.

2.4 GENERALIZED COORDINATES

We have defined the degrees of freedom of a dynamical system in Sec. 1.1.
To repeat, it is the number of independent quantities required to completely
specify, in a unique way, the position of the particles of the system. In the
absence of any constraints, an N-particle system has 3N degrees of freedom
and the number of independent coordinates required to specify the
configuration of the system is 3N.
Note that the dynamical degrees of freedom of a system do not depend on the
choice of coordinates being used to describe the system. The only restriction
54 is that the number of independent coordinates should be equal to the number
Unit 2 Constrained Motion and the D’Alembert’s Principle
of degrees of freedom. So far, we have not really discussed what these
independent coordinates should be, that describe the configuration (position of
each particle) of the system uniquely. However just knowing the configuration
of the system at any instant of time, does not tell us its configuration at all
times. For that we need to solve the equation of motion derived from the
second law of motion, central to the Newtonian formulation of mechanics:
 
mr  F
There is no restriction in the choice of independent coordinates that can be
used to describe the motion of the system, but this differential equation is most
conveniently expressed in rectangular Cartesian coordinates. For example,
for single particle motion, using a rectangular Cartesian coordinate system,
this equation reduces to the following three equations for the coordinates
x(t ), y (t ), z(t ) :
mx(t )  Fx ; my(t )  Fy ; mz(t )  Fz (2.23)

In general, these are not independent equations and may be coupled through
the force depending on all three position coordinates and/or the components
of the velocity. Recollect, further, that in your earlier courses in mechanics
you have often solved problems in mechanics using non-Cartesian
coordinates because it was more convenient to do so given the nature of
force, etc: for example in central force motion.
The general approach while using non-Cartesian coordinates has always been
the following:
i) Write the transformation equations connecting the Cartesian (x,y,z)
coordinates and the non-Cartesian coordinates say (u1,u2,u3):
x (t )  x (u1(t ), u 2 (t ), u 3 (t )) (2.24a)
y (t )  y (u1(t ), u 2 (t ), u 3 (t )) (2.24b)
z(t )  z(u1(t ), u 2 (t ), u 3 (t )) (2.24c)
For example, to describe the motion of a particle in a plane, the position
can be specified either in rectangular Cartesian coordinates (x,y) or by
in plane polar coordinates(r,). In three dimensions the we could use
either rectangular Cartesian coordinates (x,y,z), or cylindrical
coordinates (ρ,ϕ,z) or spherical polar coordinates (r, ,ϕ). The
transformation equations connecting the Cartesian coordinates to the
spherical polar coordinates are:
x (t )  r sin  cos  (2.25a)
y (t )  r sin  sin  (2.25b)
z(t )  r cos  (2.25c)
ii) Obtain the equations of motion in the new coordinates. For that we first
determine the differential elements of the Cartesian coordinates
(dx, dy , dz ) in terms of the differential elements of the new coordinates
(du1, du 2 , du3 ) and then obtain  x(t ), y(t ), z(t ) in terms of
u1(t ),u2 (t ),u3 (t ).
 
