Example for ibm + cylinder with the current periodic poisson solver

[mathrack]

No description provided.

[ia267]

Approved - I was running this case and it diverged around iteration 3170. I think for now we could set the n_iters = 3000

[mathrack]

Approved - I was running this case and it diverged around iteration 3170. I think for now we could set the n_iters = 3000

I have reproduced the issue on CPU. The number of time steps is now limited to 3000 to avoid divergence. If one wants to perform a periodic simulation with a longer simulation time, I think there are two solutions :

1. Extend the domain. A domain twice longer would allow a simulation twice longer.
2. Add a buffer layer to damp fluctuations at the outflow
3. Reduce the time step. The IBM might absorb the incoming velocity fluctuations with a smaller time step

Of course, the best solution would be to use inflow-outflow instead of periodicity. Hopefully, this will be ready shortly.

[ia267]

Make sense to limit it to 3000