CONTENTS

Module -4: Elementary problems of elasticity................................................................................. 2

1. Stretching of a Prismatical Bar by Its Own Weight. ............................................................. 2

2. Twist of Circular Shafts of Constant Cross Section. ............................................................ 5

3. Torsion of non-circular sections ........................................................................................... 7

4. General solution of the torsion problem ............................................................................... 8

5. Boundary conditions ........................................................................................................... 11

6. Stress function method ........................................................................................................ 13

7. Prandtl’s membrane analogy............................................................................................... 16

8. Torsion of thin-walled multiple-cell closed sections .......................................................... 18

9. Numerical examples............................................................................................................ 20

1
Dr. Naveenkumar D T
 MODULE -4: ELEMENTARY PROBLEMS OF ELASTICITY

1. Stretching of a Prismatical Bar by Its Own Weight.

If 𝜌𝑔 is the weight per unit volume of the bar (Fig. 139), the body forees are
𝑋 = 𝑌 = 0, 𝑍 = −𝜌𝑔 (a)

The differential equations of equilibrium (127) are satisfied by putting

𝜎𝑧 = 𝜌𝑔𝑧𝑟 𝜎𝑥 = 𝜎𝑦 = 𝜏𝑥𝑦 = 𝜏𝑦𝑧 = 𝜏𝑥𝑧 = 0 (b)

i.e., by assuming that on each cross section we have a uniform tension

produced by the weight of the lower portion of the bar. It can easily be
seen that the boundary conditions (128) at the Iatera] surface, which is
free from forces, are satisfied. The boundary conditions give zero stresses
for the lower end of the bar, and, for the upper end, the uniformly
distributed tensile stress 𝜎2 = 𝜌𝑔𝑙, in which 𝑙 is the length of the bar.

The compatibility equations (130) are also satisfied by the solution (𝑏), hence it is the correct
solution of the problem for a uniform distribution of forces at the top. It coincides with the solution
which is usually given in elementary books on the strength of materials.

Let us consider now the displacements (see Art. 78). From Hooke's law, using Eqs. (3) and (6), we
find

∂𝑤 𝜎𝑧 𝜌𝑔𝑧
𝜖𝑧 = = = (𝑐)
∂𝑧 𝐸 𝐸
∂𝑢 ∂𝑣 𝜌𝑔𝑧
𝜖𝑥 = 𝜖𝑦 = = = −𝜈 (𝑑)
∂𝑥 ∂𝑦 𝐸′
∂𝑢 ∂𝑣 ∂𝑢 ∂𝑤 ∂𝑣 ∂𝑤
𝛾𝑥𝑦 = 𝛾𝑥𝑧 = 𝛾𝑦𝑧 = + = + = + =0 (e)
∂𝑦 ∂𝑥 ∂𝑧 ∂𝑥 ∂𝑧 ∂𝑦

The displacements 𝑢, 𝑣, 𝑤 can now be found by integrating Eqs. (𝑐), (𝑑), and (𝑒). Integration of
Eq. (𝑐) gives

𝜌𝑔𝑧 2
𝑤= + 𝑤0 (f)
2𝐸

where 𝑤0 is a function of 𝑥 and 𝑦, to be determined later. Substituting (𝑓) in the second and third
of Eqs. (c), we find

2
Dr. Naveenkumar D T
 ∂𝑤0 ∂𝑢 ∂𝑤0 ∂𝑣
+ = 0, + =0
∂𝑥 ∂𝑧 ∂𝑦 ∂𝑧

Integrating above equations with respect to z, from which

∂𝑤0 ∂𝑤0
𝑢 = −𝑧 + 𝑢0 , 𝑣 = −𝑧 + 𝑣0 (g)
∂𝑥 ∂𝑦

in which 𝑢0 and 𝑣0 are functions of 𝑥 and 𝑦 only. Substituting expressions (𝑔) into Eqs. (𝑑), we
find

∂2 𝑤0 ∂𝑢0 𝜌𝑔𝑧 ∂2 𝑤0 ∂𝑣0 𝜌𝑔𝑧

−𝑧 + = −𝜈 , −𝑧 + = −𝜈 (h)
∂𝑥 2 ∂𝑥 𝐸 ∂𝑦 2 ∂𝑦 𝐸

Remembering that 𝑢0 and 𝑣0 do not depend on 𝑧, Eqs. (ℎ) can be satisfied only if

∂𝑢0 ∂𝑣0 ∂2 𝑤0 ∂2 𝑤0 𝜈𝜌𝑔

= = 0, = = (k)
∂𝑥 ∂𝑦 ∂𝑥 2 ∂𝑦 2 𝐸

Substitutsng expressions (𝑔) for 𝑢 and 𝑣 into the first of Eqs. (𝑒), wo find

∂2 𝑤 ∂𝑢0 ∂𝑣0
−2𝑧 ∂𝑥 ∂𝑦0 + + =0 (l)
∂𝑦 ∂𝑥

and, since 𝑢0 and 𝑣0 do not depend on 𝑧, we must have

∂2 𝑤0 ∂𝑢0 ∂𝑣0
= 0, + =0
∂𝑥 ∂𝑦 ∂𝑦 ∂𝑥

From Eqs. ( 𝑘) and (𝑙) general expressions can now be written for the functions 𝑢0 , 𝑣0, 𝑤0. It is
easy to show that all these equations are satisfied by

𝑢0 = 𝛿𝑦 + 𝛿1
𝑣0 = −𝛿𝑥 + 𝛾1
𝑣𝜌𝑔
𝑤0 = 2𝐸 (𝑥 2 + 𝑦 2 ) + 𝛼𝑥 + 𝛽𝑦 + 𝛾

in which 𝛼, 𝛽, 𝛾, 𝛿, 𝛿1 , 𝛾1 are arbitrary constants. Now, from Eqs. (f) and (𝑔), the general
expressions for the displacements are

𝜈𝜌𝑔𝑥𝑧
𝑢=− − 𝛼𝑧 + 𝛿𝑦 + 𝛿1
𝐸
𝜈𝜌𝑔𝑦𝑧
𝑣=− − 𝛽𝑧 − 𝛿𝑥 + 𝛾1 (m)
𝐸
𝜌𝑔𝑧 2 𝜈𝜌𝑔
𝑤= + (𝑥 2 + 𝑦 2 ) + 𝛼𝑥 + 𝛽𝑦 + 𝛾
2𝐸 2𝐸

3
Dr. Naveenkumar D T
The six arbitrary constants must be determined from the conditions at the support The support must
be such as to prevent any movement of the bar as a rigid body. To prevent a translatory motion of
the bar, let us fix the centroid 𝐴 of the upper end of the bar so that 𝑢 = 𝑣 = 𝑤 = 0 for 𝑥 = 𝑦 = 0
and 𝑧 = 𝑙. To eliminate rotation of the bar about axes through the point 𝐴, parallel to the 𝑥 - and
𝑦-axes, let us fix an element of the 𝑧-axis at 𝐴. Then ∂𝑢/ ∂𝑧 = ∂𝑣/ ∂𝑧 = 0 at that point. The
possibility of rotation about the z-axis is eliminated by fixıng an elemental area through 𝐴, parallel
to the 𝑧𝑥-plane. Then ∂𝑣/𝑑𝑥 = 0 at the point 𝐴. Using Eqs (𝑚) the above six conditions at the
point 𝐴 become