iii) Next we rewrite mr  F in terms of the new cordinates using the
transformation equations and the expressions for  x(t ), y(t ), z(t ) in terms
of u1(t ),u2 (t ),u3 (t ) .
55
Block 1 The Lagrangian Formulation of Mechanics
Unlike Cartesian coordinates, these new coordinates do not all have the
dimension of length, although they can be used to specify position in a
unique way.
At this stage we introduce the term generalized coordinates. If a dynamical
system has n degrees of freedom, then any set of n independent coordinates
that completely define the position of each particle in the system are called its
generalized coordinates. The term generalized is used to emphasize that
these coordinates need not be the usual Cartesian coordinates which are
the position coordinates having units of length. The generalized coordinates
are not necessarily lengths or angles, but may be any quantity appropriate to
the description of the position of the system. The only restrictions on the
generalized coordinates are
 The generalized coordinates are independent of each other
 The number of generalized coordinates is equal in number to the degrees
of freedom of the system.
For an unconstrained N particle system, we could define a set of 3N
generalized coordinates q1, q2 ,.......q3N connected to the 3N Cartesian
coordinates.
In defining the At this point, for the sake of simplicity in notation, we write the Cartesian
Cartesian coordinates
we have shifted from
coordinates of the N-particle system as:
 x, y , z  to an  x1(t ), x2 (t ), x3 (t ),  x 4 (t ), x5 (t ), x 6 (t ),.....,  x3N  2 (t ), x3N 1(t ), x3N (t )
 x1, x2 , x3  notation in place of the familiar:
for simplicity. So the x1(t ), y 1(t ), z1(t ), x 2 (t ), y 2 (t ), z2 (t ),....., x N (t ), y N (t ), zN (t ) .
Cartesian coordinates
of particle 1 are So now we have 3N Cartersian coordinates for the N-particle system
 x1, x2 , x3  , that of { x i }, i  1,2....3N (read the margin remark) which are related to the 3N
particle 2 are generalized coordinates q1, q2 ,.......q3N as:
 x4 , x5 , x6  and so on.
x1  x1q1, q2 ,.......q3N , t 

x 2  x 2 q1, q 2 ,.......q3N , t 

x3  x3 q1, q 2 ,.......q3N , t  (2.26)


x3N  x3N q1, q2 ,.......q3N , t 

Notice that the number of generalized coordinated is equal to the number of

rectangular Cartesian coordinates in Eq. (2.26).
You have seen how the number of degrees of freedom reduces with the
presence of constraints in the dynamical system. This also reduces the
number of coordinates needed to completely specify the system. Let us see
how.
The condition for a set of 3N variables (say x1, x2,..., x3N ) to be independent
is the following:
3N
 c i dx i  0  c1  c 2  ...  c 3N  0 (2.27)
i 1

56
Unit 2 Constrained Motion and the D’Alembert’s Principle
In other words, it is not possible to write any one differential element as a
linear combination of the other differential elements. This is possible only
when there is no (constraint) relationship connecting these variables.
For example, for the simple pendulum, the constraint equation is
x 2  y 2  L2  dx  dy  0 or dx  dy (2.28)

So x and y are no longer independent variables, and the constraint equation

can be used to eliminate either one of the variables (x or y) so that we are left
with just one independent variable. The system therefore does not have two
degrees of freedom, but just one. The constraint therefore has the effect of
reducing the number of degrees of freedom in the system.
In general, suppose that in an N particle system, there are a number k of
constraints that can be expressed as equations of the form given by Eq.(2.2),
i.e, we consider a dynamical system which has k holonomic constraints.
We can use the constraint equations to eliminate k out of the 3N coordinates
we started with, so that we are left with just 3N  k coordinates, which are
now independent. Effectively, the system will then have 3N  k degrees of
freedom.
The next question is: how does one choose these 3N  k independent
coordinates? One obvious way is to pick a set of 3N  k coordinates from the
3N Cartesian coordinates we started with. In this case the 3N  k Cartesian
coordinates would be the generalized coordinates for the system ( fewer in
number than the original 3N Cartesian coordinates). However this choice of
coordinates is not binding. We can also choose any other set of 3N  k
coordinates q1, q 2 ,.......q 3N k , and each of the original 3N Cartesian
coordinates may be expressed in terms of these coordinates as follows:
x1  x1q1, q2 ,.......q3N  k , t 

x 2  x 2 q1, q 2 ,.......q3N  k , t 

 (2.29)
x3N  x3N q1, q2 ,.......q3N  k , t 

For the constrained system therefore, the minimum number of

independent coordinates (Cartesian or otherwise) needed to define
completely the configuration of the entire system, while taking
advantage of the constraints on the system are the generalized
coordinates. The generalized coordinates need not be the rectangular
Cartesian coordinates of the system.
In Sec 2.2.1 we have already explained that for a simple pendulum with a
string of length L0 , the system has two constraints z  0 and x 2  y 2  L0 2 ,
both of which as you can see are holonomic constraints. Therefore the simple
pendulum has just one (number of position coordinates of the corresponding
unconstrained system  number of holonomic constraints =3-2=1) dynamical
degree of freedom, and therefore we need just one generalized coordinate to
describe its motion. We could select either of the Cartesian coordinates x or y
as the generlized coordinate (not z, since z  0 ). In that case the
transformation equations (Eq. 2.29) would be: 57
 Block 1 The Lagrangian Formulation of Mechanics
2
x  x; y  L0  x 2 ; z  0 (2.30a)

