Regardless of the ultimate outcome of the investigation into the March 19 FlyDubai Flight 981 (FZ981) accident, the timing of the crash—during an attempted go-around—is sure to reignite pressure from safety advocates for better pilot training on the complex but rarely performed procedures.
The final frames of imagery captured by several surveillance cameras of the Boeing 737-800 with 62 passengers and crew plunging into the ground in a steep nose-down attitude at the Rostov-on-Don Airport are eerily similar to the final seconds of another recent accident following an attempted go-around: Tatarstan Airlines Flight 363 (U9-363)—a Boeing 737-500 with 50 passengers and crew onboard—as it attempted to land in Kazan, Russia, the night of Nov. 17, 2013. Like FZ981, U9-363 plunged into the ground at high speed after the second attempt at an instrument approach and in similar weather conditions, leaving only small fragments at the airport crash zone.
 
Russia’s Interstate Aviation Committee is investigating the FZ981 crash with the help of the United Arab Emirates (UAE), Boeing, the U.S.NTSB and others. Teams were able to recover and download information from both the flight data and cockpit voice recorders and are in the process of collecting other relevant facts. To date, investigators have said only that there were no systems or component failures. Factors influencing the Kazan crash are sure to be considered, however.

 
Among the Russian Interstate Aviation Committee’s findings in the Tatarstan final report, published in December, was that the crew was task-saturated and did not notice that the autopilot had disengaged on the high-power go-around, which ultimately led to a nose-high upset. The disoriented crew responded by pushing the yoke fully forward, which put the aircraft into a steep descent that was not recoverable at the aircraft’s low altitude.
Data from the Flightradar 24 website show that FZ981 arrived in the Rostov area at approximately 2:00 a.m. local time after a 4-hr. flight from Dubai. The first attempt at the instrument landing system (ILS) approach to Runway 22 ended with a missed approach initiated at more than 1,700 ft. above ground, well above the 480 ft. minimum. It is possible that wind shear conditions, which the approach chart states can be “expected on the final” arrival to the runway, may have played a role in the abort. Wind at the airport was blowing at approximately 25 kt. and gusting to 35 kt. or higher, but relatively aligned with the runway, reducing the potential for crosswind control issues.

Climb rates during the first go-around averaged about 2,250 ft./min. up to 8,000 ft., where the aircraft maneuvered in the local area for 25 min. FZ981 later climbed to 15,000 ft., where it held for nearly 1 hr. until attempting a second instrument landing to Runway 22 at approximately 3:30 a.m.

Air traffic control recordings, circulating on the web and apparently legitimate, revealed that the crew had called the Rostov tower multiple times during the holding pattern to discuss whether the visibility and wind conditions were improving. According to Flightradar24, an Aeroflot Sukhoi Superjet 100 had tried three times to land at Rostov while FZ981 was holding, ultimately deciding to fly to an alternate airport.
On the recording, voices from the cockpit appear to be calm throughout the ensuing approach, informing the tower controller that the aircraft would climb to 8,000 ft. in case of another miss. The pilots reported being “established on the localizer,” meaning the aircraft had locked in its lateral guidance on the ILS. As it descended through approximately 1,500 ft. however, the pilot radioed the tower, “Going around, 981.”
Flightradar24 data show a steeper climb out on the second go-around, with an average climb rate of 3,200 ft./min. but peaking as high as almost 5,000 ft./min. Ground speed (airspeed plus or minus a wind component) data were roughly the same during both go-arounds, raising doubts as to whether an aerodynamic stall ended the climb. Approximately 40 sec. after the pilots initiated the go-around, at an altitude of 4,000 ft., the aircraft began a dive, with the descent rate quickly increasing to more than negative 15,000 ft./min., according to Flightradar24. Surveillance video from a location south of the crash site correlates with the Flightradar24 data, clearly showing the 737 initiating the go-around, entering an overcast cloud deck on the climb out, and 34 sec. later,falling back through the overcast in a steep nose-down attitude.
A likely contributor to the Tatarstan crash was spatial disorientation linked to somatogravic illusions during the go-around. Under positive acceleration—a lightly loaded 737 under a high-power go-around for example—somatogravic illusions, if not overridden by the aircraft’s attitude instruments, can cause a pilot to believe the aircraft is climbing.
“During go-around or takeoff phases in low-visibility conditions, while the aircraft is accelerating, pilots may try to counteract this perception of climb by pitching down the aircraft’s nose until the dive counterbalances the apparent backward tilt caused by the acceleration, which may end in impact with the ground.” said the French aviation safety agency, BEA, in a report published four months before the crash.
The Study on Aeroplane State Awareness during Go-Around (Asaga) labels somatogravic illusions as a possible factor in a number of accidents, including a Gulf Air Airbus A320 in 2000, an Armavia A320 in 2006, and a Kenya Airways Airbus A330 in 2007. BEA found that the difference between the actual pitch attitude and perceived pitch attitude during a go-around can be as much as 25 deg.
BEA concluded that pilots are ill-prepared for go-arounds, relatively rare events where many actions must be completed in a short time, leaving little margin for error in handling automation and control of the aircraft. Failure to handle either can lead to a loss of control. The agency says an Asaga-type accident is marked by inadequate management by the flight crew of the relationship between pitch attitude and thrust in any situation at low altitude. Along with calling for somatogravic illusions to be incorporated into simulators, the BEA also recommended more training for go-arounds, particularly with both engines operating, and installation of devices to limit thrust during go-arounds.
Other factors mentioned in the Tatarstan crash final report, which may be relevant for FlyDubai: Simulator training did not include missed approaches at higher altitudes with both engines operating; pilots did not receive effective upset recovery training, and there were issues with the airline’s crew work and rest hours.
According to Russia’s RT News, a former FlyDubai pilot says its pilots, particularly junior pilots, are fatigued due to work schedules. The whistle-blower apparently has provided documents to RT showing that the first officer of the ill-fated flight had worked 11 of the previous 12 days before the accident, and was transferred between day and night duty without adequate time to adjust. He noted that the schedules were legal on FlyDubai’s part, but “not healthy.”
FlyDubai is currently advertising for a crew scheduling officer responsible for making sure the carrier meets all UAE requirements, but also to “develop crew scheduling strategies to achieve increased levels of productivity without compromising on the quality of output.”
Editor's note: This article was updated to include new information.