Model Discrimination via Designed Experiments: Discriminating Between the Terminal and Penultimate Models on the Basis of Weight Average Chain Length

Abstract

Simulations have been used to study the application of weight average chain length for discriminating between the terminal and the penultimate copolymerization models. A brief description of the model discrimination method and of the copolymerization models is given. The problem of parameter observability is discussed, showing that weight average chain length on its own leads to highly correlated parameter estimates. Combining weight average chain length with triad fractions on the other hand, yields a powerful combination of measurements for model discrimination. In the simulation studies the correct model was selected in each of a variety of different scenarios used. Compared to using triad fractions alone, including weight average chain length has several advantages, notable among which is the ability to estimate radical reactivity ratios, s1, and s2, which are not observable from triad fraction data alone. Also, weight average chain length is more sensitive to changes in the rate of polymerization.