From another thread -
Article published in a UK magazine. Relevant to the subject of the opening post on the Kenya Airlines B737-800 crash.
George Has Control - or has he?
Once known as ‘George’ the autopilot has come a long way since the first one flew in 1912. But has an over-reliance on automatic systems degraded many pilots’ basic handling skills? .
. The Jeppesen ILS approach chart was for Noumea Runway 11, New Caledonia, ICAO designation NWWW. The MSA to the north-east is 7000 ft and to the south 5500 ft. An NDB, callsign NW, is on the extended centerline 8 miles from touchdown along with another NDB and VOR at the airfield. Following the recommended procedures published in the Boeing flight crew training manual, the crew made full use of the automatic pilot systems. As the aircraft turned to final approach a fault occurred with the No 1 engine thrust lever clutch motor. Thus, when the automatics commanded a thrust increase to maintain a set airspeed, the throttle lever of that engine would not move from the last setting prior to actuation of the fault. The last setting was throttles closed idle thrust. Manual throttle operation was not affected.
The autopilot was programmed to descend the aircraft to 1800 ft until intercepting the glide path. The captain called for the flaps to be extended to slow the aircraft to approach speed and the first officer set the correct speed for the flap setting into the autothrottle mode control panel. As drag from the landing gear and flap extension took effect the autothrottle system increased power to maintain the demanded airspeed but because of the autothrottle defect only one throttle moved to increase power.
With a large split between the throttles, the autopilot tried to compensate by using a combination of aileron and spoiler controls to maintain directional control. At this point both pilots were concentrating solely on monitoring the autopilot mode status, unaware of the asymmetric thrust and the steadily increasing control wheel angle. The landing check list was being read when the automatic pilot suddenly disconnected accompanied by the aural disconnect warning. Unconstrained by the autopilot, the aircraft rolled rapidly beyond 45 degrees, and losing height below the glide slope. By now the GPWS below glide-slope warning sounded and it was then for the first time the crew saw the marked difference between the throttle settings.
Observing the No 1 (left) throttle was against the idle stop and wrongly assuming the engine had flamed out, the captain called for the engine failure and shut down checklist. The first officer hurriedly reached down to locate the Quick Reference Handbook (QRH). Still scrabbling to find the right page, he never saw his attitude direction indicator (ADI) hard over beyond 60 degrees of bank and the altimeter winding down in a blur. In real life, a searching helicopter would have located wreckage at a position five miles from the airport and a quarter of a mile to the north of the ILS inbound track. There would have been no survivors. At the time of estimated impact the simulator made a loud graunching noise and the windscreen turned red with simulated blood (a macabre modification by the technicians to the windscreen view to signify Heaven’s open door. So great was the impact, a re-boot of the simulator was necessary.
Modern airliners are equipped with superbly reliable automatic flight control and navigation systems. And certainly the remarkable reduction in CFIT over the years, attest to this. Yet there is a downside. One sage pilot described this perfectly when he said that automation robs situational awareness through absence of physical/sensory cues of flying the machine and as a result atrophies flying
and thinking skills. He added that when automation degrades it can overwhelm even a highly competent well-trained crew. Other than that, automation used as intended - and not as an aircraft babysitter - was a tool for accuracy, predictability and timeliness of action.
Automatic throttle control systems have featured in several fatal crashes. In 1992, a China Airlines Boeing 737-300 crashed 15 miles from its destination. During the visual approach, with the autopilot and autothrottle engaged, the left thrust lever advanced while the right one remained at the idle position. The crew apparently failed to notice this asymmetrical power condition and allowed the aircraft to roll into a spiral dive.
Three years later in 1995, an Airbus A310 crashed in a steep nose-down attitude three minutes after take off. According to the accident report, the crash apparently resulted from a malfunction in the aircraft’s autothrottle system when power was reduced from takeoff to climb thrust. Mechanical jamming of the right throttle resulted in this engine remaining at full-power setting, causing an asymmetrical thrust condition that led to loss of control. The crew was slow to recognize the malfunction, which occurred over a period of about 40 seconds beginning when the aircraft had reached 2000 ft and continuing after it was cleared by ATC for a left turn. During this turn, the left bank steepened, with the Airbus over-banking into an 80 degree spiral dive and 290 knots before impact. Recommendations issued by the French Director General of Aviation in the wake of this accident, included the careful monitoring of throttle behaviour when the autothrottle is engaged, and its immediate disconnection and reversion to manual control if a malfunction occurs. Of course that was only saying the bleeding obvious, but crashes due to crew mishandling still occur.
In 1998 another Airbus crashed. This time at Taiwan, during a go-around from a high approach. Instead of disconnecting the autopilot when it was apparent that it was not coping quickly enough with thrust changes, the crew persisted with autopilot operation only to stall the aircraft due to an excessive out of trim situation, killing all aboard. Four years earlier an almost identical crash had occurred at Nagoya, Japan where investigators determined an extreme out of trim condition caused the aircraft to pitch up uncontrollably until it stalled.
In 1985, a China Air Boeing 747 flying at 41,000 feet over the Pacific in cloud and in moderate turbulence, experienced an autothrottle defect that caused one throttle to stay at idle thrust. Initially, there was no action taken to turn off the autothrottle system and adjust the power manually. In spite of having several minutes to act, the pilot failed to control the yawing moment and the 747 rolled inverted into a spiral dive. It reached Mach 1.0 and 5g, losing 30,000 feet before recovery was made below cloud at 9000 feet. .
Fast forward to 2009 where a Flight International headline reported
Thomsonfly stall pinned on autothrottle.. No one was hurt when in September 2007 the Boeing 737-300 came close to stalling during a night ILS approach to Bournemouth airport, UK. The autopilot and autothrottle had been engaged at the start of the approach. However, an undetected malfunction caused both thrust levers to stay closed instead of holding approach power with flaps down. The crew did not notice the subsequent loss in airspeed as the autopilot tried to hold the ILS glide slope. The captain conducted a belated go-around with the aircraft reaching a pitch attitude of 33 degrees and speed of 82 knots before he was able to get the situation under control.
Not so fortunate, was Turkish Airlines Boeing 737-800 Flight 1951 that crashed in February 2009 during an ILS approach to Amsterdam. A defective radio altimeter fed wrong information to the autothrottle system causing the throttles to close to idle thrust. Like the Bournemouth incident, the autopilot tried to maintain the ILS glide slope by pitching up and trimming back. For over a minute the crew noticed the throttles were closed yet apparently failed to take corrective action until too late to prevent the aircraft stalling into a field, killing several passengers and crew members.
In most of the cases mentioned, prompt action by the crew to notice the worsening situation followed by immediate action to disconnect the automatic pilot system and fly the aircraft manually, may well have prevented each accident. As one pilot warned “Do you control the automated cockpit or does it dominate you?” While aircraft manufacturers encourage full use of automation, it is surely with the unspoken caveat that if automation is distracting your flight operation, early intervention should prevent things from getting out of hand. The Boeing 737 flight crew training manual covers this by recommending reduction of the level of automation as far as manual flight, to ensure proper control of the aircraft is maintained.
That last paragraph,
” reducing the level of automation as far as manual flight” reveals the importance placed by manufacturers - and indeed airline operators, on the flight safety value of auto flight systems. Yet, the countervailing argument that manual flight skills are just as essential as automation is rarely mentioned in official manuals.
Perhaps a similar misguided philosophy led to the death of all 112 on board when an Adam Air Boeing 737-400 crashed into the Makassas Straits off Selawisi Island, Indonesia. The accident summary said the cause of the accident was failure of the pilots to monitor their flight instruments while trouble-shooting the navigation system. While in cloud, the automatic pilot disconnected as the crew were engaged in discussion of an apparent defect in the inertial navigation system. Unnoticed by either pilot, the 737 gradually entered a steep spiral dive. When the penny dropped, and the captain realized his aircraft was rapidly getting out of control, he made matters worse by pulling back hard on the control column instead of using the correct procedure for recovery from an unusual attitude. The aircraft broke up in the air.
We need a happy ending for a change, so you are invited to contrast the actions of the ill-fated Adam Air crew with those of a Mystere 20 Falcon crew faced with an unexpected runaway stabilizer electric trim on short final to runway 16 at Sydney, NSW, Australia. This is an edited extract from the Bureau of Air Safety Investigation of the incident. Keep in mind, the aircraft was flying over built up areas immediately surrounding the airport because this report will make your hair stand on end. The captain had 14,100 flying hours experience while the copilot had 8000 hours.
Following a normal flight the aircraft was established on final approach for landing. At about 140 feet above the ground, the approach angle flattened and the nose pitched up. The co-pilot, who was flying the aircraft, believed that a wind-shear situation had been encountered, and he initiated a go-around. As full power was applied the nose pitched up violently. The crew did not observe any cockpit indications as to the reason for this reaction.
