convergence in cylindrical coordinates
 

Hi all! I have a convergence problem in my code about 3D cylindrical coordinate and compressible unsteady flow.I use the PISO algorithm. In the pressure term d(P)/d(thita)/r the problem is the r term.In the begining I have forgotten this term and the pressure term was just d(p)/d(thita) and the code had no problem.But when I discovered this mistake and I fixed it, the code does not converge.In fact the residuals remain around 1-2 (very large) and sometimes they become extremely large. I have checked the pressure correction part but I haven't found any mistake.Has anybody faced something like that? Any suggestions? Thank you

Re: convergence in cylindrical coordinates
 

First you should verify that d(P)/d(thita)->0 when r->0, you can verify this by trying to use d(P)/d(thita)/(r+epsilon) with epsilon a small number (10-3 or simply your radial grid size near the axis of symmetry)...If you have good results then it was the origin of the problem. Then perform calculation with smaller and smaller epsilon...

secondly tries to give more freedom to the pressure (high under-relaxation factor) and low freedom for the velocity... it often helps in flow driven by a rotating walll...please try.