9.

5 CASTIGLIANO’S THEOREM 363

9.5 Castigliano’s Theorem

In 1879 Alberto Castigliano, an Italian railroad engineer, published a
book in which he outlined a method for determining the deflection or
slope at a point in a structure, be it a truss, beam, or frame. This method,
which is referred to as Castigliano’s second theorem, or the method of
least work, applies only to structures that have constant temperature,
unyielding supports, and linear elastic material response. If the
displacement of a point is to be determined, the theorem states that it
is equal to the first partial derivative of the strain energy in the
structure with respect to a force acting at the point and in the direction
of displacement. In a similar manner, the slope at a point in a structure
is equal to the first partial derivative of the strain energy in the
structure with respect to a couple moment acting at the point and in
the direction of rotation.
To derive Castigliano’s second theorem, consider a body (structure) of
any arbitrary shape which is subjected to a series of n forces P1, P2, . . . , Pn.
Since the external work done by these loads is equal to the internal strain
energy stored in the body, we can write

Ui = Ue

The external work is a function of the external loads ( Ue = ! 1 P dx ) .

Thus,

Ui = Ue = f ( P1, P2, g, Pn )

Now, if any one of the forces, say Pi, is increased by a differential amount
dPi, the internal work is also increased such that the new strain energy
becomes
0Ui
Ui + dUi = Ui + dP (9–18)
0Pi i
This value, however, should not depend on the sequence in which the n
forces are applied to the body. For example, if we apply dPi to the body first,
then this will cause the body to be displaced a differential amount d" i in the
9
direction of dPi. By Eq. 9–3 1 Ue = 21 P" 2 , the increment of strain energy
would be 21 dPi d" i. This quantity, however, is a second-order differential and
may be neglected. Further application of the loads P1, P2, . . . , Pn, which
displace the body " 1, " 2, . . . , " n, yields the strain energy.
Ui + dUi = Ui + dPi" i (9–19)

Here, as before, Ui is the internal strain energy in the body, caused by the
loads P1, P2, . . . , Pn, and dUi = dPi" i is the additional strain energy caused
by dPi. (Eq. 9–4, Ue = P"#.)
In summary, then, Eq. 9–18 represents the strain energy in the body
determined by first applying the loads P1, P2, . . . , Pn, then dPi, and
Eq. 9–19 represents the strain energy determined by first applying dPi and
 364 CHAPTER 9 DEFLECTIONS USING ENERGY METHODS

then the loads P1, P2, . . . , Pn. Since these two equations must be equal,
we require
0Ui
"i = (9–20)
0Pi
which proves the theorem; i.e., the displacement " i in the direction of Pi is
equal to the first partial derivative of the strain energy with respect to Pi.*
It should be noted that Eq. 9–20 is a statement regarding the structure’s
compatibility. Also, the above derivation requires that only conservative
forces be considered for the analysis. These forces do work that is
independent of the path and therefore create no energy loss. Since forces
causing a linear elastic response are conservative, the theorem is
restricted to linear elastic behavior of the material. This is unlike the
method of virtual force discussed in the previous section, which applied
to both elastic and inelastic behavior.

9.6 Castigliano’s Theorem for Trusses

The strain energy for a member of a truss is given by Eq. 9–9,
Ui = N 2L >2A E. Substituting this equation into Eq. 9–20 and omitting the
subscript i, we have

0P a 2A E
0 N 2L
" =

It is generally easier to perform the differentiation prior to summation. In

the general case L, A, and E are constant for a given member, and
therefore we may write

" = aNa b
0N L
(9–21)
0P A E

where
" = external joint displacement of the truss.
9 P = external force applied to the truss joint in the direction of ".
N = internal force in a member caused by both the force P and the loads
on the truss.
L = length of a member.
A = cross-sectional area of a member.
E = modulus of elasticity of a member.