The choice of generalized coordinates for a dynamical system is not unique.

We could also choose the angle  (Fig. 2.1) as the generalized coordinate, in
which case the transformation equations are:
x  L0 sin ; y  L0 cos ; z  0 (2.30b)

Now note that:

A O B i) The number of generalized coordinates for a constrained system is less
x than the number of independent coordinates used to define the position
a
vectors for each particle in the corresponding unconstrained system. Also
y2 in an N particle system, it will not be possible, in general to club the
y1 generalized coordinates in sets of three, with one such set for each
different particle.
m2
y ii) To solve for the dynamical variables of the system of particles, we now
set up equations of motion in the generalized coordinates and solve them.
m1
Then we write down the position vectors for each particle using the
(a) transformation equations of Eq. (2.29).

O Let us now try to write down the generalized coordinates for some of the
x
A mechanical systems you have studied in this chapter so far.
LA
yA Example 2.5
yB
xA Identify generalized coordinates for the following systems:
A
B LB a) An Atwood’s machine (Fig 2.14a) where the radius of the pulley is a and
the length of the string connecting the two masses is l.
B
xB b) A double pendulum as defined in Example 2.1 (Fig. 2.14b).
c) A block sliding down an inclined plane (Fig. 2.14c).
y
(b) Solution : a) There are 5 holonomic constraints:
y
x1  x 2  z1  z2  0

s and y1  y 2  a  l
 where  a is length of the string between the points A and B. So the system
a
has one degree of freedom (3 2  5=1) and hence we need just one

generalized coordinate which we can chose to be y1 or y 2 with the
transformation equations being y 2  l  a  y1 and y1  l  a  y 2.
b x
b) The four holonomic constraints defined in Eqs. (i), (ii) and (iii) of
z
(c) Example 2.1 indicate that there are just two dynamical degrees of freedom
(3 2  4=2) and hence there should be just two generalized coordinates.
Fig. 2.14: a) Atwood’s Let us choose these coordinates to be,  A and  B . Thus
machine; b) double
pendulum; c) block on x A  L A sin  A ; y A  L A cos  A ; z A  0
an inclined plane. and
xB  LA sin  A  LB sin B ;

y B  LA cos  A  LB cos B ; zB  0

58
Unit 2 Constrained Motion and the D’Alembert’s Principle
c) The block (coordinates x,y,z) follows a straight line path down the incline in
the xy-plane( z  0 ).
y
And also : tan  
bx
ay
Alternatively, we could write tan   . In either case, the system has
x
two holonomic constraints and hence the block has just one degree of
freedom and one generalized coordinate, which we can choose as s
(Fig. 2.14c), the distance of the block from the top of the incline. We get:
x  s cos  ; y  a  s sin  ; z  0

Remember that although the choice of generalized coordinates is not unique,

yet a certain set of generalized coordinates may sometimes make the
solution simpler.
You may like to write down the generalized coordinates for a few systems on
your own.