𝜌𝑔𝑙 2
−𝛼𝑙 + 𝛿1 = 0, −𝛽𝑙 + 𝛾1 = 0, +𝛾 =0
2𝐸
𝛼 = 0, 𝛽 = 0, 𝛿 = 0

Hence

𝜌𝑔𝑙 2
𝛿1 = 0, 𝛾1 = 0, 𝛾 = −
2𝐸

and the final expressions for the displacements are

𝜈𝜌𝑔𝑥𝑧
𝑢=−
𝐸
𝑣𝜌𝑔𝑦𝑧
𝑣=−
𝐸
𝜌𝑔𝑧 2 𝑣𝜌𝑔 2 2)
𝜌𝑔𝑙 2
𝑤= + (𝑥 +𝑦 −
2𝐸 2𝐸 2𝐸

It may be seen that points on the 𝑧-axss have only vertical displacements

𝜌𝑔 2
𝑤=− (𝑙 − 𝑧 2 )
2𝐸

Other points of the bar, on account of lateral contraction, have not only vertical but also horizontal
displacements. Lines which were parallel to the z-axis before deformation become inclined to this
axis after deformation, and the form of the bar after deformation is as indicated in Fig. 139 by
dotted lines. Cross sections of the bar pcrpendicular to the 𝑧-axis after deformation are curved to
the surface of 𝑢 paraboloid Points on the cross section 𝑧 = 𝑐, for instance, after deformation will
be on the surface

𝜌𝑔𝑐 2 𝜈𝜌𝑔 2 𝜌𝑔𝐿2

𝑧 =𝑐+𝑤 = 𝑐+ + (𝑥 + 𝑦 2 ) −
2𝐸 2𝐸 2𝐸

4
Dr. Naveenkumar D T
This surface is perpendicular to all longitudinal fibers of the bar, these being snelined to the 𝑧-axis
after deformation, so that there is no shearing strain 𝛾𝑓𝑦 or 𝛾𝑥𝑧 .

2. Twist of Circular Shafts of Constant Cross Section.

The elementary theory of twist of circular shafts states that the shearing stress 𝜏 at any point of the
cross section (Fig. 140) is perpendicular to the radius 𝑟 and proportional to the length 𝑟 and to the
angle of twist 𝜃 per unit length of the shaft

𝜏 = 𝐺𝜃𝑟 (a)

where 𝐺 is the modulus of rigidity. Resolving this stress into two

components parallel to the 𝑥 - and 𝑦-axes, we find

𝑥
𝜏𝑣𝑧 = 𝐺𝜃𝑟 = 𝐺𝜃𝑥
𝑟
𝑦 (b)
𝜏𝑥𝑧 = −𝐺𝜃𝑟 ⋅ = −𝐺𝜃𝑦
𝑟

The elementary theory also assumes that

𝜎𝑥 = 𝜎𝑣 = 𝜎𝑧 = 𝜏𝑥𝑦 = 0

Equilibrium equations

∂𝜎𝑥 ∂𝜏𝑥𝑦 ∂𝜏𝑥𝑧

+ + + 𝐹𝑥 = 0
∂𝑥 ∂𝑦 ∂𝑧
∂𝜎𝑦 ∂𝜏𝑥𝑦 ∂𝜏𝑦𝑧
+ + + 𝐹𝑦 = 0
∂𝑦 ∂𝑥 ∂𝑧
∂𝜎𝑧 ∂𝜏𝑥𝑧 ∂𝜏𝑦𝑧
+ + + 𝐹𝑧 = 0
∂𝑧 ∂𝑥 ∂𝑦

Boundary conditions

𝑇𝑥 = 𝜎𝑥 𝑙 + 𝜏𝑥𝑦 𝑚 + 𝜏𝑥𝑧 𝑛
𝑇𝑦 = 𝜏𝑥𝑦 𝑙 + 𝜎𝑦 𝑚 + 𝜏𝑦𝑧 𝑛
𝑇𝑧 = 𝜏𝑥𝑧 𝑙 + 𝜏𝑦𝑧 𝑚 + 𝜎𝑧 𝑛

We can show that this elementary solution is the exact solution under certain conditions. Since the
stress components are all either linear functions of the coordinates or zero, the equations of
compatibility are satisfied, and it is only necessary to consider the equations of equilibrium and the
boundary conditions. Substituting the above expressions for stress components into Eqs. of
equilibrium, we find that these equations are satisfied, provided there are no body forces. The

5
Dr. Naveenkumar D T
lateral surface of the shaft is free from forces, and the boundary conditions, remembering that for
the cylindrical surface cos(𝑁𝑧) = 𝑛 = 0, reduce to

0 = 𝜏𝑥𝑧 cos(𝑁𝑥) + 𝜏𝑦𝑧 cos(𝑁𝑦) (c)

For the case of a circular cylinder we have also

𝑥 𝑦
cos(𝑁𝑥) = 𝑟 , cos(𝑁𝑦) = (d)
𝑟

Substituting these and expressions (b) for the stress components into Eq. (c) it is evident that this
equatron is satisfied. It is also evident that for cross sections other than circular, for which Eqs. (𝑑)
do not hold, the stress components (b) do not satisfy the boundary condition (c), and therefore
solution (a) cannot be applied These more complicated problems of twist will be considered later.

Considering now the boundary conditions for the ends of the shaft, we see that the surface shearing
forces must be distributed in exactly the same manner as the stresses 𝜏𝑥𝑧 and 𝜏𝑦𝑧 over any
intermediate cross section of the shaft Only for this case is the stress distribution given

by Eqs (b) an exact solution of the problem. But the practical application of the solution is not
limited to such cases. From Saint Venant's principle it can be concluded that in a long twisted bar,
at a sufficient distance from the ends, the stresses depend only on the magnitude of the torque 𝑀1
and are practically independent of the manner in which the forces are distributed over the ends.

The displacements for this case can be found in the same manner as in the previous article.
Assuming the same condition of constraint at the point 𝐴 as in the previous problem we find

𝑢 = −𝜃𝑦𝑧, 𝑣 = 𝜃𝑥𝑧, 𝑤 = 0

6
Dr. Naveenkumar D T
This means that the assumption that cross sections remain plane and radii remain straight, which
is usually made in the elementary derivation of the theory of twist, is correct

3. Torsion of non-circular sections

From the study of elementary strength of materials, two important expressions related to Λ the
torsion of circular bars were developed. They are

𝑀𝑡 𝑟
𝜏=
𝐽
1 𝑀𝑡 𝑑𝑧
and 𝜃 = ∫   
𝐿 𝐿 𝐺𝐽

Here 𝜏 represents the shear stress, 𝑀𝑡 the applied torque, 𝑟 the radius at which the stress is required,
𝐺 the shear modulus, 𝜃 the angle of twist per unit longitudinal length, 𝐿 the length, and 𝑧 the axial
co-ordinate.
Also, 𝐽 = Polar moment of inertia which is defined by ∫𝐴  𝑟 2 𝑑𝐴.

The following are the assumptions associated with the elementary approach in deriving (7.1) and
(7.2).

1 The material is homogeneous and obeys Hooke's Law.

2 All plane sections perpendicular to the longitudinal axis remain plane following the
application of a torque, i.e., points in a given cross-sectional plane remain in that plane
after twisting.
3 Subsequent to twisting, cross-sections are undistorted in their individual planes, i.e., the
shearing strain varies linearly with the distance from the central axis.
4 Angle of twist per unit length is constant.
In most cases, the members that transmit torque, such as propeller shaft and torque tubes
of power equipment, are circular or turbular in cross-section.
But in some cases, slender members with other than circular cross-sections are used. These are
shown in the Figure 7.0.

7
Dr. Naveenkumar D T
While treating non-circular prismatic bars, initially plane cross-sections [Figure 7.0 (a)] experience
out-of-plane deformation or "Warping" [Figure 7.0(b)] and therefore assumptions 2. and 3. are no
longer appropriate. Consequently, a different analytical approach is employed, using theory of
elasticity.