*Castigliano’s first theorem is similar to his second theorem; however, it relates the load Pi to
the partial derivative of the strain energy with respect to the corresponding displacement, that
is, Pi = 0Ui >0" i. The proof is similar to that given above and, like the method of virtual
displacement, Castigliano’s first theorem applies to both elastic and inelastic material behavior.
This theorem is another way of expressing the equilibrium requirements for a structure, and
since it has very limited use in structural analysis, it will not be discussed in this book.
 9.6 CASTIGLIANO’S THEOREM FOR TRUSSES 365

This equation is similar to that used for the method of virtual work,
Eq. 9–15 ( 1 # " = !nNL >A E ) , except n is replaced by 0N >0P. Notice that
in order to determine this partial derivative it will be necessary to treat P
as a variable (not a specific numerical quantity), and furthermore, each
member force N must be expressed as a function of P. As a result,
computing 0N >0P generally requires slightly more calculation than that
required to compute each n force directly. These terms will of course be
the same, since n or 0N >0P is simply the change of the internal member
force with respect to the load P, or the change in member force per
unit load.

Procedure for Analysis

The following procedure provides a method that may be used to

determine the displacement of any joint of a truss using Castigliano’s
theorem.

External Force P
Place a force P on the truss at the joint where the desired
displacement is to be determined. This force is assumed to have a
variable magnitude in order to obtain the change 0N >0P. Be sure
P is directed along the line of action of the displacement.

Internal Forces N
Determine the force N in each member caused by both the real
(numerical) loads and the variable force P. Assume tensile forces
are positive and compressive forces are negative.
Compute the respective partial derivative 0N >0P for each
member.
After N and 0N >0P have been determined, assign P its numerical
value if it has replaced a real force on the truss. Otherwise, set P 9
equal to zero.

Castigliano’s Theorem
Apply Castigliano’s theorem to determine the desired
displacement ". It is important to retain the algebraic signs for
corresponding values of N and 0N >0P when substituting these
terms into the equation.
If the resultant sum !N ( 0N >0P) L >A E is positive, " is in the same
direction as P. If a negative value results, " is opposite to P.
 366 CHAPTER 9 DEFLECTIONS USING ENERGY METHODS

EXAMPLE 9.4
C Determine the vertical displacement of joint C of the truss shown in
4 kN
Fig. 9–11a. The cross-sectional area of each member is A = 400 mm2
3m
and E = 200 GPa.

A
B SOLUTION
4m 4m
External Force P. A vertical force P is applied to the truss at joint C,
since this is where the vertical displacement is to be determined,
(a)
Fig. 9–11b.
P Internal Forces N. The reactions at the truss supports at A and B
C
are determined and the results are shown in Fig. 9–11b. Using the
4 kN method of joints, the N forces in each member are determined,
Fig. 9–11c.* For convenience, these results along with the partial
3m
derivatives 0N >0P are listed in tabular form as follows:
4 kN A
B
4m 4m 0N 0N
Member N N(P = 0) L Na bL
0.5P " 1.5 kN 0.5P ! 1.5 kN 0P 0P
AB 0.667P + 2 0.667 2 8 10.67
(b)
AC -(0.833P - 2.5) - 0.833 2.5 5 -10.42
BC -(0.833P + 2.5) - 0.833 - 2.5 5 10.42
! = 10.67 kN # m

NAC " 0.833P ! 2.5 kN Since P does not actually exist as a real load on the truss, we require
5 3 P = 0 in the table above.
A 4
4 kN
NAB " 0.667P # 2 kN Castigliano’s Theorem. Applying Eq. 9–21, we have

10.67 kN # m
" Cv = a Na b
0.5P ! 1.5 kN 0N L
NBC ! 0.833P " 2.5 kN =
5
0P AE AE
3
4 B
NAB ! 0.667 P " 2 kN Substituting A = 400 mm2 = 400 ( 10-6 ) m2, E = 200 GPa = 200(109) Pa,
9
and converting the units of N from kN to N, we have

10.67 ( 103 ) N # m
0.5P " 1.5 kN

(c) " Cv = = 0.000133 m = 0.133 mm Ans.

400 ( 10-6 ) m2 ( 200 ( 109 ) N>m2 )
Fig. 9–11
This solution should be compared with the virtual-work method of
Example 9.2.

*It may be more convenient to analyze the truss with just the 4-kN load on it, then
analyze the truss with the P load on it. The results can then be added together to give
the N forces.
 9.6 CASTIGLIANO’S THEOREM FOR TRUSSES 367

EXAMPLE 9.5
Determine the horizontal displacement of joint D of the truss shown
in Fig. 9–12a. Take E = 29 ( 103 ) ksi. The cross-sectional area of each
member is indicated in the figure.

