American B738 at Kingston on Dec 22nd 2009, overran runway on landing

Jamaica's Civil Aviation Authority (JCAA) have released their final report concluding the probable causes of the accident were:

The investigation determined that the most probable cause of this accident was that the aircraft touched down 4,100 feet beyond the threshold, and could not be stopped on the remaining runway. The flight crew’s decision to land on a wet runway in a 14 knot tailwind, their reduced situational awareness and failure to conduct a go-around after the aircraft floated longer than usual contributed to the accident.

The investigation released following findings:

Findings as to Causes and Contributing Factors

(Definition: “Each Finding identifies an element that has been shown, through the results of thorough analysis, to have operated in the occurrence or to have almost certainly have operated in the occurrence. These Findings are related to the unsafe acts, unsafe conditions or safety deficiencies which are associated with the safety significant events that played a major role in causing or contributing to the occurrence”).

1) In the dispatch document, the flight crew was not provided with an accurate and current report on the runway condition at Kingston, nor was it required.

2) The flight crew did not review the Page 10-7X or the approach options, and consequently was not aware of the standing water warning at Kingston on Page 10-7X, and was not aware of the RNAV (GPS) Rwy 30 approach at Kingston.

3) ATC did not offer AA331 the option of the RNAV (GPS) Rwy 30 approach (the flight plan showed the aircraft was RNAV capable).

4) The flight crew’s Situational Awareness was incomplete in that they did not realize that the standing water warning of Page 10-7X, the heavy rain, the weather reports they were receiving and the lack of runway condition reports or braking action reports indicated that a Medium/Fair braking action condition was a possibility, and hence was the worst case scenario.

5) The flight crew decided to land in heavy rain on a wet runway in a tailwind close to the tailwind landing limit.

6) The flight crew did not perform an adequate landing distance assessment.

7) The flight crew did not use the RNAV (GPS) Rwy 30 approach, and land into wind on runway 30.

8) The flight crew initially briefed to land with autobrake 2, then changed this to autobrake 3 on final approach, whereas “MAX autobrakes or manual braking” was the recommended American Airlines procedure for the conditions.

9) The flight crew did not plan for “the most adverse conditions”, as instructed in the American Airlines B737 Aircraft Operating Manual.

10) The flight crew elected to land with flap 30, rather than the flap 40 recommended for short field and tailwind wind landing in the AA B737 Operating Manual (See 1.17.1.1.8).

11) The flight crew did not adjust their landing plan to the rapidly changing weather conditions.

12) The flight crew did not select the most suitable runway for landing.

13) The captain did not disengage the autothrottle when he disengaged the autopilot, nor was this required per AA SOPs, although it was part of Boeing recommended procedures.

14) The aircraft crossed the runway threshold 20 feet above the ideal height, and landed long.

15) The captain did not follow the company SOPs for landing technique and go-around.

16) The captain did not follow company recommendations for landing configuration, for landing with a tailwind and on a wet runway, or for the landing profile to be flown.

17) The captain pitched the nose of the aircraft up when passing over the threshold, resulting in the aircraft floating in the flare, then landing long.

18) The first officer did not provide all the necessary and appropriate monitoring and CRM input during the flight, especially during the final stages of the landing.

19) The aircraft touched down at 4,100 feet from the runway threshold.

20) The flight crew planned to land the aircraft without determining the runway surface condition and the braking action.

21) The aircraft did not land within the desired touchdown point or within the touchdown zone.

22) There was a 14 knot tailwind component when the aircraft landed.

23) The aircraft touched down at Vref plus 5 knots, thus increasing the landing distance required.

24) The flight crew did not conduct a go-around when the long landing made this necessary.

25) There was evidence of heavy rainfall and reduced visibility at NMIA before and during the landing of AA331.

26) There was reduced friction on the runway, as evidenced by the longitudinal deceleration rate recorded on the FDR. The presence of melted rubber balls on some of the aircraft’s main landing gear tires could also be an indication of reduced friction.

27) There was evidence of water on the runway at NMIA, from the measured rainfall, ATC runway report, weather reports, FDR data and the captain’s statement.

28) The flight crew did not apply maximum manual braking until the aircraft was more than 6,800 feet from the threshold.

29) The application of speed brakes, maximum manual braking and full reverse thrust was not sufficient to stop the aircraft before the runway end.

