John,
we used to incorporate all three hardover possibilities in the S92 simulator - SAS actuator, parrallel actuator and boost actuator. (I say 'used to' meaning they still do, but I'm no longer in that function.) SAS can more or less be discounted as there is only 5% authority and it is normally cancelled out before the crew notice, but the other two failure had markedly different outcomes depending on whether in VMC or IMC, the specific axis and the crew competence. I was always curious how the boost actuators got through certification as they clearly don't comply with the normal 29.1329 requirements (in terms of pilot recognition and reaction times), so I assume they were deemed separate from the main AFCS control system - despite being required for the AFCS to function.
Sometimes the failures were given for their own right, sometimes as a means of generating an unusual attitude and sometimes to create a particular situation - especially removing the ability to use some or all upper modes. They are also good tools for communications skills, as initially only the PF knows what is going on and he has to communicate it to the PM who only knows that there has been a control upset (with a collective boost hard-up, accompanied by overtorque and low Nr if not caught quickly). If having difficulties controlling the aircraft, the PF can find it difficult to say clearly what is happening and what he'd like doing about it!
Trim failures wherein the trim runs away at whatever its rate limited value is, and the pilot has to wait the requisite 3 seconds before reacting. That whole repertoire of "fun"?
Of course, in training the expectation is the pilot will react as quickly as they can - not sit and wait like in certification. I remember Ron telling me about the joys of the triple axis hardover testing - he especially liked the nose down/yaw left/roll right combo, with the 3 second wait.....