SAQ 5
Identify generalized coordinates for the following dynamical systems:
a) A simple pendulum of mass m suspended from a support of mass M which
is moving to the right .
b) A block of mass m sliding down a frictionless incline which in turn is sliding
on a smooth horizontal plane which has a mass M.
c) A spherical pendulum as defined in SAQ 1d.
A O B
x
Note that the entire discussion on generalized coordinates and effective
degrees of freedom is related to holonomic constraints only. y2

Once we introduce the concept of generalized coordinates, D'Alembert’s y1 T
principle can provide us with a complete solution of the dynamics of a
 m2
mechanical system. Let us see how, by solving two examples using 
T y
D'Alembert’s principle. y 2

Example 2.6 m1 m2 g ˆj

y1
Using D'Alembert’s Principle:

a) The Atwood’s machine (Example 2.4b). m1g ˆj

Fig: 2.15
b) Derive the equations of motion for the block and the plane for the system
of SAQ 5b

Solution : a) The system has just one degree of freedom. Let the
generalized coordinate be y1 which is the downward displacement of the

mass m1 (Fig. 2.15). The virtual displacement of the two blocks are y 1
and y 2 . Note that while the mass m1 moves down by y1 the mass

59
Block 1 The Lagrangian Formulation of Mechanics
 
m2 move up by y1 therefore y 1  y 1 jˆ  y 2 The applied forces are
 
F A  m1g ˆj acting an m1 and F A  m 2 g ˆj acting an m 2 .
1 2

 D'Alembert’s principle gives us:

     
(F1  p 1 ). y 1  (F2  p 2 ). y 2  0 (i)
 
where p 1  m1y1 ˆj and p 2  m 2 y1 ˆj (ii)
So we get: (m1g  m 2 g  m1y1  m 2 y1 ) y 1  0
which gives us:
 m  m2 
y1   1 g
 m1  m 2 

b) We have already described the generalized coordinates for this system in

SAQ 5(b). Since there are two generalized coordinates,

s and X, there are

two independent virtual displacements s and X (Fig. 2.16). The virtual
 
work done by the forces of constraint is N1.s  N 2 .X  0.

y

s N1
 
A s a

Fblock s
X   
X X
N2 
x
A
Fincline

Fig. 2.16: Block sliding down a moving incline.


The displacement s is an increase in s by s , at a fixed time and
involves only the motion of the block down the incline. The virtual

displacement of the block is s . The applied force acting on the block is

the gravitational force acting on the block, F A  mg ˆj and the inertial
block
  
force acting on the block is p block  ma where a is the acceleration of

the block. a is the vector sum of the acceleration of the block down the

incline, s , and the acceleration of the block along the flat plane on which
      
the incline is moving, X (Fig. 2.16). So a  s  X and p block  m s  X .  
From D'Alembert’s principle we get:
     
(F A  p block ). s    mg ˆj  m s  X  . s  0
block   
We get the following equation for the acceleration of the block down the
incline:
 cos   s  g sin   X
mg sin   ms  mX  cos 

Notice that for a stationary inclined plane  X

  0  , this reduces to

s  g sin 

We next consider the virtual displacement X in which X increases by X ,
60 and there is a displacement of the system comprising the block and the
Unit 2 Constrained Motion and the D’Alembert’s Principle
incline by X . The applied force is the gravitational force,
A
ˆ
Fblock  incline  (M  m )g j and the inertial force is
  
p block  incline  ms  (m  M ) X . D'Alembert’s Principle then gives us:
 
 mg ˆj  Mg ˆj  m s  X  
  MX  . X  0
   

So the the acceleration X  is

MX   m X
  m s cos   0  X    m s cos 
(M  m )

In the next Unit we will study how using the concept of generalized
coordinates in the D Álembert’s principle, leads us to the Lagrangian
formulation of mechanics.

2.5 SUMMARY
 Constraints

When the motion of a dynamical system is constrained in some way,

the particle or system of particles is restricted to move in a particular
way, such that the coordinates of the particle satisfy certain given
conditions.

The prescribed path or surface along which motion takes place is

called a constraint The equations that define the path or surface
along which the motion takes place are called the constraint
equations.

The forces that are responsible for creating these constraints on the
motion of the particle/particles in the system are called the forces of
constraint.

Constraints reduce the number of dynamical degrees of freedom of a

system.

