September 3, 2003 16:35:04 P.

Gutierrez

Physics 5153 Classical Mechanics

Principle of Virtual Work
1 Introduction
The first variational principle we encounter in mechanics is the principle of virtual work. It estab-
lishes the equilibrium condition of a mechanical system, and is fundamental for the later develop-
ment of analytical mechanics (Lagrangian and Hamiltonian methods).
The concept of virtual work is centered on the idea of calculating the amount of work done on
a system of particles through a virtual displacement. We will start by defining what we mean by
a virtual displacement, then discuss virtual work, state the principle of virtual work, and consider
an example of its use.

1.1 Virtual Displacement

To define what we mean by a virtual displacement, let’s consider a system composed of N particles,
possibly subject to some set of constraints, defined by 3N Cartesian coordinates (x i ) relative to
an inertial frame. Let’s assume that at some instant of time the system undergoes infinitesimal
displacements that are virtual in the sense that they occur without the passage of time (instanta-
neous), do not necessarily conform to the constraints. This change, δ~xi , in the configuration of the
system is known as a virtual displacement.
In the usual case, a virtual displacement conforms to the instantaneous constraints, that is,
moving constraints are assumed stopped during the displacement. For example, consider a system
subject to n holonomic constraints

fi (x1 , . . . , x3N , t) = 0 (1)

A total derivative corresponds to a infinitesimal displacement of the system and is give by

X ∂fi ∂fi
dfi = dxj + dt = 0 (2)
∂xi ∂t
j

notice that this gives both a spacial and a temporal displacement. In the case of a virtual dis-
placement, we assume that the temporal displacement is zero, therefore the constraint changes
by
X ∂fi
δfi = δxj = 0 (3)
∂xi
j

It is important to note the difference, the displacement occurs in zero time.

One question that may be asked, are there any conditions under which a real and virtual
displacement are the same? The answer can be seen by comparing Eqs. 2 and 3. If the constraint
equation is scleronomic, a virtual displacement is the same as a real displacement. Therefore, in
the general case virtual and real displacements are not the same, but in the scleronomic case they
are.

Principle of Virtual Work-1 lect 03.tex

September 3, 2003 16:35:04 P. Gutierrez

Before concluding our discussion of virtual displacements, let’s consider the nonholonomic case
where the constraint is given in terms of derivatives. Assume n constraint equations on a system
of 3N degrees of freedom X
aji dxi + ajt dt = 0 (4)
i

where j corresponds to the j thconstraint. Based on our definition of a virtual displacement con-
sistent with the constraints, a virtual displacement for nonholonomic constraints is given by
X
aji δxi = 0 (5)
i

This equation will become important later when we discuss calculating forces of constraint through
the Lagrange multiplier method.
So far we have considered virtual displacements in terms of Cartesian coordinates. Virtual
displacements in terms of generalized coordinates are also possible. Simply transform the Cartesian
constraint equations to the generalized coordinates. The form of the constraint equation is given
by X
aji dqi + ajt dt = 0 (6)
i
where replacing the a with
∂fj ∂fj
and (7)
∂qi ∂t
gives the holonomic constraint. For a virtual displacement, the constraint equation becomes
X
aji δqi = 0 (8)
i

Therefore the form is the same using any set of coordinates.

1.2 Virtual Work

Let’s again consider a system of N particles with 3N degrees of freedom whose configuration is
given by the Cartesian coordinates x1 . . . x3N . In addition, suppose that the forces F1 . . . F3N are
acting on the particles at the corresponding coordinates in a positive sense. The virtual work is
given by X X
δW = Fi δxi = ~ i · δ~ri
F (9)
i i
The second equality implies that the virtual work is independent of coordinates used. The equation
can be transformed as follows to any set of generalized coordinates
à !
X X ∂xi
δW = Fi δqj (10)
∂qj
j i

From this equation, we define the generalized force as

à !
X ∂xi X
Qj = Fi ⇒ δW = Qj δqj (11)
∂qj
i j

Principle of Virtual Work-2 lect 03.tex

September 3, 2003 16:35:04 P. Gutierrez

where we note that the generalized force does not have to have units of a force, just like the
generalized coordinates do not have to have units of a length. But, the product of generalized force
and coordinates has the units of work (energy).
In the expression for virtual work, the forces are assumed to remain constant throughout the vir-
tual displacement. This is true even if the forces vary drastically over a infinitesimal displacement.
A sudden change of force with position can occur in certain nonlinear systems.
Now assume that the system is subject to constraints. The force can be separated into applied
~ a and constraint forces F
forces F ~ c . The virtual work of the constraint forces in terms of generalized
coordinates is given by X
δWc = Qci δqi (12)
i

