jb747
Enthusiast
- Joined
- Mar 9, 2010
- Posts
- 12,959
Report: Air France A320 at Tel Aviv on Apr 3rd 2012, approach to stall on turning final results in Alpha Floor and flaps overspeed
Does the Airbus design contribute to this particular incident or is it a training problem ?
It's an interesting incident, but it caused me to look at another 320 accident from a few years ago, Gulf Air at Bahrain, which is even more eye watering.
Anyway ...this is the AVHerald report:
An Air France Airbus A320-200, registration F-HEPE performing flight AF-1620 from Paris Charles de Gaulle (France) to Tel Aviv (Israel) with 149 passengers and 6 crew, was on a RNAV Visual approach to Tel Aviv Ben Gurion Airport's runway 26. According to the standard instrument arrival procedure via waypoints DOVER and KEREN the crew had been instructed about 10nm prior to DOVER to reduce to minimum clean speed, shortly before overflying DOVER the crew was cleared to descend to 3000 feet and reduce to 180 knots, which the crew complied with utilizing automation in LNAV and open descent. The aircraft overflew KEREN at 3280 feet and 180 KIAS, the pilot flying however felt they were high on the approach but did not share his concern with the pilot monitoring.
How about actually doing something about it! According to the approach chart (available on the avherald site), he should have been AT 3000' overhead KEREN. Certainly fixable, but already not in accordance with the procedure. From that point he has 10.5 miles to run, so energy-wise there is no particular issue.
While on downwind the crew changed to managed speed, selected 1000 feet into the altitude window and activated open descent, the engines reduced to idle thrust. Full configuration is being selected, the Vapp of 138 KIAS is selected into the speed window.
The downwind leg has a descent profile of 2.7º, which is less than the normal 3º of a glideslope. Starting slightly high, if he'd selected 'flight path angle' and about -3.3º he would have arrived at the next way point pretty well exactly on profile (and as its actually a visual approach at this point, there is a lot of flexibility here). Doing it that way, the autothrust would have controlled the speed (and managed means it will go to the 'green dot' speed for each selected flap stage), so as flap is selected the aircraft pitch would be looking after the profile, and the autothrust would have handled the speed. What he has actually done has forced the aircraft to idle, and pitch control is handling speed. Selecting anything into the window means that it isn't managed speed, as that 'closes' if managed (i.e. FMC controlled) (at least in the AB that I know).
Still in the decent through about 1540 feet the autopilot gets disconnected, flight director and autothrust remain engaged. The aircraft turned final and was about to align with the extended runway centerline, still at 20 degrees of bank, the pilot flying applies nose up inputs for about 10 seconds while the flight director commands nose down inputs to maintain the target speed, the airspeed reduces from 135 KIAS to 122 KIAS with the pitch increasing from 0.7 to 10 degrees nose up, the pilot monitoring later provided testimony that he was monitoring the alignment with the runway.
The system expects you to obey the flight director. In this case, it's giving commands to handle the speed, and the autothrust is still at idle because of the 'open descent' selection. If he had turned off the flight director (which he obviously isn't following anyway) the autothrust would have reverted to 'speed'.
An automatic "Speed, Speed, Speed" call activated at Vapp-16 knots. The pilot flying decided to go around but did not call out the go-around.
As it is designed to do...There is a standard call for this...'Go around, flaps'. Obviously you need the other guy to understand what is happening, but you also need him to do a bunch of things...this call starts that procedure.
The pilot flying moved the throttle levers into the TOGA detent and applied nose up inputs, the pilot monitoring applied nose down inputs for about 2 seconds (dual input).
Pilot flying is now well out of his depth, isn't he. The pilot monitoring (who is the captain) probably should have taken over. His input looks to me to be in response to the speed...the dual input would have the effect of reducing or even cancelling the other's nose up input, as the system averages dual inputs. Of course, if the levers where physically interconnected, opposite inputs couldn't happen.
At that point Alpha Floor protection activated applying TOGA thrust and TOGA Lock, 3000 feet is being selected into the altitude window, open climb mode is being engaged, the speed returns into normal range, the pilots do not detect the "TOGA LOCK" status however.
The response to the speed call should have been the go around, but what he seems to have done is robotically selected the normal steep go around attitude. Given that he's almost at the stall, a much lower attitude would be appropriate until the aircraft gets back to the normal speed. So, TOGA...hold the attitude (with the FBW will do anyway unless you interfere), and then pitch up once the speed is back around 138 knots. ALPHA FLOOR activation is the stall protection in normal law. It is alway accompanied by the aircraft selecting TOGA (irrespective of the thrust lever position). As he's already selected TOGA there is no noticeable change to the power. I'd expect that they never noticed the FLOOR activation.
The aircraft climbs through 2000 feet, the crew recognizes difficulties in reducing the thrust.
Presumably at about 1,500 feet they've pulled the levers to CLB, and nothing has happened. So the auto throttle isn't being allowed to engage again. But, guess what, if you press the A/T disconnect, and simply pull the levers back a bit more, you have manual thrust control. To get rid of TOGA LOCK, you must disconnect the auto throttle, but even if you haven't recognised why the A/T hasn't got control, taking the system back manually fixes the issue. So, why does it behave like this? In the ALPHA FLOOR case, TOGA will be automatically selected even though the thrust levers are totally closed. As part of this, moving the levers has no effect and you MUST disconnect the A/T to regain thrust control. You can put them where you want, press the A/T engage, and then push to the CLB gate, and it goes back to normal. Of course, if the levers were actually servo driven, then this behaviour wouldn't be necessary.
Overall, it is a combination of training and design. Sadly, AB keep inventing more and more complex procedures which are in many cases necessary to overcome the poor man/machine interface. Their response to incidents is generally to invent a new procedure, or to try to make the aircraft even more automatic. And, it's interesting that whilst I'm being critical of Airbus here, it's actually two Boeings that have recently had autothrust related accidents (777 SFO and 737 AMS).
The flaps are selected to 1, the speed increases to 208 KIAS and still continues to increase, the gear is being extended in order to counter the speed increase, 2000 feet and 188 KIAS are being selected into the MCP, the aircraft climbs to 2500 feet before starting to descend again, the speed increased to 223 KIAS (flaps limit 215 KIAS), an overspeed alarm activated. In response to the overspeed alarm the pilot flying reduced the thrust levers to idle, which removed the "TOGA LOCK" mode and disengaged autothrottle, the engines spooled down. The pilot flying stabilized the aircraft, positioned for a second approach and landed safely.
Overall, it is a combination of training and design. Sadly, AB keep inventing more and more complex procedures which are in many cases necessary to overcome the poor man/machine interface. Their response to incidents is generally to invent a new procedure, or to try to make the aircraft even more automatic. And, it's interesting that whilst I'm being critical of Airbus here, it's actually two Boeings that have recently had autothrust related accidents (777 SFO and 737 AMS).
Last edited: