Interval reliability sensitivity analysis using Monte Carlo simulation and mouth brooding fish algorithm (MBF)

In reliability analysis and reliability-based design approach, sensitivity analysis establishes a relation between the variation in random variable parameters and that in reliability. The analysis of sensitivity is also employed to determine key random variables with the greatest contribution to reliability. The sensitivity analysis of a non-linear limit state function is relatively complicated and time-consuming. Therefore, a novel approach is proposed in this study to calculate the reliability sensitivity parameter in terms of an interval. In the proposed algorithm, the non-linear limit state function is initially converted to a linear limit state function by the Monte Carlo simulation method. Then, the interval random variables are introduced to include both epistemic and aleatory uncertainties so as to obtain a realistic result. In the second step of the proposed methodology, the mouth-brooding fish algorithm (MBF) is applied to capture the upper and lower bounds of the reliability sensitivity parameters. Several illustrative examples were then presented, showing the capability of the proposed approach in obtaining results accurately and efficiently. According to the results, the percentage of improvement in the sensitivity analysis of random variables in the proposed method, compared to the classical method, by using the MBF optimization algorithm to the parameters of random variables due to the variation Coefficient is more than 40 percents. This means that the length of the response interval has increased. •A new approach is proposed to calculate the reliability sensitivity parameter.•The non-linear limit state function is converted to linear limit state function by MCS.•Both epistemic and aleatory uncertainties are considered.•The GA is applied to find the upper and lower bounds of the reliability sensitivity parameters.