CFD Online Discussion Forums - Schemes and Solvers for skewed meshes

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)

- OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)

- - Schemes and Solvers for skewed meshes (https://www.cfd-online.com/Forums/openfoam-solving/163200-schemes-solvers-skewed-meshes.html)

Schemes and Solvers for skewed meshes

Consider me a newbie in OpenFoam.

I understand OpenFoam gives pretty good convergence on nice meshes with low skewness and low aspect ratio. But we cannot avoid skewed meshes when the geometry gets more complicatied and when we have O grid Face Cuts. Nether can we always have low aspect ratio hexa/prism cells in the boundar layer due to not enough CPU power or other reasons

In this case. What are the schemes that are suggested? Do they differ when we move from steady state solvers to unsteady solvers?

I am using Limiters for gradients every where 1) the gradient scheme CellLimited leastSquares 1 for both Pressure and Velocity gradients. The velocity Field converges easily when the divergence term of Phi,U is set to limitedLinear 1.

Interpolation Scheme: default linear ,
SnGradscheme: Corrected
and Laplcian terms Gauss linear Corrected.

My questions are:

1) What is the ideal settings for the best Convergence of Pressure equation when you have a skewed mesh ( Max Skewness > 70)

2) Does the choice of Solver PCG or GAMG effect the solution? I always found recommendation to use GAMG but PCG is giving better convergence in my case.

Quick reply: I don't have explicit answers, but I vaguely remember being part of or reading about some discussion threads on this topic. After a bit of Googling:

Quote:

Originally Posted by wyldckat (Post 576483)

Quick reply: I don't have explicit answers, but I vaguely remember being part of or reading about some discussion threads on this topic. After a bit of Googling:

Thanks for the threads. I have been able to improve the mesh and achieve convergence.
However, the new problem I am facing is the time step size. Beyond Stall Angle for an Airfoil with Vortex shedding at High Reynolds Number the time step required for accurate convergence is about ( 1e-4 to 1e-07) seconds i.e. ( courant number < 5). I recall that in Commercial Softwares I have used 0.01 seconds to 0.2 seconds to capture the same phenomenon.
On the other hand, I do believe that open source softwares does not come with everything that you dreamed for.

I have been using pisoFoam and pimpleFoam for my simulations. I would like to believe that may be some one has brought alterations/implementations to pimple algorithm so that it can support courant number > 5 ? If so can any one suggest links, references, codes? or if not, can any one guide me to a implicit stable method may be I will try implement it.

Regards

Shereez