The airplanes we operate are reliable by design, and training is so solid that we can respond to nearly all situations without allowing them to escalate beyond control. However, perfectly good airplanes can rapidly transgress from an upset state to a Loss-of-Control In-Flight (LOC-I) condition when a pilot is not trained to react properly. LOC-I is still the number one threat, and while rare, LOC-I events are likely catastrophic. The Boeing/CAST (August 2012) indicates that LOC-I killed 1648 passengers during the past ten years, and EASA’s Annual Safety Review 2012 indicates that it is also the category with the highest fatalities.

When faced with the unexpected, pilots will need to refer to their learned skills and apply best judgement within a very small window of time. It is no wonder that this subject has received focused attention during the past five years following a number of compelling accidents: Colgan Air 3407 (Q-400, Buffalo, 12 Feb. 2009), Turkish Airlines 1951 (B737, Amsterdam, 14 Mar. 2009) and Air France 447 (A330, Atlantic Ocean, 31 May 2009). While other causes of accidents have been systematically addressed through technology (CFIT, powerplant-related, mid-air collisions), LOC-I prevention requires better awareness, recognition and avoidance, and recovery training. In today’s cockpit, the pilot’s training is the final safety net to prevent LOC-I.


While the seeds for the Royal Aeronautical Society’s ICATEE (International Committee for Aviation Training in Extended Envelopes) had already been planted prior to these events, a conference in London by that society launched an earnest effort to understand the causes of these upsets and to define the best training solutions.


What triggers upsets? They can be triggered by pilot, environment or system-induced conditions as shown in the following table, based on Jacobson [2010].


Pilot-Induced Environmentally-Induced System Induced
Improper/inadequate training Weather (turbulence, icing, adverse winds, wind shear)  

Reduced envelope/mode protection

Poor energy management Wake vortices Poor energy management (systems-induced)
Changing pilot skill base Visibility (for VFR flights) Propulsion-related
Automation/mode confusion Foreign Object Damage Erroneous sensor data
Destabilized approaches Icing A/C systems failures

Improper procedures (e.g. poor monitoring)




Startle - the LOC-I Catalyst

According to Lambregts [2008], aerodynamic stall is the leading cause of fatal LOC-I accidents, contributing to 36%. Surprisingly, some crews (e.g. Colgan 3407 and AF 447) responded inappropriately to the stall warning and protection systems. Other flight crews appeared unaware of the flight condition, or reverted to maintaining altitude whatever the cost - a major systemic training deficiency. True, stall is an end product of poor energy management, inattention, inaccurate flight path monitoring, or weather-induced events. Yet, despite the escalation of the events leading to the stall, the recovery at any stage (prior to the g-break, or after loss of lift and the resulting unsteady aerodynamic conditions) remains the same: an immediate reduction of angle-of-attack is requisite.


Why then do upsets invariably become Loss of Control events? When combined with a situation that causes a pilot to become alarmed, leading to an inability to properly resolve the problem - a condition commonly known as “startle,” the upset can rapidly become a LOC-I event. Most of the LOC-I events are believed to occur when an upset provokes a startle reaction. If this is true, then unfortunately knowledge and the traditional maneuver-based training alone will not prevent LOC-I. Startling scenarios are needed.


The crew’s startle reaction is the leading catalyst that can take an upset airplane into an LOC-I condition. Can we create startle in training? Probably. Can enough startle scenarios be developed and used appropriately so that they remain effective? Possibly. Do we need to train startle management? Absolutely.


Stall Tactics

Until 2012, the aeronautical community had a misplaced emphasis on minimizing altitude loss during a stall recovery. While the safety intentions of such an emphasis may have seemed sound, it led to the inappropriate establishment of specific standards for altitude loss in proficiency checks, and that was accompanied by an unintentional hiding of stalling physics. Training events that instilled minimizing altitude loss consisted of preplanned, announced, stall recovery maneuvers. Invariably, a pilot only needed to apply power, quickly break the stall with a short nose drop, and adjust the airplane attitude to continue recovery - without compromising altitude. It was a hand-eye co-ordination task, and because the maneuver was planned, it caused no startle reaction or startle management from the pilot. However, if those are the ONLY skills that one has acquired, are they enough to prevent reoccurrences of the recent stall accidents? Perhaps not.


Light aerobatic-capable aircraft can be a beneficial learning environment, and are recommended for at least initial training at the licensing level. A good instructor can demonstrate and hone the flying skills of the pilot in UPRT and provide the bridge of knowledge that pertains to transport category aircraft. Part of the complete initial UPRT program, and the subsequent type-specific training, must rely on flight simulators with a representative flight-deck environment, even for stall training.


In a stall, the aircraft behaviour may be unpredictable as the aerodynamics are unsteady. No two airplanes or situations are the same, and no two stalls are the same. Unpredictability is not the kind of behaviour we commonly want from our simulation software, especially when qualification is required. However, with a major focus on stall training, simulator models that do represent stalls “accurately enough for the training objectives” are now becoming available. Boeing and NASA Langley Research Centre developed an accurate stall model for one aircraft type nearly a decade ago. Boeing has developed a full stall model for the 737NG using data from hundreds of flight-test stalls. “Type representative” models, depicting the needed random behaviour are also becoming available (for example, from Bihrle Applied Research), while other consortia continue to develop convincingly realistic real-time models based on wind tunnel and computational fluid dynamics data. Hence, there is no technical reason that prevents full stall training.