If the displacement is consistent with the constraint, the virtual work is zero since the force does
not act in the direction of the force
X
δWc = Qci δqi = 0 (13)
i

which is referred to as a workless constraint. These will be the type of constraint that we will deal
with most often. If the constraints are workless, then the total virtual work on the system is given
by the applied forces X
δW = Qai δqi (14)
i

1.3 Principle of Virtual Work

One of the important applications of the idea of virtual work arises in the study of static equilibrium
of mechanical system. Assume a scleronomic system of N particles. If the system is in static
equilibrium, then Newton’s laws for each of the N particles give
~a +F
F ~c = 0 (15)
i i

The virtual work for this system is given by

X
δW = ~ a · δ~ri + F
F ~ c · δ~ri = 0 (16)
i i
i

If we now assume that the constraints are workless, and the virtual displacements reversible (one
can replace δ~r with −δ~r), then the condition for static equilibrium is
X X
δW = ~ a · δ~ri = 0 ⇒ δW =
F Qai δqi = 0 (17)
i
i i

where the second equation is given using generalized coordinates. A very important point to note
here is that unlike the Newtonian approach, we do not need to know what the constraint forces
are. We only need to know the applied forces.
Now assume that the system is initially motionless, but not in equilibrium. Then one or more
of the particles has a net applied force on it, and in accord with Newton’s laws, it will start to
move in the direction of the force. Since any motion must be compatible with the constraints, the

Principle of Virtual Work-3 lect 03.tex

September 3, 2003 16:35:04 P. Gutierrez

virtual displacements can be chosen to be in the direction of the actual motion at each point. In
this case the virtual work is positive
X
δW = ~ a · δ~ri + F
F ~ c · δ~ri > 0 (18)
i i
i

Since the constraints are workless, the condition becomes

X
δW = ~ a · δ~ri > 0
F (19)
i
i

If the virtual displacements are reversed, then the virtual work is negative. None-the-less, if the
system is not in equilibrium, one can find a set of virtual displacements that will result in the
virtual work being nonzero.
These results can be summarized in the principle of virtual work: The necessary and sufficient
condition for the static equilibrium of an initially motionless scleronomic system that is subject to
workless constraints is that zero virtual work be done by the applied forces in moving through an
arbitrary virtual displacement satisfying the constraints.

1.4 Example
As a simple consider the system described in Fig. 1,where we want to determine the force F that
will keep the system in equilibrium. If we use the Newtonian approach, we require 3 equations to
solve the problem
X
Fx = 0 N1 − F = 0 (20)
X
Fy = 0 N2 − 2mg = 0
X
τ =0 mg` cos θ − N1 ` sin θ = 0

From this point it is fairly straight forward to solve the problem. One finds F = mg cot θ.
Using the principle of virtual work, we set up the equation as follows

mgδy − F δx = 0 (21)

N1 m

PSfrag replacements
mg `

θ m
F

N2

Figure 1: Two blocks on frictionless surfaces constrained by by a rod to move together.

Principle of Virtual Work-4 lect 03.tex

September 3, 2003 16:35:04 P. Gutierrez

with the constraint between x and y

) ( )
x = ` cos θ δx = −δθ` sin θ
⇒ ⇒ δx cos θ − δy sin θ = 0 (22)
y = ` sin θ δy = δθ` cos θ

since I have already assumed directions for δx and δy in Eq. 21, the sign here is dropped between
the middle and final equations. Combining the two equations

(mg cot θ − F )δx = 0 (23)

Since the displacement is arbitrary, and this equation must hold for all possible virtual displace-
ments, the quantity inside the parenthesis must be zero

F = mg cot θ (24)

the same as the Newtonian method.

The point of this example is not to show that one method is superior to the other, but that
differences in the two methods. In the Newtonian method, we required the constraint force and
a set of 3 equations to specify the problem. Using the method of virtual work we need only two
equations, one describing the work done and the second describing the constraints. We don’t need
the constraint forces.

Principle of Virtual Work-5 lect 03.tex