simulation of a wheel
Hi @Anonymous
What type of simulation are you performing? If it includes the word "Static" in the name, the problem is that your model is not statically stable, so the parts are free to fly away. At least that may be happening on some iteration during the solution when trying to find the solution. For example, what prevents the wheel from rotating? (Wheels have a tendency to do that 
) A static analysis cannot analyze a rotating wheel.
The first suggestion would be to provide some static stability to the model where possible. For the block example, one part should be fully fixed. The part that you are trying to not slide should be constrained in the direction perpendicular to the sliding direction. That will eliminate one direction of rigid body motion. You then need to hope that contact in the direction of gravity prevents the parts from passing through each other (it should in theory), and that the friction force is large enough to prevent sliding in the third direction.
Let me clarify one issue. The analysis is calculating the displacement in all directions regardless of whether there is a force in that direction or not. Constraints prevent displacement, contact does a fair job at preventing displacement, but forces do a poor job at preventing displacements. For example, the tire model may have no load that would cause the tire to rotate in theory, but mathematically a load of 0.000000001 will cause the tire to roll an infinite distance over an inifinite amount of time, and a static analysis cannot solve that. Loads will not balance perfectly because of round off errors when distributing the loads to thousands of different elements.
I suspect your understanding of the contact types is okay, but getting the model to behave mathematically is a different issue.
Feel free to ask any question you have about the different contact types. If your models are still not converging, please export the model to an F3D archive ("File > Export") and attach it to your reply.