Continuation bias factor in Boeing 737-700 runway excursion at Burbank Airport

Home Aircraft Accidents Continuation bias factor in Boeing 737-700 runway excursion at Burbank Airport
Continuation bias factor in Boeing 737-700 runway excursion at Burbank Airport

Southwest Airlines (SWA) flight 278 was en route to Bob Hope Airport (BUR), Burbank, California, when it overran the end of runway 8. The airplane came to rest about 144 ft past the departure end of runway and 71 ft into an engineered materials arresting system. Shortly before the airplane touched down on the runway, the tower controller informed the flight crew that heavy precipitation was occurring directly over the airport and that the wind was from 270° at 11 knots. BUR runway 8, which was 5,802 ft in length, was one of the shortest runways at airports where SWA operated.

Before the airplane reached the top of descent (about 0839), the flight crew requested and received a landing data report generated from the SWA performance weight and balance (PWB) system. The report indicated that maximum autobrakes should be used for landing on runway 8 and that the stopping margin—that is, the difference between the calculated landing distance (including a 15% safety factor) and the runway length available—would be 245 ft. The PWB system calculations assumed that touchdown would occur 1,500 ft from the runway threshold and that the wind would be from about 280° at 5 knots (the wind direction and speed at the time of the airport’s 0753 weather observation converted from true north to magnetic north). Both flight crewmembers expressed concern about the stopping distance given the wet runway, the tailwind that would be present, and the runway length, and they discussed that the braking action upon touchdown would need to be “pretty abrupt.” Also, even though the first officer stated, during a postincident interview, that BUR “always has a low number” for the stopping margin, the captain stated that he had not previously seen a stopping margin as low as the one for the incident flight.

While the airplane was descending through the terminal area, the controller advised the flight crew that the pilot of a King Air airplane reported a 15-knot loss of airspeed on final approach to BUR and that a corporate jet had just conducted a go-around at BUR because of the wind.

However, after the tower controller cleared the airplane to land, the controller advised the flight crew that a pilot of a Boeing 737 airplane had reported braking action as “good” 10 minutes earlier.

Shortly before landing, the tower controller reported that the wind was from 270° at 10 knots and then, less than 1 minute later, from 270° at 11 knots, The SWA B737NG Aircraft Operating Manual (AOM) stated that the tailwind limit for landing was 10 knots. After the second wind report, the first officer stated, “we got eleven knots. You want to call it good?” The captain replied, “yeah.” During a postincident interview, the captain stated that both he and the first officer assumed that the tailwind component would be about 9 to 10 knots (presumably because the wind would not be directly behind the airplane given the wind direction and the orientation of the runway).

The SWA 737 AOM also stated that, for tailwind landings, the target speed should be the reference landing speed (Vref) plus 5 knots. The PWB system determined that the Vref for the flight would be 126 knots, so the target speed should have been 131 knots. However, the aircraft performance study for this incident determined that the airplane’s airspeed over the runway threshold was 137 knots. Also, the true airspeed and groundspeed (based on flight data recorder data) showed that a tailwind of 13 to 18 knots was over the runway at that time and not the 5-knot tailwind that the PWB system calculations assumed.

Further, the aircraft performance study found that the airplane touched down about 2,500 ft past the runway threshold, which was 1,000 ft beyond the 1,500-ft touchdown point assumed in the PWB system calculations. This study finding was consistent with a BUR tower controller’s observation of the airplane touching down near the intersection of taxiway D7 and runway 8, which was about 2,600 ft from the runway threshold.

The AOM stated that, if an airplane were to touch down beyond the 1,500-ft point, the stopping margin that the PWB system calculated would be “invalid” and that, in some cases, the runway length would be “insufficient” for the airplane to stop. In this case, the longer-than-normal touchdown point, the higher-than-expected tailwind, and the faster-than-nominal approach speed increased the airplane’s required landing distance, making the 245-ft stopping margin calculation invalid.

The AOM also stated that, if the current weather conditions are significantly different than the anticipated conditions at the time of the PWB system calculations, higher-than-planned braking might be needed to account for the reduced or insufficient stopping margin. Flight data recorder data showed that the airplane’s autobrakes were activated at touchdown but then disengaged almost immediately as a result of pilot-applied pressure to the brake pedals. The maximum left and right wheel brake pressure (indicating full manual braking) was reached about 6 seconds after touchdown and remained at maximum until 1 second before the airplane came to rest.

During a postincident interview, the captain stated that he was “blending in manual brakes” because the airplane was not slowing with maximum autobrakes. However, the captain’s application of manual brakes almost immediately after touchdown was contrary to company guidance, which stated that “the intent of using the autobrake system for landing is to let the system automatically brake the aircraft to an appropriate speed, not to override the system shortly after touchdown” and “once the landing roll is stabilized, transition to manual brakes…with adverse conditions, transition at a slower speed.” According to the aircraft performance study, the maximum braking performance available from the runway was achieved even without proper use of maximum autobrakes.

