I have several questions about using XLB for wind simulation of a wind farm

[fred2019-dev]

(1) Should I set all the 5 faces(z = max, x = max, y = max, x = 0, y = 0) all to RegularizedBC ？
(2) Should I set the terrain surface to HalfwayBounceBC ?
(3) Do I always have to wait for about 1000 steps before the simulation reaches a stablized state? As I observed, the wind seems to first fill the grids in about 1000 steps, and then suddenly switch to a new and stablized state. Should I save a checkpoint of the stablized state? And if I start a new simulation of the same terrain, could I always resume from this checkpoint?

Thanks very much!

[hsalehipour]

Hi @fred2019-dev, thanks for your question. The setup for your problem seems to be closely related to the windtunnel_3d example. The exact type of boundary conditions depend a lot on your geometry and the observed stability/accuracy. But there are multiple options beyond HalfwayBounceback and/or RegularizedBC that may be suitable. We recently added a HybridBC as well which is more stable.

As for waiting to reach stabalized state: yes if you start from flow at rest, the simulation always undergoes some transient period (the exact number of iteration depends on many factors) until it reaches some steady or quasi-steady state.

Right now, we don't have an example to demonstrate restarting from a checkpoint althgouth you can easily do that yourself. Read velocity and density from your checkpoint and initialize with those, assuming equilibrium to construct your "intiial" population field.

Hopefully this is enough to get you going. If you face any bugs/problems, feel free to open another issue with a reproducable code/ gemoetry. Thanks