30) The flight crew’s situational awareness became degraded as the flight progressed.

31) The flight crew members were possibly fatigued after being on duty for nearly 12 hours, and awake for more than 14 hours.

32) The CRM in the cockpit was not adequate, and the first officer, as “pilot monitoring” did not call for go-around when the aircraft was landing long.

33) The American Airlines staff at Kingston did not follow the American Airlines, Kingston Station Manual procedures regarding runway condition reporting to AA Dispatch during inclement weather.

34) The “Field Report” of “0.10 IN WATER” by AA staff at Kingston was not the result of any measurement or inspection, but only indicated that there was water on the runway.

35) The AA dispatcher was unaware of the Page 10-7X standing water warning at MKJP.

36) AA had not made mandatory to B737 flight crews in Bulletin 737-07 the FAA recommendation in SAFO 06012 that flight crews conduct landing performance assessments before landing, although AA Flight Safety Programs management expected that all AAB737 flight crews would conform to Bulletin 737-07.

37) The use of the “advance analysis” in place of conformance to Bulletin 737-07, was considered to be acceptable by AA Flight Safety Programs management.

38) The “advance analysis” used by the AA331 flight crew was not adequate as it did not stipulate flap setting, braking technique, planned touchdown point, use of reverse thrust or runway condition/braking action.

39) The “advance analysis” used by the AA331 flight crew did not meet the requirements of a landing performance assessment.

40) AA stated “Note: no AA documents or training materials specifically define the ‘advance analysis’ concept that the JCAA cites in the draft report. AA flight crews are not trained to use, required to use, or discouraged from using this method.”

41) The advice in FAA AC 91-79, to assume standing water when rain was falling on the runway, was not included in the American Airlines operating procedures, nor was it required to be included.

42) There was no evidence that AA B737 flight crews had received ground or flight training related to the techniques and considerations for landing in tailwind conditions, and the attendant hazards, even when the B737 tailwind landing limit was recently increased from 10 knots to 15 knots.

43) The flight crew did not follow the requirements of SPC MSG NBR 9482, which stated “As always, pilots must ensure the reported tailwind component complies with airplane performance requirements for the runway in use.”

44) ATC runway surface condition reporting did not fully conform to the ICAO recommendations.

45) The flight crew did not request a runway condition report or a braking action report from ATC.

46) ATC did not alert the crew that no braking report had been received, as required by ATS MANOPS.

47) ATC did not inform AA331 that the runway was wet until less than five minutes before the aircraft landed.

48) ATC did not inform AA331 of the reported “heavy rain”.

49) ATC did not assign runway 30, the into-wind runway, as the active runway, as required by ATS MANOPS.

50) ATC did not follow the ATS MANOPS in terms of active runway assignment, placing of Weather Standby, reporting of weather, and giving the arriving traffic a braking action report.

51) The Enroute and Approach controllers gave the AA331 flight crew estimated weather reports, and did not state that this was ATC observed weather, not official weather reports.

52) Neither NMIA nor ATC had any specific procedures for conducting runway condition inspections during inclement weather, and disseminating this information to landing traffic, contrary to ICAO recommendations.

53) The recommended ICAO terminology for describing water on a runway was not used by ATC, NMIA or American Airlines.

54) There was a lack of consistency in the terminology used for runway condition reporting by ICAO, ATC, NMIA and American Airlines.

55) Runways 30 and 12 at NMIA did not have embedded centre-line or touchdown zone lighting, and the painted runway markings did not contain reflective material, as recommended by ICAO.

56) NMIA did not perform a runway surface condition inspection before AA331 landed, nor was this part of their procedures.

57) There was no Runway End Safety Area (RESA) at the end of runway 12.

58) The investigation indicated that there may have been some weaknesses in the FAA oversight of the AA Boeing 737 operations related to tailwind landing training, and approval of an increased tailwind landing limit.

59) The investigation indicated that there were some gaps in the management and operational control of American Airlines, specifically runway condition reporting, and flight crews following recommended procedures.

60) There was no MOU between NMIA and ATS that bound NMIA to keep ATS currently informed of runway conditions, in accordance with ICAO Annex 11, Chapter 7, 7.2, and ICAO Annex 14 2.9.2.

3.2 Findings as to Risk

1) The L1 door slide deployed prematurely, causing the door to be jammed and unusable for evacuation.

