I’m trying to develop a filter drag model with MFiX-Exa since it’s C++ based so can be well coupled with the open-source library libtorch. Now that we put our focus on filter model, It’s necessary to get the grid volume for traversal. In MFiX, it can be easily gained from the subroutine geometry by the code “VOL(IJK)”, but the same vector doesn’t seem to exist in MFiX-Exa. So how can I get this value exactly? Hope you can guide me please.
a. The only thing MFIX-Exa and MFiX share is the name (and FWIW my vote was to drop the mfix moniker all together to avoid this confusion)
b. AMR is still developmental, so everything you would run is on level 0 and the volume is dx^3. if you have cut-cells, then it is vfrac*dx^3. To protect in the case that we eventually allow for non-uniform grids, this would be something like dx[0]*dx[1]*dx[2]*vfrac(i,j,k)
c. We also have a postprocessing tool named filterML that post-processes a plt file and spits out the most relevant data for filtered drag modeling, i.e., volume fraction, slip velocity, heterogeneous index, drift flux (or vel?) and streamwise pressure gradient, in every filtered volume that is a constant number of the fluid mesh.
d. We already have a ton of this data for triply-periodic cluster-induced turbulence. I would need to confer with coauthors @jmusser@aike, but is this something you would be interested in?
Thank you.
Actually we’ve verified that there comes some differences(about 5% relative error) when fixed the dx value as a real number. Anyway, I want to fetch the value of dx while iterating every cell as shown in the code in attachment. usr_drag.f (5.0 KB)