Quantitative Risk Assessment - An Application of Life Data Analysis to Component Failure Forecasts

To ensure product safety, reliability, and availability, fleet management decision making must be based on accurate forecasts of component failures, availability, and demand. With the objective of preventing safety risk from exceeding a specified threshold while maintaining a required level of availability, component failure forecast is usually performed on life data that has only a few field failures. This paper presents an approach called Quantitative Risk Assessment (QRA) for forecasting component future failures based on application of life data analysis methodology. Various statistical models for component lives are hypothesized, reviewed, and evaluated. The methodology to select the best statistical model is also discussed and several life data methodologies for dealing with failure time uncertainty are also reviewed. To account for missing or partial information, Monte Carlo simulation is employed to generate complete data sets. Random generators in Monte Carlo are evaluated and selected according to various criteria. For a more complex risk assessments; those with recurrent inspections, probability of detection information, critical crack length, crack propagation rates, and retirement lives, a simulation method that evaluates scenarios under consideration is proposed to forecast future failures.