General Aviation Safety

As part of the PEGASAS COE, we are helping to make general aviation safer. Our long-term objective is to identify and develop innovative ways to significantly and demonstrably reduce GA risk. 

Presenting Risk to Pilots

The objective of this research is to improve General Aviation safety by providing pilots with effective feedback that helps them correct their mistakes and fly safer. Commercial products that leverage the addition of technology in the cockpits of small aircraft to collect flight data and help pilots visualize their flights are becoming more prevalent. The commercial products currently available refrain from any discussion of risk or flight safety. Post-flight debrief may help facilitate risk management in subsequent flights, by alerting pilots to potentially hazardous situations. However, it is not clear whether presentation format matters among pilots, and whether there are any pilot attributes that affect how pilots perceive feedback.

In this work, we evaluate how pilots respond to safety feedback presented to them in different ways during post-flight debrief. The tasks can be divided in three parts: we first collect flight data and process data from various sources (flight data recorders, smartphones, ADS-B equipment) and supplement it with additional information that may be available to us, such as weather in the area of the flight. Using all available data, we detect events in the flight that could potentially be improved. All of these events consist of behaviors that have appeared in historical accidents–we define a set of hazardous states (periods of time where the system, consisting of the aircraft and pilot, exhibits a particular behavior that may lead to an accident) and triggers (events that cause the system to transition from one state to another) using all historical accident causes and factors. The last part is informing the pilot that there is something in their flight that can be improved, in a way that motivates them to go back and seek more knowledge and try to change something about the way they are flying that may make them safer. To evaluate how to best present the pilots with such information, we are using a web-based survey that will ask pilots to self-debrief sample flights that contain safety information in different representation formats and ask them to assess the risk in the flight.

Runway Incursion Reporting Tool

The objective of this research is to design a tool that will help improve the quality of data entered into a runway incursion incident database. A runway incursion occurs when an aircraft is incorrectly present on a runway. While a runway incursion accidents are not common, runway incursion incidents are frequent, and pose a threat to runway safety. To understand how and why runway incursions occur, we analyzed root causes from past accidents and incidents. While most accident reports gave us a deep understanding of underlying issues, incident reports lacked these characteristics. These incident reports are created by air traffic controllers on duty, who may not be trained investigators. While filling out a form to report the incident, they might not always provide the information required to unveil the underlying causes of the incident. The tool will guide the controller through a series of questions which will help elicit detailed information of how and why the incident occurred, leveraging the fact that runway incursions can occur only in a limited number of ways.

Modeling Rotorcraft Accidents

High Risk Occurrence Chains in Helicopter Accidents

We developed an approach to identify high-risk sequences of events (or occurrence chains) in General Aviation (GA) accidents using historical accident data. We demonstrate our approach on the 6180 civil helicopter accidents that occurred in the United States between 1982 and 2015. We begin by providing a detailed guide to the NTSB accident database including of the pre- and post-2008 coding systems. We compare the most frequently occurring chains for different mission types and injury severity levels. The single node inflight loss of control (LOC) occurrence chain appeared in 12.5% of helicopter accidents between 1982 and 2015, appearing most frequently in both fatal and non-fatal accidents. Students’ (and sometimes instructors’) inability to perform safe autorotations or improper autorotation recoveries most often resulted in hard landings. As a first step to reducing these unfortunate training accidents, we recommend conducting an in-depth study to identify high-risk training maneuvers and hazardous situations faced by student pilots. We provide recommendations to improve the NTSB accident database and recording system that could potentially help mitigate the top causes for GA accidents. 

Comparing Hazardous States and Trigger Events in Fatal and Non-Fatal Helicopter Accidents

Helicopter safety literature generally focuses on identifying the causes and contributing factors for fatal accidents. It identifies the top mechanical and crew-related problems that lead to fatal accidents, and intervention strategies to reduce fatal accidents and improve cash survivability. Data from non-fatal accidents is an underused resource that could be used to better understand the reasons for several types of fatal accidents. If fatal and non-fatal accidents share similar stories, we can leverage information in non-fatal accident reports to reduce not just the number of fatal accidents, but rotorcraft accidents overall. We seek to identify and compare the top hazardous states (e.g., loss of control) and trigger events (e.g., incorrect cyclic input) in fatal and non-fatal accidents. We analyzed 5051 helicopter accidents that occurred in 1982–2008. Fatal accidents constitute a small proportion of accidents (15.2%), while the remaining 84.8% were non-fatal. Inflight loss of control (LOC), controlled flight into terrain/object (CFIT), weather, failure to maintain physical clearance/altitude from objects, and system/component failure comprised the top five hazardous states, which accounted for nearly 90% of fatal accidents. Poor inflight planning/decision making by pilots was most likely to trigger both fatal and non-fatal accidents LOC accidents. Decision-making errors also frequently triggered fatal CFIT accidents. The high proportion of non-fatal accidents provides an opportunity for in-depth interviews of pilots to understand the rationale behind the decisions and the actions that followed. Seven out of eight system/component failure trigger events involved the main rotor system and the rotor drive system, highlighting the importance of these systems for rotorcraft operations. Critical flight components such as tail rotor blades and gearbox, and transmission drives were equally likely to fail in both fatal and non-fatal accidents. In some cases the nature of terrain, phase of operation, and on-board protective gear were the difference between life and death.