Modified Integral Procedure (MIP) as a reliable short-cut method in mechanistic based ODE kinetic model estimation : Non-isothermal and (semi-)batch process cases

Abstract

In both small and large-scale investigations, a reliable short-cut procedure to estimate the approximate parameters is very useful for the successive rapid checking of different Kinetic Model (KM) structures for their adaptation to current process data. An improved quality of the initial parameter guess improves also the reliability and the convergence rate for a subsequent exact Nonlinear Least Squares (NLS) regression technique applied for fitting the final model. The recent proposed Modified Integral transformation Procedure (MIP) short-cut method of Maria and Rippin (1994) adds supplementary elements of similarity analysis to the classical Integral transformation Procedure (IP). By exploiting the kinetic model structure and interactive use of prior information stored in a kinetic model-data-bank, the MIP makes rapid adaptation of a KM structure and parameters, describing an already studied process, to a similar process under study. The problem decomposition and the term-by-term sensitivity and estimability analysis of the model for various experimental data sets increase the reliability of the MIP in reaching the global KM parameter solution region and improve the estimate quality for isothermal data cases. These results are extrapolated in the present work for other cases, including non-isothermal linear kinetics and/or on-line recursive kinetics estimation in (semi-)batch processes. The MIP results are compared with classical short-cut methods, extended Kalman Filter (EKF)-based recursive estimators of different complexity and exact NLS estimators.