HYBRID_SOLVER - FEM solfec bodies at boundary of parmec/solfec model

[bencannell]

Would it be possible to handle deformable FEM solfec bodies at the boundary of the solfec/parmec region of a hybrid model? It may be beneficial for the solfec bodies to be able to deform locally during collisions with adjacent solfec bodies not located on the boundary, to ensure the collision restitutions are appropriate. Currently, solfec outputs the following error message if FEM bodies are defined within the dictionary parmec2solfec.

hys.c: 342 => ERROR: Solfec-Parmec boundary body with Solfec id = 1 is not rigid

Alternatively, this may not make sense from a modelling compatibility perspective. If this is the case, it would be useful if the RIGID_TO_FEM and INITIALIZE_STATE commands could be modified to accept a list of body objects (or body.ids?) rather than solfec objects so we can initialise the position of a mixture of RIGID and FEM bodies within the same solfec object/model (i.e. RIGID bodies around the boundary, and FEM bodies within the centre of the solfec region). I don't think this is currently possible using these commands?

[tkoziara]

Hi Ben,
Currently bodies at the boundary of the Solfec-Parmec region need to be rigid: both codes are handling this region simultaneously to exchange velocities and forces and this way this is simpler to implement.
I think this could be generalised to having FE bodies on the Solfec side and rigid bodies on the Parmec side of the model. So let's keep this issue open to record the progress of such generalisation.

In the meantime - to test the hybrid solver framework - I could modify RIGID_TO_FEM and INITIALIZE_STATE commands by adding an optional argument subset that would work similarly to that implemented for the XDMF_EXPORT command. Would that be OK?
Tomek

[bencannell]

Hi Tomek,

Thank you for your response. Yes, your suggested modifications to RIGID_TO_FEM and INITIALIZE_STATE would suit our needs well.

Thanks,

Ben

[bencannell]

As an aside, is there a parmec equivalent to the solfec INITIALIZE_STATE command?

I expect (!) that in the future we will be aiming to run a short hybrid analysis using PENALTY_SOLVER, then initialise the same hybrid model using the INITIALIZE_STATE (and RIGID_TO_FEM) commands and continue running the analysis using the NEWTON_SOLVER.

[tkoziara]

Simply register this as parmec issue:) Let's call this command INITIALIZE by the way - parmec command names can be brief:) Also - along with registering - suggestions as to the expected range of functionality are welcome; Now - let's keep this thread focused on FE bodies on solfec-parmec boundary; I will clean our side comments up at some later point:)