A nosewheel’s hard impact during a landing last year shows how pattern analysis can lead to safer practices. In this event, the analysis followed the costly incident, but the broad lesson about "data-driven safety" applies nonetheless. As the saying goes, unexamined data constitutes a "data morgue."
The incident involved a Monarch Airlines B757 landing May 22, 2002, at Gibraltar. After the main gear tires touched the tarmac, the captain put the nosewheel down hard–not enough to break the strut or burst tires but sharp enough to inflict significant damage to the forward fuselage in the nose gear area. The airplane was out of service two months for repairs.
The touchdown triggered an investigation by the UK’s Air Accidents Investigation Branch (AAIB). Details of the incident are contained in the AAIB’s January 2003 bulletin (No. 1/2003).
After an uneventful flight from Luton, the airplane touched down at 132 knots, wings level, and as the air/ground logic shifted to the ground mode, the speedbrake handle moved from the armed to the deployed condition. Nothing exceptional–there are tens of thousands of landings like it.
Over the next second, however, the elevator position changed from +15.6 degrees nose up to —20 degrees nose down (the elevator has a range from +30.5 degrees up to —20.5 degrees down). As a result of the imparted momentum, the nose gear struck the tarmac at a rate of 10 degrees per second, well above the design limit of 7 degrees per second. The resulting impact caused considerable damage:
The B757 nose gear, mounted in a rectangular box sometimes known as the "dog house," is located in the underside of the forward fuselage. The walls and roof of the dog house did not evidence visible distortion. But the skin on both sides of the dog house was wrinkled considerably, and several rivets had been pulled by the strain of the impact through the skin.
A bulkhead coincident with the rear wall of the dog house was buckled, too. Inside the cabin, the cockpit door was binding on the floor.
The captain, who was the pilot flying, told investigators he thought he’d been "holding the stick back."
This is where data can serve as a useful means of measuring such impressions. The airplane was equipped with the statutory digital flight data recorder (DFDR) and also with a quick access recorder (QAR). Data was being fed weekly on all flights to the operator’s flight data monitoring (FDM) program, also known as a flight operations quality assurance (FOQA) program.
Records of 74 previous flights were examined, 18 of which involved landings at Gibraltar. The incident pilot had flown three of these other landings and one at Luton. In all four of these landings, the data showed the pilot had applied full nose-down elevator shortly after touchdown.
An Unconscious Habit
The AAIB report goes into some detail as to how the pilot might have acquired this habit. His training records gave no hint of a non-standard landing technique. The first officer on the incident flight hadn’t noticed the captain’s application of full nose-down elevator. Upon deeper analysis, it appeared that the captain’s technique for landing at Gibraltar differed from that at other destinations, with an evident, if unconscious, propensity to apply full nose-down elevator right after main landing gear touchdown. On the incident flight, the nose attitude at touchdown was higher than three previous landings of interest. A higher angle meant more room to build up downward momentum.
Although the captain professed no concern about the landing at Gibraltar that day, the AAIB suggested that a "a number of factors may have had a subconscious effect." These included:
Gibraltar Airport had a reputation for difficult wind conditions. Indeed, Monarch had imposed minimum experience levels on captains operating into Gibraltar.
The runway at Gibraltar is relatively short, although it was adequate for a B757.
The runway has no overrun, bounded as it is by the sea at each end.
During the incident flight, the lack of the No. 1 engine thrust reverser may have contributed to a slightly more demanding landing.
The collective conditions may have reinforced what had become an acquired–albeit unrecognized by the pilot–handling habit.
The Larger Implications
Why wasn’t this pilot’s propensity to aggressively lower the nosewheel caught previously? Both tendency and trend were apparent in the retrieved historic FDM data. For some months the captain had occasionally been using full nose-down elevator, and "the frequency of use had increased toward the date of the accident," the AAIB noted.
Typically, FDM programs employ computerized searches of QAR flight records for specified variances. Out-of-bounds occurrences (e.g., hard landings, excessive speed for flap deployment) are flagged as "events" for further analysis.
In this case, the pilot’s propensity had not been noticed because excessive nose-down elevator during landing had not been programmed as an "event" in the FDM software. A Monarch official explains that there are "hundreds of thousands of possible combinations" of data which have no significant safety implications and, if selected, "would simply clog up the system and render it useless as a safety tool.