The pilot in command took control of the aircraft and initiated a rolling turn. This had the effect of reducing the pitch angle, and by manipulating power levers and rolling, the pilot was able to regain control of the aircraft at about 6500 feet. A safe landing was subsequently carried out. Witnesses later reported that they observed the aircraft had reached pitch and bank angles of 60 and 40 degrees respectively. Because there had been no indication of the cause of the initial pitch up, the crew had difficulty in determining the appropriate course of action. Although the investigation revealed the initiating factor was probably a full nose-up trim situation, the crew believed the problem was caused by a structural failure or a thrust reverser malfunction
This brief official description of the incident did not tell the full story, however. In his report, the pilot stated (edited for brevity):
At 140 feet above runway 16 the copilot stated that things were wrong and he was going around. Power was applied and loss of pitch control followed immediately. I took over control of the aircraft and I can still recall the IAS drop below 107 knots. I had both hands holding the control column fully forward trying to limit the rearing of the body angle. My concern was that we were going over on our back which would have been the end, as this manoeuvre started at about 80-90 feet AGL. In my mind was that the vertical lift of the aircraft had to be minimized so I rolled it to the left in a very steep turn and had the copilot close the throttles to assist me in getting the nose of the aircraft to come below the horizon. This turned out to be successful and we recovered prior to ground contact. As power was restored the nose pitch attitude reached again the 60 to 80 degrees (as some witnesses in the control tower claimed)
During the first go-around the flaps and gear were retracted and we did a series of the above manoeuvres, using roll and power off and an application technique as described above, each time gaining some height above ground level although we had very limited heading control. We did this for about 5 or 6 times and obtained about 6500 feet AMSL During this time I was unable to take either hand off the control column or relax the force required to limit the rear of the body angle. The copilot was completely occupied with the closing and opening of the throttles to help me semi-control the nose position. The aircraft eventually flew out of the problem and returned to normal operation with no further input from the crew.
Once the first recovery had been achieved I knew that I could control the aircraft to a certain extent whilst the fuel lasted. This could not have been possible without the excellent crew co-ordination between PF and PNF. A help to me was the knowledge of aerodynamics instilled by instructors during my early training on Tiger Moths.
During the technical investigation of the incident no fault could be found in the stabilizer actuator although another fault which was considered responsible may have been intermittent in nature. The captain of the Mystere Falcon exhibited superb airmanship in recovering the aircraft from a deadly situation and if a military pilot would no doubt have been recommended for the award of the Air Force Cross. This was pure flying skill at its best.
But all that happened many years ago. The game has changed since then, and now we have the situation almost world-wide where low cost airlines recruit low experience pilots as second in command of large airliners. While this may be partly due to a perceived shortage of experienced pilots, it is more likely aimed at cost-cutting where pilots are willing to pay for their own training on passenger jets. It is common to see newly graduated commercial pilots self funding a type rating on a Boeing or Airbus. Next thing is to find an airline happy to offer them a first officer position. No problem there, with agencies keen to place them (at a price, of course) with an airline willing to hire them in the right hand seat for six months. Captains are expected to look after the new boy or girl by giving them take off’s and landings in fair weather as the youngsters build up their jet hours to qualify for full-time jobs.
Because of the low experience of many new first officers, it is with some relief their mentors can plug in the automatics from take off to landing and sit back and relax while the first officer merely monitors the automatic pilot on his leg. With airline policy dictating the full use of automatics during revenue flying the exposure to the real world of manual flying for skill retention, is neatly side-stepped.
In this article we have read of fatal accidents caused by crew mis-handling the automatic systems – particularly while flying at night or in IMC. Statistics reveal Loss of Control as the new major cause of aircraft fatal accidents. This was highlighted during a recent Asia Pacific Airline Training Symposium held in Hong Kong where the Airbus representative urged the aviation industry to confront the issue of how to ensure long-haul pilots maintain basic flying skills in the face of ever-increasing aircraft reliability and cockpit automation.
“I think that at a certain point in time we need to bring back a little bit of handling”. He advocated more simulator time for pilots to hone their basic skills.
“We put people into our training today who have forgotten how to fly, basically” That would surely be the understatement of the year.
Of course not
all pilots have forgotten how to fly. But consider this true story of a 2000 hour newly recruited first officer at the controls (on automatics, of course!) of a Boeing 737. With the airport in sight at 15 miles, the captain suggested it would be a good opportunity for his Number One to disconnect the automatics and fly by hand. This perfectly reasonable suggestion was met with a shocked look and shaking of the tousled head as the first officer said no way was she going to hand-fly when the automatics did a better job – passenger comfort and all that jazz. Now it was the captain’s turn for the shocked look and shaking of head. Another time, a keen first officer asked his captain if he (the first officer) could turn off his flight director during a VMC climb so he could practice raw data manual flying. The captain refused; saying, leave that to the simulator.
