Grid adaption for CHT
 

Dear CFD community,
I am currently working on CHT simulation which is quite classical (water circulating in a heated solid copper). I work roughly always on the same design but sometime i have to change slightly the solid part and the thermalhydraulic conditions (mainly the velocity).
Of course, it implies that i have to perform a new mesh independency study to validate my results and this part is very time consuming. That is why I thought about mesh adaption possibilities and how could I use this to obtain the best results from my simulation (and thus obtain mesh independent results with a few meshs). However, I have tried to use this tool based on residuals (momentum and energy) but the refinement of the cells which presents the highest residuals seems to not change a lot the results... Damn !!!!
That is why I need some help about that. How could I use mesh adaption in Fluent to obtain highest convergence ? Should I increase the numbers of refined cells ? Maybe is it a better idea to adapt the mesh based on cells gradient ? But in this case which order of magnitude should I consider for the refinement threshld ?
As you can see, I am looking for some guidelines to use mesh adaption tools to improve my simulations convergence.

Thank you for your help,
Best regards,
Pierre