While reading your account of the accident to AA587 accident (Flight International, 20-26 November). I was struck by how similar it sounded to one which befell a US Navy S3 Viking in the early 1990s. That aircraft was being operated on a test flight from the US Navy's test centre at Patuxent River. The purpose of the flight was to gather data to refine a flight-simulator model, and the test plan called for the pilot to make rudder inputs in phase with the aircraft's yawing motion. Unfortunately, while the test plan had looked at the safety aspects of generating steady-state sideslip, it had not considered dynamic effects. The sideslip angle which was generated dynamically was much larger than had been expected: the resulting sideforce exceeded the design strength of the S3's vertical tail, which broke off; the crew ejected and the aircraft crashed.
If, as the National Transportation Safety Board is quoted as saying, the large sideforces generated on the American Airlines Airbus "were coincident with...large rudder deflections...which appear to be consistent with pilot control inputs" then the accident may well have much more to do with aircraft-pilot coupling (APC) than with wake-turbulence. I also note from your report that the pilots of the Airbus had carried out the airline's "recovery from extreme attitudes" training programme, but I bet that didn't include highly dynamic dutch-roll manoeuvres.
I do not wish to criticise the AA587 pilots, who were probably faced with a situation the like of which they had never encountered before. But I would like to make the point that a vital element in any kind of APC event is the pilot being "in the loop" - remove the pilot's control inputs and the APC will stop. It is, of course, too early to postulate about the cause of the AA587 accident, but I would like to make a suggestion for dealing with APC (PIO) events: if the aircraft is doing something which you really aren't expecting - let go!
Tim Price Salisbury, Wilts, UKSource: Flight International
