Hierarchical coupling schemes in preCICE for 3 participants

Hello all,
I am trying to work on a problem which involves coupling of 3D-FSI model (Elastic-3D-Tube tutorial) and (Elastic-1D-Tube tutorial (monolithic)).

Now if I want to use the multi:coupling scheme for this problem then the solution diverges in the first time-window. So, to get around this problem I am thinking of a following setup:

I let the elastic-tube-3d-tutorial run in serial implicit coupling, for a particular time-window when it converges, then fluid from elastic-tube-3d-tutorial should communicate it’s values (pressure, velocity) from outlet (which is called as auto2 in this tutorial) to monilitic-elastic-1d-tube in serial explicit fashion. That way I don’t have to worry about convergence at 3d-1d FSI coupling interface and can observe the propagation of pressure pulse in monolithic-elastic-tube-1d. For better understanding, following is the visualization:

Now my question is that, how do I realize this type of coupling scenario in preCICE such that at first it carry out elastic-3d-tube computations and after it converges for a particular time-step then we couple it with elastic-1d-tube-monolithic explicitly and then we move on to next time window.


(I have already discussed this with @arslansadiq, documenting the discussion here)

In general, the idea of waiting for a coupling scheme to converge before continuing with another coupling scheme, and the ability to combine coupling schemes in a hierarchical way, would be quite interesting.

Looking again at this figure, I don’t think we really need to call it hierarchical, but it would be a combination of two coupling schemes: an implicit coupling scheme for the 3D FSI and a uni-directional, explicit scheme for the 3D Fluid to 1D Fluid.

This reminds me of the setup of the ExaFSA project (see, e.g., this paper), where there was a (I guess implicit) coupling between a structure and a fluid/acoustics (near field), and a coupling between the near-field acoustics to the far-field acoustics.
Maybe @fsimonis or @uekerman could comment on how that simulation was actually configured.

You get this behavior by just defining both coupling schemes above each other:


I don’t think that a hierarchical coupling scheme would be better than our multi-coupling in general. We also discussed this briefly in this conference paper.

How does the multi coupling diverge? Could you share the convergence file and the preCICE config. Maybe it needs some tweaking.

1 Like