Hi SaiSan,
doubling the thickness change would correspond to doubling the time step. Did you try to use the adaptive time stepping? For example:
md.timestepping=timesteppingadaptive();
md.timestepping.time_step_max=5;
md.timestepping.time_step_min=0.01;
md.timestepping.start_time=0;
md.timestepping.final_time=1000;
You should also try to use an iterative solver if they converge in your case:
md.toolkits.DefaultAnalysis=bcgslbjacobioptions();
I hope this helps
Mathieu