4. General solution of the torsion problem

The correct solution of the problem of torsion of bars by couples applied at the ends was given by
Saint-Venant. He used the semi-inverse method. In the beginning, he made certain assumptions for
the deformation of the twisted bar and showed that these assumptions could satisfy the equations
of equilibrium given by

∂𝜎𝑥 ∂𝜏𝑥𝑦 ∂𝜏𝑥𝑧

+ + + 𝐹𝑥 = 0
∂𝑥 ∂𝑦 ∂𝑧
∂𝜎𝑦 ∂𝜏𝑥𝑦 ∂𝜏𝑦𝑧
+ + + 𝐹𝑦 = 0
∂𝑦 ∂𝑥 ∂𝑧
∂𝜎𝑧 ∂𝜏𝑥𝑧 ∂𝜏𝑦𝑧
+ + + 𝐹𝑧 = 0
∂𝑧 ∂𝑥 ∂𝑦

and the boundary conditions such as

𝑋‾ = 𝜎𝑥 𝑙 + 𝜏𝑥𝑦 𝑚 + 𝜏𝑥𝑧 𝑛
𝑌‾ = 𝜎𝑦 𝑚 + 𝜏𝑦𝑧 𝑚 + 𝜏𝑥𝑦 𝑙
𝑍‾ = 𝜎𝑧 𝑛 + 𝜏𝑥𝑧 𝑙 + 𝜏𝑦𝑧 𝑚

in which 𝐹𝑥 , 𝐹𝑦 , 𝐹𝑧 are the body forces, 𝑋, 𝑌, 𝑍 are the components of the surface forces per unit
area and 𝑙, 𝑚, 𝑛 are the direction cosines.

Figure 7.1 Bars subjected to torsion

8
Dr. Naveenkumar D T
Also from the uniqueness of solutions of the elasticity equations, it follows that the torques on the
ends are applied as shear stress in exactly the manner required by the solution itself.
Now, consider a prismatic bar of constant arbitrary cross-section subjected to equal and opposite
twisting moments applied at the ends, as shown in the Figure 7.1(a).

Saint-Venant assumes that the deformation of the twisted shaft consists of

Rotations of cross-sections of the shaft as in the case of a circular shaft and warping of the cross-
sections that is the same for all cross-sections.
The origin of 𝑥, 𝑦, 𝑧 in the figure is located at the center of the twist of the cross-section, about
which the cross-section rotates during twisting. Figure 7.1 (b) shows the partial end view of the
bar (and could represent any section). An arbitrary point on the cross-section, point 𝑃(𝑥, 𝑦), located
a distance 𝑟 from centre of twist 𝐴, has moved to 𝑃′ (𝑥 − 𝑢, 𝑦 + 𝑣) as a result of torsion. Assuming
that no rotation occurs at end 𝑧 = 0 and that 𝜃 is small, the 𝑥 and 𝑦 displacements of 𝑃 are
respectively:
𝑢 = −(𝑟𝜃𝑧)sin 𝛼

But sin 𝛼 = 𝑦/𝑟

Therefore, 𝑢 = −(𝑟𝜃𝑧)𝑦/𝑟 = −𝑦𝜃𝑧 (a)

(𝑟𝜃𝑧 )𝑥
Similarly, 𝑣 = (𝑟𝜃𝑧) cos 𝛼 = = 𝑥𝜃𝑧 (b)
𝑟

where 𝜃𝑧 is the angle of rotation of the cross-section at a distance 𝑧 from the origin,

The warping of cross-sections is defined by a function 𝜓 as

𝑤 = 𝜃𝜓(𝑥, 𝑦) (c)

Here, the equations (a) and (b) specify the rigid body rotation of any cross-section through a small
angle 𝜃𝑧 . However, with the assumed displacements (a), (b) and (c), we calculate the components
of strain from the equations given below.

∂𝑢 ∂𝑣 ∂𝑤
𝜀𝑥 = , 𝜀𝑦 = , 𝜀𝑧 =
∂𝑥 ∂𝑦 ∂𝑧
∂𝑢 ∂𝑣 ∂𝑣 ∂𝑤
𝛾𝑥𝑦 = + , 𝛾𝑦𝑧 = +
∂𝑦 ∂𝑥 ∂𝑧 ∂𝑦
∂𝑤 ∂𝑢
and 𝛾𝑧𝑥 = +
∂𝑥 ∂𝑧

9
Dr. Naveenkumar D T
Substituting (a), (b) and (c) in the above equations, we obtain

𝜀𝑥 = 𝜀𝑦 = 𝜀𝑧 = 𝛾𝑥𝑦 = 0
∂𝑤 ∂𝜓
𝛾𝑥𝑧 = − 𝑦𝜃 = (𝜃 − 𝑦𝜃)
∂𝑥 ∂𝑥
∂𝜓
or 𝛾xz = 𝜃 ( − 𝑦)
∂𝑥

∂𝑤 ∂𝜓
and 𝛾yz = + 𝑥𝜃 = (𝜃 ∂𝑦 + 𝑥𝜃)
∂𝑦

∂𝜓
or 𝛾𝑦𝑧 = 𝜃 ( ∂𝑦 + 𝑥)

Also, by Hooke's Law, the stress-strain relationships are given by

𝜎𝑥 = 2𝐺𝜀𝑥 + 𝜆𝑒, 𝜏𝑥𝑦 = 𝐺𝛾𝑥𝑦

𝜎𝑦 = 2𝐺𝜀𝑦 + 𝜆𝑒, 𝜏𝑦𝑧 = 𝐺𝛾𝑦𝑧
𝜎𝑧 = 2𝐺𝜀𝑧 + 𝜆𝑒, 𝜏𝑥𝑧 = 𝛾𝑥𝑧

where 𝑒 = 𝜀𝑥 + 𝜀𝑦 + 𝜀𝑧

𝑣𝐸
and 𝜆 = (1+𝑣)(1−2𝑣)

Substituting (a), (b) and (c) in the above equations, we obtain

𝜎𝑥 = 𝜎𝑦 = 𝜎𝑧 = 𝜏𝑥𝑦 = 0
∂𝑤 ∂𝜓 (d)
𝜏𝑥𝑧 = 𝐺 ( ∂𝑥 − 𝑦𝜃) = 𝐺𝜃 ( ∂𝑥 − 𝑦)

∂𝑤 ∂𝜓
𝜏𝑦𝑧 = 𝐺 ( ∂𝑦 + 𝑥𝜃) = 𝐺𝜃 ( ∂𝑦 + 𝑥) (e)

It can be observed that with the assumptions (a), (b) and (c) regarding deformation, there will be
no normal stresses acting between the longitudinal fibers of the shaft or in the longitudinal direction
of those fibers. Also, there will be no distortion in the planes of cross-sections, since 𝜀𝑥 , 𝜀𝑦 and 𝛾𝑥𝑦
vanish. We have at each point, pure shear defined by the components 𝜏𝑥𝑧 and 𝜏𝑦𝑧

However, the stress components should satisfy the equations of equilibrium given by:

10
Dr. Naveenkumar D T
 ∂𝜎𝑥 ∂𝜏𝑥𝑦 ∂𝜏𝑥𝑧
+ + + 𝐹𝑥 = 0
∂𝑥 ∂𝑦 ∂𝑧
∂𝜎𝑦 ∂𝜏𝑥𝑦 ∂𝜏𝑦𝑧
+ + + 𝐹𝑦 = 0
∂𝑦 ∂𝑥 ∂𝑧
∂𝜎𝑧 ∂𝜏𝑥𝑧 ∂𝜏𝑦𝑧
+ + + 𝐹𝑧 = 0
∂𝑧 ∂𝑥 ∂𝑦

