CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   convergence in pimpleFoam (http://www.cfd-online.com/Forums/openfoam/97164-convergence-pimplefoam.html)

Andrea_85 February 10, 2012 06:29

convergence in pimpleFoam
 
Hi all,
I normally work with multiphase flow and interFoam, but i need some single phase simulations of flow in a channel with obstacles. My geometry is very simple: inlet, outlet and walls (2D). The simulation type is laminar. I'm using fixedVelocity at the inlet, fixedPressure at the outlet and zero velocity on the walls, the others BC are set to zeroGradient. I found less convergence to a stable solution when the injection velocity is lower. For u=1e-2 m/s and u=1e-3 the solution becomes stable in a short time and the residuals are good, but for u=1e-4 m/s the pressure solution is oscillating and residual doesn't converge. Any ideas of why? at most I would expect the opposite.


regards
andrea

niaz February 10, 2012 07:57

Dear Andrea
you decrease your velocity at inlet so the field of velocity and pressure, which you have solved, have lower value. you should decrease you residual limit one or two orders. check your fvsolution.

Andrea_85 February 10, 2012 10:59

Hi and thanks for reply.
this is my fvSolution for u=1e-4 m/s

U
{
solver PBiCG;
preconditioner DILU;
tolerance 1e-9;
relTol 0.001;
}

UFinal
{
solver PBiCG;
preconditioner DILU;
tolerance 1e-9;
relTol 0;
}


p
{
solver GAMG;
tolerance 1e-9;
relTol 0.001;
smoother GaussSeidel;
cacheAgglomeration true;
nCellsInCoarsestLevel 10;
agglomerator faceAreaPair;
mergeLevels 1;
}

pFinal
{
solver GAMG;
tolerance 1e-9;
relTol 0;
smoother GaussSeidel;
cacheAgglomeration true;
nCellsInCoarsestLevel 10;
agglomerator faceAreaPair;
mergeLevels 1;
}



}

PIMPLE
{
nOuterCorrectors 2;
nCorrectors 2;
nNonOrthogonalCorrectors 0;
pRefCell 0;
pRefValue 0;
}


I've obtained stable solution for u=1e-2 m/s after 0.06s with tolerance set to 1e-5. I would expect almost immediate convergence with u=1e-4, or not? lower tolerance has the effect of making the pressure solution more stable (smaller or no oscillation) but after 5s of simulations the resisual are still high. What is your opinion?



best
andrea

niaz February 10, 2012 14:20

Dear Andrea
if you found your stable solution when you set residuals in 10e-9 for u=10e-2 and 10e-3. it is better to reduce your residual to 10e-11 or -12 for u=10e-4.
try it

Andrea_85 February 10, 2012 15:23

Dear niaz,
sorry i have been not clear. I found stable solution for u=1e-2/1e-3 setting residuals to 1e-5, so i guess 1e-9 is ok for u=1e-4. btw the solution doesn't converge even with 1e-12.
it's really strange!:confused:

regards
andrea

niaz February 11, 2012 03:32

Dear Andrea
run your test case and save your residual report in a file them check that the residuals decrease to your settings or not?
if not, change the value of loop corrector in the fvsolution.


All times are GMT -4. The time now is 14:43.