Convergence issues with level-set conservative form

Hello Everyone,

I am trying level-set conservative form for droplet generation in flow focusing device. I have tried the following:

(a) non-conservative form: it converges well but mass conservation is an issue, droplet size keeps reducing with time and in some cases vanishes just before the outlet.

(b) conservative form: convergence problems; the error shown is 'may have reached singularity'

(c) I have tried finer meshes; smaller time steps with both conservative and non-conservative forms but same results as case (a) and (b)

(d) It looks like the level set parameters epsilon (related to interface thickness) and gamma (reinitialization parameters) could be the key but other than heuristics given in Comsol documentation, i have no clue about how to tune them. Though i tried few trial and error cases for tuning these parameters with no success.

Please share your experience/ thoughts/ suggestions about improving convergence or for suitable tuning of level set parameters.

Thanks in advance.