Assuming negligible body forces, and substituting the stress components into equilibrium
equations, we obtain

∂𝜏𝑥𝑧 ∂𝜏𝑧𝑦 ∂𝜏𝑥𝑧 ∂𝜏𝑧𝑦

= 0, = 0, + =0 (7.3)
∂𝑧 ∂𝑧 ∂𝑥 ∂𝑦

∂ ∂𝜓 ∂ ∂𝜓
(𝐺𝜃 ( − 𝑦)) + (𝐺𝜃 ( − 𝑦)) = 0
∂𝑥 ∂𝑥 ∂𝑦 ∂𝑥

Therefore, we find that the function 𝜓 must satisfy the equation

∂2 𝜓 ∂2 𝜓
+ ∂𝑦 2 = 0 (7.3a)
∂𝑥 2

Now, differentiating equation (d) with respect to 𝑦 and the equation (e) with respect to 𝑥, and
subtracting we get an equation of compatibility

∂𝜏𝑥𝑧
Hence, = −𝐺𝜃
∂𝑦

∂𝜏𝑦𝑧
= 𝐺𝜃
∂𝑥
∂𝜏𝑥𝑧 ∂𝜏𝑦𝑧
− = −𝐺𝜃 − 𝐺𝜃 = −2𝐺𝜃 = 𝐻
∂𝑦 ∂𝑥

∂𝜏𝑥𝑧 ∂𝜏𝑦𝑧
Therefore, − =𝐻 (7.40)
∂𝑦 ∂𝑥

Therefore, the stress in a bar of arbitrary section may be determined by solving Equations (7.3) and
(7.4) along with the given boundary conditions.

5. Boundary conditions
Now, consider the boundary conditions given by

𝑋‾ = 𝜎𝑥 𝑙 + 𝜏𝑥𝑦 𝑚 + 𝜏𝑥𝑧 𝑛
𝑌‾ = 𝜎𝑦 𝑚 + 𝜏𝑦𝑧 𝑛 + 𝜏𝑥𝑦 𝑙
𝑍‾ = 𝜎𝑧 𝑛 + 𝜏𝑥𝑧 𝑙 + 𝜏𝑦𝑧 𝑚
11
Dr. Naveenkumar D T
For the lateral surface of the bar, which is free from external forces acting on the boundary and the
normal 𝑛 to the surface is perpendicular to the z-axis, we have 𝑋 = 𝑌 = 𝑍 = 0 and 𝑛 = 0. The
first two equations are identically satisfied and the third gives,

𝜏𝑥𝑧 𝑙 + 𝜏𝑦𝑧 𝑚 = 0 (7.5)

which means that the resultant shearing stress at the boundary is directed along the tangent to the
boundary, as shown in the Figure 7.2.

Figure 7.2 Cross-section of the bar & Boundary conditions.

Considering an infinitesimal element 𝑎𝑏𝑐 at the boundary and assuming that 𝑆 is increasing in the
direction from 𝑐 to 𝑎,

𝑑𝑦
𝑙 = cos(𝑁, 𝑥) =
𝑑𝑆
𝑑𝑥
𝑚 = cos(𝑁, 𝑦) = −
𝑑𝑆

∴ Equation (7.5) becomes

𝑑𝑦 𝑑𝑥
𝜏𝑥𝑧 ( ) − 𝜏𝑦𝑧 ( ) = 0
𝑑𝑆 𝑑𝑆
∂𝜓 𝑑𝑦 ∂𝜓 𝑑𝑥
or ( − 𝑦) ( ) − ( + 𝑥) ( ) = 0
∂𝑥 𝑑𝑆 ∂𝑦 𝑑𝑆

Thus each problem of torsion is reduced to the problem of finding a function 𝜓 satisfying equation
(7.3a) and the boundary condition (7.6).

12
Dr. Naveenkumar D T
6. Stress function method
Based on the result of the torsion of the circular shaft, let the non-vanishing components be 𝜏𝑧𝑥
and 𝜏𝑦𝑧 . The remaining stress components 𝜎𝑥 , 𝜎𝑦 and 𝜎𝑧 and 𝜏𝑥𝑦 are assumed to be zero. In order
∂𝜏𝑥𝑧 ∂𝜏𝑦𝑧 ∂𝜏𝑥𝑧 ∂𝜏𝑦𝑧
to satisfy the equations of equilibrium, we should have = 0, = 0, + =0
∂𝑧 ∂𝑧 ∂𝑥 ∂𝑦

The first two are already satisfied since 𝜏𝑥𝑧 and 𝜏𝑦𝑧 , as given by Equations (d) and (e) are
independent of 𝑧.

In order to satisfy the third condition, we assume a function 𝜙(𝑥, 𝑦) called Prandtl stress function
such that

∂𝜙 ∂𝜙
𝜏𝑥𝑧 = , 𝜏𝑦𝑧 = − (7.7)
∂𝑦 ∂𝑥

With this stress function, (called Prandtl torsion stress function), the third condition is also
satisfied. The assumed stress have to satisfy the compatibility conditions.

From Equations (7.7), (d) and (e), we have

∂𝜙 ∂𝜓 ∂𝜙 ∂𝜓
= 𝐺𝜃 ( − 𝑦) , − = 𝐺𝜃 ( + 𝑥)
∂𝑦 ∂𝑥 ∂𝑥 ∂𝑦
Eliminating 𝜓 by differentiating the first with respect to 𝑦, the second with respect to 𝑥, and
subtracting from the first, we find that the stress function must satisfy the differential equation
∂2 𝜙 ∂2 𝜙
+ = −2𝐺𝜃
∂𝑥 2 ∂𝑦 2
∂2 𝜙 ∂2 𝜙
or + ∂𝑦 2 = 𝐻, where 𝐻 = −2𝐺𝜃 (7.8)
∂𝑥 2

The boundary condition (7.5) becomes, introducing Equation. (7.7)

∂𝜙 𝑑𝑦 ∂𝜙 𝑑𝑥 𝑑𝜙
+ ∂𝑥 𝑑𝑆 = =0 (7.9)
∂𝑦 𝑑𝑆 𝑑𝑆

The determination of stress distribution over a cross-section of a twisted bar is used in finding the
function 𝜙 that satisfies Equation (7.8) and is zero at the boundary.

Conditions at the Ends of the Twisted bar

On the two end faces, the resultants in 𝑥 and 𝑦 directions should vanish, and the moment about 𝐴
should be equal to the applied torque 𝑀𝑡 . The resultant in the 𝑥-direction is

13
Dr. Naveenkumar D T
 ∂𝜙 ∂𝜙
∬ 𝜏𝑥𝑧 𝑑𝑥𝑑𝑦 = ∬ 𝑑𝑥𝑑𝑦 = ∫ 𝑑𝑥 ∫ 𝑑𝑦
∂𝑦 ∂𝑦

Therefore, ∬ 𝜏𝑥𝑧 𝑑𝑥𝑑𝑦 = 0 (7.10)

Since 𝜙 is constant around the boundary. Similarly, the resultant in the 𝑦-direction is

∂𝜙
∬ 𝜏𝑦𝑧 𝑑𝑥𝑑𝑦 = −∬ 𝑑𝑥𝑑𝑦
∂𝑥
∂𝜙
= − ∫ 𝑑𝑦 ∫ 𝑑𝑥
∂𝑥

hence, ∬ 𝜏𝑦𝑧 𝑑𝑥𝑑𝑦 = 0 (7.11)

Thus the resultant of the forces distributed over the ends of the bar is zero, and these forces
represent a couple the magnitude of which is

