Torsion 249

Therefore, the resultant force in z direction due to tension F is

⎡ 2 ⎤
− F ∆ y ∂ z + F ∆y ⎢ ∂ z + ∂ z2 ∆ x ⎥ − F ∆ x ∂ z
∂x ⎣ ∂ x ∂ x ⎦ ∂y

⎡∂ z ∂2z ⎤
+ F ∆x ⎢ + ∆y ⎥
⎣ ∂ y ∂y 2
⎦

⎛ ∂2z ∂2z ⎞
= F ⎜ 2 + 2 ⎟ ∆ x ∆y
⎝∂x ∂y ⎠
The force p acting upward on the membrane element ABCD is p D x D y, assuming
that the membrane deflection is small. For equilibrium, therefore
⎛ 2 2 ⎞
F ⎜⎜ ∂ 2z + ∂ z2 ⎟⎟ = − p
⎝∂ x ∂y ⎠

∂ 2z + ∂ 2z = − p
or (7.43)
∂ x2 ∂ y 2 F
Now, if we adjust the membrane tension F or the air pressure p such that p/F
becomes numerically equal to 2Gq, then Eq. (7.43) of the membrane becomes
identical to Eq. (7.25) of the torsion stress function f. Further, if the membrane
height z remains zero at the boundary contour of the section, then the height z of
the membrane becomes numerically equal to the torsion stress function [Eq. (7.23)].
The slopes of the membrane are then equal to the shear stresses and these are in
a direction perpendicular to that of the slope. The twisting moment is numerically
equivalent to twice the volume under the membrane [Eq. (7.24)].

7.8 TORSION OF THIN-WALLED TUBES

Consider a thin-walled tube subjected to torsion. The thickness of the tube need
not be uniform (Fig. 7.11). Since the thickness is small and the boundaries are free,
the shear stresses will be essentially parallel to the boundary. Let t be the magni-
tude of the shear stress and t the thickness.
Consider the equilibrium of an element of length Dl, as shown. The areas of cut
faces AB and CD are respectively t1 Dl and t2 Dl. The shear stresses (complemen-
tary shears) are t1 and t2. For equilibrium in z direction we should have
−τ1 t1 ∆l + τ 2 t2 ∆l = 0

Dl
A B
A
t1

B
D t1
C Z
D t2
C

Fig. 7.11 Torsion of a thin-walled tube

250 Advanced Mechanics of Solids

or τ1 t1 = τ 2 t2 = q, a constant (7.44)
Hence, the quantity t t is a constant. This is called the shear flow q, since the
equation is similar to the flow of an incompressible liquid in a tube of varying
area. For continuity, we should have V1A1 = V2A2, where A is the area and V the
corresponding velocity of the fluid there.
Consider next the torque of the shear about point O [Fig. 7.12(a)].
DF = q Ds

Dl
Ds

d
O

(b)

(a)
Fig. 7.12 Cross-section of a thin-walled tube and torque due to shear

The force acting on an elementary length Ds of the tube is

DF = τ t ∆ s = q ∆ s
The moment arm about O is h and hence, the torque is
DT = q ∆ sh = 2q ∆ A
where DA is the area of the triangle enclosed at O by the base s. Hence, the total
torque is
T = Σ 2q ∆ A = 2qA (7.45)
Where A is the area enclosed by the centre line of the tube. Equation (7.45) is
generally known as the Bredt–Batho formula.
To determine the twist of the tube, we make use of Castigliano’s theorem.
Referring to Fig. 7.12(b), the shear force on the element is t t Ds = q Ds. Because
of shear strain g , the force does work equal to

DU = 1 (τ t ∆ s ) δ
2
1
= (τ t ∆ s ) γ ∆l
2

=
1
(τ t ∆ s ) ∆ l
τ
2 G

q 2 ∆l ∆ s
= (7.46)
2G t

T 2 ∆l ∆ s
= (7.47)
8 A2 G t
 Torsion 251

using Eq. (7.45). The total elastic strain energy is therefore

T 2 ∆l ds
U= ∫ t (7.48)
8 A2G
Hence, the twist or the rotation per unit length (Dl = 1) is
∂U
= T ds
∂ T 4 A2G ∫ t
q= (7.49)

Using once again Eq. (7.45)

q ds
q=
2 AG ∫ t (7.50)

7.9 TORSION OF THIN-WALLED MULTIPLE-CELL

CLOSED SECTIONS
We can extend the analysis of the previous section to torsion of multiple-cell
sections. Consider the two-cell section shown in Fig. 7.13.

