What can be seen here, is that the even though the
t3bd case has a finer mesh, with more elements, the results seem worse.
What could be seen in the command prompt, though, reveals what’s going on:
m : 147: 3.15E-05 ...
*** CONVERGENCE *** 7.71197723E-04
This in contrast to
m : 200: 1.60E-04 ...
Number of requested steps done. 200
Simply put, this is caused by the latter mesh containing almost triple the amount of elements in comparison to the first. This requires more iterations to come to a proper solution.
In other words, the second simulation has not yet found a final solution.