𝑀𝑡 = ∬ (𝑥𝜏𝑦𝑧 − 𝑦𝜏𝑥𝑧 )𝑑𝑥𝑑𝑦

∂𝜙 ∂𝜙
= − ∬ (𝑥 + 𝑦 ) 𝑑𝑥𝑑𝑦
∂𝑥 ∂𝑦

Therefore,

∂𝜙 ∂𝜙
𝑀𝑡 = − ∬ 𝑥 𝑑𝑥𝑑𝑦 − ∬ 𝑦 𝑑𝑥𝑑𝑦
∂𝑥 ∂𝑦

Integrating by parts

𝑑𝑢 𝑑𝑣
∫𝑣 𝑑𝑥 = 𝑢𝑣 − ∫ 𝑢 𝑑𝑥
𝑑𝑥 𝑑𝑥

∂𝑥 ∂𝑦
𝑀𝑡 = − [∫ x𝜙 dy − ∬ 𝜙 𝑑𝑥𝑑𝑦 + ∫ 𝑦𝜙 dx − ∬ 𝜙 𝑑𝑥𝑑𝑦]
∂𝑥 ∂𝑦

∂𝑥 ∂𝑦
𝑀𝑡 = − [∫ x𝜙 dy − ∬ 𝜙 𝑑𝑥𝑑𝑦 + ∫ 𝑦𝜙 dx − ∬ 𝜙 𝑑𝑥𝑑𝑦]
∂𝑥 ∂𝑦

Observing that 𝜙 = 0 at the boundary, we get

𝑀𝑡 = ∬ 𝜙𝑑𝑥𝑑𝑦 + ∬ 𝜙𝑑𝑥𝑑𝑦 (7.13)

∴ 𝑀𝑡 = 2 ∬ 𝜙𝑑𝑥𝑑𝑦

14
Dr. Naveenkumar D T
Hence, we observe that each of the integrals in Equation (7.13) contributing one half of the torque
due to 𝜏𝑥𝑧 and the other half due to 𝜏𝑦𝑧 .

𝒙𝟐 𝒚𝟐
Show that 𝝓 = 𝒎 [𝒂𝟐 + 𝒃𝟐 − 𝟏] is a stress function for the solution of a torsion problem of

elliptical cross-section. Find the value of 𝒎 and the torsional rigidity.

for x=a, y=0 𝜙 = 0

for x=0, y=b 𝜙 = 0

also 𝜙 = 0 for any combination of x and y. therefore it is a possible stress function for the torsion
problem

∂2 𝜙 ∂2 𝜙
+ = −2𝐺𝜃
∂𝑥 2 ∂𝑦 2

∂2 𝒙 𝟐 𝒚𝟐 ∂2 𝒙 𝟐 𝒚𝟐
[𝒎 [ + − 𝟏]] + [𝒎 [ + − 𝟏]] = −2𝐺𝜃
∂𝑥 2 𝒂𝟐 𝒃𝟐 ∂𝑦 2 𝒂𝟐 𝒃𝟐

𝟐 𝟐
𝒎[ 𝟐
+ 𝟐 ] = −2𝐺𝜃
𝒂 𝒃

𝒃𝟐 + 𝒂𝟐
𝟐𝒎 [ ] = −2𝐺𝜃
𝒂𝟐 𝒃𝟐

𝐺𝜃𝒂𝟐 𝒃𝟐
𝒎=−
𝒃𝟐 + 𝒂𝟐

𝐺𝜃𝒂𝟐 𝒃𝟐 𝒙𝟐 𝒚𝟐
𝝓=− [ + − 𝟏]
𝒃𝟐 + 𝒂𝟐 𝒂𝟐 𝒃𝟐

𝐺𝜃𝒂𝟐 𝒃𝟐 𝒙𝟐 𝒚𝟐
𝑀𝑡 = 2 ∬ 𝜙𝑑𝑥𝑑𝑦=2 ∬ {− [𝒂𝟐 + 𝒃𝟐 − 𝟏]} 𝑑𝑥𝑑𝑦
𝒃𝟐 +𝒂𝟐

2𝐺𝜃𝒂𝟐 𝒃𝟐 𝒙 𝟐 𝒚𝟐
𝑀𝑡 = − 𝟐 ∬ [ 𝟐 + 𝟐 − 𝟏] 𝑑𝑥𝑑𝑦
𝒃 + 𝒂𝟐 𝒂 𝒃

2𝐺𝜃𝒂𝟐 𝒃𝟐 𝟏 𝟐
𝟏
𝑀𝑡 = − 𝟐 [ ∬ 𝒙 𝑑𝑥𝑑𝑦 + ∬ 𝐲𝑑𝑥𝑑𝑦 − ∬ 𝑑𝑥𝑑𝑦]
𝒃 + 𝒂𝟐 𝒂𝟐 𝒃𝟐

For elliptic cross-section,

15
Dr. Naveenkumar D T
 𝜋𝑎3 𝑏
𝐼𝑥 = ∬ 𝒙𝟐 𝑑𝑥𝑑𝑦 =
4

𝜋𝑎𝑏 3
𝐼𝑦 = ∬ 𝒚𝟐 𝑑𝑥𝑑𝑦 =
4

∬ 𝑑𝑥𝑑𝑦 = 𝜋𝑎𝑏

2𝐺𝜃𝒂𝟐 𝒃𝟐 𝟏 𝜋𝑎3 𝑏 𝟏 𝜋𝑎𝑏 3

𝑀𝑡 = − [ + 𝟐 − 𝜋𝑎𝑏]
𝒃𝟐 + 𝒂𝟐 𝒂𝟐 4 𝒃 4

2𝐺𝜃𝒂𝟐 𝒃𝟐 𝜋𝑎𝑏 𝜋𝑎𝑏

𝑀𝑡 = − [ + − 𝜋𝑎𝑏]
𝒃𝟐 + 𝒂𝟐 4 4

2𝐺𝜃𝒂𝟐 𝒃𝟐 1 1 2𝐺𝜃𝒂𝟐 𝒃𝟐 1
𝑀𝑡 = − 𝜋𝑎𝑏 [ + − 1] = − 𝜋𝑎𝑏.
𝒃𝟐 + 𝒂𝟐 4 4 𝒃𝟐 + 𝒂𝟐 2

𝜋𝐺𝜃𝒂𝟑 𝒃𝟑
𝑀𝑡 = 𝟐
𝒃 + 𝒂𝟐

𝑀𝑡 𝒃𝟐 + 𝒂𝟐
𝜃=
𝐺 𝜋𝒂𝟑 𝒃𝟑

∂𝜙 ∂ 𝐺𝜃𝒂𝟐 𝒃𝟐 𝒙𝟐 𝒚𝟐 𝐺𝜃𝒂𝟐 𝒃𝟐 𝟐𝒚 2𝐺𝒂𝟐 𝒚

𝜏𝑥𝑧 = = [− [ + − 𝟏]] = − [ 𝟐] = − 𝜃
∂𝑦 ∂𝑦 𝒃𝟐 +𝒂𝟐 𝒂𝟐 𝒃𝟐 𝒃𝟐 +𝒂𝟐 𝒃 𝒃𝟐 +𝒂𝟐

2𝐺𝒂𝟐 𝒚 𝑀𝑡 𝒃𝟐 +𝒂𝟐 2𝑀 𝒚
𝜏𝑥𝑧 = − 𝒃𝟐 +𝒂𝟐 = − 𝜋𝑎𝒃𝑡 𝟑
𝐺 𝜋𝒂𝟑 𝒃𝟑

∂𝜙 ∂ 𝐺𝜃𝒂𝟐 𝒃𝟐 𝒙𝟐 𝒚𝟐 𝐺𝜃𝒂𝟐 𝒃𝟐 𝟐𝒙 2𝐺𝒃𝟐 𝒙

𝜏𝑦𝑧 = − ∂𝑥 =− ∂x [− [𝒂𝟐 + 𝒃𝟐 − 𝟏]] = [𝒂𝟐 ] = 𝒃𝟐 +𝒂𝟐 𝜃
𝒃𝟐 +𝒂𝟐 𝒃𝟐 +𝒂𝟐

2𝐺𝒃𝟐 𝒙 𝑀𝑡 𝒃𝟐 +𝒂𝟐 2𝑀 𝒙
𝜏𝑦𝑧 = 𝒃𝟐 +𝒂𝟐 = 𝜋𝑎3𝑡 𝐛
𝐺 𝜋𝒂𝟑 𝒃𝟑

7. Prandtl’s membrane analogy

It becomes evident that for bars with more complicated cross-sectional shapes, more analytical
solutions are involved and hence become difficult. In such situations, it is desirable to use other
techniques – experimental or otherwise. The membrane analogy introduced by Prandtl has proved
very valuable in this regard.

