Multiplicative sensitivity analysis and its role in development of simulation models

Abstract A procedure of extended sensitivity analysis for simulation models is developed which is applicable to the situation in which input uncertainties are specified in a multiplicative (rather than an additive) way. This type of specification usually occurs when error estimates are large relative to the parameter values. The input parameter errors are assumed to have log-normal distributions and the basis for this assumption is explained along with the advantages it provides. The new procedure, multiplicative sensitivity analysis, is developed for the purposes of internal model validation and identification of priorities for specific experimental and theoretical studies needed for a particular system being modelled. The applicability of the procedure in the very early stages of a research program is its main advantage and further expands the role of simulation models.