Study of effect of R-ratio and overload on fatigue crack growth using artificial neural network

Growth of a crack in a components or structures subjected to cyclic loading conditions are a prime concern as it limits the life of the component. Number of experimental and analytical works is available in the literature for prediction of fatigue crack growth (FCG). The experimental results are developed by extensive experiments on a material. The analytical models are based on some material constants and are valid only under certain conditions. To use these analytical models for prediction of fatigue crack growth, one has to perform experiments to know the material constants. There is a need of a quantitative predictive method which may predict FCG for a range of materials with available material properties. In this paper, FCG rate data available in the literature have been used to make an Artificial Neural Network (ANN) based model to predict Fatigue Crack Growth Rate (FCGR) for different materials at different R-ratio. With the help of neural network fitting of data can be achieved without making prior assumptions about the relationship to which the data are fitted. The aim was to study FCGR of different metal alloys like Steel alloy, Aluminium alloy and Titanium alloy for providing the FCG rate and the effect of load ratio (R) on these materials with known low cycle fatigue and other mechanical properties. The effect of overload was also studied for aluminium alloys like AA7020-T6 and AA2024-T3 by using ANN with Bayesian Regularization algorithm. The study well revealed that ANN can be well utilized to predict the fatigue crack growth using known material properties under different R-ratio and crack retardation behavior under overload situations.