Reply To: Waveguide bending Loss using Lumerical FDE and FEEM

[Guilin Sun]

If the loss is small, it will be sensitive.

You may need to do converge test with fixed mesh size, not the fixed mesh points in FDE. In FEEM you may need to use higher polynominal orders, or finer edge length, or both. Since the two solvers use very different meshing techniques, it is understandable that if the mesh is not fine enough and the simulation region is not large enough, there are some discrapancy, in particular the small loss.  In addition, make sure the material data is exactly the same.