Proposta e análise de diferentes metodologias para estimação de parâmetros cinéticos por meio do ajuste de perfis dinâmicos de concentração

Abstract

Purpose: The estimation of kinetic parameters is an essential step in the design of reactors, allowing the temporal prediction of the reaction yield and the identification of optimal operating conditions. This manuscript presents a comparison of three different methodologies for parameter estimation using a case study: the Henkel process. Methods: The first methodology is based on multiple linear regression. The second and third are based on nonlinear regression solved by employing optimization and the objective function of the proposed optimization problem uses reaction rates and concentration values, respectively. Results: The best values of the estimated parameters, considering both physical consistency and statistical significance, are those obtained using nonlinear regression using the experimental values of concentration. The results did not show any bias, although some of the parameters had large confidence intervals. Conclusions: The analytical solution of a linearized nonlinear problem does not achieve global minima, therefore it is a discouraged to use. The use of preprocessed variables in the optimization problem, in this case, the reaction rate, is not recommended due to larger error accumulation compared to errors from numerical integration.