For all but one test, we ran the tanks dry with a 700-800 lb fuel imbalance (~4-5 mins), all hands off. The RAT fires up first, and the APU came on line about 20 seconds later. In one sim, we had a successful auto restart of one main engine that lasted about 2 seconds. Very violent reaction by the aircraft. Airplane banked about 80 degrees. We also tested the case for simultaneous flameouts with similar results. We found that it does not matter which engine flames out first. The plane always starts a post fuel exhaustion turn based on the pre fuel exhaustion rudder trim position dialed into the TAC.
It should be noted that we conducted no tests for any cases where it was assumed that a pilot was in control. We were focused on the ghost plane scenarios. Obviously, if a pilot was alive and engaged, then there are an infinite set of possibilities to simulate. Could have augured in very fast, changed course or glided for 80NM. But for the pilotless scenario, the sim's make it pretty clear that a fast, steep spiral descent is highly likely, ending in a very high energy impact. Such a scenario is consistent with the final BFO values that correspond to ~4500 ft/min descent rate at 00:19:29 and 15,000 ft/min at 00:19:37. At these descent rates, a very high airspeed is required, consistent with possible flutter and loss of the flaperon at some altitude before the main impact. Loss of the flaperon in flight could best explain the lack of impact damage. The flaperon could have “floated down” and made a relatively benign impact while the rest of the airplane hit very hard.