My question is:
Is there an existing capability to have the fluid act back against the geometry? For example, modelling a rigid ball/bouy/ship bobbing on a fluid surface? Is the reaction of the fluid against the pump already being modelled (in the sense that it affects the rotation of the pump, not just the motion of the fluid).
I’ve not managed to find any such examples so far, apologies if I’ve just not searched properly.
I attended the latest Palabos training session last month and asked this very question there. The answer was that this is currently not implemented; a separate solver would need to be programmed for the rigid object, and such a solver does not exist at this time. What you see in the examples is just rigid bodies moving unaffected according to the preset parameters, but they do have an effect on the surrounding fluid through the immersed boundary conditions.
If you do have both the interest and the time to program such a solver, then please let me know.
Thanks for the info AriH, I was afraid that may be the case.
Going forward my options are either look into another technique (perhaps Smooth Particle Hydrodynamics) or, like you said, write a solver around Palabos for the rigid body. If I go for the latter option I’ll be sure to let you know here. Out of interest, are you attempting the same solution?
Edit: Also, was there any mention at the course of this feature being a planned future development? Seems like a logical next step forwards to me.
The SPH programs I have tried do indeed have this “floating” functionality, but they are often lacking in other respects (unstable, not parallelizable, etc.) There’s no single program that will do everything well - it’s always going to be a tradeoff based on what your specific application is going to be. So if two-way dynamics is essential for your project, then you might consider SPH - for me, it would be a nice addition, but it’s not absolutely necessary, and so I’m afraid I’m not spending any time on trying to implement it in Palabos. I’m only in it for the parallelization.
There was no mention of a two-way dynamics algorithm being implemented in the near future. But Palabos is open-source, so if someone does program it, it will become part of the repository. Likewise, if one of FlowKit’s consulting clients requests it, it will also eventually become part of the main program. Until then, we’re on our own…
Yes, parallelisation was what drew me to Palabos and LBM so it would be a shame to throw it out. My ideal solution is going to be coding rigid body physics into Palabos I guess… though, easier said than done!