Unless you're actually trying to determine "how did it look/feel to the crew?" or something similar, where you need a pilot involved, then use of a flight simulator may not add much, or even be misleading.
Use of simulation modelling - the models which are used for both engineering design and which 'drive' a flight simulator - is usually far more fruitful, because it's possible to better control the variables in the investigation.
Lets say you were investigating an accident at rotation. In the sim you could keep doing takeoffs and get a 'feel' for things, but each takeoff would have different crew behaviour and reaction times and so on.
With a (desktop) model, one could keep a standard takeoff technique, say, and vary other factors to determine their relative input to the end event. Or you could vary pilot inputs in a controlled fashion (vary the column force in fixed increments, etc.). It's a far more scientific and analytical approach, and likely to give a greater understanding, IMO.
I even query the use of a flight sim to "get inside the accident pilot's head" - the test crew KNOW they are in a sim, they know why (generally) and quite simply are different people with different experiences, reactions, biases, etc. It still ends up guesswork.