[ponghe]

Hi Everyone!

I have been running a RANS k-epsilon simpleFOAM case with a porous media using the solver rhoPorousSimpleFoam. I generally do not have trouble getting all residuals in the 1e-4 range. For whatever reason, the pressure residual is just not coming down in the latest case that I am working on. The mesh did fail one check but did not cause much issues during the run. The checkMesh log is as follow:

#############################################
Checking geometry...
Overall domain bounding box (-2.38773e-07 0 -1.0354e-06) (1.1684 4.5847 5.0927)
Mesh (non-empty, non-wedge) directions (1 1 1)
Mesh (non-empty) directions (1 1 1)
Boundary openness (-2.663151e-15 -1.4056313e-14 -3.7989542e-15) OK.
Max cell openness = 4.6642222e-16 OK.
Max aspect ratio = 14.440033 OK.
Minimum face area = 4.2669832e-08. Maximum face area = 0.00018051869. Face area magnitudes OK.
Min volume = 1.5063985e-10. Max volume = 2.1712054e-06. Total volume = 5.5016469. Cell volumes OK.
Mesh non-orthogonality Max: 65.716777 average: 10.440301
Non-orthogonality check OK.
Face pyramids OK.
***Max skewness = 12.461076, 270 highly skew faces detected which may impair the quality of the results
<<Writing 270 skew faces to set skewFaces
Coupled point location match (average 0) OK.

Failed 1 mesh checks.
#############################################

The residual plot is as follow which shows p being much higher than other scalars:



Anyone have experience or had encountered similar situation before? Much appreciated if you can let me know!

Thanks!

[Gerry Kan]

Hi Ponghe:

The residual behavior looks rather healthy, though I would not use only k and epsilon alone as comparison. How did the residuals for Ux, Uy, and Uz evolve over the iterations?

I take it this is a polyhedral mesh (as oppoed to Cartesian)? Depending on the complexity of your geometry, getting 1e-3 / 1e-4 for residuals is reasonable, though pressure could be a little bit lower (say, just below 1e-3). My hunch is a group of cells (perhaps the highly skewed group reported) refused to converge.

In addition to residuals, you should also add probe points at key regions and monitor whether the flow characteristics converged or oscillating, which in some cases could imply a grid problem, when such oscillation is not expected.

Hope that helps, Gerry.

[ponghe]

Thanks Gerry! This is very helpful. The Ux/Uy/Uz seem reasonable between 1e-3 and 1e-4. I'll try to add probe points to see if the flow converges locally.

Thanks again.