When pilots abort an approach and perform a go-around, the action compresses and accelerates the elements of airmanship and automation into a 1-min. window, leaving little margin for error in either.

A new study published last month by the French aviation safety agency, BEA, concludes that pilots are ill-prepared for the rare events, which are often initiated by air traffic control or the non-flying pilot who is monitoring the approach. In either case, the pilot at the controls can be startled by the order, initiating a chain of events that can lead to loss of control after a go-around is initiated.

In a BEA questionnaire answered by 831 pilots, 60% said they had encountered difficulties with go-arounds in the air or in a simulator. The predominant reasons for initiating go-arounds were weather, including tailwinds, unstabilized approaches or air traffic control commands. Troubles during the maneuvers included the higher-than-normal acceleration with takeoff/go-around (TOGA) thrust and all engines operating due to the lower weight of the aircraft at its destination, unnoticed disengagement of the automatic systems, and unexpected air traffic control requests, increasing workload and stress.

Based on its findings, the agency has made 34 recommendations to regulators and airframers, including a call for additional initial and recurrent go-around training with higher-fidelity simulators, modifying the aircraft's go-around thrust to account for actual flight conditions, and studying “the means required to detect and correct erroneous mode selection during a go-around.”

BEA says that between 1985 and 2010, there were least 25 accidents or serious incidents attributed to situations the agency now defines generically as errors in aircraft state awareness during go-around (Asaga), claiming 954 lives. Three fatal accidents the BEA investigated in 2009 and 2010—involving a Yemenia Airbus A310 at Moroni, Comoros; an AeroUnion A300 freighter in Monterrey, Mexico; and an Afriqiyah A330 in Tripoli, Libya—alerted the agency to potential issues with go-arounds and prompted the study.

BEA says the loss-of-control crashes occurred either during go-arounds or at full-thrust configuration with high nose-up angle, considered Asaga-type events. More broadly, BEA 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.

On average, Asaga-type events accounted for 3.67% of airline fatalities per year from 1985-2010, though in 2009 and 2010 the number was significantly higher at 20% and 27%, respectively. “Thus, even though the number of Asaga-type events is relatively low, each of them produces a high number of casualties, which justifies taking specific prevention measures,” says the BEA.

Included in the forensics the BEA used for the study were the August 2000 crash of a Gulf Air A320 in Bahrain and a problematic, previously unreported go-around of an A380 at John F. Kennedy International Airport in New York on Oct. 11, 2010. According to the study, the first officer was at the controls of the A380 during a poorly executed visual approach to Runway 31L, with the captain performing the monitoring function. At 480 ft. altitude and 1 mi. from the runway, the captain ordered a go-around as the aircraft was flying too high and too fast to meet standard operating procedures for a stabilized approach. The first officer initiated the go-around, pushing throttles forward to the TOGA position. Due to a relatively low initial go-around altitude requirement of 1,000 ft. and the first officer's inability to properly control thrust, the aircraft experienced several flap overspeed warnings and experienced vertical speeds as high as 4,200 ft. per min. during the 45-sec. ordeal.

Along with the questionnaires, the 13-year effort included a broad survey of accidents in BEA and International Civil Aviation Organization (ICAO) databases, a detailed analysis of 16 accidents or incidents, including six narrative descriptions of events, and 13 full-flight sessions on the Boeing 777 and Airbus A330 simulators.

Pilots in BEA's questionnaire revealed that go-arounds are relatively rare and training is not comprehensive or representative. There are approximately 2-4 go-arounds per 1,000 flights, which translate to about one per year for short and medium-haul pilots and one every 5-10 years for long-haul pilots.

Combined with the rarity of the event is a training environment that does not provide physiological realism. According to providers that BEA interviewed, go-around training for a type rating “takes place in an environment that imposes very few constraints,” including no or limited interaction with air traffic control, no major system malfunctions and no environmentally sensitive flight paths. A physiological aspect of go-arounds—somatogravic illusions—is not “systematically covered during the training” and the effects are not modeled in simulators. BEA is recommending that the effect be incorporated into simulators.

Somatogravic illusions occur when acceleration along the flight path is combined with a nose-up pitching moment, both of which take place when an aircraft with under-wing engines begins a go-around. Pilots may counteract the perceived nose-up pitch with a strong desire to put in a nose-down control input. BEA's analysis shows that the difference between the actual pitch attitude and perceived pitch attitude during a go-around can be as much as 25 deg. “Pilots may try to counteract this perception of a climb by pitching down the aircraft nose until the dive counterbalances the apparent backward tilt caused by the acceleration,” says the BEA, noting that the effect can be compounded by a false visible horizon, for example a shoreline or unlit background terrain.

Given that go-around training for initial type rating takes place in simulators and most recurrent training simulates a go-around with a failed engine (reducing the acceleration effects that worsen somatogravic illusions), BEA is concerned that pilots are not experiencing the phenomena until it is too late. “Although no objective standard exists for somatogravic illusions, the major simulator manufacturer indicated that this issue has been studied, but never implemented,” says the BEA, not identifying the simulator manufacturer it had talked to. “However, it seems possible to tailor the motion cues to the specific context of a go-around.” The idea would be to use an exaggerated nose-up angle to simulate the somatogravic illusion during a go-around. BEA says the pitch capability likely exists given that the typical operational pitch range is 10-15 deg. nose up or down, though the simulator can move as far as 20-25 deg. in pitch. “This should be researched so that illusions can be simulated more accurately, and to define the associated regulatory criteria for the qualification of simulators,” the BEA says. Of concern to the BEA is that “some American airlines are putting pressure on the FAA” to authorize pilot training with fixed-base simulators to save money. “In view of the simulator issues discussed, this would appear to be inappropriate for this phase of flight,” says the BEA.

Reducing go-around thrust levels could help with Asaga-type incidents, in part by reducing the acceleration levels that could lead to nose-up illusions.

According to BEA, Airbus recently certified a reduced-thrust TOGA mode called “GA SOFT” on the A380, and is working to certify the same mode in the A330 and A340. Boeing on the 777 and 747 already has a reduced thrust mode that limits vertical speed to 2,000 fpm. “The main objective of this thrust limitation is to limit the effects of somatogravic illusions,” says BEA. “But by also inducing a vertical speed limitation, it can give extra time for the completion of the go-around.”