 Classification of Constraints

i) Holonomic and Nonholonomic Constraints

For an N-particle system described by the set of position vectors

  
r1, r 2 ,......, rN , if the constraint is defined by an equation connecting
the coordinates of the particles (and possibly time), of the following
form:
  
f  (r1, r 2 ,......, r N , t )  0,   1, 2, 3....,3N  k (2.6)

the constraint is said to be holonomic constraint. Holonomic

constraints are independent of the velocities. If the constraint
relation is not holonomic, it is said to be nonholonomic. A 61
Block 1 The Lagrangian Formulation of Mechanics
nonholomic constraint may depend on the velocities of the particles
in the system or be expressed as an inequality.

ii) Scleronomic and Rheonomic Constraints

Constraints defined by conditions that do not depend explicitly on

time are called scleronomic constraints. Constraints defined by
conditions that depend explicitly on time are rheonomic constraints.

 Virtual Displacement

A virtual diplacement, denoted by r , is an infinitesimal change in the
configuration of a system resulting from any arbitrary infinitesimal
change in the coordinates of the system, at any given instant of time.
This displacement is consistent with all the forces and constraints
present in the system at that instant of time.

 Virtual Work

If Fi is the net force acting on the ith particle of the system, we define
 
the virtual work done by force Fi for the virtual displacement r i as
the quantity:
 
W i  Fi .r i
 
The total virtual work in the system is  W i   Fi .r i  0.
i i

In a constrained system the net force is the sum of the applied force
and the force of constraint.
The Principle of Virtual Work states that:
The necessary condition for the static equilibrium of a
dynamical system, is that the virtual work done by all the
applied forces on the system is zero, provided that the virtual
work done by all the constraint forces is zero:

 
 Fi A .r i 0 (2.17)
i

where Fi A is the applied force on the ith particle of the system.

 D'Alembert’s Principle
D'Alembert’s Principle states that for an N-particle system where the
virtual work done by all the constraint forces is zero we can write:

 Fi A  p i . ri
N   
0
i 1

where  p i may be thought of as an effective reverse force, also called
the inertial force.
 Generalized Coordinates
62
Unit 2 Constrained Motion and the D’Alembert’s Principle
If a dynamical system has n degrees of freedom, then any set of n
independent coordinates that completely define the position of each
particle in the system are called its generalized coordinates.
An N particle system with k holonomic constraints has 3N  k
degrees of freedom and hence 3N  k generalized coordinates.  

2.6 TERMINAL QUESTIONS Fig. 2.17: TQ 2.

y
1. Calculate the virtual work for a small block of mass m moving down an
frictionless inclined plane of mass M moving to the right on a smooth m
surface(Example 2.6). The inclined plane makes an angle  with the
horizontal. 
 m F
x
2. Two unequal masses m1 and m2 are constrained to move on two smooth
inclined planes and they are connected by an inextensible string of length Fig. 2.18: TQ 3.
L which passes over a fixed smooth pulley P (Fig. 2.17). Determine the
condition of static equilibrium.
3. Two frictionless blocks of mass m each are connected by a massless rigid
rod of length L. The system is constrained
 to move in the vertical plane
under the action of an applied force F as shown in Fig. 2.18. Calculate the
condition for static equilibrium.
4. Use D'Alembert’s Principle to write the equations of motion (i) a body of
mass m falling freely under gravity, (ii) a simple pendulum and (iii) a box of y
mass m on a stationary inclined plane.
5. Use D'Alembert’s Principle to write the equations of motion for the system
described in TQ2.
( x, y , z )

