If you really want to experiment with antennas, i'd suggest downloading the eval versions of CST Studio, ANSYS HFSS, or Altair HW Feko. You can find full hacked versions if you dig around.
I use all three, in full licensed form - under lifetime eval licenses i negotiated for as I do a lot of non-commercial research and help alpha/beta test new builds.
Why i recommend those, assuming you'd use the eval versions for curiosity, is they aren't NEC engine based - aka wire simulation.
They are built around CAD/CAM engineering type design and modelling using actual material based testing, so assuming you the right solver for the type of antenna and fairly accurately model your design, you get results which leaves NEC modelling solution dead in the water.
Mind you, the downside is even with a fast quad core computer, single workstation computing isn't fast due to the sheer modelling solver complexity. Not a prob here as my laptop is a 12 core item, with a high end mobile GPU (yes, the software also supports GPU additional accelerated compute as well) and a bank of redundant quad, hex and octa core towers use for distributed computing as you'd usually use with such software were you use the DC rig to do the heavy lifting.
But even on a quad core single computer, you can get a feel of the potential and really research, model and have pretty accurate plans to physically prototype once you get your ideas near perfect on the simulation - and if you did it right, if you construct accurately - very little residual mods are needed to the final physical build to get them resonant and acting properly.
Just remember, no matter what software solution you use, NEC based or otherwise, don't rely on virtual ground and GP's, build actual GP's and ground areas and mast substitutes into the model and use the correct method for the software to simulate proper termination to the FP's. Then you'll get something applicable to an installation model rather than a pointless free-space model.