Hi!

How can I define if the mesh I'm using is good enough? I've already made few mesh refinements, but each of them gives more and more strange results.

Thanks,

Julie

Unfortunately for turbulent flows the only way to determine a good enough grid size is to compare different mesh densities and hope that at some point the changes will be small enough to conclude that a certain mesh density is 'good enough'. Additionally, it is always also a play between accuracy and computational time.

Good luck !

Hi, Ossi!

The problem is that I've already tried 3 different meshes (~3000, ~9000 and ~34000 (!) cells) and each of them gives different (on my opinion, more and more strange) results. The only explanation I have that the choosen geometry is not good for the problem beeing solved.

Thanks anyway,

Julie

I do not know what kind of a problem you are dealing with but if it is a 3-D geometry, then the 30 000 cells do not sound too much.

perhaps you do not have enough cells, or tou may be trying to solve your case steady state and is in fact transient, or how do you know your case is converged in eash run, are the residuals still dropping even though you have reached the default convergence criteria, if so try setting the convergence criteria to 1e-6 in your models. Also try comparing flow fields from different iteration numbers in your solution and see if they are changing. You can actually subtract flowfield variables from different iteration numbers by using the operate command.

cheers

allan

Hi, Ossi!

The geometry I deal with is a sector with one cell layer in the angular direction.

Cheers,

Julie