support for nonunifrom grids

[biao-geng-me]

I read in the paper that WaterLily uses uniform grids. While this is straightforward to setup, it limits the size of the flow domain. The grids are commonly stretched in Cartesian grid + Immersed boundary method type of CFD code.

[biao-geng-me]

Just to follow up with the issue. Could you share some insights on the choice of uniform grids in WaterLily? Any comment on implementing nonuniform grids in WaterLily? Thank you!

[b-fg]

There is some work in progress for reducing flow domains using Biot Savart boundary conditions: BiotSavartBCs.jl, arXiv. There is no short-term plan for non-uniform grids, but PRs are welcomed!

[biao-geng-me]

Thank you for the information. I did some test cases using the BiotSavartBCs.jl. It seems that its performance is not good. The biot_mom_step! function was from 10 to 100 times slower than the regular mom_step! function. In the end there was little speedup compared to using regular B.C.s and a large domain.

I'm considering working on adding non-uniform grids in WaterLily, but while I have some experience with CFD algorithms, Julia is totally new to me. How much effort do you think it will be to add support for non-uniform grids in WaterLily?