Different Timesteps in FSI

Hi everyone,

the “Future Work” part in the FASTv8 readme file contains this passage

“Upgrade the loose coupling algorithm of the glue code to allow each module to have its own time step, including time steps larger than the glue
code time step.”

I am currently working on a partitioned FSI method coupling a potential theory based fluid solver and a multibody system solver for the structure part. It would be really nice to be able to use different timesteps for each solver, particularly a larger timestep for the fluid solver. Unfortunately, after moderate search, I have not been able to find any literature on the subject. I plan to use FAST v8 for validation and while studying the readme file I am across the cited passage, which led me to believe there should be some information on this.

Could you comment on the problem or recommend literature?

Best Regards,

Tobias Lampe

Dear Tobias,

We wrote a paper that explains our coupling algorithm for mixed time steps, including modules with time steps larger than the glue code. You can find that paper here: nrel.gov/docs/fy16osti/63203.pdf. However, after implementing the algorithm for some simple example coupling problems (as explained in the paper), we didn’t find enough benefit to warrant full implementation of the algorithm within FAST. So currently in FAST, the coupling algorithm only allows modules to have time steps smaller than the glue-code time step (i.e. time step subcycling). We may review/update our algorithm at some point in the future, including options for mixed time steps, but we are not currently working on that.

In our SOWFA high-fidelity wind farm aerodynamics tool, we use SOWFA as a driver for FAST, which allows the much more computationally expense CFD solver (OpenFOAM) to take larger time steps than the structural solver (FAST). Effectively, FAST is subcycled within OpenFOAM.

Best regards,

Many thanks for pointing out the document! I’ll read into it and see if I can gain knowledge regarding my problem.