DOI: 10.1007/s10915-005-9028-x
Journal of Scientific Computing, Vol. 27, Nos. 1–3, June 2006 (© 2005)
Pseudo Spectral Methods Applied to Problems
in Elasticity
Chris Talbot
1
and Andrew Crampton
1
Received October 14, 2004; accepted (in revised form) September 13, 2005; Published online January 4, 2006
Pseudo spectral methods offer an attractive alternative to finite element proce-
dures for the solution of problems in elasticity. Especially for simple domains,
questions involving both two and three dimensional elasticity (Navier’s Equa-
tions or their non-linear generalisations) would seem to be reasonable candidates
for a pseudo spectral approach. This paper examines some simple vibrational
eigenvalue type problems, demonstrating how Navier’s equations can be recast
into pseudo- spectral format, including first derivative boundary conditions
representing zero traction. Fourier–Chebyshev methods are shown to give solu-
tions with typical spectral accuracy, with the addition of pole conditions being
necessary for the case of a two dimensional disc. There is also consideration
given to time-stepping solutions of elastodynamic problems, especially those
involving non-linear friction effects, the authors particular interest being the
study of disc brake noise. It is shown that, at least for relatively simple cases, it
is possible to model systems in such a way that animated graphical output can
be provided as the system of partial differential equations is numerically inte-
grated. This provides a useful tool for engineers to rapidly examine the effect
of parameter changes on a system model.
KEY WORDS: Pseudo spectral; elasticity; brake noise; vibration.
1. INTRODUCTION
For simpler problems on regular domains, problems in solid mechanics
can be modelled by Fourier–Chebyshev collocation methods that enable
both eigenvalue and time-stepping problems to be solved efficiently with
simple code. Application of pseudo-spectral (PS) methods to elasticity,
using the 2D or 3D Navier equations with periodic boundary conditions,
1
School of Computing and Engineering, University of Huddersfield, Huddersfield, UK.
E-mail: c.j.talbot@hud.ac.uk
443
0885-7474/06/0600-0443/0 © 2005 Springer Science+Business Media, Inc.