The airplane initially reached and maintained a deceleration rate between about 0.3 and 0.4 G, but the deceleration rate decreased to between 0.15 and 0.20 G after the airplane crossed over the left edge of the runway from a grooved to a smooth paved surface. (This maneuver is consistent with the captain’s statement about potentially turning onto taxiway D1 and his application of left rudder and the tiller.) The decrease in deceleration is consistent with a lower friction coefficient on the smooth paved surface compared with that on the grooved runway surface. Once the airplane entered the engineered materials arresting system, the deceleration rate increased to a maximum of about 0.6 G until the airplane stopped.

The flight crew had multiple opportunities to assess whether to continue the approach to a landing, but none of those opportunities resulted in a decision to go around. Although the flight crew received automatic terminal information service (ATIS) information Hotel before the airplane reached the top of descent, as required by SWA procedures, the crew failed to consider that the information might no longer be valid after receiving reports of changing weather conditions. Also, when the flight crew received the report from the tower controller indicating that the wind was from 270° at 10 knots, the crew recognized that the tailwind would be at SWA’s limit. Shortly afterward, the crew learned that the wind speed had increased to 11 knots. However, neither of these wind reports led to flight crew recognition that the PWB system-calculated stopping margin was no longer valid. Specifically, the crew did not discuss that the wind in the landing data report (which was issued about 37 minutes earlier) no longer reflected the current wind or that an updated landing data report from the PWB system was needed, thus missing another opportunity to better understand and address the deteriorating weather situation.

Both ATIS information Juliet and Kilo, which became effective about 21 and 9 minutes, respectively, before touchdown, included windshear advisories, and SWA procedures stated that pilots should not continue an approach if known windshear existed. Although the approach controller did not notify the incident flight crew about the more recent ATIS reports, which also discussed the significantly increased (11-knot) tailwind, the crew should have recognized the potential threat for windshear after receiving and acknowledging the reports of an airplane that had a 15-knot loss of airspeed on final approach and an airplane whose pilot conducted a go-around because of wind, and the crew should have discontinued the approach after receiving these reports.

During a postincident interview, the captain stated that pilots could request multiple landing data reports from the PWB system but that the same data might be received if the weather information in the system had not been updated. In this case, the weather information would have been updated because, after the BUR hourly observation at 0753 (which was the basis for the original PWB landing data calculations), two additional weather reports (a special weather observation and an hourly weather observation) were generated at 0841 and 0853, respectively. (These reports became the basis for ATIS information Juliet and Kilo). In addition, the flight crew had the option to manually enter the wind information. With a 10- or an 11-knot tailwind (as reported in the 0841 and 0853 weather observations), updated PWB system calculations would likely have indicated that the airplane could not safely land on runway 8, especially given that the previously calculated stopping margin (when the tailwind was 5 knots) was only 245 ft.

The National Transportation Safety Board (NTSB) evaluated why the flight crew continued the approach to a landing. The captain reported that he had flown into BUR between 80 and 100 times. The captain also reported that his previous flights into BUR occurred in visual flight rules flight and without any significant precipitation. The first officer reported that he had flown into BUR at least 100 times and estimated that he encountered a tailwind or rain from 5% to 10% of time. Although the flight crewmembers had experience landing on the short runway, their lack of substantial exposure to adverse weather at the airport resulted in pilot mental models for landing that did not fully account for the environmental challenges that compounded the short runway challenges.

In addition, the flight crew’s decision to continue the approach to a landing was consistent with a psychological concept referred to as plan continuation bias, which is an unconscious cognitive bias to continue with an original plan despite changing conditions. After hearing the wind report indicating an 11-knot tailwind, the flight crew justified continuing the approach, even though the tailwind component would be “right on the edge” of the company’s 10-knot limit. (Specifically, the 270° wind was 10° off from a direct tailwind for runway 8; thus, the 11-knot wind speed would result in a tailwind between 10 and 11 knots.) The crewmembers did not consider taking another action, such as performing a go-around to allow them time to reassess the situation, which would have been consistent with company guidance that instructed pilots to go around if a landing appeared unsafe. Thus, the flight crew’s decision to land on a short runway with the reported 11-knot wind almost directly on the airplane’s tail was intentional due to plan continuation bias, and the decision was inappropriate.

