I was just wondering what everyone else is using for discretisation schemes for their coupled models and what kind of performance you are getting? The "standard" ones that come from the tutorial cases (and also often found in OF-related literature for pure CFD cases) are:
gradSchemes: Gauss linear
divSchemes: Gauss linear or Gauss limitedLinear 1
laplacianSchemes: Gauss linear corrected
For my situation, I have convection driven flow, with some decent contribution from gravity on the particles. The mesh is nicely orthogonal and uniform. Also, I am using cfdemSolverPiso.
So I use "uncorrected" for snGrad terms. For the gradient terms, I find leastSquares performs just as well as Gauss linear, but seems to give less fluctuations on the flow. Has anyone else found this? Using upwind for the divergence schemes also give much smoother results (as expected), but the results do not show the right trends as with Gauss linear/limitedLinear.
Does anyone have comments on what works best for them?