GVFlyer
Well-known member
- Joined
- Feb 22, 2002
- Posts
- 1,461
The Toulouse - Blagnac accident occured with factory test pilots flying the aircraft. I learned about the Airbus design philosphy through a Society of Experimental Test Pilots (SETP) review of this mishap.xrated said:
Here's a Der Spiegle comment and review session on Airbus accidents conducted with Bernard Ziegler Technical Director of Airbus Industrie for flight test.
Interview with Bernard Ziegler, described as Technical Director of Airbus Industrie, responsible for flight test and certification (`Zulassung') of all
Airbus aircraft.
The interview focuses on the reliability of Airbus aircraft, in the light of the following crashes: Bangalore, Feb 90 (A320: landed short of the runway in clear weather, 92 dead); Strasbourg, Jan 92 (A320 descended into a hill in clouds on a backcourse approach to the airport, 87 dead); Warsaw, Sep 93(A320, landing in a thunderstorm, overran the runway, 2 dead, many injured); Nagoya, Apr 94 (A300, copilot and autopilot in control conflict, eventually nose rose at an extreme angle and the plane stalled, crashing tail first onto the ground, 246 dead); Toulouse-Blagnac Jun 94 (A330, testing engine-outgo-arounds, stalled and crashed, 7 dead including the Airbus chief test pilot). The Habsheim A320 accident is not mentioned. The header to the intro says:
"Airbus Industrie is under pressure. Twelve total-losses since 1987 with 815 dead have awakened doubts about the concept of airplanes dependent on electronics. Do technical failures contribute to the series of accidents? Or are pilots overextended by the`flying computers'?"
Ziegler says they've had a lot of bad luck recently. He says there's no reason to change the Airbus `philosophy' of taking over some of the pilot's tasks by computer, pointing out that all of the new technology developed by Airbus, from `glass cockpit' to new types of autopilot, has been followed by `all the others'. And, ` the pilot still has the last decision. Whoever suggests the contrary doesn't know what they're talking about.'
They discuss the problems in Warsaw concerning the late deployment of airbrakes and thrust reverse, concerning which he points out that (a) it's a requirement for all modern airplanes that deployment is not enabled until the plane is firmly on the ground; and (b) there are particular limits on landing, for example not when a tailwind is stronger than 10 knots, or when the landing speed is too high. In Warsaw, these boundaries, which were carefully ascertained in test flights, were crossed. When asked why the Polish investigators singled out late deployment of airbrakes and reversers... he said that the level of the compression sensors on the landing gear, and the landing logic, has been changed for Lufthansa at the request of the client, but that only an expert can tell the difference between the old and the new landing logic.
There follows a discussion about computers vs other kinds of flight control, during which he says that there is in principle no difference between more traditional methods of control and the fly-by-wire of the A320, and that it's an illusion to believe that there's ever a direct connection between the pilot's hand and the behavior of an airplane - flying is in this sense something artificial.
He is asked about the involvement of the autopilot in Nagoya, and about a prima facie similar problem with an autopilot in 1991 in Moscow (an A310), and why Airbus had not modified all the autopilots of these types. He replies that requiring expensive modifications is not a simple matter, and must first be thoroughly investigated to see if they cause more problems than they solve. He says that in conjunction with the certification authorities, Airbus had developed an autopilot modification and recommended that A300-600 clients perform it, and after the Moscow incident had notified everyone officially of the correct use of the autopilot mod. After the Nagoya accident, Airbus decided that the modifications they had recommended to A300-600 and A310 aircraft should be mandatory.
Concerning the A330 test flight crash in Toulouse, he points out that it was a difficult but not dangerous test, and in response to a question concerning entering the right autopilot `flight level', he points out that it was mistakenly left at 2000ft but should have been at 7000ft according to the checklist. He says that the fundamental error was that the crew let the nose-high, low-speed situation persist too long, and speculates why: because they took the nose-high situation for an anomaly and they wanted to see what would develop [according to the preliminary report, it was pilot commanded. They were confused as to which mode the flight control was in]; because the test engineer trusted the pilot to know when to return to normal; and Nick Warner [chief test pilot Airbus] had been critised before by test engineers for retaking control too quickly, and maybe was sensitive to potential criticism in this case also. It was a question just of two seconds delay.
The consequences, he says, will be that automatic protection will be developed that will rule out such extremely unlikely accidents, and that the A330 and A340 are the first aircraft to be protected automatically against the development of such a flight condition.
A few comments -
Warsaw: Ziegler correctly points out regulations concerning thrust reverse and airbrakes. However, no mention was made by the interviewer or Ziegler of the wheel brakes themselves. The wheels did not spin up on landing to the required speed to allow the anti-skid system to function as designed. Ziegler's selection of the tailwind for commentary raises some hypothetical considerations. At the given landing speed, with the tail wind, the wings were developing less lift than they would have been without the tail wind, making it more likely that the braking functions would have been enabled by
the sensors. On the other hand, had there been no tailwind, the pilots would have landed at the same indicated airspeed, which would have given them 10 kts slower ground speed, but the same amount of lift preventing the sensors from indicating ground contact. For similar problems not to have occurred in this situation, the wheels would have not to have aquaplaned at this slower landing speed. But in the accident situation, they did not appear to spin up to speed until the ground speed was well below this, and much more of the airplane weight was on the wheels.
It's a simple consequence of the landing logic that braking systems did not deploy under the landing condition in Warsaw, as may be seen from an inspection of the description of the logic in the Flight Crew Operating Manual.
Bangalore: it appears the pilots were confused as to which control mode the airplane was in. Under the particular conditions of flight, the engines went to flight-idle and the airplane descended rapidly into the ground while the
pilots were trying to figure out what was going on.
Nagoya: The autopilot appears to have been engaged and in `go-around' mode (`abort landing, gain altitude quickly'). The copilot, who was flying, was pushing hard forward on the control column trying to land the airplane. The autopilot was counteracting this by configuring the airplane aerodynamically for full nose-up.When the copilot eventually let go of the column, the airplane's nose rapidly rotated upwards to an extremely high angle and the speed decayed severely, causing the aircraft to stall nose-high, close to the ground. It hit the ground tail-first. The standard procedure in which pilots are trained is to disconnect the autopilot and ensure it is disconnected if they want to hand-fly the plane onto the runway. There are numerous puzzles concerning this accident.
Toulouse: Under the correct checklist settings, the pitch of the aircraft on takeoff would have been automatically controlled when the autopilot was engaged. The co-pilot who was flying rotated on take-off to a high angle. Meanwhile, Warner engaged the autopilot (which took three tries) and `failed' the left engine. It's surmised they were expecting the autopilot to return the aircraft to a precise pitch as it handled the situation, as planned. The aircraft was flying in a different control regime than planned due to the mistaken altitude-capture setting of 2000ft rather than 7000ft on the autopilot. Pitch was not `protected' by the autopilot in this regime. Speed decayed rapidly since the nose did not go down, the aircraft was unable to maintain lateral control when it was below the airspeed required to do so, and yawed and rolled. After this situation had developed, Warner throttled back the right engine to regain lateral control, as well as regaining wings-level and nose-level. When control was regained, the ground was just a little too close.
What do these mishaps have in common?
GV
~
Last edited:
