The flight crew of an American Airlines Boeing 737-800 could not stop the aircraft on the remaining runway after it touched down far beyond the approach threshold at the airport in Kingston, Jamaica, the night of Dec. 22, 2009. Fourteen passengers were seriously injured and the aircraft was destroyed when it overran the runway onto a rocky beach.
In a recently published report on the mishap, the Jamaica Civil Aviation Authority (JCAA) concluded that “the flight crew’s decision to land on a wet runway with a 14-knot tail wind, their reduced situational awareness and failure to conduct a go-around after the aircraft floated longer than usual contributed to the accident.”
The report said that in the last seconds of the approach to Runway 12 at Kingston’s Norman Manley International Airport, the captain might have been affected by a visual illusion that caused him to perceive that the aircraft was too low and to make control inputs that prolonged the touchdown.
Contributing to this “black hole” illusion were the isolated location of the airport; darkness and heavy rain; the absence of runway approach lights, touchdown zone lights and centerline lights; and the nonreflective paint that had been used for the runway markings at the airport.
Moreover, the crew had decided to conduct a straight-in ILS (instrument landing system) approach to Runway 12 despite the tail wind and rainfall. The decision was based primarily on the reported ceiling, which was close to the minimum altitude prescribed for a circling approach and a landing into the wind on Runway 30.
The pilots were not aware that a suitable global positioning system (GPS) approach was available to Runway 30. The report noted that the chart for the GPS approach was in the flight deck library and that the procedure was in the flight management system database. Investigators found “no obvious reason” why the crew was not aware of the approach.
Another possible factor was fatigue: “Although the flight crew had just had three days’ rest and their flight/duty/rest times were within the required limits at the time of the accident, they were at the end of the third flight of a long duty day,” the report said.
The pilots had flown a round-trip between Miami and Baltimore and had changed aircraft in Miami for the flight to Kingston.
“The flight crew had been on duty for nearly 12 hours and awake for more than 14 hours, and it was almost ‘bedtime’ in their recent diurnal cycle,” the report said. “The flight crew [was] possibly fatigued; however, the extent to which this affected their performance could not be determined.”
The captain, 49, had 11,147 flight hours, including 2,727 hours in type. After earning an aeronautics degree, he worked as a flight instructor and then as a pilot and operations director for a charter company. He was employed by American Airlines in 1986 as a 727 flight engineer. He progressed as a 727 and McDonnell Douglas MD-11 first officer, and then as a 727 and 737 captain.
“He was familiar with Caribbean routes and had landed at Kingston on Runway 12 before during rainy weather,” the report said.
The report did not specify the first officer’s age but noted that he had 6,120 flight hours, including 5,027 hours in type. After graduating from an aviation college, he worked as a flight instructor and charter pilot. He was employed by American Eagle as an ATR 72 copilot in 1994 and was hired by American Airlines in 1998 as a 727 flight engineer. He then progressed as a 727 and 737 first officer.
“The first officer said he had been to Kingston many times before,” the report said. “He had landed there at night and in the rain.”
The 737, being operated as American Airlines Flight AA331, departed from Miami at 0122 coordinated universal time (UTC; 2022 local time) with 148 passengers and six crewmembers. The captain was the pilot flying.
The flight proceeded uneventfully until turbulence was encountered at 37,000 ft over Cuba. The pilots described the turbulence as “fairly rough” and “real bumpy.” In-flight service was suspended several times, and the captain told the cabin crew to prepare early for landing. However, the turbulence subsided as the 737 neared Jamaica.
Scattered thunderstorms and rain had been forecast for Kingston. The airport had a single, 8,911-ft (2,716-m) runway that straddled a thin peninsula jutting into the Caribbean Sea south of Kingston. Neither Runway 12 nor Runway 30 had a runway end safety area.
The approach lights for Runway 12 had been out of service for a month due to an underwater electrical system fault. “There was not much peripheral lighting around the runway because of its location across a peninsula with sea at both ends and the absence of a nearby settlement,” the report said. “A commercial power outage had caused the airport to be operating on its standby power generator. The outage resulted in even less than normal peripheral lighting around the airport.”
At 0248 UTC, the crew received an aircraft communications addressing and reporting system (ACARS) message containing a special weather observation for Kingston that was 20 minutes old. The observation included surface winds from 310 degrees at 9 kt, 5,000 m (about 3 mi) visibility in thunderstorms and moderate rain, and a broken ceiling at 1,400 ft.
The first officer, who was handling the radios, then asked the approach controller for an update on the weather conditions at the airport. The information provided by the controller was similar to that in the ACARS message, except that the ceiling had dropped to 1,000 ft. When asked about recent arrivals, the controller said that one aircraft had landed within the past hour and “didn’t have any problems coming in.”
Jamaica’s Montego Bay Sangster International Airport had been filed as the flight’s primary alternate, but the runway at the airport was scheduled to be closed for maintenance shortly after the 737’s estimated time of arrival at Kingston. Thus, the aircraft was carrying extra fuel for the possibility of diverting to the secondary alternate, Grand Cayman in the Cayman Islands, about 270 nm (500 km) northeast of Kingston.
“This brought the calculated landing weight on arrival at Kingston very close to the aircraft’s maximum landing weight,” the report said.
Complicating the situation was that the departure from Miami had been delayed by the necessities of offloading baggage belonging to a passenger who had not boarded the aircraft and then, after leaving the gate, of coordinating with maintenance personnel on deferring a malfunctioning air-conditioning pack temperature controller according to the provisions of the minimum equipment list.
Thus, the flight was behind schedule. Just before beginning the descent to Kingston, the crew was told by company dispatch that Montego Bay had closed. “Because of the fuel needed to fly to [Grand Cayman], the crew discussed making one approach into Kingston, and, if this approach missed, they would proceed directly to their alternate,” the report said.
In 1964, Boeing began design work on a twin-turbine, short-range, narrowbody airliner that could compete with the Douglas DC-9, the British Aircraft Corp. One-Eleven and the Sud Aviation Caravelle. The result was the 737, which shared many components and assemblies with the 727 trijet, and differed from its competitors by having wing-mounted engines, a shorter fuselage and a wider cabin seating six abreast. It was the first Boeing airliner to have a two-pilot flight deck.
The first model, the 737-100, entered service in 1968, followed shortly by the -200, both with Pratt & Whitney JT8D engines. Continuous improvement over the years has increased the airplane’s size, power, payload and performance.
The “Classic” series — the -300, -400 and -500 models — was introduced in the 1980s. Boeing then began work on the “Next-Generation” 737s, which have larger wings, higher cruise speeds, transcontinental range, CFM56 engines and glass cockpits. The -600, -700 and -800 models entered service in the 1990s, and the -900 and -900ER models followed in the next decade.
The 737-800 entered service in 1998 with 27,300-lb (12,383-kg) thrust CFM56-7 engines and accommodations for 162 passengers in a two-class cabin configuration or for 189 passengers in single-class. Maximum weights are 174,200 lb (79,017 kg) for takeoff and 144,000 lb (65,318 kg) for landing. Typical cruise speed is 0.789 Mach. Maximum cruise altitude is 41,000 ft, and maximum range is 3,115 nm (5,769 km).
About 7,900 737s had been delivered by the end of 2013. The -600, -700, -800 and -900 models, as well as the business jet version, are currently in production, and a new model called the 737 Max is scheduled to begin service in 2017.
Sources: Boeing, The Boeing 737 Technical Guide, The Encyclopedia of Civil Aircraft
While briefing for the arrival at Kingston, the crew decided that due to the reported 1,000-ft ceiling, conducting a straight-in ILS approach to Runway 12 with the tail wind was a better option than circling to land on Runway 30. The published decision height for the straight-in approach was 278 ft, and the minimum decision altitude for circling was 1,150 ft (1,140 ft above airport elevation).
During post-accident interviews, the first officer told investigators that both the captain and he had conducted the ILS approach to Runway 12 many times. “In their interviews, both [pilots] said there was no instrument approach to Runway 30 at Kingston,” the report said.
The aircraft configuration chosen by the crew did not follow the recommendations of the 737 operating manual, the report said. The manual states that a flaps 40 setting should be used when landing with a tail wind, but the crew decided to use flaps 30. They also decided initially to use the autobrake 2 setting, although the conditions at Kingston dictated using either the maximum autobrake setting or manual braking.
Landing performance data provided by the operating manual for a flaps 30 landing show that at maximum landing weight, the 737-800 requires 6,795 ft (2,071 m) on a dry runway, 7,814 ft (2,382 m) on a runway reported as wet or as having good braking action, 8,440 ft (2,573 m) when braking action is fair or medium, and 11,090 ft (3,380 m) when braking action is poor.
The data also indicate that each knot of a tail wind component increases landing distance by about 191 ft (58 m). Thus, a 14-kt tail wind would increase landing distance by 2,674 ft (815 m).
At about 0304, the approach controller told the crew, “You may have to circle to land. The wind, uh, three zero degrees at one zero knots.” Notably, the controller did not mention the GPS approach to Runway 30 as an option.
“Understand that,” the first officer replied. “We can go ahead and, uh, take a straight-in.”
Despite the reports of moderate rain at the airport, the crew at no time requested, and the controllers did not provide, a runway condition report or a braking action report.
The airport “lacked operational procedures for the conduct of runway surface inspections during inclement weather,” the report said. “And the lack of agreements between the airport, air traffic service and other users for the furnishing and distribution of inspection results precluded flight crews from being apprised of the most recent runway conditions prior to arrival.”
At 0314, the approach controller cleared the crew to conduct the straight-in ILS approach to Runway 12 and advised that the runway was wet and that the surface winds were now from 320 degrees at 15 kt. The report noted that this was the crew’s second indication of an increasing tail wind.
The controller asked, “Are you able to still land, uh, make a straight-in approach runway one two?”
“We copy the wind, and we can go straight-in to one two,” the first officer replied.
The cockpit voice recording contained no discussion between the pilots about the tail wind, the rainfall or the runway conditions. Although the airline required pilots to conduct a landing distance assessment when conditions change or deteriorate after a flight is dispatched, the flight crew did not do so.
Investigators found that pilots arriving from the north generally preferred to land straight-in on Runway 12 because “it afforded a quicker and more convenient procedure … and required less taxi time to reach the terminal building.”
Postaccident interviews revealed that the 737 pilots previously had landed with tail winds on wet runways. A factor that might have influenced their decision making was that at no time during the approach did the tail wind exceed the company’s limitation of 15 kt. (The limit had only recently been increased from 10 kt.)
“The flight crew was concerned mainly with the tail wind being 15 knots or less, and the controllers appeared more concerned with the tail wind conditions than was the flight crew,” the report said. “The tail wind reports … should have triggered a serious warning for the flight crew. … However, they continued the approach without any discussion or extra briefing, and at each wind check from ATC [air traffic control] of less than 15 knots, they immediately responded that they would land with the tail wind, despite the heavy rain and the controllers’ querying of the intention to land on Runway 12.”
At 0317, the approach controller handed off the flight to the airport tower controller. On initial contact, the crew reported inbound on the ILS at 2,800 ft.
The tower controller reminded the crew that the winds were from 320 degrees at 15 kt and said, “Confirm still requesting runway one two.”
“That’s affirmative,” the first officer said. “Runway one two.”
The tower controller cleared the crew to land on Runway 12 and advised that the runway was wet. The report said that this was the first runway-condition advisory received by the crew and likely was based solely on the presence of heavy rain at the airport.
According to International Civil Aviation Organization guidance, a runway reported simply as “wet” means that the “surface is soaked, but there is no standing water.” The 737 operating manual indicates that a runway reported simply as being “wet” means that braking action can be considered “good.”
However, approach-briefing material provided to the crew contained a caution about the potential for standing water on the runway when it rains at the Kingston airport. Investigators found that neither company dispatchers nor the 737 crew were aware of this.
The report said that braking action at the time of the accident likely corresponded to the airline’s characterization of “fair/medium,” rather than “wet/good.”
Nevertheless, the controller’s advisory prompted the first officer to ask the captain if they should use a higher autobrake setting. “Runway’s wet,” he said. “You want to go to brakes three perhaps?” The captain agreed that they should select the autobrake 3 setting.
‘Feeling for the Runway’
The aircraft encountered heavy rain but no turbulence during the approach and broke out of the clouds at about 1,000 ft. The captain disengaged the autopilot at about 550 ft but kept the autothrottles engaged.
Recorded flight data showed that the captain made several nose-up pitch-control inputs as the aircraft crossed the runway threshold at 70 ft radio altitude, or about 20 ft higher than the ideal crossing height (Figure 1).
The autothrottle system was in the speed mode and holding 148 kt (the reference landing speed plus 5 kt), resulting in groundspeed of 162 kt. Flight data recorder (FDR) data confirmed a tail wind component of 14 kt, as well as a left crosswind component of 7 kt.
The captain disengaged the autothrottles at about 35 ft and manually moved the throttles to flight idle about 14 seconds after the aircraft crossed the threshold. At this point, the 737 was 3,800 ft (1,158 m) past the threshold and “floating” down the runway in a shallow rate of descent.
“The FDR showed pitch-control motions indicative of the captain ‘feeling for the runway,’ and this prolonged the flare,” the report said. “The captain did not appear to realize that the landing was going to be long [and] the first officer made no comment.”
The 737 touched down 1,130 ft (344 m) beyond the touchdown zone and 4,811 ft (1,466 m) from the end of the runway. The aircraft bounced on touchdown and settled again about 200 ft (61 m) farther down the runway. The spoilers and thrust reversers deployed, and the autobrake system activated about 4,311 ft (1,314 m) from the departure threshold.
“The captain stated that the aircraft was not decelerating as expected using autobrake 3, and he overrode the autobrake system, applying maximum manual braking … and selecting maximum reverse thrust,” the report said. “He was joined by the first officer simultaneously applying maximum manual braking.”
The captain later told investigators that he had considered initiating a go-around when he first perceived that the aircraft was not decelerating properly but expected that it eventually would slow down. There were 2,111 ft (643 m) of runway remaining when maximum braking was applied.
Investigators found no sign that the aircraft hydroplaned before it overran the runway at a groundspeed of 62 kt at 0322. “The aircraft broke through a fence, crossed above a road below the runway level and came to an abrupt stop on the sand dunes and rocks between the road and the waterline of the Caribbean Sea,” the report said.
The 737’s fuselage broke into three pieces, the nose landing gear and the left main landing gear collapsed, and the right main landing gear and the right engine were torn off the wing. Fuel leaked from the right wing tanks onto the sand, but, probably due to the heavy rain, there was no fire.
The report said that none of the serious injuries to the 14 passengers was life-threatening. The other 134 passengers either were unhurt or sustained minor injuries. “None of the flight crew and cabin crew was seriously injured, and they were able to assist the passengers during the evacuation,” the report said.
The JCAA issued several recommendations based on the findings of the accident investigation (ASW, 7–8/14, p. 8). Among them was that flight crews of all transport category aircraft should be required to conduct landing performance assessments before each landing and that the assessments should include at least a 15 percent safety margin.
The authority also recommended that operators provide flight and ground training on the hazards of tail wind landings and “firmly discourage” tail wind landings on contaminated runways or when heavy rain is falling.
This article is based on Jamaica Civil Aviation Authority Aviation Accident Investigation Report JA-2009-09, “Runway Overrun on Landing, American Airlines Flight AA331, Boeing 737-823, United States Registration N977AN, Norman Manley International Airport, Kingston, Jamaica, (MKJP), 22 December 2009.” The report is available at <jcaa.gov.jm>.