Abstract
An improved error propagation technique was needed for the analysis of nuclear reactor safety systems. In response, a computer program for calculating the first four moments of a second-order function was prepared and validated. This program, called SOERP, can accomodate up to 30 statistically independent random variables. Thus, the moments for any function that can be expanded in a multivariable Taylor series up to the second order can be estimated. These moments can then be used to determine a probabilty density function describing the dependent variable. The above two steps represent a method for evaluating a complex system's performance fluctuations that arise from random variability in the behavior of system components. For example, tolerance limits on the performance of electrical circuits or on transit times in a repair facility may be evaluated more precisely. Another application is the evaluation of profitability calculations in which the input data are subject to uncertainty.

This publication has 2 references indexed in Scilit: