Discussion Forum

Hello,

I would appreciate a lot if someone could help me with a problem I have when performing a polarization curve simulating a PEM fuel cell.
Specifically, the problem lays in the fact that I am unable to reach low potentials (mass transfer region). I perform a polarization curve by setting a parametric solver in which the cell potential is the parameter to be progressively changed. I think that the reason is that, as potential gets lower, oxygen is consumed on the cathode side and water is generated. Therefore, oxygen mass fraction gets lower whereas water mass fraction grows. Reached a certain point, COMSOL calculates negative values for the oxygen mass fraction and values over 1 for water mass fraction and this causes a convergence problem when taking the last solution as the initial approximation for the next iteration (undefined value found for w_O2 and w_H2O).
I could avoid this by changing the discretization from linear to quadratic in the mass transport equations, by reducing the relative tolerance and setting a lower step in the parametric solver, but it only works for a very simple geometry. When I try to solve the model for a real geometry (much more complex), which is my actual interest, I still have the same convergence problem due to that COMSOL calculates mass fractions below 0 and over 1.

Can anyone suggest something to solve this problem??

Thank you very much