Dl
q1 = t1t1 Dl

A1 t1
q3 t 2t 1
q2 = t2t2 A2 t3
t2

(a) t3
(b)
Fig. 7.13 Torsion of a thin-walled multiple cell closed section
Consider the equilibrium of an element at the junction, as shown in
Fig. 7.13(b). In the direction of the axis of the tube
− τ1 t1 ∆l + τ 2t2 ∆l + τ 3t3 ∆l = 0
or t 1t1 = τ 2 t2 + τ 3t3
i.e., q1 = q2 + q3 (7.51)
This is again equivalent to a fluid flow
Cell 1
dividing itself into two streams.
Cell 2 Choose any moment axis, such as
A2 A1 q1
q2 q1 point O (Fig. 7.14).
q2
O The shear flow in the web can be
considered to be made up of q1 and
–q2, since q3 = q1 – q2. The moment
Fig. 7.14 Section of a thin-walled mul- about O due to q1 flowing in cell 1
tiple cell beam and moment (with web included) is [Eq. (7.45)]
axis T1 = 2q1 A1

Chapter_07.pmd 251 7/3/2008, 5:56 AM

 Energy Methods 155
it is important to note that dn is a linear displacement if Fn is a concentrated load,
or an angular displacement if Fn is a couple or a torque. Further, we must express
the strain energy in terms of the forces (including moments and couples) since it
is the partial derivative with respect to a particular force that gives the corre-
sponding displacement. In the next section, expressions for strain energies in
terms of forces will be obtained.

5.10 EXPRESSIONS FOR STRAIN ENERGY

In this section we shall develop expressions for strain energy when an elastic
member is subjected to axial force, shear force, bending moment and torsion.
Figure 5.10(a) shows an elastic member subjected to several forces. Consider a
section of the member at C. In general, this section will be subjected to three
forces Fx, Fy and Fz and three moments Mx, My and Mz (Fig. 5.10(b)). The force Fx
is the axial force and forces Fy and Fz are the shear forces across the section.
Moment M x is the torque T and moments My and Mz are the bending moments
about the y and z axes respectively. Let Ds be an elementary length of the
member; then when Ds is very small, we can assume that these forces and
moments remain constant over Ds. At the left-hand section of this elementary
member, the forces and moments have opposite signs. During the deformation
caused by the axial force Fx alone, the remaining forces and moments do no
work. Similarly, during the twist caused by the torque T = Mx, no work is as-
sumed to be done (since the deformations are extremely small) by the other
forces and moments.
Consequently, the work done by each of these forces and moments can be
determined individually and added together to determine the total elastic strain
energy stored by Ds while it undergoes deformation. We shall make use of the
formulas available from elementary strength of materials.

A My
C
F3
DS Fy
C Fx
F1 F2

Mx
Fz
Mz
(a) (b)
Fig. 5.10 Reactive forces at a general cross-section
156 Advanced Mechanics of Solids
(i) Elastic energy due to axial force: If dx is the axial extension due to Fx, then

DU = 1 Fxδ x
2
F
= 1 Fx ⋅ x ∆ s
2 AE
using Hooke's law.
Fx2
\ DU = ∆s (5.17)
2 AE
A is the cross-sectional area and E is Young's modulus.
(ii) Elastic energy due to shear force: The shear force Fy (or Fz) is distrib-
uted across the section in a complicated manner depending on the
shape of the cross-section. If we assume that the shear force is distrib-
uted uniformly across the section (which is not strictly correct), the
shear displacement will be (from Fig. 5.11) D s Dg and the work done by
Fy will be
Dg DU = 1 Fy ∆s ∆γ
Ds Dg 2
From Hooke's law,
Fy
Dg =
Ds AG
Fig. 5.11 Displacement due to where A is the cross-sectional area
shear force and G is the shear modulus. Sub-
stituting this
F
DU = 1 Fy ∆ s y
2 AG

Fy2
or DU = ∆s
2 AG
It will be shown that the strain energy due to shear deformation is ex-
tremely small, which is often ignored. Hence, the error caused in assuming
uniform distribution of the shear force across the section will be very
small. However, to take into account the different cross-sections and non-
uniform distribution, a factor k is introduced. With this
k Fy2
DU = ∆s (5.18)
2 AG
A similar expression is obtained for the shear force Fz.
(iii) Elastic energy due to bending moment: Making reference to Fig. 5.12, if Df
is the angle of rotation due to the moment Mz(or My), the work done is

DU = 1 M z ∆φ
2
 Energy Methods 157
From the elementary flexure formula, we
have
Df Mz
= E
Iz R
R
1 Mz
Mz or =
Ds R EI z

Mz where R is the radius of curvature and lz

is the area moment of inertia about the z
Fig. 5.12 Displacement due axis. Hence
to bending moment
∆ s Mz
Df = = ∆s
R EI z
Substituting this
M z2
DU = ∆s (5.19)
2 EI z
A similar expression can be obtained for the moment My.
(iv) Elastic energy due to torque : Because of the torque T, the elementary
member rotates through an angle Dq according to the formula for a
circular section
T = G ∆θ
Ip ∆s

T
i.e. Dq = ∆s
GIp
lp is the polar moment of inertia. The work done due to this twist is,