16
Dr. Naveenkumar D T
Let a thin homogeneous membrane, like a thin rubber sheet be stretched with uniform tension fixed
at it’s edge which is a given curve (the cross-section of the shaft) in the xy-plane as shown in the
figure 7.5.
When the membrane is subjected to a uniform lateral pressure 𝑝, it undergoes a small displacement
𝑧 where 𝑧 is a function of 𝑥 and 𝑦

Consider the equilibrium of an infinitesimal element ABCD of the membrane after deformation.
Let 𝐹 be the uniform tension per unit length of the membrane. The value of the initial tension 𝐹 is
large enough to ignore its change when the membrane is blown up by the small pressure 𝑝. On the
face 𝐴𝐶, the force acting is 𝐹. 𝑑𝑦. This is inclined at an angle 𝛽 to the 𝑥-axis. Also, tan 𝛽 is the
∂𝑧 ∂𝑧
slope of the face 𝐴𝐵 and is equal to ∂𝑥. Hence the component of 𝐹𝑑𝑦 in 𝑧-direction is (−𝐹𝑑𝑦 ∂𝑥).

The force on face BC is also 𝐹𝑑𝑦 but is inclined at an angle (𝛽 + Δ𝛽) to the 𝑥-axis. Its slope is,
therefore,

∂𝑧 ∂ ∂𝑧
+ ( ) 𝑑𝑥
∂𝑥 ∂𝑥 ∂𝑥

Figure 7.5 Stretching of a membrane

∂𝑧 ∂ ∂𝑧
and the component of the force in the 𝑧-direction is 𝐹𝑑𝑦 [∂𝑥 + ∂𝑥 (∂𝑥) 𝑑𝑥]

17
Dr. Naveenkumar D T
 ∂𝑧
Similarly, the components of the forces 𝐹𝑑𝑥 acting on face AB and CD are −𝐹𝑑𝑥 ∂𝑦 and
∂𝑧 ∂ ∂𝑧
𝐹𝑑𝑥 [∂𝑦 + ∂𝑦 (∂𝑦) 𝑑𝑦]

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

∂𝑧 ∂𝑧 ∂2 𝑧 ∂𝑧 ∂𝑧 ∂2 𝑧
= −𝐹𝑑𝑦 + 𝐹𝑑𝑦 [ + 2 𝑑𝑥] − 𝐹𝑑𝑥 + 𝐹𝑑𝑥 [ + 2 𝑑𝑦]
∂𝑥 ∂𝑥 ∂𝑥 ∂𝑦 ∂𝑦 ∂𝑦
2 2
∂ 𝑧 ∂ 𝑧
= 𝐹 ( 2 + 2 ) 𝑑𝑥𝑑𝑦
∂𝑥 ∂𝑦

But the force 𝑝 acting upward on the membrane element ABCD is 𝑝𝑑𝑥𝑑𝑦, assuming that the
membrane deflection is small.

Hence, for equilibrium,

∂2 𝑧 ∂2 𝑧
𝐹( + ) = −𝑝
∂𝑥 2 ∂𝑦 2
∂2 𝑧 ∂2 𝐳
or + = −𝑝/𝐹
∂x 2 ∂y 2

Now, if the membrane tension 𝐹 or the air pressure 𝑝 is adjusted in such a way that 𝑝/𝐹 becomes
numerically equal to 2𝐺𝜃, then above Equation of the membrane becomes identical to Equation of
the torsion stress function 𝜙.

The membrane analogy provides a useful experimental technique. It also serves as the basis for
obtaining approximate analytical solutions for bars of narrow cross-section as well as for member
of open thin walled section.

8. Torsion of thin-walled multiple-cell closed sections

18
Dr. Naveenkumar D T
Consider the two-cell section shown in the Figure. Let 𝐴1 and 𝐴2 be the areas of the cells 1 and 2
respectively. Consider the equilibrium of an element at the junction as shown in the Figure 7.8( b).
In the direction of the axis of the tube, we can write

−𝜏1 𝑡1 Δ𝑙 + 𝜏2 𝑡2 Δ𝑙 + 𝜏3 𝑡3 Δ𝑙 = 0

or 𝜏1 𝑡1 = 𝜏2 𝑡2 + 𝜏3 𝑡3

i.e., 𝑞1 = 𝑞2 + 𝑞3

This is again equivalent to a fluid flow dividing itself into two streams. Now, choose moment axis,
such as point O as shown in the Figure 7.9.

The shear flow in the web is considered to be made of 𝑞1 and −𝑞2 , since 𝑞3 = 𝑞1 − 𝑞2 . Moment
about O due to 𝑞1 flowing in cell 1 (including web) is 𝑀𝑡1 = 2𝑞1 𝐴1

Similarly, the moment about 𝑂 due to 𝑞2 flowing in cell 2 (including web) is 𝑀𝑡2 =
2𝑞2 (𝐴2 + 𝐴1 ) − 2𝑞2 𝐴1

The second term with the negative sign on the right hand side is the moment due to shear flow 𝑞2
in the middle web.

Therefore, The total torque is

𝑀𝑡 = 𝑀𝑡1 + 𝑀𝑡2
(a)
𝑀𝑡 = 2𝑞1 𝐴1 + 2𝑞2 𝐴2

To Find the Twist (𝜃)

For continuity, the twist of each cell should be the same.

19
Dr. Naveenkumar D T
We have

𝑞 𝑑𝑆
𝜃= ∮
2𝐴𝐺 𝑡

1 𝑞𝑑𝑆
or 2𝐺𝜃 = 𝐴 ∫ 𝑡

𝑑𝑆 𝑑𝑆 𝑑𝑆
Let 𝑎1 = ∮ for Cell 1 including the web 𝑎2 = ∮ for Cell 2 including the web 𝑎12 = ∮
𝑡 𝑡 𝑡

for the web only

Then for Cell 1

1
2𝐺𝜃 = 𝐴 (𝑎1 𝑞1 − 𝑎12 𝑞2 ) (b)
1

For Cell 2

1
2𝐺𝜃 = 𝐴 (𝑎2 𝑞2 − 𝑎12 𝑞1 ) (c)
2

Equations (a), (b) and (c) are sufficient to solve for 𝑞1 , 𝑞2 and 𝜃.

9. Numerical examples
A thin walled steel section shown in figure is subjected to a twisting moment 𝑻. Calculate the
shear stresses in the walls and the angle of twist per unit length of the box. Figure 𝟕. 𝟏𝟐

Ans. Let 𝐴1 and 𝐴2 be the areas of the cells (1) and (2) respectively.