6. Use D'Alembert’s Principle to write the equation of motion moving for a r
bead of mass m moving on a frictionless vertical hoop of radius R under
x
the action of gravity.
R
7. Use D'Alembert’s Principle to write the equation of motion for a spherical z
pendulum of mass m with a variable length given by r  R(1  cos t ) .
Fig. 2.19: SAQ 1a.
2.7 SOLUTIONS AND ANSWERS
z
Self-Assessment Questions ( x1, y1, z1)
O x
1. a) The particle moves on the surface of the sphere till it falls off the L0
sphere. Taking the coordinates of the particle to be (x, y, z) where the
origin of the coordinate system is the centre of the sphere(Fig. 2.19),
the equation of constraint is: ( x2, y 2 , z2 )
 y
r  R or x 2  y 2  z 2  R 2 Fig. 2.20: Simple
pendulum with a
b) Let the x-axis lie along the direction of motion of the support moving support.
Fig. (2.20). The coordinates of moving support are ( x1, y1, z1) and that
of the bob of the pendulum are x2 , y 2 , z2  .The equations of constraint
are z1  0; y1  0; z2  0 and ( x 2  x1)2  y 22  L0 2 .

63
 Block 1 The Lagrangian Formulation of Mechanics
y c) The coordinate of the bead are x, y , z  and the wire is in the xy-plane
(Fig. 2.21). The equations of constraint are:
(x, y, z) x2 y2
b z  0 and   1.
x a2 b2
a
d) The coordinates of the bob are x, y , z  (Fig. 2.22). However, the motion
z pendulum as confined to the surface of a sphere of radius L0 , so the
Fig. 2.21: SAQ 1c. equation of constraint is:

y x 2  y 2  z 2  L0 2
( x, y , z )
2. a) Let us coordinates of the two ends of the rod be  x1, y 1, z1  and
x2 , y 2 , z2  (Fig. 2.23). Since both ends of the rod must always touch
x
the inner wall of the sphere, we must have:
L
z x12  y 12  z12  A 2 ; x 22  y 22  z 22  A 2 (i)
Fig. 2.22: SAQ 1d.
And since the length of the rod is L, we must have
y
( x 2  x1 ) 2  ( y 2  y 1 ) 2  ( z 2  z1 ) 2  L2 (ii)
( x 2 , y 2 , z2 )
All three constraints are holonomic and scleronomic.
A x b) The constraints are (Fig. 2.24):
L
z ( x1  x 2 ) 2  ( y 1  y 2 )2  ( z1  z 2 ) 2  l 2
( x1, y1, z1)
( x 2  x 3 ) 2  ( y 2  y 3 ) 2  ( z 2  z3 ) 2  l 2
Fig. 2.23: SAQ 2a.
( x 3  x1 ) 2  ( y 3  y 1 )2  ( z3  z1 ) 2  l 2
y
The constraints are holonomic and scleronomic.
( x3 , y 3 , z3 )
c) For an Atwood’s machine (Fig. 2.2) with variable length, say l  l (t ) , in
l addition to the constraints z1  z2  x1  x 2  0 which are holonomic
x
( x 2 , y 2 , z2 ) and scleronomic, we also have the constraint y 1  y 2  l (t ) which is
z ( x1, y1, z1) holonomic and rheonomic, since the length depends explicitly on time.

Fig. 2.24: SAQ 2b. 3. The virtual displacement is r  x iˆ  y ˆj , (the wire is in the xy-plane,
y hence z  0 ) tangential to wire (Fig. 2.25). So the forces are the

 constraint F C which is force exerted by the wire to keep the bead on the
( x, y , z ) F C 
 wire, which is perpendicular to the virtual displacement r and the
 r  
r applied force, which is the weight of the bead : F A  mg ˆj . The virtual
b FA
x
a work done is
   
W  F A . r  F C . r   mg y
z
Fig. 2.25: SAQ 3. 4. From D'Alembert’s principle we have

 
  
Fi A  p i . ri  0
i
The virtual displacement can be any arbitrary displacement that is
consistent with the constraints in the system. Let us say that it is a real
infinitesimal displacement. So,

64
Unit 2 Constrained Motion and the D’Alembert’s Principle

 
  