2) The captain’s seat crotch belt bracket broke during the impact sequence.

3) Although most of the overhead bins remained in place, the overhead bins in the first class section through the forward break, and at the aft break, dislodged.

4) All of the passenger service units throughout the airplane became dislodged.

5) The AA KIN FIELD REPORT of “0.10 IN WATER” was based only on the observation of water on the apron, and not of water on the runway

6) The AA landing performance tables were predicated on landing at 1,000 feet from the threshold, whereas AA Standard Operating Procedures spoke to the “desired touchdown point” as being 800 – 1,500 feet, and the “touchdown zone” as being 1000 – 3000 feet, or first third of the runway.

7) Both of the cabin crew members seated at L1 reported that they were unable to retrieve their flashlights because the doors to the emergency equipment compartments under the jump seats were jammed due to the buckling of the floor.

8) Some parts of the aircraft’s emergency lighting system did not function after the accident.

9) There was a conflict between the recommendations of the NTSB and FAA for flight crews to perform landing distance assessments according to specific parameters, and the much less specific “advance analysis” used by the AA331 flight crew.

10) There was no written or recorded record of the ATIS, and none was required by ICAO, JCARs, or the Air Traffic Services Planning Manual (Doc 9426).

11) JCAA ATS lacked recurrent training and proficiency checks for controllers.

12) There was inconsistency between AC 91-79 and SAFO 06012 related to runway surface condition reporting.

13) There was no clearly defined policy in the American Airlines B737 Operating Manual to the effect that the first officer could call for a go-around, with it being compulsory for the captain to follow through with this.

The JCAA reported that the captain (49, ATPL, 11,147 hours total, 2,727 hours on type) was pilot flying, the first officer (ATPL, 6,120 hours total, 5,027 hours on type) was pilot monitoring.

The JCAA reported that it did not become clear during the investigation whether during the approach to Kingston the crew had obtained ATIS and had performed the pre-approach briefing, the cockpit voice recorder did not capture ATIS or the briefing. The crew however called the approach controller prior to leaving cruise level asking for the weather and received weather information, that was not consistent with the official weather report. The weather data provided by the approach controller indicated winds from 310 degrees at 7.5 knots, visibility 5 miles, moderate shower at the station, temperature 21, dew 20, broken cloud at 1000 feet.

While descending to 15000 feet the crew called approach again asking whether any turbulence had been reported with the reply in the negative.

After being handed off to the approach controller the crew were told to expect an ILS 12 approach and reported that the weather was unchanged with winds from 310 degrees at 10 knots (American Airlines limit of tailwind at 15 knots) and added, that the crew might need to circle to runway 30 stating winds were now from 320 degrees at 10 knots. The crew requested a straight approach in runway 12.

While the aircraft descended to 4000 feet the approach controller advised the winds were now from 320 degrees at 15 knots and queried whether the crew was still to land on runway 12 which the crew replied to in the affirmative.

So far there had been no mention of the runway condition at all.

At 2800 feet the crew intercept the localizer and was handed off to tower, tower advised winds were from 320 degrees at 14 knots and queried whether the crew still requested runway 12. The crew requested runway 12, tower cleared the aircraft to land on runway 12 and added "be advised runway wet" (which was the first mention of the runway condition, less than 5 minutes before touchdown).

The first officer later reported that during final approach the aircraft's systems indicated a tail wind of 8-9 knots, he could not recall the tail wind indication at touch down. The aircraft had been configured early, flaps 30 were selected, and was on localizer and glideslope, after breaking through the cloud with intermittent view of the ground at 2000 feet, consistent view of the ground at 1500 feet and breaking clear of cloud at 1000 feet, the visual approach slope indicators (VASI) were indicated they were "right on it".

At about 550 feet the captain disconnected the autopilot but kept autothrottle engaged.

Descending through 500 feet the first officer called "On Speed, Sink 800", in post flight interviews the first officer stated the captain flew a nice approach, was never too high or too fast. The approach however was noisy due to heavy rain, the windscreen wipers were on.

Descending through about 500 feet the captain began to align the aircraft with the runway flying a slight right hand turn.

Both pilots stated in post flight interviews the aircraft was "right in the slot" while crossing the runway threshold.

The flight data recorder indicated that the aircraft crossed the runway threshold at 70 feet, the crew therefore received a high indication on the VASI and glideslope indicator.

