criteria for result validation
 

which criteria are the most important to define a simulation good or bad?

I´am working with cfdesign which is an add-on for MSC-Nastran and i simulate the flowforces in a hydraulic valve. The flow is turbulent with a pressuredrop of 70e+05 Pa and i use the K-eps turb. model.

There are effects i don´t understand: - when a simulation is completed a take a look on the residual´s of pressure and velocity, if they are small <0,001 i´am defining the solution for convergent and the mesh for adequate.

- the second data i observe are the "y plus-values" on the walls of interest for the flowforces (shearstress*Area + pressure*Area)

=> now my problem is that i get only results that match with experimental results if the mesh is very fine, but the finer the mesh the smaller the y-plus values. The y-plus values should be in a range of 35 < y+ < 350; if they are to small the wallnear nodes are to close to the wall and do not represent the physic.

=> which criteria are most important for validating a result ?

=> are the y-plus values less important ?

=> ...

mfg

Matthias =>

Re: criteria for result validation
 

Hi Matthias,

Is it possible for you to refine your mesh everywhere except near the wall? As such, it is a clear requirement for the use of wall-functions that the first y+ should be between the range you have specified.

On the other hand, if you have the required resources to capture the sub-layer, why don't you use a low Re model right down to the wall??

chidu...

Re: criteria for result validation
 

The convergency it no measurement of the correctness of the solution.

If your mesh is too coarse then even a converged solution might be completly wrong.