A model may be a physical model (like a plastic skeleton in Biology class), a mathematical model (like the way a police radar gun measures the speed of your vehicle), a statistical model (like the "bell-shaped" curve is used to find percentiles, etc. The important thing is that it is not the real items; it represents them.
An experiment transforms "inputs" (conditions) to "outputs" (results). An example of an experiment that uses real items: every part of a new airplane might be broken in various ways to determine when it will fail; this happens before a real plane is ever built or flown. An experiment may use a model rather than real items, for example, rolling dice 1000 times to find out the probability of getting doubles.
A simulation is a substitute for an experiment; it uses a computer or special equipment that is not the real items, but gives a similar result. For example, a computer can create pseudo-random numbers and "roll" dice 1,000,000 times in a couple seconds, then report the results (different every time). Ever driven a car using a simulator?
Note that there is a difference between experimental probability and theoretical probability.
One of the wonderful benefits of math (like algebra) is that formulas may be used to model real processes (like a car crash). With a good model, we no longer have to crash real cars as much to test or to design them.