I have been working on the problem of natural convection in a 3d rectangular cavity and I have been using a Galerkin method whose trial and test functions are divergence free ans verify that are zero at the boundary. In order to get these trial functions I use the beam functions (Chandrasekar functions). The problem is that as they appear as eigenvectors of a regular Sturm-Liouville problem (instead of a singular one) I think that the convergence of the Galerkin method wouldn't be good enough. Does anybody know how to expand the 3d velocity using divergence free functions which involve Chebyshev or Legendre polynomials? Thanks in advance Dolors Puigjaner

You should contact

Jonas Holdeman <jtholdem@usit.net>

Jonas has similar interests, and I think he has some results that may help you.

Good Luck!