354 CHAPTER 9 DEFLECTIONS USING ENERGY METHODS

9.3 Principle of Virtual Work

The principle of virtual work was developed by John Bernoulli in 1717
and is sometimes referred to as the unit-load method. It provides a general
means of obtaining the displacement and slope at a specific point on a
structure, be it a beam, frame, or truss.
Before developing the principle of virtual work, it is necessary to
make some general statements regarding the principle of work and
energy, which was discussed in the previous section. If we take a
deformable structure of any shape or size and apply a series of external
loads P to it, it will cause internal loads u at points throughout the
structure. It is necessary that the external and internal loads be related by
the equations of equilibrium. As a consequence of these loadings,
external displacements ⌬ will occur at the P loads and internal
u
displacements d will occur at each point of internal load u. In general,
L these displacements do not have to be elastic, and they may not be
u related to the loads; however, the external and internal displacements
must be related by the compatibility of the displacements. In other words,
if the external displacements are known, the corresponding internal
A displacements are uniquely defined. In general, then, the principle of
P¿ ⫽ 1 work and energy states:
Apply virtual load P¿ ⫽ 1 ⌺P⌬ = ⌺ud
(a) Work of Work of (9–12)
External Loads Internal Loads

Based on this concept, the principle of virtual work will now be

P2 developed. To do this, we will consider the structure (or body) to be of
arbitrary shape as shown in Fig. 9–6b.* Suppose it is necessary to
P1 determine the displacement ⌬ of point A on the body caused by the “real
loads” P1, P2, and P3. It is to be understood that these loads cause no
dL L movement of the supports; in general, however, they can strain the
material beyond the elastic limit. Since no external load acts on the body
at A and in the direction of ⌬, the displacement ⌬ can be determined by
9 A first placing on the body a “virtual” load such that this force P⬘ acts in
the same direction as ⌬, Fig. 9–6a. For convenience, which will be
⌬ apparent later, we will choose P⬘ to have a “unit” magnitude, that is,
P3 P⬘ = 1. The term “virtual” is used to describe the load, since it is
Apply real loads P1, P2, P3
imaginary and does not actually exist as part of the real loading. The unit
(b) load (P⬘) does, however, create an internal virtual load u in a representative
element or fiber of the body, as shown in Fig. 9–6a. Here it is required
Fig. 9–6
that P⬘ and u be related by the equations of equilibrium.†

*This arbitrary shape will later represent a specific truss, beam, or frame.
†Although these loads will cause virtual displacements, we will not be concerned with their
magnitudes.
 9.3 PRINCIPLE OF VIRTUAL WORK 355

Once the virtual loadings are applied, then the body is subjected to the
real loads P1, P2, and P3, Fig. 9–6b. Point A will be displaced an amount ⌬,
causing the element to deform an amount dL. As a result, the external
virtual force P⬘ and internal virtual load u “ride along” by ⌬ and dL,
respectively, and therefore perform external virtual work of 1 # ⌬ on the
body and internal virtual work of u # dL on the element. Realizing that the
external virtual work is equal to the internal virtual work done on all
the elements of the body, we can write the virtual-work equation as

virtual loadings
1#⌬ = ⌺u # dL (9–13)
real displacements

where
P⬘ = 1 = external virtual unit load acting in the direction of ⌬.
u = internal virtual load acting on the element in the direction of dL.
⌬ = external displacement caused by the real loads.
dL = internal deformation of the element caused by the real loads.
By choosing P⬘ = 1, it can be seen that the solution for ⌬ follows directly,
since (1)⌬ = ⌺u dL.
In a similar manner, if the rotational displacement or slope of the tangent
at a point on a structure is to be determined, a virtual couple moment M⬘
having a “unit” magnitude is applied at the point. As a consequence, this
couple moment causes a virtual load uU in one of the elements of the body.
Assuming that the real loads deform the element an amount dL, the
rotation u can be found from the virtual-work equation

