Purdue University Graduate School
MS Thesis Neelakshi Majumdar IEEE Purdue V2.pdf (3.67 MB)

A State-based Approach for Modeling General Aviation Fixed-wing Accidents

Download (3.67 MB)
posted on 2019-01-16, 19:46 authored by Neelakshi MajumdarNeelakshi Majumdar

General Aviation (GA) is a category of aircraft operations, exclusive of all military and commercial operations. According to Federal Aviation Administration (FAA), fixed-wing aircraft (also known as airplanes) account for 76.2% of all the estimated registered GA fleet in the United States. Out of all the GA accidents that the National Transportation Safety Board (NTSB) investigated in 2017, 87.7% of the accidents involved fixed-wing aircraft. The NTSB reports on all GA accidents and records the accident details in their database. The NTSB database has an abundance of accident data, but the data is not always logically complete and has missing information. Many researchers have conducted several studies to provide GA fixed-wing accident causation using the NTSB accident data. The quantitative analyses conducted by the researchers focused on a chain of events approach and identified the most frequent events in accidents. However, these studies provided little insight into why the events in the accidents happened. In contrast, the qualitative analyses conducted an in-depth study of limited accidents from the NTSB database. This approach helps in providing new findings but is difficult to apply to large scale datasets. Therefore, our understanding of GA fixed-wing accident causation is limited. This research uses a state-based approach, developed by Rao (2016), to provide a potentially better understanding of causes for GA fixed-wing accidents. I analyzed 10,500 fixed-wing accidents in 1982–2017 that involved inflight loss of control (LOC-I) using the state-based approach. I investigated the causes of LOC-I using both a conventional approach and a state-based approach. I analyzed fatal, non-fatal and overall LOC-I accidents in three timeframes: 1989–1998, 1999–2008 and 2008–2017. This multi-year analysis helped in discerning changes in the causation trends in the last three decades. A mapping of the LOC-I state definition to the NTSB codes helped in identifying 2350 more accidents in the database that were not discernible using the conventional approach. The conventional analysis revealed “directional control not maintained” as the top cause for the LOC-I accidents, which provides little information about how loss of control happened in accidents. The state-based analysis highlighted some important findings that contribute to LOC-I accidents that were not discernible using the conventional approach. The state-based analysis identified preflight mechanical issue as one of the new causes for LOC-I with a presence in 5.1% of LOC-I accidents in 2009–2017. It also helped in inferring some of the missing information in the accident data by modeling the accidents in a logical order. Using the logic rules in the state-based approach, I inferred that the pilot’s tendency to hit objects or terrain caused loss of control in 19.9% of LOC-I accidents in 2009–2017. Further, the logic rules helped in inferring that 7.5% of LOC-I accidents in 2009–2017 involved hazardous condition of an aircraft before the start of flight. A comparison of the findings from state-based approach with the GAJSC (General Aviation Joint Steering Committee) safety enhancements revealed that the state-based approach encompassed all the potential issues addressed in the safety enhancements. Additionally, a state-based analyses of larger datasets of fatal and non-fatal accidents suggested some new potential issues (such as improper maintenance) that were not explicitly addressed in the GAJSC safety enhancements.


Degree Type

  • Master of Science in Aeronautics and Astronautics


  • Aeronautics and Astronautics

Campus location

  • West Lafayette

Advisor/Supervisor/Committee Chair

Dr. Karen Marais

Advisor/Supervisor/Committee co-chair

Dr. Dengfeng Sun

Additional Committee Member 2

Dr. William A. Crossley

Usage metrics



    Ref. manager