论文部分内容阅读
Reliability evaluation of critical systems such as solid rocket motors is generally a difficult and costly task in terms of the real testing.In this case, numerical analysis technique that can predict the reliability at the early stage of design can be a viable solution, from which possible design improvements to attain higher reliability can be made in a fast manner.In this study, reliability analysis procedure is developed for a solid rocket motor case, which undergoes high stresses due to the internal pressure generated by propellant during the mission flight.Failure condition is defined as that the maximum stress exceeds the yield strength of the motor case.In order to determine internal pressure, a simple 1-dimensional steady conservation equations of mass, momentum and energy are solved, given a set of propellant properties and motor dimension.In view of the maximum stress, critical part is identified as given in Fig.1.FEA for the part is performed under the applied pressure using ANSYS 13.0.Result under an initial condition is given in Fig.1.Probabilistic information due to the uncertainties in the propellant and geometry are given either from the previous knowledge or by the direct measurements.In the latter case, influences of the statistical uncertainties due to the insufficient data that are too few to estimate associated parameters are considered by employing Bayesian approach.For the evaluation of maximum stress, response surface model is constructed to reduce the costly computations in FEA.Bayesian reliability analysis is conducted accounting for the insufficient data of propellant properties, which runs conventional reliability analysis that determines failure probability under the given probability distribution as its inner loop and probability distribution of failure probability as its outer loop.For the inner loop, First Order Reliability Method (FORM) is employed using the commercial software NESSUS.The results are shown in Table 1.In the conventional reliability analysis approach, the failure probability at the head end part is 0.0143.In the Bayesian approach, the failure probability is determined in the form of distribution, which represents our degree of belief arising due to the insufficient data.For the sake of conservatism, the 95% upper confidence limit is chosen, which is 0.0194.Detailed formulation and implementation will be presented in the conference.