virtual loadings
1#u = ⌺uu # dL (9–14)
real displacements

where 9
M ⬘ = 1 = external virtual unit couple moment acting in the direction of u.
uu = internal virtual load acting on an element in the direction of dL.
u = external rotational displacement or slope in radians caused by the
real loads.
dL = internal deformation of the element caused by the real loads.
This method for applying the principle of virtual work is often referred
to as the method of virtual forces, since a virtual force is applied resulting
in the calculation of a real displacement. The equation of virtual
work in this case represents a compatibility requirement for the structure.
Although not important here, realize that we can also apply the principle
 356 CHAPTER 9 DEFLECTIONS USING ENERGY METHODS

of virtual work as a method of virtual displacements. In this case virtual

displacements are imposed on the structure while the structure is subjected
to real loadings. This method can be used to determine a force on or in a
structure,* so that the equation of virtual work is then expressed as an
equilibrium requirement.

9.4 Method of Virtual Work: Trusses

We can use the method of virtual work to determine the displacement of
a truss joint when the truss is subjected to an external loading, temperature
change, or fabrication errors. Each of these situations will now be
discussed.
B
External Loading. For the purpose of explanation let us consider
1
the vertical displacement ⌬ of joint B of the truss in Fig. 9–7a. Here a
Apply virtual unit load to B typical element of the truss would be one of its members having a length L,
(a) Fig. 9–7b. If the applied loadings P1 and P2 cause a linear elastic material
response, then this element deforms an amount ⌬L = NL >A E, where N is
P1 the normal or axial force in the member, caused by the loads. Applying
Eq. 9–13, the virtual-work equation for the truss is therefore
P2

1#⌬ = a
L nNL
(9–15)
AE
⌬ B

Apply real loads P1, P2 where

1 = external virtual unit load acting on the truss joint in the stated
(b)
direction of ⌬.
Fig. 9–7 n = internal virtual normal force in a truss member caused by the
external virtual unit load.
⌬ = external joint displacement caused by the real loads on the truss.
N = internal normal force in a truss member caused by the real loads.
L = length of a member.
A = cross-sectional area of a member.
9 E = modulus of elasticity of a member.
The formulation of this equation follows naturally from the development
in Sec. 9.3. Here the external virtual unit load creates internal virtual
forces n in each of the truss members. The real loads then cause the truss
joint to be displaced ⌬ in the same direction as the virtual unit load, and
each member is displaced NL >A E in the same direction as its respective n
force. Consequently, the external virtual work 1 # ⌬ equals the internal
virtual work or the internal (virtual) strain energy stored in all the truss
members, that is, ⌺nNL >A E.
*It was used in this manner in Sec. 6.3 with reference to the Müller-Breslau principle.
 9.4 METHOD OF VIRTUAL WORK: TRUSSES 357

Temperature. In some cases, truss members may change their

length due to temperature. If a is the coefficient of thermal expansion for
a member and ⌬T is the change in its temperature, the change in length
of a member is ⌬L = a ⌬T L. Hence, we can determine the displacement
of a selected truss joint due to this temperature change from Eq. 9–13,
written as

1 # ⌬ = ⌺na ⌬T L (9–16)

where
1 = external virtual unit load acting on the truss joint in the stated
direction of ⌬.
n = internal virtual normal force in a truss member caused by the
external virtual unit load.
⌬ = external joint displacement caused by the temperature change.
a = coefficient of thermal expansion of member.
⌬T = change in temperature of member.
L = length of member.

