REVISITING THE DESIGN OF EXPERIMENTS FOR COPOLYMER REACTIVITY RATIO ESTIMATION

Tidwell and Mortimer used the D-optimal criterion to design terminal model reactivity ratio estimation experiments. In this paper, two problems are investigated which were not addressed by Tidwell and Mortimer. The first problem is the design of reactivity ratio estimation experiments subject to composition constraints. It is shown that all the key information contained in the D-optimal criterion can be summarized in two equations: one which is a function of r1, and one which in a function of r2. These two equations can be used to design reactivity ratio estimation experiments under composition constraints which are within 3 mol % of the D-optimal feed compositions. The second problem is the design of penultimate reactivity ratio estimation experiments. The D-optimal criterion for the penultimate model was derived and optimized for two existing systems as well as a large number of hypothetical systems. The D-optimal criterion for the penultimate model was found to be much more complicated than the terminal model criterion. For this reason, the optimal penultimate reactivity ratio estimation experiments can not be approximated by simple equations. Anyone interested in designing such experiments must derive and optimize the D-optimal criterion. (C) 1993 John Wiley & Sons, Inc.