The FAA and NTSB have done commendable jobs focusing on the prevention of loss of control in general aviation flights. Both agencies have engaged general aviation alphabet groups and pilots themselves in a sustained effort to decrease GA crashes, in particular fatal crashes caused by loss of control in flight (LOC). According to the NTSB, nearly half of all GA accidents are caused by loss of control in flight. LOC remains the biggest killer in GA accidents, according to the NTSB’s data of accidents from 2008 to 2014. The FAA’s data shows similar results regarding the impact of LOC on GA fatalities. In-flight loss of control—mainly stalls—accounts for the largest number of fatal GA accidents. While fatal GA accidents are trending down, there were still 209 fatal accidents in Fiscal Year 2017 that resulted in the deaths of 347 people.

Part of the FAA’s focus on preventing loss of control in flight has been a focus on emphasizing to GA pilots the importance of establishing and maintaining a stabilized approach and landing. In addition, the FAA has emphasized the importance of a go-around if factors for a stabilized approach are not met. These factors are worth repeating:

• maintain a specified descent rate
• maintain a specified airspeed
• complete all briefings and checklists
• configure aircraft for landing (gear, flaps, etc)
• be stabilized by 1,000 feet for IMC; 500 feet for VMC, and
• ensure only small changes in heading/pitch are necessary to maintain the correct flight path.

The FAA warns that if these factors are not met, a go-around should be initiated or “you risk landing too high, too fast, out of alignment with the runway centerline, or otherwise being unprepared for landing.” In short, you risk losing control of the aircraft.

I’m thinking of all this as I’m reading an accident report prepared by the Transportation Safety Board (TSB) of Canada, the equivalent of the U.S.’s NTSB, on

the crash of an N-registered, Mitsubishi MU-2B-60 en route to a remote island in the Gulf of St. Lawrence in Quebec, Canada. The crash garnered a lot of media attention in Canada because a former Canadian cabinet minister was killed in the crash along with four members of his family. The pilot and a “pilot passenger” were also killed. The pilot held both a U.S. private pilot certificate and a Canadian airline transport pilot certificate. He had fulfilled all special FAA requirements for flying an MU-2 as pilot-in-command. Although the aircraft is certified for single-pilot operations, it was this pilot’s practice to fly with an additional pilot referred to as a “pilot passenger.” The “pilot passenger” held both U.S. and Canadian commercial pilot certificates with multi-engine IFR ratings.

FDR Yields Useful Data

This accident investigation is notable for a tool available to investigators that is not usually available in general aviation accidents. While there was no flight data recorder (FDR) or cockpit voice recorder—and none were required by law—the aircraft was equipped with a General Aviation Safety Network Wi-Flight FDR system. According to the accident report, “The Wi-Flight GTA02 FDR is based on a smartphone, with extensive software customization options. Although this system was not designed or marketed to meet the requirements of [Canadian aviation regulations], it does record cockpit ambient sound, complete cockpit voice audio from the radio microphones, GPS information, and acceleration data. The system can automatically generate alerts after the flight, when certain parameters of the recorded flight are exceeded by either pilot inputs or unsafe flight conditions.” Investigators successfully extracted data from the Wi-Flight system.

Because of this equipment, investigators had a unique insight into what exactly happened in the aircraft in the minutes leading up to the accident. (Under Canada’s privacy laws, the cockpit voice recorder data can be used for accident investigations but not released to the public.)

The TSB did find that the pilot’s lack of experience in the MU-2B likely had an effect on his inappropriate reaction to the aircraft speed falling within a few knots of the stall speed. But I believe the series of events that led to the crash can be viewed separately from the type of aircraft flown. In other words, I believe that the pilot’s decision-making and the failure to do a go-around when the approach became unstable is applicable to pilots of any aircraft. And it is for this reason that I’m writing about this.

This is a summary of the sequence of events minutes before the crash according to the accident report:

At 1227:14, the aircraft crossed DAVAK on a heading of 114 degreesM at 4500 feet ASL—1,500 feet higher than the published procedure crossing altitude. The aircraft was descending at 1,600 fpm and at an airspeed of 238 knots—about 100 knots above the recommended approach speed of 140 kias. This resulted in the aircraft deviating significantly from the inbound course of 072 degrees and subsequently proceeding on a meandering flight path.

At this point, the pilot's workload had increased significantly. There was no time available during the approach to carry out the approach checklist or the before-landing checklist.

At 1227:36, the airspeed was 226 knots—about 85 knots above the recommended approach speed of 140 kias. The power levers were then reduced to idle, causing the gear warning horn to activate. The pilot then cancelled the gear warning horn.

At about 7 nm from the runway, as the aircraft descended from 3600 feet asl [above sea level] to 2800 feet asl, the wind shifted from a southerly wind component to a headwind component of approximately 20 to 25 knots.

At 1228:23, at 5.8 nm from the runway, the aircraft reached about 3,000 feet asl, and the pilot advised the passenger-pilot that, because the aircraft was very high, the rate of descent would have to be increased.

At 1228:45, the pilot indicated he was going to slow down to reach the flap and gear extension speed; otherwise, the aircraft would not be able to land. The pilot also commented that the aircraft was too high.

Almost immediately afterwards, the aircraft crossed IMOPA—the final approach waypoint, 4.2 nm from the runway—at 2,200 feet asl, which is 790 feet above the published crossing altitude of 1,410 feet asl. The aircraft was descending at 1900 fpm, the speed was 188 knots—about 50 knots above the recommended approach speed of 140 kias—and the power levers remained at idle.

At 1229:22, when the aircraft was 2.7 nm from the runway, the airspeed had decreased to 175 knots—35 knots above the recommended approach speed of 140 kias—and the descent rate had been reduced to 1,200 fpm. At this time, the landing gear was lowered and the flaps were set to 5 degrees. The aircraft continued to descend, and the airspeed continued to slow.

At 1229:34, the aircraft was at 1,250 feet asl; six seconds later, it was at 1,000 feet asl. The pilot indicated that the rate of descent had to be further reduced and noted that the aircraft radio altimeter was set at 600 feet agl.

At 1229:58, when the aircraft was 1.6 nm from the runway at approximately 600 feet agl, the passenger-pilot indicated he could see the ground on the right side of the aircraft. Although the pilot acknowledged this, he did not indicate that he had visual contact with the runway environment. Four seconds later, the pilot stated that he would continue the approach and fly the aircraft manually.

It was at this point that the pilot disconnected the autopilot, 500 feet above the ground and at an airspeed close to the stall speed of the aircraft. He applied power and the aircraft experienced an upset which the pilot was not able to recover from.

According to the report, during the approach the pilot never discussed the possibility of executing a go-around.

What I would like to leave you with, especially the pilots but also those who fly with them as passengers, are a few questions to consider. Have you ever found yourselves too high, too fast or in an other otherwise unstable approach and continued the landing anyway?  Do you see any point in these two minutes and 44 seconds when you would have made a different decision than this particular pilot did? 

Most of all, I would like to hear your recommendations for how to get pilots to stop this sequence of events. The accident report discusses different cognitive biases that affect pilot decision-making. Plan continuation bias—"the deep-rooted tendency of individuals to continue their original plan of action even when changing circumstances require a new plan”—is one that I have seen as a factor in many accident investigations and I believe is frequently a factor in the decision not to go around even when the approach is clearly unstable.

 The TSB also prepared a video of the accident sequence.