Introduction to Elasticity/Torsion of noncircular cylinders
From Wikiversity
Contents |
[edit] Torsion of Non-Circular Cylinders
[edit] About the problem
- Solution first found by St. Venant.
- Tractions at the ends are statically equivalent to equal and opposite torques
. - Lateral surfaces are traction-free.
[edit] Assumptions:
- An axis passes through the center of twist (
axis). - Each c.s. projection on to the
plane rotates,but remains undistorted. - The rotation of each c.s. (
) is proportional to
.
where
is the twist per unit length.
- The out-of-plane distortion (warping) is the same for each c.s. and is proportional to
.
[edit] Find:
- Torsional rigidity (
). - Maximum shear stress.
[edit] Solution:
[edit] Displacements
where
is the warping function.
If
(small strain),
[edit] Strains
Therefore,
[edit] Stresses
Therefore,
[edit] Equilibrium
Therefore,
[edit] Internal Tractions
- Normal to cross sections is
. - Normal traction
. - Projected shear traction is
. - Traction vector at a point in the cross section is tangent to the cross section.
[edit] Boundary Conditions on Lateral Surfaces
- Lateral surface traction-free.
- Unit normal to lateral surface appears as an in-plane unit normal to the boundary
.
We parameterize the boundary curve
using
The tangent vector to s is
The tractions
and
on the lateral surface are identically zero. However, to satisfy the BC
, we need
or,
[edit] Boundary Conditions on End Surfaces
The traction distribution is statically equivalent to the torque
. At
,
Therefore,
From equilibrium,
Hence,
[edit] The Green-Riemann Theorem
If
and
then
with the integration direction such that S is to the left.
Applying the Green-Riemann theorem to equation (17), and using equation (16)
Similarly, we can show that
.
since
.
The moments about the
and
axes are also zero.
The moment about the
axis is
where
is the torsion constant. Since
, we have
If
, then
, the polar moment of inertia.
[edit] Summary of the solution approach
- Find a warping function
that is harmonic. and satisfies the traction BCs. - Compatibility is not an issue since we start with displacements.
- The problem is independent of applied torque and the material properties of the cylinder.
- So it is just a geometrical problem. Once ψ is known, we can calculate
- The displacement field.
- The stress field.
- The twist per unit length.















![\begin{align}
\nabla^2{\psi} = 0 ~~\Rightarrow~~~
\psi_{,1}-x_2 & = (\psi_{,1}-x_2) + x_1(\psi_{,11} + \psi_{,22}) \\
& = \psi_{,1} + x_1\psi_{,11} - x_2 + x_1\psi_{,22} \\
& = (x_1\psi_{,1} - x_1x_2)_{,1} + (x_1\psi_{,2} + x_1x_1)_{,2} \\
& = \left[x_1(\psi_{,1} - x_2)\right]_{,1} +
\left[x_1(\psi_{,2} + x_1)\right]_{,2}
\end{align}](http://upload.wikimedia.org/math/b/5/7/b57d297680179ed36c459a903ab6ce30.png)
![\text{(17)} \qquad
F_1 = \mu\alpha\int_S\left[x_1(\psi_{,1} - x_2)\right]_{,1} +
\left[x_1(\psi_{,2} + x_1)\right]_{,2} dS](http://upload.wikimedia.org/math/f/b/a/fbac35ffdbd15cdfba233818a8573765.png)



