Discussion Forum

Hello to all! I have a problem with gradients. My goal is to simulate wetting of a cylinder (1st picture). The red wall is the wetted wall. The dotted line is the axis of symmetry. Unfortunately, I can't upload the mph files because I don't have an access to my university PC now, but I remember almost everything about the model and have some graphics to show the problem. The equations I use are similar to the COMSOL equations for wetting (phase field), but my boundary condition is different.

It looks like this:
Pt = k*(A-B),
where P is the phase field variable, B = nr*Pr, Pt is the time-derivative of P, Pr is the r-derivative of P.

I'm using COMSOL 4.3.
At the start, A is bigger than B, but at equilibrium state they should become equal A=B which means Pt = 0. The problem is... the left part [ Pt ] and the right part [ k*(A-B) ] are similiar, but not equal. (2nd picture) This slight difference accumulates, and the system can't reach equilibrium. My assumption is that Pr isn't calculated correctly (3rd picture) because it's should look like a parabola and in my calculation it only resembles parabola.

What can be the root of this problem? At first I thought of making the mesh finer, but this wasn't effective. I'm using rectangular elements. I'm almost sure it has to do with my solver/numerical settings because I'm not strong at COMSOL yet. Thanks in advance!