Apart from being a "Victim" of simulators for a good few years I had very little understanding of how these machines are built and work. I am slowly getting more knowledgeable in my current job.
This is what I Know (or think I know).
1. The best possible simulator is achieved by using the manufacturers flight data pack.
2. However, as someone already mentioned, there is little appetite amongst the Flight Test community to test dangerous failures such as TR drive and in some large MEH, full EOLs. Therefore some of the flight test data is absent to support all the failures in the RFM.
3. Contrary to many beliefs, an FFS is built to reflect the design and therefor the failure modes prescribed in the RFM. Going off piste by trying to simulate conditions not prescribed in the RFM is very problematic but far worse than that is someone trying to draw conclusions about the heli, in a sim that was never designed to meet the conditions applied.
4. The simulation of actual aircraft software behaviours such as AFCS, FADEC etc is very important in modern sophisticated helicopters so for the most part real black boxes/software should ideally be used. However, if the flying characteristics (flight loop) has not been accurately captured even this strategy becomes unrealistic.
In the end it then comes down to subjective opinion. Does it behave, feel, smell, sound right? The more experienced the source of that opinion, the better the result after subsequent tuning. IMHO the best source of that opinion is obviously the Type Certificate Test Pilot(s).
But to be honest, again IMHO, entering autorotation is often poorly trained both in FFS and the aircraft. So many times I see poor pilot performance matched only by the poor quality of the debrief points. Very very often Instructors fail to point out the relationship between collective and cyclic in respect of NR stability!
The collective sets the datum and the cyclic creates instability in that datum (when waggled around). In aircraft with good ATT AFCS mode, often a quick stab on the trim release and leaving the cyclic still for a moment, regenerates pilot capacity to process the subsequent actions. In this respect simulators are invaluable PROVIDED sufficient time on the OPC cycle is allocated to this exercise. Sadly, all too often, its squeezed in at the end of a session generally with little or almost no real training given.
I have to say that since this accident I genrally start the OPC cycle with a good hour of EOLs, TRDS, and TR malfunctions with the emphasis on Training the candidate rather than testing. Most pilots need this after 6 or so months blasting along in the cruise. I am very lucky, I have an FFS with the full OEM Datapack. Even so I am sometimes painfully aware hat I am training pilots to "Fly the sim" in respect to these exercises, where the aircraft may behave differently to some degree.