Modeling fatigue crack growth for life-extending control

This dissertation presents a nonlinear dynamic model of fatigue crack growth in the state-space setting under variable amplitude cyclic load. The model is especially suited to the needs of real-time decision-making for life-extending control. The state variables are crack length and crack opening stress. The model is capable of capturing the effects of a single-cycle overload, block loads, random loads, and irregular sequences through a fading memory algorithm. Model predictions are in good agreement with experimental data on 7075-T6 and 2024-T3 aluminum alloys. Compiled results also demonstrate that the proposed model compares well with one of the most comprehensive models, FASTRAN-II that is used by the aircraft industry. Specifically, the state-space model recursively computes the crack opening stress via a simple functional relationship based on the principle of fading memory and does not require the storage of the stress history for its execution. Therefore, savings in both computation time and memory requirements are significant.; The need for a reliable damage model for life-extending control is addressed with reference to the colossal inaccuracies that could occur in controller synthesis for a reusable rocket engine if a simplistic damage model is used under variable-amplitude load conditions. The seemingly counter-intuitive notion of overload injection could be gainfully utilized for life-extending optimization. The proof of this concept is demonstrated on a laboratory test apparatus by life-extension of test specimens with intentionally injected overload pulses at specific intervals.; A stochastic model of fatigue crack growth under variable-amplitude load is proposed using the framework of the state-space model. The stochastic model is validated with four sets of constant-amplitude load test data and a set under variable-amplitude load test. The crack growth process is observed to be nearly deterministic for a cyclic load applied to a given specimen, provided that its microstructure does not undergo any major change. Therefore, a reliable deterministic model is a prerequisite for a stochastic model of fatigue crack growth. The stochastic model of fatigue crack growth does not require expensive computations to obtain a solution to the stochastic difference equations.