The captain said in post flight interviews that the Head Up Display gave too much information and did permit to be decluttered. The HUD provided a "fake runway" consistent with the offset localizer, the real runway and the flight path vectors, which was too much information.

According to the flight data recorder the captain provided nose up inputs at about 70 feet AGL, disconnected authrottle at 35 feet AGL and pulled the thrust levers to idle. Up to that point the autothrottle had maintained Vref30+5 knots, 148 KIAS, which was 162 knots over ground.

Due to the pitch up the aircraft maintained a shallow rate of descent, the aircraft floated and touched down 4100 feet past the runway threshold and 1130 feet past the end of the touch down zone. On wheel spin up the spoilers automatically deployed.

The aircraft bounced and touched down a second time 200 feet further down the runway, automatic brakes setting 3 activated about 4600 feet past the runway threshold, the thrust reversers were deployed.

The captain stated in post flight interviews that he did not feel deceleration as would normally occur with autobrakes 3 and applied manual braking on the brakes pedals and applied maximum reverse thrust. The first officer assisted with the brakes pedals. A full brakes pressure of 3000 psi activated according to the flight data recorder indicating maximum brakes had been applied indeed.

Both flight crew said in post flight interviews that they both sensed an abnormal lack of deceleration despite autobrakes and maximum brakes and soon realized they would leave the runway.

The aircraft crossed the runway end at 62 knots over ground, passed through the chain link airport perimeter fence, crossed the road and came to a stop on the sandy and rocky shore line area.

The cockpit was dark after the aircraft came to a stop, the crew began to work the emergency evacuation checklist.

14 passengers received serious injuries, the 6 crew and 140 other passengers received no or minor injuries.

The aircraft was substantially damaged with the right main gear and right engine torn off, the left main gear and nose gear collapsed and the fuselage broken up into three parts, wings and flaps damaged and the right wing tanks ruptured.

The JCAA reported a runway survey revealed:

- There was evidence of edge damming on both sides of the runway.

- There was evidence of water flowing off both sides of the runway, then soaking away.

- Maintenance of drainage system was poor.

- There were obstacles on runway strip. (rocks, old concrete blocks and uncovered drainage cisterns).

- Runway surface had polishing in touchdown zone of runway 12.

- Surface was not deformed, and had positive transverse slopes along entire length.

- There was no cracking, apart from longitudinal joint cracks on entire length, and lateral cracks associated with construction, plus some block cracking in touchdown zone of runway 12.

- Runway was in good condition overall.

There was "no evidence of the drainage system in the

runway strip being linked to the main drainage system or drainage ducts routed under the apron".

The runway had no runway end safety area (RESA) on either side, the surface had not been grooved.

Post accident examination of the tyres revealed no evidence of hydroplaning and no evidence that any wheel had locked up during landing. However, the right hand tyres showed melted rubber balls while the left hand tyres did not. The JCAA wrote: "The rubber balls on the right main tires could possibly be an indication that the tires were turning and the brakes were working to slow them, but there was reduced friction such that the rubber balls were not removed during the braking process. With good friction, the rubber balls would normally be scrubbed off by the runway during braking. There was no ready explanation why the tires on the right gear should have melted rubber balls, and the tires on the left gear did not."

The JCAA analysed: "Although the AA331 flight crew was aware of the circle-to-land procedure to runway 30 from the ILS Runway 12 approach and that option was suggested to them by the Approach controller, they decided not to conduct that approach as it required a higher ceiling than was reported and they decided (as stated in the first officer’s post-accident interview) that the straight in to runway 12 was more appropriate than doing a circling approach, and had more chance of a successful outcome. Given the weather conditions, and the fact that the flight crew was not aware of the RNAV (GPS) Rwy 30 approach, this was an appropriate decision." and stated further: "If the flight crew had used the RNAV (GPS) Rwy 30 approach and landed on Runway 30, the accident would probably have been avoided."

The JCAA analysed: "In the latter stages of the transition to the final approach fix, some eight minutes before landing, AA331 was cleared for the ILS Rwy 12 and, in the same transmission, was advised by the Approach controller that the wind was from 320 degrees at 15 knots. The first officer read the clearance back and then the Approach controller asked if AA331 was still able to make a straight in to runway 12 and gave the wind from 320 degrees at 14 knots. The AA331 flight crew replied in the affirmative.

