Stochastic modeling of fatigue damage for on-line monitoring, prognostics and life-extending control

The primary goal of a stochastic damage monitoring, failure prognostic and life-extending control system is to achieve the desired plant performance with enhanced safety, reliability, and availability. This integrated decision system is potentially useful for on-line damage monitoring, predicting remaining service life and for making operation and maintenance decisions.; Modeling the stochastic nature of material degradation is necessary for failure prognostics and risk analysis of plant components. Two different stochastic models of fatigue crack growth are developed. The first model is formulated on the assumption that the crack growth rate is lognormal distributed. This model structure allows construction of a filter for estimating the current damage state and predicting the remaining service life based on the underlying principle of extended Kalman filtering. This allows evaluation of the first two moments of crack length where the computation time is at least two orders of magnitude less than that required for solving the Kolmogorov equation. The second model is based on the assumption that crack length is lognormal-distributed. This model does not require an explicit solution of stochastic differential equations and is ideally suited for on-line damage monitoring. Both models allow the expected value of crack length to be matched with the deterministic prediction. The variance of crack length derived from both models is verified with experimental fatigue crack growth data.; The goal of life-extending control is to achieve an optimized trade-off between dynamic performance and structural durability of the plant under control. Enhancement of structural durability is achieved using nonlinear optimization to generate a sequence of open loop commands that maneuver the plant from a known initial state to a desired final state subject to constraints on the rate of damage accumulation in critical components. Further, a methodology for the development of a robust feedforward/feedback control policy for high performance and life extension of mechanical structures is developed. This concept is experimentally verified on a laboratory testbed. Test results demonstrate that the fatigue life of test specimens can be substantially extended with no appreciable degradation in the dynamic performance of the mechanical system.