CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Convergence with coupled implicit solver (http://www.cfd-online.com/Forums/fluent/38424-convergence-coupled-implicit-solver.html)

Henrik Ström October 28, 2005 08:47

Convergence with coupled implicit solver
 
Hello!

I have an unsteady simulation that I have run earlier versions of using the segregated solver, and that worked just fine. However, now I want to add a very fast chemical reaction to my system, which will make the problem stiff. This causes the segregated solver to have problems as the reaction starts after a few hundred timesteps. I therefore want to change to the coupled (implicit) solver and also possibly use the special stiff-chemistry ODE solver that is built-in in the coupled solver.

My problem is, though, that when I change to the coupled solver, the solution diverges already in the first timestep (that is, before the reaction even takes place and for a situation where the segregated solver still works OK)... So, what can I do to get it to work? I have tried very low Courant numbers, but that does not seem to help (only slow down the rate at which it diverges). Are there any other parameters that I should experiment with? (FAS in the explicit formulation is not applicable since it is an unsteady simulation).

Any help or hints is greatly appreciated!

Thanks, Henrik

nasser October 29, 2005 03:57

Re: Convergence with coupled implicit solver
 
Hi 1-use very small time step for first itration 2-use very small cournt number in first 3-enable stiff model 4- refine your mesh goodbye


All times are GMT -4. The time now is 03:01.