Discussion Forum

Dear all,

I am trying to do a mesh independence test on a time dependent Alkaline Water Electrolysis 2D simulation (Similar to the application libraries model). The physics involved are Euler Euler multiphase model and secondary current distribution. The file with the coarse mesh converges very quickly, but when I refine the mesh to check the mesh independence, the free time stepping starts taking very small timesteps after a certain time, and the simulation stalls with negligible progress. If I switch off multiphase, the model converges on both the meshes.So, I believe the non-linearity introduced by the Euler Euler model is causing this issue. I tried to check the model solution at this time for any signs of unphysical results/high residuals, but did not find any reason as to why the timestep size reduce. In the logs, I can see that the NLFAIL column increments before the time steps reduce, indicating the nonlinear solver is failing at the given timestep. The model possibly has a singularity at the junction point of the electrode and the insulating wall, which causes some oscillations in the electrolyte current density and disperse phase volume fraction, but even then the model converges at some conditions of inlet velocity, mesh size etc. I cannot pinpoint the exact reason for this behaviour. I had to clear all solutions due to size restrictions, please run the simulations to check the behaviour I described. Will be really grateful for any answers/suggestions, as I am new to comsol and attempting to solve this issue since more than a month. Please let me know in case any further info on this case is required.

Best Regards Amit