Type of solver and convergence problem

Hi there,

I'm trying to validate the RANS solver by doing a simple NACA0012 airfoil in a transonic flow. This case is different from the tutorial because i'm using a mesh with boundary layer and I'm also using Spalart-Almaras model. The mesh is provided by the NASA and theoretically, convergence should be ideal.

Here my questions:

1) What type of solver Caedium is using when the option "Compressible flow" is turned on? It is a compressible scheme (rhoCentralFoam) or it still based on the pressure ?

2) Is there a trick to help convergence ? Even in linear upwind scheme with relaxation factors reduced at minimum, I'm still diverging after 1400 iterations. Before the divergence, cp curve looks good but it diverge anyway.

3) Where can I found the info correlating Caedium and openfoam, i.e. that the function I turn "on" in Caedium is equal to a specific command line in openfoam.


Transonic Flow

In answer to your questions:

1) Subsonic and Transonic (i.e., boundary conditions with subsonic velocity) steady-state simulations use rhoSimpleFoam, which is a pressure-based solver.

2) If you are using high aspect ratio cells (say great than 5) within the boundary layer then that can cause convergence problems. I have found the k-omega SST turbulence model with wall functions to be the most robust configuration. If with the Spalart-Almaras turbulence model you are using y+ values ~= 1 then that can cause convergence issues.

3) We do not document the relationship between Caedium and OpenFOAM settings. In the main the options presented in Caedium are similar in name and function to those in OpenFOAM. However, there is not always a one-to-one correspondence, e.g., a low-Reynolds number turbulence model uses different boundary conditions compared to a wall function turbulence model. If you want to see the explicit OpenFOAM case then use File->Export to export to OpenFOAM (.ofm).

For another perspective see "Tutorial: transonic flow over NACA 0012".