LVDH wrote:No, we are doing approximated physics, thanks to the coarse meshes, a rather rough estimation of physics.
And you might not see it, but you are changing far more than one variable at a time and driving conclusions out of a far too small data set.
My previous experience with Khamsin, although not very long, showed a reliable behaviour of the solver: the system reacted as one would logically expect to both boundary condition changes and cars's geometry changes.
An example concerning moving floor: I did try the difference between a stationary floor and a moving one (moving at a half the windspeed); the moving floor did imply an increase -quite a small one- of downforce in the areas where i expected that increase to happen.
Therefore, yes, an approximation of reality, but a decently good one...
That's why, and i'm repeating it once again, i'm trying to find a logical cause for the discrepancy without attributing to the CFD approximation the issue. If the approximation happens to be the actual cause, then the challenge would loose its meaning, as we should should heavily bend our cars design to what the CFD dictates rather than to the laws of physics.
Having CFD experience yourself, have you got some data about the influence of rotating wheels VS non rotating on similar cars?
Or can you give your opinion about k-epsilon/k-omega? Should they present similar outputs in the case nothing stalls?
