| The objective of this research was to develop a comprehensive probabilistic methodology such that Federal Aviation Administration (FAA) engineers can conduct a risk assessment of a General Aviation structural issue in support of policy decisions.;After some wing fatigue cracks were found in the fleet of twin Cessnas, the FAA started to be concerned about this unsafe condition and therefore advised a risk assessment. A risk assessment of the continued operational safety of the General Aviation fleet can provide important insight to the criticality and severity of a potentially serious structural issue. As such, the methodology to address risk assessment and risk management of General Aviation structural issues is needed. This information will provide an approach to enable a nonbiased review of data to assure airworthiness.;This work aimed to examine the different components, including airplane-to-airplane and flight-to-flight variations to develop a probabilistic methodology that can perform risk assessment in small airplanes. Requisite supporting technology and data issues were investigated. In particular, probability distributions of relevant inputs were developed so that a realistic risk assessment could be obtained. Representative sensitivity studies were also executed in order to demonstrate and validate the methodology.;After developing a deterministic code using Fortran, the number of flights-to-failure was evaluated for two different case studies: single engine unpressurization basic instructional usage and single engine unpressurization aerobatic usage, the last one presenting a more severe usage.;Using a probabilistic code written in Fortran and Monte Carlo sampling, a statistical representation of the flights-to-failure was developed. These statistics were then post-processed to determine a probability distribution function, cumulative distribution function, and survival function of aircraft life. Qualitative and quantitative evaluations of different variables involved in the model were performed. Results showed that variables such as Sink Rate does not play an important role on fatigue evaluation and further investigation in this variable would not be necessary; however, variables as Flight Duration, Miner's Coefficient, and Gust and Maneuver loading are important to the problem. Correlation factors and stepwise linear regression supported the importance of the different variables in the problem. Finally using linear regression, a predictive model (response surface) was developed to calculate flights-to-failure. Another linear regression model was developed for early failures with better correlation. Statistics about the goodness of these models was developed. |
