[Getdp] Travelling Wave Boundary Conditions

[David Eimerl]

To GetDP Users/ FDTD Experts!:

 

I am setting up a vector electromagnetic wave problem with travelling wave
boundary conditions on the surfaces of a cube. 

The E-field on the boundary is solution of the full vector Maxwell's
equations for a wave travelling in a general direction (i.e. not normal to
any cube surface)

The cube side is a few wavelengths.

 

I'm trying a FunctionSpace for Form1,  with BF_Edge basis functions, and a
Galerkin resolution. The problems I run into include:

 

(a)  Using a constraint with a TimeFunction that depends on the corrdinates
as well as $Time, does not compile.

(b)  HarmonicToTime presumes the phase of the field on the boundary is
everywhere the same, so that it works for a standing wave i.e. for two
travelling waves, but not for a single travelling wave. This forces me to
use a Newmark scheme, I think.

(c)  Adding time-dependent terms in NxH and following the MWE.pro examples
reproduces the HarmonicToTime result, (if I did it right).

(d)  getDP seems to generate the resonant modes of the cube quite easily
with several different boundary conditions. However, a full complement of 3D
PML layers with R~0.001 does not remove this resonant behavior. In fact they
apparently has no effect.

 

Any suggestions? Or examples? Say from a general 3D waveguide problem?

 

DavidE

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.geuz.org/pipermail/getdp/attachments/20130215/8facce45/attachment.html>