Aviation Accidents and Lessons Unlearned LXXXII: Trans World Airlines Flight 843
On July 30, 1992, Trans World Airlines (TWA) Flight 843 (TWA843), a Lockheed L-1011, registration number N11002, was taking off from John F. Kennedy Airport when the captain aborted the takeoff. The aircraft returned to the runway but ran out of stopping distance; the plane veered left off Runway 13R, where the aircraft came to rest in the grass. The aircraft was evacuated safely, but was destroyed by fire. The National Transportation Safety Board (NTSB) generated accident report AAR-94/04 and assigned accident number DCA92MA044 to this accident. Despite a search, there was no information for this accident in the NTSB archive.
The NTSB determined the probable causes of this accident, “… were design deficiencies in the stall warning system that permitted a defect to go undetected, the failure of TWA’s maintenance program to correct a repetitive malfunction of the stall warning system, and inadequate crew coordination between the captain and first officer that resulted in their inappropriate response to a false stall warning.”
The most disappointing part of analyzing NTSB accident investigation reports is that NTSB reports have always been disappointing, that the NTSB solves … nothing. In TWA843, the NTSB failed to find valid causes, instead they focused on non-issues in their investigation. The problem with NTSB reports was – and continues to be – the irrelevant information that takes up space needed for crucial information.
To be clear, there were no design deficiencies as the probable cause stated, no defects in the stall warning system. In flight crew interviews, the captain verified, “… feeling and hearing the stick-shaker stall warning …” However, in report section 1.11.1 Cockpit Voice Recorder the NTSB reasoned and later chose to make the non-argument that the stick-shaker rattle should’ve been captured by the cockpit voice recorder (CVR). So what? Why did NTSB investigators feel this was important?
NTSB interviews for surviving pilots are conducted with company-supplied lawyers and union (if applicable) representation present. They are advised and protected. Did the NTSB think all three flight crew members lied about the stick-shakers? Stick-shakers really aren’t that loud, especially in a busy environment. It was more likely the rattle was masked by other noises from two spinning nose tires, noisy cockpit conversations, airstream, airframe sounds, and instrument sounds drowning out the rattle. It wasn’t pilot lies or design issues. The investigators wasted time on irrelevant items.
Finding 12 determined, “the captain made a split-second decision to reject takeoff … His decision was very likely, in part …” “Very likely”? That’s not fact but opinion. Why did the NSTB resort to assuming what the captain decided? Didn’t investigators interview him? What did he say his decision was due to? TWA843’s CVR transcript became an NTSB focus, useless trivialities were raised; the NTSB began interpreting the transcript and the interviews. But transcripts and recorders are tools – not facts. When the NTSB defers to ambiguous information, it leads to fabrication.
What lay beyond the departure end of runway 13R
In the “split second” decision, the captain reviewed many possibilities, such as the hundreds of citizens in Inwood, NY, just beyond the runway. Why didn’t investigators ask questions referring to the indicators, the instruments? Had the captain looked at his gauges before deciding to abort? Did he cross-check with the first officer’s gauges or the standby gauges to see if they indicated a possible stall? Had he monitored the takeoff roll, mindful of his instruments? These are questions that demanded answers, not questions invented to provide opinion.
In section 1.17.1 TWA Procedures, page 46, stated, “… TWA does not require a verbal pre-takeoff briefing regarding of abnormal or emergency events on takeoff.” What was being stated here? Did the NTSB suggest the pilots didn’t perform a pre-takeoff briefing? If not, why make the statement in an official report? How would we, the industry, know the crew didn’t conduct a verbal pre-takeoff briefing?
This is important. The CVR transcript contained 31 minutes and 46 seconds of the final cockpit conversation; it began at the captain’s greeting to the passengers, well after the point when the pilots would’ve had such a discussion. The pre-takeoff briefing, where the flying pilots discuss emergency procedures specific to each flight, such as what runway to return to; alternates; who’ll man the radios; these were never documented in the transcript. In other words, the NTSB never provided proof that the briefing didn’t happen. The captain had 2397 flight hours (FH) on the L-1011, the first officer (FO), 2953 FH. The captain had 15,854 FH; the FO had 13,793 FH in comparable equipment.
This meant both flying pilots were experienced – not just on the L-1011 – but other airliner equipment. Pre-takeoff briefings are what these pilots have been doing for years in their professional career. That NTSB investigators would make a casual off-the-cuff remark that the pilots – and other TWA pilots – didn’t do their job correctly is irresponsible, more so because it’s inserted in an official government document. How would the pilots or TWA airlines defend themselves from such a reckless insinuation? This is what happens when NTSB investigators steer away from facts and wander into personal opinion.
