A few FEKO questions about simulating a fat dipole antenna

I have been using FEKO to try to simulate a fat dipole antenna for wideband ground penetrating applications. Using the infinite green's function to model the substrate, air, then sand (in that order), I get very good results in the far field, 8 decibels max and very directive. However, when I try to use a finite substrate (with no air or sand), the results aren't nearly as good (low gain, not directive). If I try to model the substrate, the air, and the sand with relatively large cuboids, I either get memory errors (with MOM) or overflow errors (with MLFMM). What's the best way to represent the infinite green's function for substrates, but make it finite instead of infinite?

A more general question about fat dipoles: if we have a ground plane directly under a substrate of length lambda/4, shouldn't that give the most gain in the desired direction (away from the ground)? Because the wave will travel a full 360 degrees and return in phase.

I have also been modeling waveguides. The results seem accurate using the default MOM/FEM, but I cannot use the MLFMM for them. An array of waveguides at multiple frequencies can take a very long time, but the MLFMM doesn't even finish.

I guess more generally: does anyone have a link to a site that describes the different CEM methods and which antennas they are useful for?

Thanks in advance.

I do not know the solution to your problem, but I had a question based on your simulation. For your fat dipole, are you using a line model or a cylinder model.

If you are using cylinder, how do you excite the cylinder. For example, with wire, you use the wire port.

Thanks,

Are there any warnings given? Also, you mention that you left out the sand and air, so you have made changes to the model. Can one expect then to get the same result as with the Green's Functions?

You are looking for a truncated Green's Function. This can be done, but AFAIK not possible in FEKO :(

It should, yes. The total one-way electrical length to the ground plane should be lambda/4 - did you consider the wavelength in the substrate above the ground?

(my, you have quite a shopping list of questions, don' you? hehe)

Are you using a dielectric loaded waveguide? Then the FEM/MOM would be good for this, yes.
How many elements in the array? If the coupling between the waveguides is small, you could compute the pattern of one waveguide and make an array of waveguides using the radiation pattern point source approximation (AR card if you are an EDITFEKO nerd)

Well, the methods in FEKO is described on their site:
http://www.feko.info/feko-product-info/technical