Hi Amritayan,
I apologize for the late reply, I was out of office. It can be a bit tricky to accurately calculate the bending losses due to the radiation. So, convergence testing is recommended. As your results have converged for the larger FDE span (which is expected), I wouldn't change the advanced PML parameters.
For the anisotropic material, it would consider the different indices along the different axes.
Regards,
Amrita