A Training Study to Improve Monitoring During A Go-Around

As part of an FAA program to improve go-around (GA) safety, we were asked to determine if we could improve the performance of the Pilot Monitoring (PM) during a GA maneuver. In particular, we focused on the PM’s ability to monitor and advise the Pilot Flying (PF) during maneuver performance.
Project objectives were to
1. Develop a set of GA scenarios that capture important aspects of the GAs that are encountered in line operations,
2. Develop a description of skilled performance for monitoring during a GA,
3. Develop training that could help pilots learn to monitor more effectively and, therefore, support the GA maneuver as PM, and
4. Use full-flight simulators to evaluate whether monitoring performance improved after training.
We worked with two experienced pilots to develop a description of skilled monitoring for a GA in a Boeing 737NG; it included:
• a rapid assessment of the situation prior to initiating the GA,
• an understanding of how the airplane and the autoflight system will behave during the maneuver, including the time-course of those changes, in order to establish strong expectations for monitoring flight deck indications,
• attention to indications that are the most relevant for guiding control inputs as the maneuver progresses, and
• a knowledge of specific thresholds for triggering actions on controls.
Guided by this description, we developed an Experimental training module and a Control training module. The Control training was similar to how GA training is currently done, focusing on the procedural callouts and required actions during the maneuver.
We then studied PM performance during GA maneuvers in a full-flight simulator with 16 737 First Officers serving as the PM; 8 pilots in each training condition. Each participant flew eight GA scenarios (Set A), completed a training module, and then flew eight more GA scenarios in a parallel set (Set B). A confederate pilot served as PF for each scenario. The PF flew half of each set of GA scenarios well and flew the other half poorly (“bad” scenarios), meaning that the PM needed to detect performance problems and speak up.
Performance analysis addressed only the “bad” GA maneuvers. The primary focus of analysis was on performance differences tied to training condition and set, and importantly, the interaction of training group and set. Specifically, the question asked was, “Did the experimental training condition change performance from Set A to Set B more than did the control training condition?”
Despite the Control group having, on average, significantly more total flight time, the Experimental group significantly outperformed the Control group on five measures of effective PM performance:
• Calling out the PF’s maneuver performance
• The period of time when the PF flew poorly
• Two measures of time related to managing post-maneuver airspeed
• GA knowledge
We will discuss the findings in detail as well as other findings from the study tied to the perceived value of various types of training for this maneuver.