DU = 1 T ∆θ
2
T 2 ∆s
= (5.20)
2GI p

Equations (5.17)-(5.20) give important expressions for the strain energy

stored in the elementary length Ds of the elastic member. The elastic
energy for the entire member is therefore
S F2
x
(i) Due to axial force U1 = ∫ 2 AE ds (5.21)
0
S k F2
y y
(ii) Due to shear force U2 = ∫ 2 AG ds (5.22)
0
S k z Fz2
U3 = ∫ 2 AG ds (5.23)
0
S M y2
(iii) Due to bending moment U4 = ∫ 2 EI ds (5.24)
0 y
158 Advanced Mechanics of Solids
S M z2
U5 = ∫ 2EI ds (5.25)
0 z
S 2
T
(iv) Due to torque U6 = ∫ 2GI ds (5.26)
0 p

Example 5.5 Determine the deflection at end A of the cantilever beam shown in
Fig. 5.13.
Solution The bending moment
x
at any section x is
B A
L M = Px
P The elastic energy due to bend-
Fig. 5.13 Example 5.5 ing moment is, therefore, from
Eq. (5.24)

U1 = ∫ ( )
2
L Px dx P 2 L3
=
0
2 EI 6 EI
The elastic energy due to shear from Eq. (5.22) is (putting k1 = 1)
L
P 2 P2 L
U2 = ∫ 2 AG dx = 2 AG
0
One can now show that U2 is small as compared to U1. If the beam is of a
rectangular section

A = bd , I = 1 bd 3
12
and 2G ª E
Substituting these
U2 P2 L 3
= ⋅ 6bd2 3 ⋅ 2G
U1 2bdG 12 P L
d2
=
2 L2
For a member to be designated as beam, the length must be fairly large com-
pared to the cross-sectional dimension. Hence, L > d and the above ratio is
extremely small. Consequently, one can neglect shear energy as compared to
bending energy. With
P 2 L3
U=
6 EI
we get

∂ U P L3
= = δA
∂ P 3EI
which agrees with the solution from elementary strength of materials.
 160 Advanced Mechanics of Solids

Example 5.8 For the structure shown in Fig. 5.16, what is the vertical deflection at
end A?

Solution The moment at any section q of the

curved part is Pr (1 - cos q ). The bending mo-
r
ment for the vertical part of the structure is a
q A constant equal to 2Pr. The bending energy
therefore is

⎡⎣ Pr (1 − cos θ )⎤⎦ L ( 2 Pr )2
2
P π
∫ r dθ + ∫ dx
L 0
2 EI 0
2 EI

We neglect the energy due to the axial force.

Then
π P2 r 3 2P2 r 2 L
U= 3 +
Fig. 5.16 Example 5.8 4 EI EI

(
\ dA = ∂ U = 3 π r + 4 L Pr
)
2

∂P 2 EI

Example 5.9 The end of the semi-circular member shown in Fig. 5.17, is subjected
to torque T. What is the twist of end A? The member is circular in section.

y Solution The torque is a mo-

ment in the xy plane and can be
O represented by vector T, as
T shown. At any section q, this
A q r vector can be resolved into two
x
components T cosq and T sinq.
T The component T cosq acts as
A torque and the component T sinq
q q O as a moment.
sin
T

T The energy due to torque

co
sq

is, from Eq. (5.26),

U1 = ∫ (
π T cos θ ) 2
(a) (b) Plan View r dθ
0
2GI P
Fig. 5.17 Example 5.9
= π rT
2

4GI P
The energy due to bending is, from Eq. (5.24),
π (T sin θ )2
U2 = ∫ r dθ
0
2 EI

= π rT
2

4 EI

Chapter_05.pmd 160 7/3/2008, 7:55 AM

 Energy Methods 161
Ip is the polar moment of inertia. For a circular member
4
Ip = 2 I = π r
2
Substituting, the total energy is
2 ⎛ ⎞
U = U1 + U 2 = π rT ⎜ 1 + 1 ⎟
4 ⎝ GI p EI ⎠
Hence, the twist is

(
q = ∂ U = π rT 1 + 1 2 4
∂T 2 2G E π r )
= 1 ( 1 + 1 )T
r 2G E
3

5.11 FICTITIOUS LOAD METHOD

Castigliano's first theorem described above helps us to determine the displace-
ment at a point corresponding to the force acting there. Situations arise where it
may be desirable to determine the displacement (either linear or angular) at a point
where there is no force (concentrated load or a couple) acting. In such situations,
we assume a small fictitious or dummy load to be acting at the point where the
displacement is required. Castigliano's theorem is then applied, and in the final
result, the fictitious load is put equal to zero. The following example will describe
the technique.

Example 5.10 Determine the slope at end A of the cantilever in Fig. 5.18 which
is subjected to load P.
P Solution To determine the
slope by Castigliano's method
L we have to determine U and
A M take its partial derivative with
respect to the corresponding
force, i.e. a moment. But no
Fig. 5.18 Example 5.10 moment is acting at A. So, we
assume a fictitious moment M
to be acting at A and determine the slope caused by P and M. Since the magnitude of
M is actually zero, in the final result, M is equated to zero.
The energy due to P and M is,
L ( Px + M )2
U = ∫ dx
0
2 EI

2 3 2 2
= P L + M L + MPL
6 EI 2 EI 2 EI