Advanced Stochastic Techniques for Jet Engine Component Life Prediction

The paper addresses significant aspects of stochastic modeling for jet engine component life prediction. Probabilistic life prediction for gas turbine engine components represents a very difficult engineering problem involving stochastic modeling of multiple, complex random phenomena. A key aspect for developing a probabilistic life prediction tool is to incorporate, and to be open to modeling advances related to dynamic complex random phenomena, including space-time random variabilities of mission environment and material parameters, aero-elastic interactions, friction at contact interfaces, multi-site fatigue, progressive damage mechanism, including loading interactions, etc.. The paper addresses the main aspects involved in stochastic modeling of component fatigue life prediction for jet engine rotating components, specifically fan blades. The paper highlights the need of the use of stochastic process and field models for including space-time varying random aspects. Mission speed profiles produced by pilot’s random maneuvers are modeled by pulse non-Gaussian stochastic processes. These pulse processes are approximated using linear recursive models when the cluster effects are not significant. A more general approach, useful when cluster effects are significant, based on a combination of two pulse processes is used. Aero-pressure distribution on blade as well as blade surface geometry deviations due to manufacturing are idealized by using factorable stochastic field models. Also, stochastic field models are used for modeling strain-life and damage accumulation curves. Stochastic damage accumulation models are based on randomized stress-dependent models (nonlinear damage rule models). The paper also addresses mathematical modeling of stochastic nonlinear responses in multidimensional parameter spaces. Stochastic response surface techniques based on factorable stochastic fields or optimum stochastic models are suggested. An illustrative example of a jet engine blade is used for discussion and to show the consequences of different modeling assumptions.