"Usually elevator position following main gear touchdown was not selected as an ‘event’ in our system, as nobody had anticipated the type of control inputs that caused [this] incident," the official adds.
Immediately after the incident, Monarch modified its FDM software to include nose-down elevator as an "event."
The airline also modified its training programs to improve awareness of the damage potential from excessive nosewheel pitchdown. The UK’s Civil Aviation Authority (CAA) advised other operators to look at their programs accordingly.
The AAIB sees a larger implication: The day is fast approaching when operators of all transport-category aircraft weighing more than 60,000 pounds (27,000 kg) must have FDM programs in place. If the potential for FDM programs "is to be fully exploited," the AAIB intoned, then the software needs to be customized to suit each carrier’s aircraft, standard operating procedures and route structure.
This bland bureaucratic admonishment indicates that relevant performance parameters need to be chosen to maximize the potential for pre-emptive pattern analysis. Or, more bluntly, it’s preferable to close the barn doors of safety before the horses of hazard get out and cause accidents. (The full AAIB bulletin may be viewed at www.aaib.dft.gov.uk/bulletin/jan03/htm.)
Other Issues
From one accident, a range of direct and ancillary issues emerge. The May 22, 2002, incident involving a Boeing 757 operated by Monarch Airlines links to the following:
Flight data monitoring (FDM) programs: As of Jan. 1, 2002, FDM programs were declared a "recommended practice" by the International Civil Aviation Organization (ICAO) for operators of aircraft weighing more than 44,000 pounds (20,000 kg). FDM programs become a "standard," effective Jan. 1, 2005, for operators of aircraft weighing more than 60,000 pounds (27,000 kg).
Minimizing vulnerability: The British Air Accidents Investigation Branch (AAIB) noted that cabinets containing electrical power distribution panels had been driven upwards into the floor beams by the force of impact. This upwards deflection was not enough to contact or crush many flight control, engine and braking system cables that run to the flight deck. However, the AAIB cited a similar September 1999 nose landing gear impact during a B757 landing in Gerona, Spain, where "cables were severed or jammed, resulting in the engines being rendered uncontrollable."
"It should be noted that nose landing gears are not subject to the provisions of JAR 25.721a; these apply to the main landing gears and deal with fuse pins that allow the gears to break away in the event of a heavy impact so that damage to fuel tanks is minimised," the AAIB noted.
Recall also the Sept. 23, 1999, landing overrun of a Qantas B747-400 at Bangkok, in which the nosewheel gear collapsed and the structural dog house was damaged, causing loss of the public address system (May 2002, page 47). In its April 2001 report of this accident, the Australian Transport Safety Bureau (ATSB) noted that the main and alternate public address (PA) control systems were located in the main equipment center. During the impact sequence, the nose landing gear collapsed rearward and upward right into this area, knocking out the PA system. The ATSB said the loss illustrated an inherent "design weakness" by not separating the primary and alternate systems:
"The loss of the PA and cabin interphone systems was potentially hazardous … [contributing] to various communication difficulties and key personnel leaving their stations.
"Nose landing gear collapse is likely to be part of…any runway excursion involving collision or rough terrain. In B747-400 aircraft, the loss of cabin PA…would likely be a consequence of such a sequence. This highlights a design weakness inherent in having both the normal and alternate system co-located in the equipment center in the lower fuselage behind the nose landing gear bay." See www.atsb.gov.au/aviation/acci/ojh/vh-ojh.pdf, page 64.
Data filtering: Readers will recall that U.S. National Transportation Safety Board (NTSB) officials expressed their dismay with the discovery of filtered DFDR data in their investigation into the fatal Nov. 12, 2001, crash of an American Airlines A300-600 (October 2002, page 45). The AAIB noted that the absence of filtered data facilitated its investigation into the Gibraltar incident:
"Early Boeing 767 and Boeing 757 DFDRs used EICAS [engine indicating and crew alerting system] data as the source for recording elevator position. Unfortunately, these values had been filtered and smoothed and were not representative of the actual position during changes in surface deflection.
"This anomaly, highlighted in previous NTSB and AAIB reports, resulted in the development of a modification that provided raw surface position to the FDR whilst retaining the smoothed/filtered values for EICAS presentation.
"This modification had been effected [on the accident aircraft] with the result that the elevator positions stated are correctly correlated with timings of other events."