𝜋𝑎2
∴ 𝐴1 =
2
𝐴2 = (2𝑎 × 2𝑎) = 4𝑎2

For Cell (1),

20
Dr. Naveenkumar D T
 𝑑𝑠
𝑎1 = ∮ (Including the web)
𝑡
𝜋𝑎 + 2𝑎
𝑎1 = ( )
𝑡

For Cell (2),

𝑑𝑠
𝑎2 = ∮
𝑡

2𝑎 2𝑎 2𝑎 2𝑎
= + + +
𝑡 𝑡 𝑡 𝑡
8𝑎
∴ 𝑎2 = ( )
𝑡

For web,

2𝑎
𝑎12 = ( )
𝑡

Now,
For Cell (1),

1
2𝐺𝜃 = (𝑎 𝑞 − 𝑎12 𝑞2 )
𝐴1 1 1
2 (𝜋𝑎 + 2𝑎) 2𝑎
= 2[ 𝑞1 − ( ) 𝑞2 ]
𝜋𝑎 𝑡 𝑡
2𝑎
= [(2 + 𝜋)𝑞1 − 2𝑞2 ]
𝜋𝑡𝑎2
2
∴ 2𝐺𝜃 = [(𝜋 + 2)𝑞1 − 2𝑞2 ]
𝜋𝑎𝑡

For Cell (2),

1
2𝐺𝜃 = (𝑎 𝑞 − 𝑎12 𝑞1 )
𝐴2 2 2
1 8𝑎 2𝑎
= 2 [ 𝑞2 − 𝑞 ]
4𝑎 𝑡 𝑡 1
2𝑎
= 2 [4𝑞2 − 𝑞1 ]
4𝑎 𝑡
1
∴ 2𝐺𝜃 = [4𝑞2 − 𝑞1 ]
2𝑎𝑡

Equating (1) and (2), we get,

21
Dr. Naveenkumar D T
 2 1
[(𝜋 + 2)𝑞1 − 2𝑞2 ] = [4𝑞2 − 𝑞1 ]
𝜋𝑎𝑡 2𝑎𝑡
2 1
or [(𝜋 + 2)𝑞1 − 2𝑞2 ] = [4𝑞2 − 𝑞1 ]
𝜋 2

4
[(𝜋 + 2)𝑞1 − 2𝑞2 ] = [4𝑞2 − 𝑞1 ]
𝜋
4(𝜋 + 2) 8
∴ 𝑞1 − 𝑞2 − 4𝑞2 + 𝑞1 = 0
𝜋 𝜋
4(𝜋 + 2) 8
[ + 1] 𝑞1 − [ + 4] 𝑞2 = 0
𝜋 𝜋
4(𝜋 + 2) + 𝜋 8 + 4𝜋
[ ] 𝑞1 − [ ] 𝑞2 = 0
𝜋 𝜋
or (4𝜋 + 8 + 𝜋)𝑞1 = (8 + 4𝜋)𝑞2
5𝜋 + 8
∴ 𝑞2 = ( )𝑞
4𝜋 + 8 1

But the torque due to shear flows should be equal to the applied torque. i.e., 𝑇 = 2𝑞1 𝐴1 + 2𝑞2 𝐴2
Substituting the values of 𝑞2 , 𝐴1 and 𝐴2 in (3), we get,

𝜋𝑎2 5𝜋 + 8
𝑇 = 2𝑞1 ( ) + 2( ) 𝑞 ⋅ 4𝑎2
2 4𝜋 + 8 1
5𝜋 + 8
= 𝜋𝑎2 𝑞1 + 8𝑎2 ( )𝑞
4𝜋 + 8 1
𝑎2 (𝜋 2 + 12𝜋 + 16)
∴𝑇=[ ] 𝑞1
(𝜋 + 2)
(𝜋 + 2)𝑇
∴ 𝑞1 = 2 2
𝑎 (𝜋 + 12𝜋 + 16)

Now, from equation (1), we have,

2 (𝜋 + 2)𝑇 5𝜋 + 8 (𝜋 + 2)𝑇
2𝐺𝜃 = [(𝜋 + 2) 2 2 − 2( ) 2 2 ]
𝜋𝑎𝑡 𝑎 (𝜋 + 12𝜋 + 16) 4𝜋 + 8 𝑎 (𝜋 + 12𝜋 + 16)

(2𝜋+3)𝑇
Simplifying, we get the twist as 𝜃 = [2𝐺𝑎3 𝑡(𝜋2 +12𝜋+16)]

22
Dr. Naveenkumar D T
Ans. (a) For the box section, we have

𝑇 = 2𝑞𝐴
= 2. 𝜏. 𝑡. 𝐴
𝑇 = 2. 𝜏. 𝑡. 2𝑎 × 𝑎
𝑇
∴𝜏 = 2
4𝑎 𝑡

Now, For solid circular section, we have

𝑇 𝜏
=
𝐼𝑝 𝑟

Where 𝐼𝑝 = Polar moment of inertia

𝑇 𝜏
∴ = 𝑎
𝜋𝑎4
( 32 ) (2)

32𝑇 2𝜏
or 𝜋𝑎4 = 𝑎

16𝑇
∴𝜏=( )
𝜋𝑎3

Equating (a) and (b), we get

𝑇 16𝑇
= ∴ 64𝑎2 𝑡𝑇 = 𝜋𝑎3 𝑇
4𝑎 𝑡 𝜋𝑎3
2
𝜋𝑎
∴𝑡=
64

(b) The stiffness of the box section is given by

23
Dr. Naveenkumar D T
 𝑞 𝑑𝑠
𝜃= ∮
2𝐺𝐴 𝑡
𝑇
Here 𝑇 = 2𝑞𝐴 ∴ 𝑞 =
2𝐴
𝑇 𝑎 2𝑎 𝑎 2𝑎
∴𝜃= [ + + + ]
4𝐺𝐴2 𝑡 𝑡 𝑡 𝑡
6𝑎𝑇
=
4𝐺𝐴2 𝑡
6𝑎𝑇
=
4𝐺(2𝑎2 )2 𝑡
6𝑎𝑇
∴𝜃=
16𝑎4 𝐺𝑡

The stiffness of the Solid Circular Section is

𝑇 𝑇 32𝑇
𝜃= = 4 =
𝐺𝐼𝑝 𝜋𝑎 𝐺𝜋𝑎4
𝐺 ( 32 )

Equating (c) and (d), we get

6𝑎𝑇 32𝑇
=
16𝑎 𝐺𝑡 𝐺𝜋𝑎4
4

6𝑎 32
=
16𝑡 𝜋
6𝜋𝑎
∴𝑡=
16 × 32
3 𝜋𝑎
∴𝑡= ( )
4 64

A two-cell tubular section shown in Fig.Q.8(b) is formed by conventional air voids shapes
having an interior web twisting moment 𝟏𝟎, 𝟎𝟎𝟎 𝐍 − 𝐦. Determine the shear for distribution
and shear stress area of cell 𝟏 = 𝟔𝟖𝟎 𝐜𝐦𝟐 and area of cell 𝟐 = 𝟐𝟎𝟎𝟎 𝐜𝐦𝟐 .

