Ever since computers and airplanes came together, modal confusion and pilot perplexity have gone hand in hand. Now and again, they hook up with grave consequences.
Case in point is the Jan. 3, 2004, crash of a Flash Airlines Boeing 737-300 at Sharm el Sheikh, Egypt. It disappeared off ground radar at 4:44 a.m., Cairo time, about three minutes after takeoff. The airplane was bound for Cairo and was slated to go on to Paris. Most of those killed among the 135 passengers were French tourists, along with eight flight attendants and the cockpit crew of three: the captain, first officer and an observer in the jumpseat.
The aircraft took off, climbed normally and began a left climbing turn, as scheduled. But at 2,000 feet the turn slowly inverted to the right. The aircraft progressively rolled until it was banked 90 degrees at about 5,460 feet. It then rapidly lost height and dived into the Red Sea with no distress call.
The Egyptians produced a fact-finding report of the accident. Senior Egyptian investigator Shaker Kelada said at a press conference that it was too early to determine what had happened. "In the next stage, we will cross-examine the data we have collected, and we will then observe [whether it] was the result of a faulty automatic pilot or human error by the pilot."
However, while the investigation winds its stately course, it is possible to make a few informed judgments from the facts at hand.
The captain was the pilot flying. He was experienced, with 7,444 flying hours, although only 474 hours on the B737. An element of that experience may prove significant. He had much time in the Egyptian Air Force MiG-21, where the Russian-based artificial horizon presentation is different from the standard display in Western-made instruments. Specifically, in the Russian display, the "wings" of the aircraft symbol move, presenting the bank angle on the outside lower portion of the instrument. The pilot sees the aircraft’s actual position with respect to the horizon, as if he were observing from outside and in trail behind the aircraft. The horizon line also moves with the aircraft.
On the Western-type artificial horizon, the roll of the aircraft is shown by the slope of the horizon line, and pitch is represented by the position of the horizon line with respect to the nose of the aircraft. In the Russian display, pitch is decoupled, presented by a moving index that is separate from display of the aircraft’s bank angle.
It is well known that in stressful situations one tends to fall back on previous experience, and in this case the horizon line may for a moment have been presumed to be the bank angle, hence the pilot’s tendency to impart a right bank. Is such confusion possible? In the Swiss investigation of a Cross Air crash of a Saab 340 on Jan. 10, 2000, investigators theorized that the captain was "confused about the precise attitude and flight path."
"It suggests…at least temporary misinterpretation of the FD [flight director] display as an attitude display, by analogy with the aircraft symbol in the Russian horizon, occurred," the Swiss investigators theorized. The captain came from Moldavia and had been trained initially on Russian aircraft.
The second element of the Egyptian case involves the first officer. He was relatively junior, with just 788 hours of experience and only 242 hours on the B737. From the cockpit voice recording transcript, he shows little understanding of his vital intervention role. This despite having a good appreciation of the developing situation–although not fully understanding its cause and effect. This lethal hesitancy was, in equal parts, cross-cockpit authority gradient and lack of the moral fortitude to intervene. Cockpit resource management (CRM) trainers perhaps need to teach junior first officers to identify the ramifications of not acting responsibly and assertively.
The third element to the case may be the most significant. The B737 autopilot will not engage if there is any pressure being applied to the control column at the time. If one tries to engage autopilot in a turn, it takes a bit of practice to unload, or center the column, and then engage the autopilot. Older B737-300s were equipped with very positive paddle switches that had a magnetic lock when engaged. If they didn’t engage, they very positively sprang back to the OFF position. By this means, pilots knew the autopilot had failed to engage.
Later models of the B737 were equipped with a press-button, comparable to the other switches on the mode control panel. This was a giant leap backward: in order to verify that the autopilot has engaged, the mode annunciator panel must be cross-checked. It wouldn’t be difficult for a crew, especially a tired one–which may be the case here–to push the button while a control input was in, believe the autopilot was engaged, and miss the lack of the green annunciator on the upper EADI display.
The Egyptian Ministry of Civil Aviation’s factual report on the Flash Airlines crash informs us that, "If certain required conditions are met, the selected autopilot will synchronize the roll channel autopilot servo to the current position of the ailerons [emphasis added]." The captain called for autopilot after takeoff, but then said, "Not yet," most likely because he was still in (or rolling out of) a turn.
Unfortunately, the first officer, possibly distracted by simultaneous transmissions from an inbound aircraft, went ahead and engaged the autopilot. It kicked back due to nonsynchronization, and the disconnect warbler sounded. A radio transmission may have drowned out the sound, but someone (either pilot) reflexively cancelled the cavalry charge warbler, and the situation became as though it had never sounded at all.