The CVR contained no spontaneous discussion between the flight crew in response to this weather information that was given by the Approach controller, which was the second indication to the flight crew that the tailwind had increased. This did not prompt further discussion between the flight crew regarding adjusting the landing technique (using additional flap or braking) for the increased tailwind, which would be expected. This lack of discussion may indicate that their CRM was not adequate. It is possible that the flight crew may have been experiencing some attention tunneling due to distraction by the heavy rain they were flying through, and concentration on the weather picture on the radar between the aircraft and the runway. The early flap selections and speed reductions for fuel still to be burned off for landing, and announcing HUD settings and presentation may have also diverted their attention. As far as they were concerned, the wind was still within the company limit for landing. Less than five minutes prior to landing, the runway was reported as “Wet” by the Tower controller. This resulted in the first officer suggesting a change from autobrakes setting 2 to setting 3, to which the captain agreed."

The JCAA analysed:

Just as the aircraft flew over the runway threshold at about 70 feet RA, the DFDR indicates that the captain made several slight pitch up inputs. At this point the first officer made no comment, and the captain continued with the landing. The following night visual illusion factors and depth perception impediments may have contributed to the pitch up inputs and the subsequent long landing:

1. The landing was conducted in the hours of darkness, with visibility as low as

2,200 metres in heavy rain, as recorded in the SPECI taken three minutes after the accident.

2. Due to the absence of touch down zone and centre line lighting, and the absence of reflective material in the runway marking paint on runway 12, most of the light from the landing lights was reflected away from the runway surface and away from the aircraft instead of back to the cockpit; this, together with the halo effect caused by the diffusion of light by water when viewing runway edge lights through a wet windshield, may have limited the captain’s depth perception cues, and made it difficult for him to judge exactly the position of the touchdown zone.

3. The captain’s visibility may have been impeded by the heavy rain on the windshield, and the rapid movement of the windshield wipers.

4. The captain did not have a clear far visual horizon for horizontal and vertical reference due to the absence of lights beyond the runway and the reduced visibility in the rain.

5. The possibility that the captain experienced a visual illusion that the aircraft was lower than it actually was because its higher groundspeed, caused by the tailwind, resulted in the runway lights going by faster, may have caused the captain to prematurely make the pitch up inputs to flare for landing.

The GCAA analyzed that the use of the autothrottle after the autopilot was disconnected was a factor contributing to the accident stating: "it is stated in the Boeing 737 NG Flight Crew Training Manual, page 1.34, that auto-throttle use is recommended only when autopilot is engaged in CMD."

The JCAA analysed: "Thus, in accordance with the requirements of Bulletin 737-07, as the weather had not significantly changed when the landing briefing was probably completed at top of descent, a landing distance assessment was not required at that time; however, when the wind reports increased to a 15 knot tailwind at about 03:15 UTC, about seven minutes before landing, and the increased wind speed would have required a landing distance assessment, it was too late in the approach for the flight crew to do so. Nevertheless, there was no obligation for the flight crew to continue with the landing under these circumstances, and it would have been prudent for them to have performed a go-around."

The airline had introduced a new 15 knots tailwind limit together with the requirement to determine landing performance when landing with tailwind, the document stated: "PILOTS MUST ENSURE THE REPORTED

TAILWIND COMPONENT COMPLIES WITH AIRPLANE PERFORMANCE REQUIREMENTS FOR THE RUNWAY IN USE."

The JCAA analysed: "The evidence indicated that AA informed the Boeing 737-800 flight crews of this new increased tailwind limitation without providing them with any additional training, or making any restrictions for wet or contaminated runway operations, or un-grooved runway operations. It also indicated that the persons in AA who had responsibility for training may have been complacent regarding the hazards of tail wind landings. 122 The same concern applies regarding the FAA personnel who provided oversight for American Airlines, and who approved the increase in tailwind landing limits from 10 to 15 knots, without requiring a training program for tailwind landing, although this may have been because the FARs did not require it to be trained or tested."

The combination of these factors may have made the captain uncertain of the position of the aircraft in relation to the touchdown zone while descending for landing, and caused him to inadvertently pitch the nose of the aircraft up."

Aerial Overview of runway and wreckage (Photo: JCAA):

Landing Sequence Key Events (Graphics: JCAA):