[Zen]

Hello,

When I try to use the Roe-turkel scheme to accelerate convergence the solution does not converge, whereas if I use a normal Roe scheme without preconditioning I get a converged solution but after a long time.

In particular, if I use FGMRES I get the following error:

!!! Error: The FGMRES solver has diverged. Now exiting... !!!

If I use BCGSTAB I get non-phyiscal points in the solution and the solver stops.

Any idea why?

Z

[hlk]

Originally Posted by Zen

Hello,

When I try to use the Roe-turkel scheme to accelerate convergence the solution does not converge, whereas if I use a normal Roe scheme without preconditioning I get a converged solution but after a long time.

In particular, if I use FGMRES I get the following error:

!!! Error: The FGMRES solver has diverged. Now exiting... !!!

If I use BCGSTAB I get non-phyiscal points in the solution and the solver stops.

Any idea why?

Z

Thank you for your interest in SU2
It is difficult to tell from little information, but you may want to check the quality of your mesh. It is also a good idea to double-check the boundary conditions.
Please refer to the test cases (the su2code/TestCases github repository) for examples of config files for various conditions.