Simple method based on sensitivity coefficient for stochastic uncertainty analysis in probabilistic risk assessment

For the analysis of stochastic uncertainty in probabilistic risk assessment, a simple method based on the sensitivity coefficient was developed. The sensitivity coefficient can be defined as the importance of the parameter included in the risk assessment model to the output such as the probability of the target event. When the contribution of the parameter to the output is assumed to be linear, the sensitivity coefficient equals Fussell-Vesely importance. The present method does not require a lot of calculation cost and can treat the covariance of the parameters included in the risk assessment directly. The result obtained by the present method was compared with that obtained by other methods such as the Monte Carlo method in the analysis of the simple fault tree model. The results of the present method agree well with Monte Carlo method in the analysis of the fault tree model with beta factor method and that with the Multiple Greek Letter method.

Automated summary

The sensitivity coefficient is a simple yet effective method for evaluating the importance of parameters in a risk assessment model. This method is cost-effective and can directly handle the covariance of parameters, making it comparable to other methods such as Monte Carlo simulations in analyzing fault tree models.