FSI with interFoam and CalculiX: Convergence

[Alphaoo1]

(moved from FSI with interFoam and CalculiX)

I am trying to solve the divergence issue with interFoam and CalculiX.
(FSI Flap case modified for interFoam).

I modified a few things with the OpenFoam settings and increased the time it takes for the simulation to diverge.
Initially ‘interFoam’ diverged way before the falling water column hit the flap.

But after these modifications, as soon as the water column hits the flap it diverges rapidly.
I have tried both, explicit and implicit coupling as well to try and increase the time at which the solution diverges, but no help as implicit diverges before explicit. (As per my understanding implicit should be comparably more stable than explicit)
Fluid_interFoam_explicit.log (119.3 KB)
Fluid_implicit_lastParts.log (78.6 KB)

Here, I tried reducing the water column size, but there is a similar results as above.

Any hints, advice are very welcome.

Thank you.

Edit: Here my OpenFoam files.fvSchemes.txt (1.2 KB) fvSolution.txt (2.0 KB)

I am using the following settings in controlDict:

maxCo           2;
maxAlphaCo      1;
nAlphaSubCycles=5;
maxDeltaT       1;

[uekerman]

Could be just a very tough situation for the quasi-Newton coupling due to the sudden change.
Compare this post: IQN fails for quickly changing forces during wetting in a free-surface FSI problem

[Ulrich]

Dear Alphaoo1,
did you modify the OF-adapter according the suggestions of Moaxm https://github.com/precice/openfoam-adapter/issues/127 ?
After this we received a good agreement for the elastic dam break (https://www.dhcae-tools.com/preCICE.html).
Best regards

Ulrich

[Alphaoo1]

Dear @Ulrich,
Yes, I did modify the OF adapter according to the suggestion of Moaxm.
And I was able to run the dam break case with the Help of Kyle Davis.

Later on, it was a matter of fine-tuning the Implicit mapping.

After this we received a good agreement for the elastic dam break (preCICE).

I actually started the work on this after looking at the LinkedIn postby DHCAE team.

Edit: Here I have posted the results.

[Sun]

Dear @Alphaoo1,
Did you do your work by using CastNet? Is it free for us to download? I also want to do FSI works with interFoam and Calculix. Therefore, could you help me about how to debug the OF adapter and set up case?

[Alphaoo1]

Hello @Sun,

I have not used CastNet, I directly installed all the adapters needed and ran the case.
I think with the new updates to the OpenFoam adapter the interFoam + CalculiX case should run without any modifications to the adapter.

You can find the case here. The precice-config file is already set and should work without any issues.

[Sun]

Dear @Alphaoo1,

Thanks for your help. I’ll do it and learn from it.

[Sun]

Dear Alphaoo1,
I’m sorry to send the private message to you. Now, I’ll keep the conservation going on the topic.
I checked my output of solid and then found there was noting in the folders flap.frd and flap.cvg. The part of solid didn’t run at all. But there is an error in Fulid.log.

[Alphaoo1]

Dear @Sun,

Check your Fluid/system/controlDict.
The error

Unknown function type preciceAdapterFunctionObject

means that OpenFOAM cannot find OpenFOAM adapter lib.
You have changed the name of the adapter to libpreciceAdapterFunctionObject_interFoam. Check if the file is there in $FOAM_USER_LIBBIN (/home/OpenFOAM/platforms/version/lib)

This might be caused because

• You compiled OpenFOAM adapter with a name preciceAdapterFunctionObject (or some other name) but you are using libpreciceAdapterFunctionObject_interFoam
• The OpenFOAM adapter was not compiled at all (failed due to some errors)

[Stan]

Dear @Ulrich ,
I am doing the same elastic dam break case using olaFlow (a toolbox based on interFoam) and our in-house solid solver. However，when the water reached the right and upper boundary after impacting the dam, the simulation of fluid domain crashed suddenly.

run.log (802.0 KB)
fvSchemes.txt (2.3 KB)
fvSolution.txt (2.2 KB)
The simulation log, fvScheme file, and fvSolution file are attached. Could you please give me any hints on how to deal with such condition?
Best regards
Stan

[Ulrich]

I used IsoAdvector (ESI OpenFOAM) for this benchmark. But it seems your time set is too high. I would try to reduce the time step to have a interface Co-number below 0.5
Hope it helps
Ulrich

[Stan]

Thank you for your reply. I’ll try IsoAdvector with smaller time steps.
I also have a question about the Poisson’s ratio adopted in the benchmark. In Vilmar Fuchs’s dissertation (Link Ref 2), the Poisson’s ratio of the solid is set as 0.5. However, in this thread, it is said that CalculiX will not run if the Poisson’s ratio is set as 0.5 (also in our solid solver). Could you please provide the Poisson’s ratio you adopted in the benchmark?