| A detailed discussion is presented examining the purpose of sensitivity analysis and its most prominent utilities in probabilistic engineering design. In this effort, there will be an examination of the work being conducting in probabilistic sensitivity methods. Development of the probabilistic sensitivities is considered an essential aspect of probabilistic analysis and is often revered as the most critical factor in design modifications and failure assessment. Probability of failure and response sensitivities with respect to the mean, standard deviation, and correlation coefficients are calculated using the kernel functions to assess each parameter's significance on the probability of failure (POF). Sensitivities arrive from the partial derivative of a system's response or POF. The method of approximating sensitivities through use of the expected value operator is known as the score function or likelihood ratio and is utilized herein. The score function is used so that the existing samples from a probabilistic analysis can be reused to obtain the partial derivatives of the probability-of-failure or response moments as a function of the input distributions means, standard deviations and correlation coefficients with typically negligible cost. The method is general so that any sampling routine is applicable. |