When the NTSB makes these types of subliminal suggestions, they damage the pilots’ careers and TWA’s market reputation. The statements are made in ignorance, which is worse because the NTSB investigators don’t even realize they’re causing damage. For some reason the NTSB’s opinions carry weight with the media, who are also ignorant but in a sensationalistic way, an almost absence of malice without the absence. This major misconception is because NTSB investigators felt the need to go off script. Instead of saying they didn’t know, they assumed without facts. This was seen in the ValuJet 592 accident report when an NTSB investigator – a pilot – cast aspersions on aircraft maintenance persons in the industry with his opinion, making it sound like fact. Terms like ‘most likely’ and ‘it is possible’ don’t belong in accident reports that are meant to be based on facts.
The NTSB has always dismissed aircraft maintenance as a waste of resources, even when I worked at the NTSB. This was/is out of ignorance of aircraft maintenance’s crucial part in investigations. It’s a major exploitable weakness of every NTSB investigation; every airline and every manufacturer know the NTSB has this blind spot. In almost 58 years of existence, the NTSB still fails to understand what accident investigations are meant to find, continuing to apply non-industry principles into major industry investigations. The NTSB failed to employ maintenance-experienced investigators in TWA843, those who know the difference between a system failing and the failure of a system.
Efforts during this entire investigation should’ve been focused on the stall warning and instrument panel. However, the NTSB didn’t/doesn’t/won’t hire for experience those who can investigate maintenance; they leave maintenance issues to those without career knowledge. Maintenance-specific interviews wouldn’t be conducted for another ten years; the quality of maintenance investigating was non-existent.
Findings 15 and 16 addressed a stall warning component, but neither finding made sense. #15 described a ‘deficiency’ in the Angle of Attack (AOA) component, a mechanical sensor that references aircraft angle to longitudinal zero degrees; an automatic monitoring system.
L-1011 Angle of Attack Vane from accident report
In Finding #15, NTSB investigators stated, “the intermittent malfunction of the right AOA sensor was not detectable during preflight system tests… permitted the malfunctioning sensor to cause a false warning when the air-ground sensor on the landing gear went to the ‘air status’ on takeoff.” The problem with intermittent failures is they don’t always show up; they’re … intermittent, even during preflight system tests. For example, a ‘gear-in-transit’ light might illuminate during preflight checks but a half hour later, during gear retraction, the bulb could fail. It happens. Tests try to simulate an inflight mode but wiring issues can be unpredictable. Refer to Eastern Airlines flight 401 in 1972.
#16 started out okay but quickly dissolved into a trend analysis mess. “The repetitive malfunctions were not detected by the TWA quality assurance trend monitoring program because the program used a calendar day, rather than flight hour basis to detect trends.” What? One has nothing to do with the other. Periods of time are determined by what is being tracked; trend analysis on the AOA sensor was determined by its manufacturer and/or the operator.
But assuming finding #16 made sense, that the AOA was trending repetitively, the investigators didn’t show AOA failure documentation referred to in section 1.16.6.2 FDR Data as Used in the Performance Study that proved a trend existed; times shown weren’t consistent or within identifiable parameters. TWA’s AOA trend wouldn’t have shown patterns in any trend analysis program. For one, the AOA in question moved around different L-1011s and it’s unclear if the AOA was part of a parts-pool with other airlines. NTSB investigators assumed a lot, such as that AOA vanes received individual treatment or that trends were that obvious. They’re not. When failure events for a particular component are spread out over weeks or months, the failures don’t show up as a trend, especially if the reason for removal changes.
Documented under section 1.16.4 TWA AOA Sensor Reliability Control, the investigators failed to explain how they believed the reliability program went wrong. As stated, “TWA personnel reported that similar specifications are used on other airlines’ airplanes for which TWA provides maintenance services.” This is a repetitive problem because of NTSB’s unfamiliarity with aircraft maintenance. NTSB investigators do not understand aircraft maintenance programs. For example, while TWA works on other airlines’ airplanes, those airlines must monitor trends per their own maintenance programs – not TWA’s. Trivial? No, the NTSB created a problem that didn’t exist; the NTSB pushed a non-existent problem, not just onto TWA, but onto the industry.
Even the NTSB’s attention to items not associated with TWA843 was unnecessary. Recommendations, like A-93-16 through A-93-18, had nothing to do with the accident, therefore NTSB should’ve talked to the FAA off-line. Fire and Rescue? Write another report. Unless Fire and Rescue had a direct effect on the accident cause, it should’ve been moved to another investigation and/or report. AAR-93/04 is an accident report, not a collection of problems that are added randomly.
Attention to root cause, not quantity of recommendations over causal quality, was the goal. The NTSB had a chance to work an accident as accessible as it can get; everything was there for investigators to see. Instead the NTSB focused too much on ambiguity and opinion and ignored the very direction that was practically laid out for them.