When is a solution mesh independent? I adapt my mesh and creat a new solution. But how can I decide if my new solution is the same like the old one? Thanks Armin

For example, (1)create a point or a plane in your system and set your own variables such as velocity, temperature, concentration etc. (2)Compare initial and adapted value at your point or plane. (3) If there is no change in your values, that means grid independent solution. (4) Ideally, if you want to get a 100% grid independent solution, try (1)-(3) steps at whole grid points but that's not an easy problem. (5) I personally think, to some extent, to determine a grid independent soln. is what users have to do.

Get a solution of the relevant parameters to your problem, e.g Cp, and plot them on one mesh. Re-run the simulation on a grid having twice as many nodes, and do the same plot. Obviously you might want to refine your grid in a sensible way, i.e. not necessarily uniformally. Compare the plots gain and estimate the error using a standard measure. The lower the measure the more mesh independent your parameter (problem) is becoming. This should also point to where mesh refinement is needed.

Read papers by Roache (1989), "Need for Numerical Accuracy", J. Spacecraft and Rockets, Vol. 27(2), pp.98-102 or the 1998 AIAA guide Guide for the Verification and validation of CFD Simulations.