Fabrication Errors and Camber. Occasionally, errors in

fabricating the lengths of the members of a truss may occur. Also, in
some cases truss members must be made slightly longer or shorter
in order to give the truss a camber. Camber is often built into a bridge
truss so that the bottom cord will curve upward by an amount
equivalent to the downward deflection of the cord when subjected to
the bridge’s full dead weight. If a truss member is shorter or longer
than intended, the displacement of a truss joint from its expected
position can be determined from direct application of Eq. 9–13,
written as

1 # ⌬ = ⌺n ⌬L (9–17)

where
1 = external virtual unit load acting on the truss joint in the stated 9
direction of ⌬.
n = internal virtual normal force in a truss member caused by the
external virtual unit load.
⌬ = external joint displacement caused by the fabrication errors.
⌬L = difference in length of the member from its intended size as caused
by a fabrication error.

A combination of the right sides of Eqs. 9–15 through 9–17 will be

necessary if both external loads act on the truss and some of the members
undergo a thermal change or have been fabricated with the wrong
dimensions.
 358 CHAPTER 9 DEFLECTIONS USING ENERGY METHODS

Procedure for Analysis

The following procedure may be used to determine a specific

displacement of any joint on a truss using the method of virtual work.

Virtual Forces n
r Place the unit load on the truss at the joint where the desired
displacement is to be determined. The load should be in the same
direction as the specified displacement, e.g., horizontal or vertical.

r With the unit load so placed, and all the real loads removed from
the truss, use the method of joints or the method of sections and
calculate the internal n force in each truss member. Assume that
tensile forces are positive and compressive forces are negative.

Real Forces N
r Use the method of sections or the method of joints to determine
the N force in each member. These forces are caused only by the
real loads acting on the truss. Again, assume tensile forces are
positive and compressive forces are negative.

Virtual-Work Equation
r Apply the equation of virtual work, to determine the desired
displacement. It is important to retain the algebraic sign for each
of the corresponding n and N forces when substituting these
terms into the equation.

r If the resultant sum ⌺nNL >A E is positive, the displacement ⌬ is

in the same direction as the unit load. If a negative value results,
⌬ is opposite to the unit load.

r When applying 1 # ⌬ = ⌺na ⌬TL, realize that if any of the

members undergoes an increase in temperature, ⌬T will be
9
positive, whereas a decrease in temperature results in a negative
value for ⌬T.

r For 1 # ⌬ = ⌺n ⌬L, when a fabrication error increases the length of

a member, ⌬L is positive, whereas a decrease in length is negative.
r When applying any formula, attention should be paid to the units
of each numerical quantity. In particular, the virtual unit load can
either be assigned no units, or have any arbitrary unit (lb, kip, N,
etc.), since the n forces will have these same units, and as a result
the units for both the virtual unit load and the n forces will cancel
from both sides of the equation.
 9.4 METHOD OF VIRTUAL WORK: TRUSSES 359

EXAMPLE 9.1
Determine the vertical displacement of joint C of the steel truss shown
in Fig. 9–8a. The cross-sectional area of each member is A = 0.5 in2
and E = 29 ( 103 ) ksi.

F E

10 ft
A B C D

10 ft 10 ft 10 ft
4k 4k
(a)

SOLUTION
Virtual Forces n. Only a vertical 1-k load is placed at joint C, and ⫺0.333 k
the force in each member is calculated using the method of joints. The

0.333 k
k

⫺
1

0.
47
results are shown in Fig. 9–8b. Positive numbers indicate tensile forces k

94
0.
1 1k

3
47

k
and negative numbers indicate compressive forces. 0.
⫺
Real Forces N. The real forces in the members are calculated using 0.333 k 0.667 k 0.667 k
the method of joints. The results are shown in Fig. 9–8c. 0.333 k
1k 0.667 k
Virtual-Work Equation. Arranging the data in tabular form, we have virtual forces n

Member n (k) N (k) L (ft) nNL (k2 # ft) (b)

AB 0.333 4 10 13.33
BC 0.667 4 10 26.67
CD 0.667 4 10 26.67 ⫺4 k
DE - 0.943 - 5.66 14.14 75.42 ⫺
k 5.
FE - 0.333 -4 10 13.33 .6
6 4k 66
5 0 4k k
EB - 0.471 0 14.14 0 ⫺
BF 0.333 4 10 13.33
4k 4k 4k
AF - 0.471 - 5.66 14.14 37.71 9
4k 4k 4k 4k
CE 1 4 10 40
⌺246.47 real forces N

