You are on page 1of 2

3 TORSION

Most students will already be familiar with the simple theory of


torsion of circular shafts, which is based on the simplifying
assumption that plane cross-sections of the shaft remain plane and
parallel to one another during straining. In turn, this means that
radial straight lines drawn on the cross-section remain straight
during straining, and that the shear strain is proportional to
radius.

A little thought will show that the above assumption simply is


not true for shafts of non-circular cross-section. For example, a
square prism of rubber when twisted in the hands would have its
originally plane end faces warped, showing that out of plane
displacements had resulted from the twisting. Torsion theory for
a three-dimensional body which allows for 'warping' of the cross-
section is developed here. A set of displacements is assumed which
means that compatibility conditions need not be considered, and
after satisfying the equilibrium equation and boundary conditions a
full solution is obtained. The original solution is due to St.
Venant.

3.1 Displacements, Strains and Stresses


y
y

Fig. 3.1

The shaft under consideration is assumed to be prismatic. The


origin of coordinates is at the fixed end and a couple T is applied
to the free end in the sense illustrated in figure 3.la. The angle
of twist per unit length is a so that the total twist for a shaft
of length i is ia. Consider a cross-section of the shaft at z from
the origin so that the angular displacement of a line OA drawn on
the cross-section is az, see figure 3.lb. The displacement of A .is
to A' where AA' is raz and the x component of displacement is

u = -AA'sine = -azy
74

P. R. Lancaster et al., Advanced Solid Mechanics


© P.R. Lancaster and D. Mitchell 1980
since

sine .L
r

Similarly the y component of AA' is

v = AA'cose = azx
since
X
cose
r

The displacement w in the z direction is less easy to specify.


However, it seems reasonable to assume that w will be independent
of z but will depend on position within the cross-section. It is
also reasonable to assume that w will depend directly on the angle
of twist.

Therefore, the full set of displacements within the boundary of


the shaft are assumed to be

u = -azy
v = azy

w = al/J(x, y)
) (3.1)

where 1/J = 1/!(x, y) is some function of x andy (to be determined) and


is called the warping function. Using these displacements and the
definitions of chapter 1, the strains are easily deduced as

au 0, e av = 0 e aw = 0
e ay , z az
X ax y

au :lv
Yxy -+ 0
ay ax
aw= a(:ll/J - y)
:lu+ -

)
Yxz = -
:lz ax :lx
(3. 2)
= :lv + :lw = a(~ + x)
Yyz az :ly :ly

and the non-zero stresses are simply

Gy XZ Ga(~ - y)

}
T
xz ax
(3. 3)

T Gyyz Ga(~
:ly
+ x)
yz

75

You might also like