24
Dr. Naveenkumar D T
 (08 Marks)
𝑑𝑆
Ans. For Cell 1, 𝑎1 = ∮ (including the web)
𝑡

67 33
= +
0.06 0.09

therefore, 𝑎1 = 1483.33

For Cell 2 ,

33 63 48 67
𝑎2 = + + +
0.09 0.09 0.09 0.08

Therefore, 𝑎2 = 2437.5

For web,

33
𝑎12 = 0.09 = 366.67

Now, for Cell 1,

1
2𝐺𝜃 = (𝑎 𝑞 − 𝑎12 𝑞2 )
𝐴1 1 1

1
= (1483𝑞1 − 366𝑞2 )
680

Therefore, 2𝐺𝜃 = 2.189𝑞1 − 0.54𝑞2 For Cell 2,

25
Dr. Naveenkumar D T
 1
2𝐺𝜃 = (𝑎 𝑞 − 𝑎12 𝑞1 )
𝐴2 2 2
1
= (2409𝑞2 − 366𝑞1 )
2000

Therefore, 2𝐺𝜃 = 1.20𝑞2 − 0.18𝑞1

Equating (i) and (ii), we get

2.18𝑞1 − 0.54𝑞2 = 1.20𝑞2 − 0.18𝑞1

or 2.36𝑞1 − 1.74𝑞2 = 0
or 𝑞2 = 1.36𝑞1
The torque due to shear flows should be equal to the applied torque

Hence, from Equation (a),

𝑀𝑡 = 2𝑞1 𝐴1 + 2𝑞2 𝐴2
10,000 × 100 = 2𝑞1 × 680 + 2𝑞2 × 2000
= 1360𝑞1 + 4000𝑞2

Substituting for 𝑞2 , we get

10000 × 100 = 1360𝑞1 + 4000 × 1.36𝑞1

Therefore,

𝑞1 = 147 N and 𝑞2 = 200 N

A two cell tube as shown in the Fig. Q8 (b) below is subjected to a torque of 𝟏𝟎𝐤𝐍 − 𝐦.
Determine the shear stress in each part and angle of twist per metre length. Take modulus of
rigidity of the material as 𝟖𝟑𝐤𝐍/𝐦𝐦𝟐 .

Ans. For Cell 1

26
Dr. Naveenkumar D T
Area of the Cell = 𝐴1 = 150 × 100 = 15000 mm2

𝑑𝑠
𝑎1 =∮ (including web)
𝑡
150 100 150 100
= + + +
5 5 2.5 5
= 130

For Cell 2
1
Area of the cell = 𝐴2 = × 150 × √(125)2 − (75)2
2
= 7500 mm2

𝑑𝑠
∴ 𝑎2 =∮ (including web)
𝑡
150 125 125
= + +
2.5 2.5 2.5
∴ 𝑎2 = 160

For the web,

150
𝑎12 = = 60
2.5

For Cell (1)

1
2𝐺𝜃 = (𝑎 𝑞 − 𝑎12 𝑞2 )
𝐴1 1 1
1
∴ 2𝐺𝜃 = (130𝑞1 − 60𝑞2 )
15000

For Cell (2)

1
2𝐺𝜃 = (𝑎 𝑞 − 𝑎12 𝑞1 )
𝐴2 2 2
1
= (160𝑞2 − 60𝑞1 )
7500

Equating (a) and (b), we get

1 1
(130𝑞1 − 60𝑞2 ) = (160𝑞2 − 60𝑞1 )
15000 7500

Solving, 𝑞1 = 1.52𝑞2

Now, the torque due to shear flows should be equal to the applied torque.
27
Dr. Naveenkumar D T
 i.e., 𝑀𝑡 = 2𝑞1 𝐴1 + 2𝑞2 𝐴2
10 × 106 = 2𝑞1 (15000) + 2𝑞2 (7500)

Substituting (c) in (d), we get

10 × 106 = 2 × 15000(1.52𝑞2 ) + 2𝑞2 (7500)

∴ 𝑞2 = 165.02𝑁

∴ 𝑞1 = 1.52 × 165.02 = 250.83 N

Shear flow in the web = 𝑞3 = (𝑞1 − 𝑞2 ) = (250.83 − 165.02)

∴ 𝑞3 = 85.81 N

𝑞1 250.83
∴ 𝜏1 = = = 50.17 N/mm2
𝑡1 5
𝑞2 165.02
𝜏2 = = = 66.01 N/mm2
𝑡2 2.5
𝑞3 85.81
𝜏3 = = = 34.32 N/mm2
𝑡3 2.5

Now, the twist 𝜃 is computed by substituting the values of 𝑞1 and 𝑞2 in equation (a)

1
i.e., 2𝐺𝜃 = 15000 [130 × 250.83 × 60 × 165.02]

1 22706.7
∴ 𝜃 = 15000 × 83×1000 = 1.824 × 10−5 radians / mm length or 𝜃 = 1.04 degrees/m length

A hollow alluminium tube of rectangular C/S shown in below Fig.Q.8(a) is subjected to a

torque of 𝟓𝟔, 𝟓𝟎𝟎 𝐍-m along longitudinal axis. Determine the shearing stress and the angle
of twist. Assume 𝑮 = 𝟐𝟕. 𝟔 × 𝟏𝟎𝟗 𝐍/𝐦𝟐

Ans. The above figure shows the membrane surface 𝐴𝐵𝐶𝐷

28
Dr. Naveenkumar D T
Now, the Applied torque = 𝑀𝑡 = 2𝑞𝐴

56,500 = 2𝑞(0.5 × 0.25)

56,500 = 0.25𝑞

hence, 𝑞 = 226000 N/m.

Now, the shearing stresses are

𝑞 226000
𝜏1 = = = 18.833 × 106 N/m2
𝑡1 0.012
𝑞 226000
𝜏2 = = = 37.667 × 106 N/m2
𝑡2 0.006
226000
𝜏3 = = 22.6 × 106 N/m2
0.01

Now, the angle of twist per unit length is

𝑞 𝑑𝑠
𝜃= ∮
2𝐺𝐴 𝑡

Therefore,

226000 0.25 0.5 0.25

𝜃= 9
[ + (2) + ]
2 × 27.6 × 10 × 0.125 0.012 0.006 0.01

or 𝜃 = 0.00696014rad/m

A hallow circular torsion member has an outside diameter of 𝟐𝟐 𝐦𝐦 and inside diameter of
𝟏𝟖 𝐦𝐦, wlith mean diameter 𝐃 = 𝟐𝟎 𝐦𝐦 and 𝐭/𝐃 = 𝟎. 𝟏𝟎. Calculate the torque and angle
of twist per unit length if shearing stress at mean diameter is 𝟕𝟎𝐌𝐏𝐚. Calculate these values
if a cut is made through the wall thickness along the entire length. 𝐆 = 𝟕𝟕. 𝟓𝐆𝐏𝐚.

𝜋𝐷2 𝜋x400
𝐴= = = 100𝜋mm2
4 4

𝐭
= 𝟎. 𝟏𝟎
𝐃

𝑇 = 2𝑞𝐴
= 2. 𝜏. 𝑡. 𝐴
𝜋
𝑇 = 2x70x(0.1x20)x 4 (202 ) = 87976 N − mm

29
Dr. Naveenkumar D T
 𝑞 𝑑𝑠
𝜃= ∮
2𝐺𝐴 𝑡

Therefore,

70x0.1x20 𝜋𝑥20
𝜃= 𝜋 [ ]
2x77500x 4 (202 ) 0.1𝑥20

rad rad
or 𝜃 = 0.00009 mm = 0.09 m

When a cut is made, we have the torsion member becomes a narrow rectangle of width

πD 0.1×20
𝑏= = 10𝜋 mm, ℎ = 𝑡/2 = 0.1𝐷/2 = = 1 mm
2 2

𝜏max
𝜃 =
2𝐺ℎ
70
= = 0.00045 rad/mm = 0.45N/m
2 × 77.5 × 103 × 1
8𝑏ℎ2 𝜏max
𝑇 =
3
8 × 10𝜋 × 12 × 70
= = 5865 N. mm = 5.865 N m
3

30
Dr. Naveenkumar D T
 31
Dr. Naveenkumar D T