AMR with RANS
 

Hello All,

I have a conceptual question that I would like to discuss with you. The definition of the sub-grid scale (SGS) is the difference between the actual field and the resolved field. Then, how can we use the SGS criterion in applying AMR for RANS simulations where there is not a resolved part of the turbulence energy? Isn't it more logical, or physically and mathematically acceptable, to use the Value-Based AMR with RANS?

By my understanding, SGS actually quantifies how under-resolved a certain part of the flow field is, and that's where you want to have finer mesh, to make the difference between resolved field (numerical result) and actual physical field (ground truth) as small as possible. In this regard, although RANS is modeling the turbulent energy, SGS is still valid to quantify such difference, thus it can still be regarded as the criterion to determine where we should add more meshes.

Hi AEAli,



Thanks for your question. Yes, you are right that RANS does not resolve the turbulence field, but models it. However, unless your are using DNS, turbulence field is always modeled.



Here, CONVERGE gives SGS options only for velocity/temperature/pressure/species/void, which is coupled with turbulence, by using the turbulent viscosity calculated by the RANS model. The second spatial derivative difference between these resolved variables (velocity/temperature/pressure/species/void) and the actual field are then used for AMR. So, SGS is acceptable even when we are using RANS.



For your second question, the difference between SGS and value-based AMR is not decided by which turbulence model is used. They should be used based on the user's specific need. For instance, for a VOF simulation, if you know the interface void fraction between two phases is ~0.6, then you can specify the value-based void AMR to be 0.5-0.7. For flame propagation, if you know the flame temperature is around ~2000K, then you can use value-based temperature AMR. For other situations where the flow is highly mixed, unsteady, and unpredictable, you may wanna use SGS AMR.



Please let me know if you have any other questions.


Thanks,
--Mingyi

-----------------------------------------------------------------------------------------


Mingyi Liang


Research Engineer, Application Team
Convergent Science