- #1
ashesh
- 3
- 0
I am having some serious problems in solving for the propagation constant of a inhomogeneously filled rectangular waveguide...
The dimensions that I am using are a = 2.22(along x-axis) and b =
1.0(along y axis) and the dielectric filled in half the region has a
dielectric constant of 2.45, this is the configuration that was used by Harrington in TimeHarmonic Electromagnetics Fields, page.161.
I divide the rectangular waveguide in triangular elements such that there exist elements with edges along the interface.
My stiffness matrices involve terms that have the dielectric contant.
Since one side of the edge on the interface has er=1.0 and on the other er=2.45, is there some special way I should be handling the elements involving these elements. From the literature, it is clear that in FDTD technique when ever this situation arises, we take the average value of dielectric constant. But no where I could find any such information as to how the situation is handled in FEM.
As of now, my program works fine only if number of divisions along x is restricted to 2 with any number of divisions in y-direction. When I increase the divisions along x-direction ,the result converges, but to a completely wrong value.
Please refer me to appropriate material or solution that can help me
handle the above situation.
Thanks in advance...
The dimensions that I am using are a = 2.22(along x-axis) and b =
1.0(along y axis) and the dielectric filled in half the region has a
dielectric constant of 2.45, this is the configuration that was used by Harrington in TimeHarmonic Electromagnetics Fields, page.161.
I divide the rectangular waveguide in triangular elements such that there exist elements with edges along the interface.
My stiffness matrices involve terms that have the dielectric contant.
Since one side of the edge on the interface has er=1.0 and on the other er=2.45, is there some special way I should be handling the elements involving these elements. From the literature, it is clear that in FDTD technique when ever this situation arises, we take the average value of dielectric constant. But no where I could find any such information as to how the situation is handled in FEM.
As of now, my program works fine only if number of divisions along x is restricted to 2 with any number of divisions in y-direction. When I increase the divisions along x-direction ,the result converges, but to a completely wrong value.
Please refer me to appropriate material or solution that can help me
handle the above situation.
Thanks in advance...