too bad convergence
 

Hi everybody, I'm in trouble with a finite difference code to resolve a Poisson equation in a channel.

I have a 3D problem with Neumann condition in one direction and the others are periodic.

In regular mesh, it works perfectly but as soon as I use an stretched mesh ( even very very smooth ), my SOR ( or other relaxtion method ) converge on large residual ( about 0.1 ).

I have no idea how to get a good convergence.

Thanks for your help.

Re: too bad convergence
 

Have you seen this?

http://www.cfd-online.com/Forum/main...cgi/read/18519

Re: too bad convergence
 

Thanks for your response, I've already seen it but it doesn't correspond to my problem. In my case, Pressure equation is inviscid, and it doesn't converge ( nor diverge )even for very smooth stretch.