TOPICS
Search

Warping Function


A warping function in Saint-Venant torsion is a harmonic function on a plane cross section that describes axial displacement during uniform twisting of a homogeneous, isotropic, linearly elastic prismatic shaft. If beta is the twist per unit length, the axial displacement is betaw(x,y) (Lipton 1998, pp. 86-87). The warping function is determined by the geometry and enters the calculation of torsional rigidity.

Let Omega be a bounded, connected plane domain whose boundary consists of disjoint smooth curves, and let n=(n_x,n_y) be its outward unit normal vector. The defining boundary-value problem is

del ^2w=0 in Omega
(1)
(partialw)/(partialn)=yn_x-xn_y on partialOmega,
(2)

where del ^2 is the two-dimensional Laplacian and partialw/partialn=del w·n. The Neumann boundary conditions express the absence of traction on the lateral surface (Lipton 1998, p. 87). They apply to every boundary component, including those surrounding holes. The divergence theorem gives the compatibility condition

 int_(partialOmega)(yn_x-xn_y)ds=int_Omega((partialy)/(partialx)-(partialx)/(partialy))dxdy=0.
(3)

A solution exists and is unique up to an additive constant. The normalization

 int_Omegawdxdy=0
(4)

selects a unique warping function.

The variational characterization uses the functional

 E[v]=int_Omega[((partialv)/(partialx)-y)^2+((partialv)/(partialy)+x)^2]dxdy.
(5)

The warping function minimizes E[v] over the Sobolev space H^1(Omega) of square-integrable real-valued functions whose first weak partial derivatives are also square integrable. No boundary values are prescribed in this minimization. The minimum is the geometric multiplier K in the torsional rigidity GK, where G is the shear modulus (Lipton 1998, p. 89). Its weak formulation is

 int_Omegadel w·del vdxdy=int_Omega(y(partialv)/(partialx)-x(partialv)/(partialy))dxdy
(6)

for every v in H^1(Omega). Expanding the squares and using this identity yields

 E[w+v]=E[w]+int_Omega||del v||^2dxdy.
(7)

This proves minimality and shows that all minimizers differ by a constant. In particular, testing with v=w gives

 K=E[w]=I_p-int_Omega||del w||^2dxdy,
(8)

where I_p=int_Omega(x^2+y^2)dxdy is the polar area moment of inertia about the origin. Consequently K<=I_p, with equality iff w is constant. Any trial function v in H^1(Omega) therefore gives the upper bound K<=E[v].

For a disk x^2+y^2<a^2 centered at the origin, the normal derivative is zero and the normalized solution is w=0. Thus

 K=(pia^4)/2.
(9)

The same solution holds for a concentric annulus b^2<x^2+y^2<a^2, with 0<b<a, giving

 K=(pi(a^4-b^4))/2.
(10)

These cross sections undergo no axial warping. For the ellipse x^2/a^2+y^2/b^2<1, with positive semiaxes a and b, the solution is

 w(x,y)=(b^2-a^2)/(a^2+b^2)xy,
(11)

and its energy is

 K=(pia^3b^3)/(a^2+b^2).
(12)

These formulas agree with the displacement and torsional rigidity given by Timoshenko and Goodier (1951, pp. 264-265), after dividing their axial displacement by beta. For a!=b, the nonzero level curves of w lie on rectangular hyperbolas. The circular limit a=b recovers w=0.

The harmonic displacement warping function differs from the Prandtl stress function used in the Poisson problem formulation of torsional rigidity. For a simply connected domain, define the normalized Prandtl stress function f by

(partialf)/(partialx)=-2((partialw)/(partialy)+x)
(13)
(partialf)/(partialy)=2((partialw)/(partialx)-y).
(14)

Then del ^2f=-4, and the additive constant can be chosen to impose zero Dirichlet boundary conditions. With this normalization K=int_Omegafdxdy. For a domain with holes, f is constant on each boundary component, but these constants need not be equal.


See also

Harmonic Function, Neumann Boundary Conditions, Prandtl Stress Function, Torsional Rigidity, Weak Formulation

Explore with Wolfram|Alpha

References

Lipton, R. "Optimal Fiber Configurations for Maximum Torsional Rigidity." Arch. Rational Mech. Anal. 144, 79-106, 1998. https://doi.org/10.1007/s002050050113.Timoshenko, S. and Goodier, J. N. Theory of Elasticity, 2nd ed. New York: McGraw-Hill, 1951.

Cite this as:

Weisstein, Eric W. "Warping Function." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/WarpingFunction.html

Subject classifications