The airplane touched down on the runway 1,000 ft beyond the 1,500-ft touchdown point assumed in the PWB system calculations and specified in SWA procedures. The horizontal distance from the runway threshold that was required for the airplane to descend and touch down was substantially increased by the airplane’s higher-than-anticipated groundspeed. Contributing to the increased groundspeed were the higher-than-expected tailwind and the airplane’s faster-than-nominal approach speed as it crossed the threshold. (The target speed was 131 knots, Vref plus 5 knots. The airplane crossed the threshold at 137 knots, Vref plus 11 knots.) Although the wind and the excess airspeed both contributed to the airplane’s higher groundspeed (which led to the longer-than-normal touchdown), the wind played more of a role than the excess airspeed.

The captain estimated that the airplane touched down between 1,300 and 1,500 ft from the runway threshold. This estimate was based on the expected timing between 10 ft above ground level (when the airplane began to flare) and touchdown and was not based on external cues such as the painted runway markings and relevant taxiway intersections. Also, the first officer was “pretty confident” that the airplane touched down by 1,500 ft. However, the tailwind increased the groundspeed and thus the distance traveled in a given time, causing the airplane to touch down farther on the runway than both flight crewmembers assumed.

The NTSB recognizes that, according to their postincident interviews, both flight crewmembers did not realize that the airplane touched down 2,500 ft from the runway threshold. The captain’s failure to correct the excess approach speed and both flight crewmembers’ lack of awareness of the long touchdown were consistent with the crew experiencing a narrowing of attention. This phenomenon occurs when certain information is overlooked as individuals focus on a narrow field of attention perceived to be the most threatening or salient (in this case, the tailwind and the need for aggressive braking upon touchdown).

The NTSB considered whether the pavement change on the runway (the first 500 ft beyond the runway 8 threshold was constructed of light-colored concrete, and the remainder of the runway was constructed with dark-colored asphalt) created a visual illusion that affected the flight crew’s assessment of where the airplane touched down. Because the captain stated that he was following guidance from the heads-up display and precision approach path indicator lights to the runway, the pavement change was likely not a factor in the crew’s assessment of the touchdown point.

According to the AOM, a go-around should be performed if “the pilot determines that a landing in the touchdown zone cannot be safely accomplished because…the aircraft touches down beyond 1500 ft. with an insufficient PWB System-computed stopping margin.” The AOM also stated that a landing on a runway with a reduced or an insufficient stopping margin “becomes more critical on shorter runways” (such as runway 8 at BUR) and that a go-around would be “the better option” compared with continuing the landing. The crewmembers’ recognition that the airplane had flown beyond the touchdown point would have been another trigger to conduct a go-around and reassess the landing conditions. (SWA procedures allowed flight crews to go around until the thrust reverser levers were raised.)

As previously stated, the flight crew should have recognized that the landing data report that the PWB system calculated (which provided a relatively short stopping margin that concerned the flight crew) might no longer be accurate. Thus, even with the stressful, fast-paced, and dynamic situation that was occurring, the flight crew should have called for and executed a go-around.

After the incident, SWA issued a flight operations bulletin that discussed that company pilots must execute a go-around if an airplane touches down beyond the first one-third of the available runway length, the first 3,000 ft of the available runway length, or 1,500 ft plus the planned PWB system-calculated stopping margin, whichever is the most restrictive. For the incident flight, the first and last criteria would have applied, with the last criterion being the most restrictive because it would have required a go-around for a touchdown occurring at or beyond 1,745 ft—about 760 ft before the touchdown point during the incident landing. (As previously stated, at the time of the incident, SWA required a go-around if landing in the touchdown zone could not be safely accomplished because the airplane touched down beyond 1,500 ft with an insufficient PWB system-computed stopping margin.)

SWA also provided the NTSB with a list (dated June 2020) of other safety improvements that resulted from this incident. However, none of those safety improvements required company flight crews to reassess whether the information in a PWB landing data report remained valid in changing conditions. This incident demonstrated that landing conditions can change during a flight and that the landing data report that a flight crew receives before the top of descent might not be sufficient to ensure a safe stopping distance at the time of arrival. SWA procedures did not specify the conditions that would warrant obtaining updated landing data reports from the PWB system (such as when a controller-reported wind differs significantly from the wind used in a landing data report). In addition, SWA did not instruct its flight crews to verify the PWB system performance data as an airplane gets closer to its destination, which would take little time to accomplish.

Probable Cause:
(1) The flight crewmembers’ decision, due to plan continuation bias, to continue the approach despite indications of windshear and a higher-than-expected tailwind and
(2) the flight crew’s misperception of the airplane’s touchdown point, which was farther down the runway than the crew assumed because of the faster-than-expected groundspeed. Contributing to the accident was Southwest Airlines’ lack of guidance to prompt flight crews to reassess operator-provided landing data when arrival weather conditions differ from those used in the original landing data calculation.

Click Here to see the Full/Original Article

Leave a Reply