[niels]

I'm programing on direct-agglomeration for multigrid on unstructured grid, now using NACA0012 for 2D test case. It works well for invicid transonic flow, but when considering viscous flow and turbulence model, just first order solver works, second order doesn't converge.
Prolong method is volume average, just first order method is used for interpolation.
How can I get the convergence result for second order flow solver, is there any good suggestion?
Thanks

[ogloth]

Hi,

there is a number of things which might help. First a question: do you have an anisotropic boundary layer mesh? If yes, did you only coarsen the mesh in one direction until you reach a roughly isotropic grid (semi-coarsening)?

A pragmatic approach is to only use a second order discretisation on the fine grid and a more "convergence friendly" scheme on the agglomerated levels. Once the solver has converged, you will end up with the fine grid solution (i.e. a second order solution).

'hope this helps a little!

-- Oliver