Discriminating between the terminal and penultimate models using designed experiments: An overview

Statistical model discrimination methods have been applied to the modeling of copolymerization kinetics in order to improve our ability to correctly choose between competing models. The objectives were to quantify the benefits of using model discrimination methods, to examine available copolymer measurements, and to rigorously compare model discrimination techniques. Simulations were developed (for three representative free-radical systems) capable of generating data based on either the terminal or penultimate model. Using the simulations to generate data, three model discrimination methods were applied to the three copolymer systems alternatively using copolymer composition, triad fraction data, and the combination of copolymer composition and rate as the measured variable(s), The results showed that model discrimination methods can correctly discriminate often using fewer experiments than have been used in the past and that they should be able to detect smaller differences between models.