246.47 k2 # ft
1 k # ⌬ Cv = a
nNL (c)
Thus =
AE AE
Fig. 9–8
Converting the units of member length to inches and substituting the
numerical values for A and E, we have
( 246.47 k2 # ft )( 12 in.>ft )
1 k # ⌬ Cv =
( 0.5 in2 )( 29 ( 103 ) k>in2 )
⌬ Cv = 0.204 in. Ans.
 360 CHAPTER 9 DEFLECTIONS USING ENERGY METHODS

EXAMPLE 9.2
The cross-sectional area of each member of the truss shown in
Fig. 9–9a is A = 400 mm2 and E = 200 GPa. (a) Determine the vertical
displacement of joint C if a 4-kN force is applied to the truss at C.
(b) If no loads act on the truss, what would be the vertical displacement
of joint C if member AB were 5 mm too short?

C
4 kN

3m
A
B

4m 4m

(a)

Fig. 9–9

SOLUTION
Part (a)
Virtual Forces n. Since the vertical displacement of joint C is to be
determined, a virtual force of 1 kN is applied at C in the vertical
direction. The units of this force are the same as those of the real
loading. The support reactions at A and B are calculated and the
n force in each member is determined by the method of joints as
shown on the free-body diagrams of joints A and B, Fig. 9–9b.

1 kN

C ⫺
9 N 0.8
33k 33 0.833 kN 0.833 kN
kN
0.8
⫺ 5 3 3 5
A B 4 4
A 0.667 kN 0.667 kN B
0.667 kN

0.5 kN 0.5 kN 0.5 kN virtual forces n 0.5 kN

(b)

Real Forces N. The joint analysis of A and B when the real load of
4 kN is applied to the truss is given in Fig. 9–9c.
 9.4 METHOD OF VIRTUAL WORK: TRUSSES 361

C
4 kN
kN ⫺
2.5 2.5 kN 2.5 kN
2.5 kN
5 3 3 5
A
4 kN B A 4 4
4 kN 2 kN 2 kN B
2 kN

1.5 kN 1.5 kN 1.5 kN real forces N 1.5 kN

(c)

Virtual-Work Equation. Since AE is constant, each of the terms

nNL can be arranged in tabular form and computed. Here positive
numbers indicate tensile forces and negative numbers indicate
compressive forces.

Member n (kN) N (kN) L (m) n NL (kN2 # m)

AB 0.667 2 8 10.67
AC - 0.833 2.5 5 -10.41
CB - 0.833 - 2.5 5 10.41
⌺10.67

Thus,
10.67 kN2 # m
1 kN # ⌬ Cy = a
nNL
=
AE AE

Substituting the values A = 400 mm2 = 400 ( 10-6 ) m2, E = 200 GPa =
200 ( 106 ) kN>m2, we have

10.67 kN2 # m
1 kN # ⌬ Cv = -6
400 ( 10 ) m2 ( 200 ( 106 ) kN>m2 )
⌬ Cv = 0.000133 m = 0.133 mm Ans.

Part (b). Here we must apply Eq. 9–17. Since the vertical
displacement of C is to be determined, we can use the results of
Fig. 9–9b. Only member AB undergoes a change in length, namely, 9
of ⌬L = -0.005 m.
Thus,

1 # ⌬ = ⌺n ⌬L
1 kN # ⌬ Cv = (0.667 kN)(-0.005 m)
⌬ Cv = -0.00333 m = -3.33 mm Ans.

The negative sign indicates joint C is displaced upward, opposite to

the 1-kN vertical load. Note that if the 4-kN load and fabrication
error are both accounted for, the resultant displacement is then
⌬ Cv = 0.133 - 3.33 = -3.20 mm (upward).