[hoehnp]

Hello together,


while integrating CFDEMcoupling as an additional fluidModel in solids4Foam, I face the bug as described in https://www.cfdem.com/forums/bug-fin...using-parallel and polyMesh::findCell slow in parallel, but meshSearch::findCell sometimes fails


I suggested I replaced meshSearch::findCell by polyMesh::findCell. Sofar the code seems to compile and run. Is there anything I should still consider in this replacement?


Thanks a lot everyone.


Patrick

[hoehnp]

Hello everyone,
during the development I am facing a second issue now. solids4foam uses unlike the original FSI toolkit regions for the fluid and the solid phase. When trying to reconstruct a decomposed problem, I found that lagrangian fields are no longer recognized. This seems to be caused by the fact, that the IOobjects in the lagrangian parts of reconstructPar include the region "fluid" in the path, while the actual folder structure would require a region "".

One of course could start completely rewriting this part of reconstructPar. Before doing this I wonder, if anybody has faced and solved similar problems before?

Thanks a lot everyone
Patrick

[bigphil]

Hi Patrick,

Yep, it seems that you need to edit reconstructPar such that it treats lagrangian fields the same whether they are in a region or not.

Philip

[hoehnp]

Hi Philip,


thank you for your answer. I am now facing a few more questions during my further implementations:
1. If I want to introduce the particle forces in the fluidSolidInterface, from my understanding I have to do this in Foam::fluidSolidInterface::transferFacesZoneToZone . Is this assumption correct?
2. How could I handle a rotation of the solid domain? Rotate the solid mesh and also put a rotation boundary on the inner patch of the fluid mesh? Only rotate the fluid mesh?
3. Could I also get write access to your bitbucket repository to share my results so far?


Best Regards,
Patrick

[bigphil]

Hi Patrick,

See my replies below:

1. If I want to introduce the particle forces in the fluidSolidInterface, from my understanding I have to do this in Foam::fluidSolidInterface::transferFacesZoneToZone . Is this assumption correct?

Sort of. The transferFacesZoneToZone function maps a field on the fluid interface to the solid interface. You can see that this function is called once in the fluidSolidInterface class in the void Foam::fluidSolidInterface::updateForce() function at line 1694. You will see that it is mapping the fluidZoneTotalTraction field to the solidZoneTotalTraction field.

For you, I would suggest that you add your particle forces to this fluidZoneTotalTraction field before it is mapped to solidZoneTotalTraction; that way, you do not need to modify the transferFacesZoneToZone function.

One thing to be careful of is that fluidZoneTotalTraction and solidZoneTotalTraction are "zone" fields, meaning that this field is the same on every processor in parallel. The globalPolyPatch class is used to transfer between "patch" fields (local to each processor) and global "zone" field.

2. How could I handle a rotation of the solid domain? Rotate the solid mesh and also put a rotation boundary on the inner patch of the fluid mesh? Only rotate the fluid mesh?

The solid domain uses a Lagrangian approach, and as such, its motion comes from solving its governing equation (remember the solid mesh is not independent of the solid motion/displacement/velocity field). Consequently, to impose a rotation on the solid, it should be done through appropriate boundary conditions on the displacement field e.g. a rotation boundary condition at the axle patch of your solid, like the fixedRotation boundary condition.

Then for the fluid, depending on the magnitude of the rotation, you may need to define a rotating region, where I guess you know the approximate rotational velocity in advance, and the fluid mesh motion solver would need to account for the interface motion relative to this rotation.

3. Could I also get write access to your bitbucket repository to share my results so far?

Sure. If you have some working tutorials showing new features, you can send me an email describing them and I can grant you permission.

Best,
Philip