Hello,
I am a Master Thesis student in Aerospace Engineering of the University of Bologna and I am working on an FSI problem involving the sliding contact between two thin deformable plates with zero clearance, made of steel and fixed on one end, under the action of mineral oil.
Since large deformations are expected, I am making use of the explicit dynamic solver in CalculiX coupled with pimpleFoam after multiple failed trials with implicit solvers. The coupling scheme I’m currently using is parallel implicit with IMVJ. I found that rbf-compact-tps-c2 and thin-plate-splines mapping lead to diverge of the force while nearest-neighbor mapping does not. Nearest-neighbor mapping however leads to unphysical solutions (ref. fig). In addition, there is a singularity in the pressure distribution.
To avoid sub-cycling I used as coupling time step of 1E-9s as dictated by CFL requirements by CalculiX. Contact is implemented as contact pair, type surface to surface (face to face penalty contact). The surface behaviour is set to linear pressure-overclosure with slope equal to 50 times the Young’s modulus as a first guess, and frictionless.
I would like to enquire if there are any known issues with the use of the explicit algorithm in CalculiX with preCICE or even the use of contact. I would greatly appreciate any help/advice from the community.
If I understand correctly, the meshes of the two participants do not maintain the same space relation over time. You would need the support for dynamic meshes feature, which is currently under development: Topics tagged dynamic-meshes
Hi,
Thanks for replying. I am not using dynamic mesh. Furthermore, looking at Configuration of the Multiple Perpendicular Flaps tutorial case I use only one solid instance that includes the stack of plates since the contact is implemented in CalculiX and I am not sure if contact can be modelled using two independent instances of CalculiX. The simulation crashes after very few time windows, with a displacement in the order of microns and no remeshing. Are there any particular incompatibilities between preCICE and CalculiX’s explicit solver?
I unfortunately do not know any more details about the CalculiX adapter other than the ones included in the documentation. Maybe one of the @calculix-users knows (join that group since you are one).
But your use case is not yet clear to me. Could you make a simplified figure, showing the domains (fluid / solid) and what fields they are exchanging with preCICE?
If you model the sliding contact directly in CalculiX, then I guess this should be fine. But I don’t know of anybody else that has already used that feature in a coupled simulation.