D
P
5 P
1.2

"(20 ! 0.75P)
2 16
in 0.5
9 ft 0.5 0.75 in2 in 2 7 ! .67
1 6.6
A 1 in2 1 in2
C P
B "13.33 "13.33
12 ft 12 ft
10 ! 0.75 P 20 ! 0.75 P 10
10 k
(a)
(b)
Fig. 9–12

SOLUTION
External Force P. Since the horizontal displacement of D is to be
determined, a horizontal variable force P is applied to joint D,
Fig. 9–12b.
Internal Forces N. Using the method of joints, the force N in each
member is computed.* Again, when applying Eq. 9–21, we set P = 0
since this force does not actually exist on the truss. The results are
shown in Fig. 9–12b. Arranging the data in tabular form, we have

0N 0N
Member N N (P = 0) L Na bL
0P 0P
AB -13.33 0 -13.33 12 0
BC -13.33 0 -13.33 12 0
CD 16.67 0 16.67 15 0
9
DA 16.67 + 1.25P 1.25 16.67 15 312.50
BD -(20 + 0.75P) -0.75 - 20 9 135.00

Castigliano’s Theorem. Applying Eq. 9–21, we have

312.50 k # ft ( 12 in.>ft ) 135.00 k # ft ( 12 in.>ft )
" Dh = a N a b
0N L
= 0 + 0 + 0 + +
0P A E ( 0.5 in2 ) 3 29 ( 103 ) k>in2 4 ( 0.75 in2 ) 3 29 ( 103 ) k>in2 4
= 0.333 in. Ans.

*As in the preceding example, it may be preferable to perform a separate analysis of

the truss loaded with 10 k and loaded with P and then superimpose the results.
 368 CHAPTER 9 DEFLECTIONS USING ENERGY METHODS

EXAMPLE 9.6
Determine the vertical displacement of joint C of the truss shown in
Fig. 9–13a. Assume that A = 0.5 in2 and E = 29 ( 103 ) ksi.

F E

10 ft

B C D
A
10 ft 10 ft 10 ft
4k 4k
(a)

F "(0.333P ! 2.667)E

"
2)

(0
77

.9
3.

43
!

P
1P
! 2.667
P

!
0.333P

47 6 P
71

B 0. .88

1.
.4

88
" 1
(0

6)
! C
"

A D
0.333P ! 2.667 0.667P ! 1.333 0.667P ! 1.333

0.333P!2.667k 4k P 0.667P!1.333k
(b)

Fig. 9–13

SOLUTION
External Force P. The 4-k force at C is replaced with a variable
9 force P at joint C, Fig. 9–13b.

Internal Forces N. The method of joints is used to determine the

force N in each member of the truss. The results are summarized in
Fig. 9–13b. Here P = 4 k when we apply Eq. 9–21. The required data
can be arranged in tabulated form as follows:
 9.6 CASTIGLIANO’S THEOREM FOR TRUSSES 369

0N 0N
Member N N (P = 4 k) L Na bL
0P 0P
AB 0.333P + 2.667 0.333 4 10 13.33
BC 0.667P + 1.333 0.667 4 10 26.67
CD 0.667P + 1.333 0.667 4 10 26.67
DE -(0.943P + 1.886) -0.943 - 5.66 14.14 75.42
EF -(0.333P + 2.667) -0.333 -4 10 13.33
FA -(0.471P + 3.771) -0.471 - 5.66 14.14 37.71
BF 0.333P + 2.667 0.333 4 10 13.33
BE -0.471P - 1.886 -0.471 0 14.14 0
CE P 1 4 10 40
! = 246.47 k # ft

Castigliano’s Theorem. Substituting the data into Eq. 9–21, we have

246.47 k # ft
" Cv = a Na b
0N L
=
0P AE AE
Converting the units of member length to inches and substituting the
numerical value for AE, we have
(246.47 k # ft) ( 12 in.>ft )
" Cv = = 0.204 in. Ans.
( 0.5 in2 )( 29 ( 103 ) k>in2 )
The similarity between this solution and that of the virtual-work
method, Example 9.1, should be noted.
9