preCICE doesn't converge with a large strain


I’m simulating a fsi problem using openfoam-precice adapter. The simulation runs successfully at first. However, after some while the simulation diverse suddenly. The simulation converges within 3 iterations in the previous time step (in file: precice-Calculix-iterations). Then at the last time step, the maximum iteration 200 is reached and stops the simulation. The maximum strain is around 1.6-2.2. Does anyone know this type of error? Is this caused by the large strain?

I checked the CFD mesh in the final step, there’s no problem. The deformation is acceptable. I also ramped the force to avoid it in one shot, but it doesn’t work.

Hi, @xiangli,

indeed, the first possible reason that came to my mind was a too large deformation, which would create very skewed cells. This could originate by a large force on the CFD side, or some other (technical or modelling) reason on the structure side.

To be sure, please double-check whether you are looking at results at the same time. You can configure OpenFOAM to write results more frequently, or you can use the preCICE exports feature to write the values that preCICE knows.

After checking the results in every time step next to the problematic time, I would try to identify if the spike starts on the CFD side or in the structure side (does the force or the displacement spike first?).

I would also try to see if this is anyhow related to checkpointing (e.g., the spike happens after reloading a checkpoint in implicit coupling), or try to identify what is special about this time.

Thanks, I’ll try to check them. By the way, if the coupling effect is very strong which leads to the simulation being very unstable, are there any measures we can take to improve this?

The usual choices here would be to use an implicit coupling scheme with a small coupling time window, but this also has limitations.