Fi A  m i ri . dri  0 (i)
i
We can simplify the second term as follows:
  d 1   dT i
 mi ri .dri  dt  2 mi ri .ri  dt   dt
dt  dT
i i i

where T is the kinetic energy of the system. Let us now assume that the
applied force in the system is conservative and can be written as the
   
gradient of a scalar quantity and so Fi A . dri  
  i V . dri  dV .
i i
Eq. (i) is then
dV  dT  d (T  V )  0  T  V  Constant
5. a) As explained in SAQ 1(b) the coordinates of the support are ( x1, y 1, z1 ) z
( x1, y1, z1)
and that of the bob of the pendulum are ( x 2 , y 2 , z 2 ) and the motion of O x
the support is along the x-axis.The system has two degrees of freedom L0

since there are four constraints and so we need two generalized
coordinates to describe the motion of the system. We can choose
these to be the x-coordinate of the moving support, x1, and the angle , ( x2, y 2, z2 )
y
shown in Fig. 2.24. The transformation equations (Eq. 2.29) are:
Fig. 2.26: SAQ 5(a).
x1  x1; y 1  0; z1  0
and
x2  x1  L0 sin ; y 2  L0 cos ; z2  0
b)
y

s

a

X

x
b
z
Fig. 2.27: Block sliding down a moving incline.
We consider the block and inclined plane system of Example 2.5(c), in
which the plane is now moving to the right along the x-axis. The
generalized coordinate for the motion of the block on the incline is s,
which is the distance of the block from the top of the incline. Since
there is no movement of the plane in the y or z direction, we need just
one generalized coordinate to describe the motion of the plane along
the x-axis, say X as shown in Fig. 2.27.
c) The system has two degrees of freedom so can choose the
generalized coordinates to be the polar angle  and azimuthal angle  .

Terminal Questions
1. We calculate the virtual work done for the motion of the block down the

inclined plane, for the virtual displacement s . As explained in Example

2.6, the displacement s is an increase in s by s , at a fixed time
(Fig.2.16) . This involves only the motion of the block down the incline. The
65
Block 1 The Lagrangian Formulation of Mechanics
 
virtual displacement of the block is rblock  s . The applied force acting

on the block is the gravitational force acting on the block, F A  mg ˆj ,
block
 
the force of constraint is the normal force F C  N1 which is
block

perpendicular to s . The virtual work done is:
    
W  (F A  F C ). s  ( mg jˆ  N1 ). s  mg sin  s
block block

2. We use the principle of virtual work to solve this problem. The virtual
 
displacements are s1 for mass m1 and s 2 for mass m2 (Fig. 2.28).The
constraint on the motion is s1  s 2  L, so the system has just one degree
of freedom and we choose the generalized coordinate to be either s1 or
s2 . y P

 
N1 N2

s1 s s2 
 1 
F1 s2
F2
 
x
Fig. 2.28: TQ 2.

The forces are the gravitational force F1  m1g jˆ and the normal reaction
 
force N1 acting on the mass m1 and the gravitational force F2  m2 g ˆj

and the normal reaction force N 2 are acting on the mass m2 .
   
Work done by forces of constraint: N1 .s1  N 2 . s 2  0 .
Therefore, using the principle of virtual work we can write:
  
F1. s1  F2. s2  0  m1g sin  s1  m2 g sin  s 2  0 (i)
Since s1  s 2  L  s1  s 2 , Eq. (i) reduces to

m1g sin   m2 g sin 

which is the required condition for static equilibrium.
3. y


 N1
r1  
F1 N2 
r2 
 F
 x
F2

Fig. 2.29: TQ 4.
 
The virtual displacements are r1  yˆj and r2  xiˆ as shown in
Fig. 2.29.
 The work done by the constraint forces is
  
zero (N1 .r1  N 2 . r2  0) , so using the principle of virtual work we get :
     
F1. r1  F2 . r2  F . r2  0
 
where F1  mg jˆ , F2  mg ˆj . So,
mg y  0  F x  0  mg  y  F  x z
66


Unit 2 Constrained Motion and the D’Alembert’s Principle

Since the length of the rod is fixed, we have x 2  y 2  L2 , which gives us: z

xx  yy  0  xx   yy


y y z F
Also:  tan  so   cot  1
x x
y
So we get F  mg cot 
x
4. i) The motion is one-dimensional and the  generalized coordinate is Fig. 2.30: TQ 4(i).
z (Fig. 2.30). The applied is force  F1  mg kˆ and the inertial force is

p  mz kˆ .

