[dimasbarile]

Hi everyone,

I'm running a case with 60 topoSet boxes distributed throughout the domain, where each set is read by the fvOptions file to introduce a source term. I want to use "scotch" method for decomposition. Because of the way the source term is written it is much faster for the simulation to achieve 2 things in mesh decomposition:

1- Ensure that none of the sets are split across processors.
2- Balance the number of sets across processors as evenly as possible.

I've already addressed the first issue by applying constraints in the decomposeParDict file. However, I'm struggling with the second point. Is there a way to modify the weight of each cell so that the cells in a topoSet have a higher weight, and the domain is decomposed accordingly? I've explored the "strategy" option in Scotch, but haven't found a solution.

Any advice or suggestions would be greatly appreciated!

Thanks!