Too Low On The Final Approach, Part 2
In Part 1, we described how a Convair 440 transport airplane crashed short of the runway at Toledo Express Airport (TOL) in Monclova, Ohio, on Sept. 1, 2019.
Using a DJI Phantom 4 quadcopter drone equipped with a GPS/GLONASS receiver and an FC6310 camera, NTSB investigators recorded the view of Runway 25 at TOL from above the wreckage location. They flew three drone flights at 4:30 AM the day after the accident to document the visual appearance of the approach to the airport and the accident area in conditions similar to those at the time of the crash. A photo taken 300 ft. above the accident site shows the runway, approach lights and PAPI lights clearly visible. As the drone descended, subsequent photos showed less clear runway light detail and obscuration of foreground lights.
FAA air traffic control radar plots showed the airplane maintaining a groundspeed of about 150 kt. at 02:18, gradually increasing to about 170 kt. as the airplane approached the field. Then, at 02:34:40, the groundspeed began to drop off dramatically. In 3 min., 20 sec. the speed had dropped to 75 kt. With winds at the airport only 4 kt., the difference between groundspeed and indicated airspeed would have been negligible. The Convair flight manual showed that 75 kt. was the stall speed of the airplane.
The pilot and copilot, aged 69 and 72, respectively, had been paired together for every flight in the airplane since March of 2018. Their flight times had to be gleaned from their most recent FAA medical records, since their logbooks could not be located. The pilot reported 8,000 total hours of flight time; the copilot had 11,287 hr. The aircraft logs showed they both flew 175.1 hours in the airplane in the preceding year.
Both pilots had completed a proficiency check in the airplane with an FAA inspector on July 22, 2019. Both held FAA Class 2 medical certificates with waivers and limitations, but those restrictions were not provided in the accident report. Interviews with their wives about their activities in the days before the flight were unremarkable. No evidence of drug or alcohol use was found.
The investigation suffered from a dearth of useful information that is usually available in a major accident investigation. The most glaring absence was the lack of a flight data recorder (FDR) or cockpit voice recorder (CVR). The old radial engine-powered cargo plane fell outside the strictures of 14 CFR 91.609, which defines which aircraft must have the recorders.
The airplane was also not equipped with ADS-B Out capability to broadcast its position. The accident happened four months before ADS-B out was mandated for most aircraft flying in controlled airspace. Without either FDR or ADS-B Out data, investigators were forced to rely on FAA radar data, which has lower definition and less detail. When the airplane slowed, for example, an FDR would have shown the power settings. With only radar data, the reason for the slowdown had to be assumed.
The lack of a CVR meant that the crew’s remarks and ambient cockpit sounds were not available. Yawns and pilot remarks about being tired are often part of the way fatigue is documented. Were the pilots silent as the airspeed fell to stall speed, or were they distracted or talking about something else at the time? We don’t know.
There was no examination of the operation of the company, Ferreteria e Implementos San Francisco. Was it truly a two-person, one-airplane operation? Was there an operations manual, operations specifications or a safety management system or documents? If not, was this because the operator was in violation of regulations or because of a lack of regulations? This would normally be part of a major investigation.
No human performance expert did a 72-hr. history of the two pilots. This history plus other details about the pilots is normally necessary to identify missed sleep opportunities or other problems that contribute to fatigue.
Like the missing pilot logbooks, it could very well be that much of the needed information was simply not available.
Conclusions and Comments
In its probable cause finding, the NTSB said: “The flight crew’s failure to maintain the proper airspeed on final approach…resulted in an inadvertent aerodynamic stall and impact with trees and terrain. Contributing to the accident was the flight crew’s fatigue due to the overnight flight schedule.” The first sentence describes what happened; the second sentence is an educated guess about why it happened.
The fatigue case is supported by the time of night the accident occurred—at 02:39—right in the middle of the flight crew’s window of circadian low. Anyone not adapted to night work would be tired at that time. Fatigue is also supported by the aerodynamic stall, which can be explained by fatigue-induced inattention. Fatigue might explain why the crew ignored the red PAPI lights that should have been visible to them.
Not addressed in the NTSB’s probable cause finding was the effect of featureless terrain illusion, sometimes called “black hole effect.” Both the FAA’s Pilots Handbook of Aeronautical Knowledge and the Airman’s Information Manual (AIM) describe this visual illusion.
To quote the AIM, Section 8-1-5, “An absence of ground features, as when landing over water, darkened areas and terrain made featureless by snow, can create the illusion that the aircraft is at a higher altitude than it actually is. The pilot who does not recognize this illusion will fly a lower approach.”
Even experienced pilots are subject to visual illusions. The Convair pilots could have been overconfident in their ability to judge their height and distance from the airfield. They apparently never tried to capture the ILS glideslope, which they could have easily done if they had any concerns about terrain clearance.
NTSB investigations have tended to overlook this phenomenon in explaining past accidents. Of 10 investigations classified as “undershoots” and taking place at night, none attributed the accident to visual illusion, even though it seemed likely to have contributed to the cause in a number of the cases.
In this data-poor investigation, the ability of accident investigators to document visual illusion was just as minimal as their documentation of fatigue. But I suspect visual illusion was at least a contributing factor.
I noticed one other aspect of this investigation that most people would miss. It was only one of 10 major accident launches conducted by the NTSB in 2019. In the 10 years from 2012 to 2021 (excluding 2019), the average number of major launches was less than 3 per year. Five of the major investigations in 2019 resulted in board-approved “blue cover” reports, which represent a huge workload for the staff. In addition, NTSB aviation staff was involved, mostly behind the scenes, in the Boeing 737 Max crashed which took place overseas in 2018 and 2019.
Highs and lows in workload are a fact of life at the NTSB. Nonetheless, 2019 had to be a blowout year for the 100 investigators and managers in the Office of Aviation Safety. It’s not surprising to me that the Convair accident investigation took so long to complete and was reduced in scope to a limited investigation.
If the Convair accident had taken place in a more typical year, I would like to think that some recommendations would have issued from the board to address deficiencies in large airplane Part 125 operations and in on-demand cargo operations as well.
Too Low On The Final Approach, Part 1: https://aviationweek.com/business-aviation/safety-ops-regulation/too-lo…