The virtual displacement is z  zkˆ .Using D'Alembert’s principle we
write:
O
 
   x
F1  p .z  0  mg  mz z  0 L0

Which give us the familiar equation of motion z  g. 

ii) The generalized coordinate is , the virtual displacement in the r
 F1
direction of motion is r  L0ˆ where ̂ is the unit vector in the
 y
direction of increasing  (Fig. 2.31). The applied force is F1  mg ˆj and
 Fig. 2.31: TQ 4(ii).
the inertial force is p  m L  . From D'Alembert’s principle we get:
0
   
F1. r  p.r  0  ( mg L0 sin   mL0 2)   0

g
The equation of motion is:    sin .
L0

iii) For a block on a fixed inclined plane (Example 2.5c) the generalized
coordinate is s, the applied force is F1  mg ˆj , where m is the mass of
 
the block and the inertial force is p  ms . The virtual displacement is

s . With D'Alembert’s principle we get
   
F1.s  p.s  0  mg sin  s  mss  0

The equation of motion is s  g sin  .

5. Continuing from the solution of TQ 2 (Fig. 2.28), we can write the inertial
   
forces on the two masses m1 and m2 as p1  m1s1 and p 2  m2 s2
respectively. From D'Alembert’s Principle we have:
       
F1.s1  F2 . s2  p1.s1  p 2 .s2  0
 
 m1 g sin  s1  m2g sin  s2  m1s1 s1  m2s2  s2  0

However, since s1  s 2  L , s2  s1 and s1  s2 . Using s1 as the

generalized coordinate, the equation of motion for s1 is

m g sin   m 2 g sin 
s1  1
m1  m2
6. Let the coordinates of the bead be ( x, y , z ) . The constraints on the motion
of the bead are:
67
 Block 1 The Lagrangian Formulation of Mechanics
i) z  0, since the loop is fixed vertically in the xy-plane.

ii) x 2  y 2  R 2

Thus there is just one generalized coordinate, which we can take as the
z angle  shown in Fig. 2.32.The applied force is F1  Mg iˆ , the inertial
 
force is p  m R ˆ and the virtual displacement is s  R   ˆ , where ̂ is
the unit vector in the direction of increasing  .

s Therefore, using D'Alembert’s principle we get,
 F1    
x F1.s  p.  s  0  (mg R sin   m R 2)    0
R g
And the equation of motion is:   sin .
R
z
7. As discussed in SAQ 5c, the generalized coordinates for a spherical
Fig. 2.32: TQ 6. pendulum of fixed length are  and  . In this question the length is
variable, so we have three generalised coordinates r , , , which are the
spherical polar coordinates. In spherical polar coordinates the acceleration
of the system is

r  ( r  r 2  r 2 sin 2 )rˆ  (r  2r  r  2 sin  cos ) ˆ

 ( r sin   2r  sin   2r  cos ) ˆ


The virtual displacement is: r  r  ˆ  r sin   ˆ ( for a virtual
displacement , dt is zero and hence r  0 ).

The applied force is F  mg cos rˆ  mg sin  ˆ

From D'Alembert’s principle

   
F .  r  m r. r  0

 ( mg cos  rˆ  mg sin  ˆ ).( r   ˆ  r sin    ˆ )

 m (r  r  2  r 2 sin 2 ) rˆ . (r   ˆ  r sin   ) ˆ

 m (r  2r   r  2 sin  cos )ˆ .( r ˆ  r sin  ˆ )

 m (r  sin   2r  sin   2r   cos ) ˆ . (r   ˆ  r sin    ˆ )  0

On simplifying these equations we get the following equations of motion for

 and  :

r   2r   r  2 sin  cos   g sin 

and r sin    2r  sin   2r cos     0

The equation of motion for r is: r   R  sin t  r  R2 cos t

68