Ellipsoid non-probabilistic reliability analysis of the crack growth fatigue of a new titanium alloy used in deep-sea manned cabin

Abstract

Reliability analysis of crack growth fatigue of titanium alloy is one of the important guarantees for the safety of deep-sea manned cabin. However, the traditional probabilistic fatigue reliability analysis methods often need accurate parameter statistics information. In this paper, under the condition of the limited test data, a first-order approximate ellipsoid non-probabilistic reliability analysis method for fatigue crack growth of a titanium alloy was proposed. A limit state equation of the crack growth correction model, considering the small crack effect, is formulated according to the fatigue life model. The uncertain parameters are constructed by the interval model, normal distribution, and Khachiyan algorithm. The different limit state equations are obtained by changing the coefficient κ. The first-order approximation and Monte Carlo methods are used to solve the reliability of the ellipsoid model. Calculations demonstrate that the results of the ellipsoid non-probabilistic reliability method are more conservative than that of the probabilistic reliability method. When κ is less than or equal to 2, the Khachiyan algorithm is preferred to build the ellipsoid model, and when κ is greater than 2, the interval model is preferred. Through parameter sensitivity analysis, the uncertain parameters KC, n, m, A, and ΔKthR are found to significantly affect the crack growth rate.