Some more impressions of the NTSB AA587 hearings and on-line documents:
(a) There were two related questions which I at first thought was crazy lawyer-talk, but which on thinking about it some more, may amount to a good suggestion. The questions were along the lines “wouldn’t it have been better if the fin had been WEAKER, so that part of the fin would have failed before the main attachments? Shouldn‘t progressive failure be a design principle?”. Sounds crazy, but then I seem to remember incidents where a 707 lost engines in a spin, a 707 lost about 12 feet of wing, and a Concorde lost part of a rudder, all landing safely. Progressive failure as design principle sounds like a good idea. All part of the general principle that we should be less concerned about preventing failures (because they will always happen eventually) than about what happens when there is a failure. I remember a Fire Chief saying “don’t worry about whether there is going to be a fire at your home, worry about when, because it will happen eventually”.
(b) Boeing
warns in an on-line document that “Elevators and ailerons are not designed for abrupt reversals from a fully displaced position.”. We have heard a lot about rudder reversals, but maybe abrupt reversals of other controls are not necessarily safe either. They certainly happened with AA587.
(c) The same document says that “The amount of roll that is generated by using the rudder is typically proportional to the amount of sideslip, NOT the amount of rudder deflection.”. So that if you do not know how much of a sideslip you are in (e.g. in a sudden cross-wind), you may be surprised by the a/c response to a rudder input.
(d) The same document also says “It is difficult to perceive sideslip and few modern transport airplanes have true sideslip indicators.” It goes on to point out that “ball” and “slip/skid” indicators are indicators of side-force and not sideslip. So you never do know how much of a sideslip you are in, and hence do not know in advance what the response to the rudder will be.
(e) The same document also says “Because sideslip must build up to generate a roll, there is a time lag between the pilot making a rudder input and the pilot perceiving a roll rate. This lag has caused some pilots to be surprised by the abrupt roll onset and in some cases to interpret the rapid onset of roll as being caused by an outside element not related to their rudder pedal input. If the pilot reacts to this this abrupt roll onset with another large input in the opposite direction, large amplitude oscillations in roll and yaw can result.”.
(f) In the hearing it was said that although right rudder eventually produces a right bank and turn, the first response, possibly not perceived by the pilot, is a slight left roll and right yaw. Presumably these occur during the “delay” period. The turn starts only after the yaw has started.