Hi all,

I'm conducting a second opinion for a simulation which is made in CFX. But I will not make a validating simulation, I will just judge the assumtions and conclusions.

I have the following point of doubt: I have allway's used simulations with rectangulair cellstructure (mainly FDS/Smokeview), but the simulation I have to judge is also built with tetrahedra. The point is that there are 2.000.000 cells with a domain wich is aproximatly 100.000 m^3. With local refinement cellstucture. The question is if this is generaly fine enough for a good simulation? In the report I cant find a validation which proves the cellstructure is fine enough.

Could anyone say if this is unusual or not?

Best regards,

Jan

Hi Jan,

It's impossible to say without seeing the case and doing a mesh refinement study. Keep in mind that there are many factors that contribute to error, mesh is only one of them. Often geometric simplifications, boundary condition selection and model selection have as significant an effect. A mesh refinement study can be done to ensure the discretization error is minimized, but the results can still be off if you don't set up the right problem.

-CycLone