Andrea,
I too don't know how PTC has implemeted its contact algorithm, so I can't answer your question. Contact is indeed complex and difficult to model. From what I hear, if you need a good contact model with accurate representation of what is happening on a detailed level, MSC Marc is state of the art. I have seen some research publications where this software has been used with success.
I rarely use contact analysis on any more advanced level than what you see in my example here above. If you need to understand what happens on a more detailed level than this, I would recommend physical testing. And even if you use FEA, you will need testing; if you use FEA at the limit of what it is capable of, you need to tune your model, compare with reality, so you understand what is going on, and how to interpret your FE-results relative to your test results.
