Obtaining precise parameter estimates for nonlinear catalytic rate models

Abstract

AbstractA method is reviewed which allows data points to be chosen in such a fashion that precise estimates of the parameters in nonlinear reaction rate models can be obtained. This method allows each future data point to be selected such that the confidence region of the estimated parameters is smaller with it than with any other possible data point within the region of experimentation. This procedure is applied for Hougen‐Watson models with hypothetical experimental data which were generated with the guidance of an example from the current chemical engineering literature. It is found that, for the same number of data points, the parameters in the model can be estimated eighteen times more precisely by using this suggested experimental design than by another commonly used design. Confidence regions are presented for the parameters of the Hougen‐Watson models with two types of designs.It is found that the positions of the data points in the well‐designed experiments are more sensitive to the functional form of the model than to the current estimates of the magnitudes of the parameter values.