A generalized probabilistic rejection test for multivariable sensitivity analysis

A probabilistic rejection test for multivariable sensitivity analysis is presented. The test is applied by randomly changing all the assumed unimportant (those having low sensitivity values) input parameters simultaneously and calculating the appropriate response. It is shown that by repeating this procedure N times, where N is much smaller than the number of input parameters, it is possible to assign a probability limit to the assumption that a high-sensitivity parameter exists. Application of the test is demonstrated in a nuclear-waste-disposal problem.