Upset Prevention and Recovery Training (UPRT) Part One
I have not had the pleasure of reading PPRUNE for a time, but am quite pleased to view this thread. I have been working on Upset and LOC-I issues for the past four years, and prior to that, on simulator fidelity issues. I am currently a line Captain flying 737's, and served about ten years as a Check-Airman. I am happy to see a few posts identifying ICATEE (International Committee for Aviation Training in the Extended Envelope), as I have been working on this committee since the summer of 2009 (a sad irony was that the first day of our first meeting was the day we lost AF 447).
I have greatly benefitted from the previous posts, as it confirms many of the assumptions we have arrived at in ICATEE. By the way, there are many other excellent working groups globally working on Upset and Stall issues (two FAA ARC's, EASA, ICAO, AIAA, SAE, etc.), and there is tremendous consensus between the groups, giving us confidence that we can identify a solution to eliminating or at least reducing LOC-I in the future.
If I may, I would like to summarize the important lessons we have learned from our activities in ICATEE. Our mission is to eliminate/reduce LOC-I through enhanced UPRT. Our method was to conduct a thorough review of the accident/incident literature, and then to seek training solutions. As much as possible, we focused on identifying the existing global training infrastructure that could contribute to enhanced UPRT. Our approach is a long-term approach; we will not be able to "fix" the current deficit in UPRT knowledge and skills of existing airline pilots overnight, but we can begin moving the needle towards more robust knowledge and skill development. And we can certainly better equip future airline pilots entering the inventory. Much of what we need to repair is in eliminating negative training, emphasizing academics, using simulators appropriately, and then providing some realistic exposure to the real g-environment and psychological/physiological benefits of training upsets/stall in real aircraft.
Examples of negative training; the practical test standards (FAA) that emphasized "minimum loss of altitude" for approach to stall and stall training (now removed with new AC from FAA, but still wrongly emphasized at some airlines and operators. Regardless, we have suffered from years of this negative training that will have to be "trained-out" by proper emphasis on immediately reducing AOA).
In-appropriate use of simulators. A FFS is very accurate in the normal flight envelope, as the aero model is based on actual flight test data. FFS lack this model beyond the normal envelope. ICATEE has worked with OEM's, NASA, etc. to derive "non-normal envelope data" to produce a more accurate and capable upset model. Flight test data does require an OEM to produce approach to stall and limited stall data, but it is derived from a very benign entry, wings-level, 1-g with 1 knot per second or so deceleration up to the stall. However, most "real-world" in-advertant stalls are dynamic. We need the capability to provide an "enhanced" aero-model in the FFS, and we are close to producing that.
For now however, FFS should only be used in accordance with their accuracy; for now that should be in approach to stall training, and not full aerodynamic stall training. If pilots and instructors attempt to perform "real stall training" in simulators now, it will likely not be accurate. For example, most (if not all) swept-wing jet transport FFS will replicate a full aerodynamic stall very much like an approach to stall; the jet (FFS) will be controllable around the pitch and roll axis. This "benign" behavior will give the wrong training impression to pilots on the criticality of full aerodynamic stall; a much different condition that an "approach to stall". In the future, we will need to adopt the improved type-specific aero-models and begin providing more "accurate" aerodynamic stall training to airline pilots (precisely because of the large difference in performance of an aircraft in aerodynamic stall vs. approach to stall).
End Part One