While some of the more enlightened operators permit manual flying in appropriate airspace and weather conditions, others discourage it. A regional operator in Australia flying turbo-prop commuter aircraft insists crews use full automation from after lift-off to short final even in CAVOK conditions, leading to the ridiculous situation where furious knob twiddling of the autopilot takes place as the aircraft tracks from downwind to base and final in the circuit area. These are the same crews that eventually move up the food chain to fly the big jets. The regulator doesn’t mind one bit as long as rules are not broken. .
The reality is that manufacturers and airline operators will not venture from their comfort zone and encourage pilots to hand fly at their discretion. Some captains will quietly encourage a spot of manual flying here and there – flight directors on, of course. No point in stretching the friendship too far, they say. One captain interviewed observed that automation complacency is so well entrenched that many pilots politely decline his suggestion they hand fly on raw data occasionally –even if a recurrent simulator session is getting close. A few months ago, a Boeing Company check pilot talked to pilots of a major Hong Kong-based airline. In his travels, he noted that some airline pilots tended to avoid the opportunity to practice hand flying skills until the very last minute before a simulator session. This was a waste of time, he said – because manual flying skills had to be maintained throughout the year and that a hand flown ILS in good weather after months on autopilots, did nothing for basic flying skills.
Following the UK report on the near stall of the Thomsonfly Boeing 737-300 mentioned earlier, Flight International quoted a CAA observation that:
Pilots familiar with operating older aircraft, which had more variable reliability, are nearing the end of their careers, and there is a generation of pilots whose only experience is of operating aircraft with highly reliable automated systems.” The writer maintained that exercises mandated in recurrent training programmes have scarcely changed since the days of the Super Constellation. So training no longer represents what crews are likely to have to deal with today.
The solution is change the mind-set of regulators and operations departments. While Loss of Control accidents will never be eradicated, training departments should learn the lessons of past accidents and apply them to more realistic simulator training. Normally in recurrent simulator training, each crew member can expect two hours per session as pilot flying. Over a 12 month period this amounts to at least eight hours of handling per year depending on the number of sessions scheduled. Most of these sessions in the simulator will be on automatic pilot, with occasional manual flying to meet regulatory requirements. One engine inoperative flying forms a good proportion of each simulator session and much of that is on automatic pilot. This does not leave much time for practicing raw data pure flying skills. That policy must change if airlines are serious about learning the hard earned lessons from past accidents.
While LOFT exercises are valuable for practicing CRM skills, they rarely provide manual handling practice. Valuable simulator time is sometimes wasted in lengthy LOFT scenarios that require complex checklist use and associated crew discussion. Often the pilot finds he has done only two landings by the end of the session. To ensure a high standard of manual flying skills are maintained throughout the year (not an unreasonable expectation), simulator sessions should be scheduled to allow equal time between automatics and raw data manual flying. If that is a problem then additional time should be scheduled. Simulator training might be expensive but peanuts compared to the cost of a crash such as those described above.
Pilots should be given the opportunity to practice raw data hand flying in sequences covering high and very low altitude stall recovery, circuits and landings in strong crosswinds, unusual attitude recoveries, GPWS pull up manoeuvres, bad weather low level circuits, forced landings following loss of both engines, flight on standby flight instruments, instrument approaches general and all flaps up landings on limiting length runways. Additional sequences could include black-hole approaches over terrain devoid of ground lighting, and slippery runway short field landings. For the purpose of all these exercises, the automatic pilot, autothrottle and flight directors should be left off. To keep these pure flying sessions within reasonable time limits, the instructor should use the freeze and re-positioning facility of the simulator as well as common-sense reduction of QRH and checklist reading. Keep in mind the aim is to allow the pilot the opportunity to practice basic handling skills within the time constraints of simulator scheduling.
Earlier, mention was made of the superb skill and airmanship displayed by the captain of the Mystere 20 Falcon at Sydney, in recovering from a series of extreme attitudes caused by a stabilizer trim malfunction. He attributed this skill to lessons instilled into him by his instructors on Tiger Moths. Who knows that one day in the future, a pilot recovering from a one in a million jet upset in a highly automated airliner will thank his simulator instructor for teaching him the art of pure flying.