Page 1 of 1

The Error when running the simulation

Posted: Mon Sep 28, 2015 11:39 am
by MartianDuan
Hello Users,

When I wan trying to run the simulation withe very thin viscous layer. K-omega-sst model was chosen and the inner cirulation was set at the inlet.

I came across the following error:
/home/martian/APP.SRC/code_saturne-4.0.2/src/alge/cs_multigrid.c:2780: Fatal error.

algebraic multigrid [Pressure]: divergence after 21 cycles:
initial residual: 2.5793e-03; current residual: 1.3719e+02.

The mesh is generated by using the Salome 7.6.0. The total height of the viscouse layer is 0.0005m, containing 15 layers with the expansion ratio as 1.2.

I did not come across the similar error when the total height of visocus layer is thicker than 0.001m. However the simualtion diverges as the yplus is higher than 1.

I would like to understand what the error is about. It si related to my mesh quality or the setting in the simualtion or both.

The listing log is attached.

Waiting for your reply.

Thanks in advance,

Yu.

Re: The Error when running the simulation

Posted: Tue Sep 29, 2015 6:55 pm
by Yvan Fournier
Hello,

The issue is probably related to your mesh, but maybe also to the CFL value. I do not see the CFL values in you listing: either you deactivated their logging, or it is not activated by default when not using the GUI. Using the GUI for this at least would make it easier to check if this value is too high or not.

Did you check your mesh quality ?

In your log, there is the following message:
@
@ @@ WARNING: Wall distance calculation
@ =========
@ The laplacian solution does not respect the maximum
@ principle. We recompute the laplacien without
@ reconstructions.
I'm not sure which option to change for this (need to check), but this happens when the mesh quality is bad, especially when you have cells that are both thin and warped (which is basically unavoidable with very thin boundary layer cells on curved surfaces, unless refinement is also done tangentially in high curvature areas).

Regards,

Yvan