NEW PROCEDURE FOR OPTIMAL-DESIGN OF SEQUENTIAL EXPERIMENTS IN KINETIC-MODELS

The high correlation between preexponential factors and activation ene rgies is a severe drawback in the estimation of kinetic parameters. Th is difficulty is generally overcome using two different techniques, i. e., reparametrization and parameter separation. It has been demonstrat ed that in sequential experimental design the former technique can onl y occasionally lead to better numerical results, with no theoretical a dvantage over the non-reparametrized model. In this work we show that the separability property can be a convenient tool also for the optimi zation of experiment planning. A numerical example is considered, and both cases of constant absolute and relative errors are examined.