Use of Statistical Experimental Design in a Kinetics Study

Abstract

A brief review of the statistical design of experiments where the objective of the experiments is to estimate parameters in a given model is presented. Methods for choosing experimental conditions that, for a given number of experiments, will result in the maximum accuracy (or the smallest confidence region) of the estimated parameters are discussed. The techniques are applied to the design of an experimental study of the rate equation for formation of phosgene from carbon monoxide and chlorine with constraints on the entering feed composition and temperature of the reactor. The resulting confidence regions for several designs are presented and show, for instance, that two seemingly good designs can differ in size of confidence regions by a factor of 30.