The captain and first officer both assumed that the autopilot was engaged. It’s an infectious mindset. The captain, selecting turn directions via control wheel steering (CWS) inputs to his yoke, quickly became confused when it fails to function as designed, i.e., limit bank angles to 30 degree and roll out on selected headings. An indication that the No. 1 slat stayed in the extended position during flap retraction (as confirmed by the flight data recorder) may have compounded the problems of engaging the autopilot and may explain the aircraft’s persistent right banking.
But this does not explain why the crew did not take over manual control when the autopilot disconnected. Instead they tried over and over again to find a solution by attempting to reengage the autopilot while outside of engagement parameters–to no avail. Try to understand the confusion of limited appreciation, ineffective cross-cockpit communication, and a manifest inability to cope with the suddenly evident non-normal event. Ponder the crutch that the autopilot has become for those whose scan and ability to function under a high workload is otherwise deficient.
An Automation Confusion Accident?
Selected timeline of the Jan. 3, 2004, crash of a Flash Airlines Boeing 737-300
Timeline/Comment
04:43:55 — the Captain calls "Autopilot."
There was no immediate response from the flight officer (F/O) or mode changes recorded on the flight data recorder (FDR) (likely due to radio traffic),
04:43:58 — Captain states, "Not yet."
i.e., "Don’t engage autopilot until I’ve rolled wings level?"
04:43:59 — FDR records the autopilot engaged, and that the Roll mode transitioned to Control While Steering (CWS-R) mode. This transition would have resulted in loss of Heading Select mode.
F/O misunderstands and engages autopilot (A/P) anyway (1 second later). Major implications ensue … on a very black night, over the Red Sea.
04:44:00 — F/O states, "Autopilot in command, sir."
04:44:01 — Captain states "EDEELO," (an Arabic exclamation expressing a sharp response of some kind).
A result of A/P being engaged while rolling (ailerons and even spoilers deflected) then "kicks" and induces a disconnect due to no synchronization with the aileron servo-actuators. In addition the left outboard (No.1) slat was faulty and may have remained extended, inducing a minor rolling moment to the right (not a major malfunction).
04:44:02 — Cockpit voice recorder (CVR) records the A/P disconnect warning and the FDR recorded the A/P disengaged. The aural warning lasted for 2.136 seconds.
But the pilot (busy exclaiming) evidently fails to register what it was conveying and reflexively cancels the autopilot warbler — then forgets about it (outta sight or outta hearing equals outta mind).
During this time, an increase in pitch and decay in airspeed were observed. Crew is unaware that neither pilot nor autopilot is now in control… aircraft is flying itself.
04:44:05 — Captain requests Heading Select mode.
Pilot still assumes that autopilot is engaged and has control.
04:44:07 — F/O states "heading select" and the FDR records Heading Select mode engaging. This mode transition would have resulted in the reappearance of the flight director roll command bar (for direct pilot guidance). During this sequence, the aircraft’s left bank continued to decrease at a slow rate until the airplane was briefly wings level. Beginning at this time, the FDR records a series of aileron motions that command a right bank and subsequent right turn.
Less than a minute to impact. F/O positive response tends to confirm A/P is doing its duty (even though F/O has only hit the Heading Select button).
Pilot thinks that his yoke input is using CWS-R and that the aircraft will roll out on its selected heading.
Pilot is making necessary CWS roll initiation inputs to the yoke.
04:44:18 — Captain says, "See what the aircraft did?" At this point the aircraft bank angle was about 12 degrees to the right. (extended No.1 slat may have been inducing a right roll).
Aircraft has gone through the ordained heading without rolling out. Captain is now confused, still convinced that autopilot is still engaged and that he’s using CWS-R.
04:44:27 — F/O states, "Turning right, sir." Three seconds later, the captain responds, "What?" At the same time, bank angle is 17 degrees to the right and the FDR records the aileron motions to increase the right bank.
Confusion by pilot between displayed bank directions (left for right) and rolls further right? Are the command bars obscuring his interpretations of aircraft roll attitude or is he more likely confusing Western and Russian-style artificial horizon (AH) displays?
04:44:31 — F/O states, "Aircraft is turning right." One second later, the captain responds, "Ah."
02:44:35 — Captain states, "Turning right." At this point, the bank angle was 23.6 degrees to the right.
F/O attempts to keep the captain informed. A misreading of a 30-degree banked right display for a 30-degree left? Has the captain reverted to "seeing" his MiG-21’s AH presentation?
04:44:37 — Captain states, "How turning right?" Bank angle was 29.7 degrees. (i.e., "How come it’s turning right?" Puzzlement.)
Up to this point, pilot is possibly using either CWS or the inoperative A/P turn controller to counter the adverse turn, or is misreading the electronic attitude display indicator’s (EADI) bank presentation and confusing it with a MiG-21’s.
04:44:41 — Captain states, "OK, come out." At this point, the bank angle was slightly more than 40 degrees right bank and the FDR records the ailerons’ returning to just beyond neutral, the high right roll-rate stopped and a momentary left roll-rate occurred resulting in a slight decrease in the right bank from 43 to 42 degrees before additional aileron movements command an increase in the right bank.
25 seconds to impact. Pilot confusion evident, he’s attempting to get the CWS back within 30? angle of bank to enable its CWS-R ability to auto-roll out on the heading bug, but he’s misreading the EADI’s depiction of bank attitude, possibly because of the presence of the FD command bars, plausibly because he’s seeing it as a MIG21 attitude instrument display.
04:44:41.5 — F/O states, "Overbank." The bank angle at this time was just beyond 50 degrees right bank. The airplane reaches its maximum altitude of just over 5,460 feet.
04:44:41.7 — Captain states, "Autopilot." He repeats the statement at 02:44:43.4.
Pilot is disoriented; a situation perhaps compounded by the FD command bars or reversion to interpreting a Russian-style AH display.
He now suspects autopilot is at fault or disconnected and wants autopilot re-engaged/reset, but it’s outside its engagement limits anyway.
04:44:44 — F/O states, "Autopilot in command." No A/P engagement was recorded on the FDR. The bank angle was approaching 60-degree right bank. Pitch angle was zero and altitude was 5,390 feet.
Only a reassurance by the F/O. He himself still believes that the A/P remains engaged. An F/O cannot visually discern any difference between manual control and CWS-R use via the yoke. But this unhelpful reassurance serves to further mislead the captain.
04:44:46 — Captain again states, "Autopilot."
Pilot still wants autopilot engaged because he feels that he’s "lost the plot." The situation now turns critical as aircraft attitude quickly goes extreme.
04:44:48 — F/O states, "Overbank, Overbank, Overbank." The bank angle was passing through 70 degrees right bank, pitch angle was 3 degrees nose down and altitude was 5,330 feet. Two seconds later, the captain responds, "OK."
The aircraft’s nose is falling naturally due to overbank. CRM dictates that an F/O should now assume disorientation and take control, but F/O also believes A/P is in control and is quite undecided. He’s naturally reassured by the captain’s throwaway acknowledgment of, "OK." (Note the perils of mutual reassurance.)
The FDR continues to record aileron motions that increase the right bank.
Pilot is disoriented, misreading the unusual beyond-his-experience EADI presentation and rolling the wrong way.
04:44:52.8 — F/O again states, "Overbank." Bank angle was approaching 90 degrees, pitch attitude was 23 degrees nose down, and the altitude was 4,860 feet.
04:44:53.4 — Captain responds, "OK, come out." The FDR records aileron motions to increase the right bank.
Pilot continues to roll the wrong way.
04:44:56 — F/O states, "No autopilot commander." Bank angle was 102 degrees, pitch attitude was 37 degrees nose down, and the altitude was 4,100 feet.
Has F/O been trying to engage the A/P, (but it’s outside its engagement parameters)? Or has he just had a dawning realization (or confirmed via its annunciators) that the A/P is off?
04:44:58 — Captain states, "Autopilot." At the same time the FDR records a large aileron motion to the left and the airplane begins rolling back toward wings level. The maximum bank angle recorded was 111 degrees right. Pitch attitude at the time was 43 degrees nose down and altitude was 3,470 feet.
Captain is confused by the EADI display and still sees the autopilot as being either his salvation (or the cause). However his recovery actions are via the yoke and now in the correct sense. He has recognized the unusual attitude and started a positive recovery, although too late to affect the outcome (8 seconds from impact).
04:44:58.8 — The observer states, "Retard power, retard power, retard power."
04:45.01.5 — Captain states "Retard power," and the FDR records both engine throttles being moved to idle. The bank angle was 51 degrees right bank, pitch attitude was 40 degrees nose down and altitude was 2,470 feet.
04:45:02.0 — The CVR records the sound of the overspeed warning. The FDR records the airspeed as 360 knots indicated airspeed (KIAS). Recovery from severe right bank and nose down pitch continues.
04:45:04.3 — Captain states, "Come out."
Bank angle is 14 degrees right, pitch attitude is 31 degrees nose down, altitude is 760 feet, and airspeed is 407 KIAS.
Unrecoverable parameters; insufficient height available. Note how fast loss of control occurs.
04:45:05 — The CVR records a sound similar to ground proximity warning. Aircraft impacts the water at 04:45:06 with bank angle 24.6 degrees to the right; pitch angle 24 degrees nose down; vertical G load 3.9g/angle of attack 5.46 degrees, speed 416 knots/HDG 316 degrees.
Source of CVR: www.civilaviation.gov.eg/flash.pdf