Interpolating Sweeps is used for a very broadband frequency sweep i.e. maxfreq/minfreq > 4. Only drawback is it doesn't produces field solution in a specific frequency.
Fast sweep should be default sweep as long as maxfreq/minfreq < 4.
In Discrete sweep the simulation time is proportional to the number of frequency points specified. It is used to calculate fields after doing an interpolating sweep.
It doesn't matters which sweep type you have choosen your results should be same for all.
Best sweep that gives good agreement between simulation and measurement is "Discrete", but to avoid long time simulation, "Fast" sweep gives a good estimate till you reach optimum parameters, then confirm the optimum parameters with "Discrete" sweep.
As far as I know, fast sweep shouldn't differ that much from discrete sweep. Please review your excitation, and boundary conditions.
If you are sure about every thing, then nothing you can do except use a powerful workstation and conduct parametric analysis.
there is some thing that may save your time, that you study the sensitivity of each parameter you need to change. This will give you a good lead to the effect of each parameter and will allow you to minimize the parametric runs.
I worked with spiral antennas, and I divided the bandwidth to steps of 2 GHz and used fast sweep, Then I divided the bandwidth to the steps of 1 GHz again with fast sweep.
For frequencies bellow 10 GHz the first sweep was reasonable but the second one gave me results that agreed with the measurements.
So I found the narrower steps in solution bands would be better !
when I simulated spiral antenna, the mesh structure for lower frequencies was big and the inner parts of spiral didn't need too much mesh; but then mesh structure for higher frequencies was small and the outer parts of spiral, despite the current on those parts was small and negligible, was meshed so densely that computational space requirement was more than my RAM.
Using a center frequency and one frequency setup in my simulation didn't give good results because its mesh was not proper for higher frequencies and inner parts.
So I got good results with dividing the wide-bandwidth into small ones.
Interpolating Sweeps is used for a very broadband frequency sweep i.e. maxfreq/minfreq > 4. Only drawback is it doesn't produces field solution in a specific frequency.
Fast sweep should be default sweep as long as maxfreq/minfreq < 4.
In Discrete sweep the simulation time is proportional to the number of frequency points specified. It is used to calculate fields after doing an interpolating sweep.
It doesn't matters which sweep type you have choosen your results should be same for all.