Discussion Forum

Hi everyone,

I am working on modeling the transport of a solid, non-spherical particle through a fluidic microchannel using the fluid-structure interaction physics in COMSOL 5.3a. This study needs to be done using FSI rather than the particle tracing module due to interests in the rotation and orientation of the particle. The inlet of the channel is given a normal inflow velocity. The coarse fluid domain mesh is automatically remeshed using a distortion condition.

I have no issues with convergence until the particle hits a wall of the channel with the errors associated with the 'non-linear solver not converging and the predicted solution guess leads to undefined function value' or 'inverted mesh error'. Ideally, I would like to incorporate a method for inelastic collision, but I am currently simply trying to resolve the issues with convergence once the particle hits the wall.

From the discussion with COMSOL support, I incorporated solid walls with defined solid mechanics parameters and introduced contact pairs for the particle boundaries and the wall boundaries with no initial contact pressure. Incorporating a small contact offset between the source and destination boundaries removes the issue with mesh inversion, but the solution still does not converge.

This seems to be an issue others have been trying to deal with: https://uk.comsol.com/forum/thread/129601/how-to-model-2d-fluid-structure-interaction-fsi-if-solid-collides-with-the-wall-?last=2016-11-21T22:50:48Z

https://uk.comsol.com/forum/thread/40105/solid-mechanics-2d-stationary-contact-does-not-converge?last=2017-02-13T03:24:12Z

Any ideas on how to address this issue? Could I perform a topology change (ramp up the viscosity when the particle is close to the wall using the wall-distance interface)? Or better yet, apply a boundary force upon contact? Would really appreciate any suggestions.