Simulator Stall Training Requirements

Is training the recovery from an approach-to-stall in a simulator sufficient? The viewpoints differ. While US Public Law 111-216 and recent Part 121 revisions require training to full stalls and upsets, some argue that this could lead to negative training transfer. While they argue that improvements in prevention alone are sufficient, others believe training should go beyond that.


As the aircraft comes close to the aerodynamic stall, the aircraft flight characteristics degrade and the controls become sluggish. Buffet cues may help the pilot to respond, and in some cases, the vibrations can be so severe that instruments become unreadable. Pilots could be drawn into a tendency to maintain the nose-up attitude or try to control bank angle at the expense of recognizing and recovering by reducing the angle-of-attack. Therefore, there is a strong argument in favour of exposing pilots to the complete threat environment in a properly controlled manner.


The FAA conducted a study in Oklahoma City in late 2013 involving 45 Boeing 737 airline pilots who had been previously “approach-to-stall” trained in their company simulators. They were all briefed on, and indicated they were familiar with, the recently published OEM Stall Recovery Template explaining how to recover at first indication of stall by applying a nose-down pitch input until the stall warning is eliminated.


During the study, the airline pilots were presented with an unexpected surprise stall situation. The result was as startling for the researchers as it was for the pilots: Only one-fourth of the pilots applied the proper stall recovery procedure correctly when surprised. Most of the pilots - for a significant length of time - applied back pressure, worsening the stall. The advanced stall models also tempted pilots to deviate more from the proper recovery technique through actions such as applying significant pedal inputs.


The bottom line of this eye-opener was that reverting to the old recovery technique was a dominant response when pilots were surprised. Clearly, the approach-to-stall maneuver-based training leaves something to be desired. Exposure to the startle effect acting as a psychophysical catalyst in combination with the stall reveals errors that simulator training can correct. Both aeronautical knowledge and exposure to the threat environment can be used to develop the confidence that is needed to avoid startle and learn to recover properly. In other words, remaining calm during an emergency can only be fully realized after one has been shown that they are indeed capable of resolving that emergency.


Using today’s technology to get the most out of UPRT is strongly recommended. While modifications to stall models and the presentation of UPRT-critical information on the instructor operating station may involve time and investment, airlines should applaud the fact that over half of the required training can be accomplished by making better use of current-day simulators, when combined with proper knowledge-based training. The Airplane Upset Recovery Training Aid (AURTA) is the distinguished source of the aeronautical knowledge, covering causes and cures for most upsets.


Investing in Risk Reduction

UPRT can provide the biggest bang for the buck when properly implemented and quality assured. The forthcoming ICAO Manual of Aeroplane Upset Prevention and Recovery, a manifestation of several international committees including ICATEE, promises to define the training elements necessary to ensure pilots develop the requisite knowledge and skills for a successful program. However, the development or revision of those programs is left to the operators and local authorities themselves.


An innovative development is taking place at South African Airways (SAA), whose underwriter has pledged assistance with the implementation of their UPRT program. According to their Chief Training Captain Johann Du Plessis, “Being faced with an upset condition which takes a pilot out of their comfort zone cannot be successfully recovered from, unless ingrained recovery techniques have been developed. Therefore, our insurers and underwriters appreciate the risk associated with loss of control in flight and have been extremely supportive with the introduction of a recognized and comprehensive upset prevention and recovery training programme”. This includes acquiring a tablet-based version of the AURTA, development of the entire training program, and a complete “Train-the-Trainer” program for their instructors utilizing on-aircraft and simulator training.


“The airline industry recognizes that LOC-I is the leading cause of fatal accidents”, states SAA’s Chief Standards Pilot, Captain Sandy Bayne. “A lot of the actions you take in recovering from an upset are really counter-intuitive.” Following completion of the course at Aviation Performance Solutions in Mesa, Arizona, Captain Bayne felt “I now have the ability to impart this knowledge, to understand the concepts behind an upset and recovering from it.”


While not every nation, airline or pilot may have the luxury to impart on-aircraft training (it is recommended though as part of CPL or MPL training), the message is clear: Properly designed and carefully instructed programs that integrate knowledge and practical exposure to the upsets - that develop the confidence pilots need to bring an airplane back into its flight envelope - can be powerful in preventing upsets and avoiding Loss of Control In Flight.


And don’t forget that important catalyst: Surprise!



Jacobson, S., “Aircraft Loss of Control Causal Factors and Mitigation Challenges”. In Proc. of AIAA Guidance, Navigation and Control Conference, Toronto, Aug 2010, AIAA-2010-8007 CP.


Proceedings of AIAA GNC Conf., Toronto, 2010

Lambregts, et al, “Airplane Upsets - Old Problem, New Issues”. In Proc. of AIAA Modeling & Simulation Technologies Conf., Honolulu, Aug 2008, AIAA-2008-6867 CP.


About the authors:

Dr. Sunjoo Advani, the owner and president of International Development of Technology b.v., is an aerospace engineer and pilot. He has chaired the ICATEE team in the development of the content that became the core of the ICAO Manual of Aeroplane Upset Prevention and Recovery.


Dr. Jeffery Schroeder is the FAA’s Chief Scientific and Technical Advisor for Flight Simulation Systems. He has served as Research & Technology Co-Chair of ICATEE.


Capt. Bryan Burks is a Seattle based pilot and UPRT content developer with Alaska Airlines. He has served as Training & Regulatory Requirements Co-Chair of ICATEE.