Probabilistic reliability evaluation of space system considering physics of fatigue failure

Mission assurance requires due diligence reliability for space systems taking into account limited accessibility, high uncertainty on the life data and high cost of failure. The methods based on physics of failure are promising approaches for durability evaluation of these systems. In this study, the reliability analysis is aimed for space structures, with the focus on fatigue failure. In this research, the deterministic fatigue simulation is conducted on space systems (satellite in orbit, low-level LEO, made of aluminum 2024-T3), using models with constant and variable amplitude loading. Walker and Forman models are preliminary utilized in this study for life prediction and were compared with the experimental results. In the case of the variable amplitude loading, the Wheeler model is used for consideration of retardation process due to the amount of plastic zone in crack tip. Deterministic crack growth simulation was numerically simulated by using the MATLAB software and has been compared with commercial AFGROW software for verification and proper match with experimental data. In the analysis of stochastic fatigue crack growth, uncertainty is analyzed by using the Monte Carlo simulation. It is observed that by increasing the crack length, uncertainty range is widening. The universal stochastic crack growth model proposed by Yang and Manning, was used for reliability analysis based on giving probabilistic method for the purpose of power and second polynomial models. In this study, these models are evaluated and three models of I) rational model, II) exponential model and III) global model are proposed. The exponential model demands less computational resources however it has a lower accuracy. The fractional model, proposed in this research, is based on the modification to Forman model. However, these models don't consider geometric factor. The Global model, another model proposed in this research, has the capability of considering this aspect. In multiplicative stochastic factor (Yang and Manning method), accuracy of the approximation is most important. By improving the accuracy of this relation, the results accuracy is enhanced. For this purpose, for increasing efficiency of this method, the accuracy of approximation must be increased by correcting prior models or provide new accurate models.