CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Understanding pimpleFoam convergence criterion (http://www.cfd-online.com/Forums/openfoam-solving/114601-understanding-pimplefoam-convergence-criterion.html)

Nucleophobe March 13, 2013 19:46

Understanding pimpleFoam convergence criterion
 
Hi, I have a question about convergence criteria for pimpleFoam. I read other relevant threads but couldn't find a clear answer.

My understanding is that with the simpleFoam solver, it is necessary to run until the 'p' residuals drop at least three orders of magnitude (1e-3) and the 'U' (Ux, Uy, & Uz) residuals drop about four to six orders of magnitude (1e-4 - 1e-6), then stop changing.

I'm now using pimpleFoam to obtain a transient solution (different boundary conditions). My understanding is that in this case, both the 'p' and 'U' residuals must drop around one order of magnitude per time step.

My questions:
1) Should the initial residuals drop below a certain threshold before the transient solution is considered accurate? I saw in other threads that this was mentioned, but I saw no discussion on how this criterion is determined. If this is true, are pimpleFoam solutions always wrong during start-up? I don't really understand this.
2) How do you determine the criterion for how much the residuals should drop per time step?

Thank you community!
-Nuc


All times are GMT -4. The time now is 16:00.