不規則過大荷重下での疲労寿命解析に対する新しい高速モンテカルロシミュレーションスキーム

抄録

A fast Monte Carlo simulation scheme is newly developed for fatigue failure under random overloads, which is based upon an importance sampling technique using a probability measure transformation. A probabilistic model proposed by the authors is applied to describe fatigue crack growth under random overloads, in which a two-criteria failure condition consisting of a fatigue fracture mode as well as a ductile fracture mode is taken into the model. Then, an importance sampling measure is newly proposed as a mixture measure of “double mode”, one is for estimating probability associated with fatigue mode and the other is for probability associated with ductile mode. Numerical examples are shown for demonstrating the proposed simulation scheme, which clarifies that (i) the proposed method can accurately estimate extremely small failure probabilities (such as order of 10^{－ 5}) and (ii) it achieves a significant improvement in computational efficiency, which can reduce the computing time to approximately 1/50 of that required by a large-scale crude Monte Carlo simulation.