Análise e modelagem do processo industrial de pré-branqueamento com oxigênio da polpa celulósica produzida pelo processo kraft

Abstract

This work presents a methodology for determining the most appropriate expressions among a set of possible empirical kinetic equations to describe the rate of delignification and presents hybrid models to concentrated parameters for quantification and prediction of Kappa number at the exit of the reactor of the pre-bleaching stage of a Kraft pulp mill. Since the Kappa number is proportionally related to the amount of lignin present in the cellulosic pulp, the proposed models are based on the differential dynamic mass balance of the Kappa number and the empirical kinetic equation of the Kappa number consumption that was proposed as a function of variables of processes. These variables are: cellulose pulp flow, oxygen flow, reactor pressure, 𝑁𝑎𝑂𝐻 solution flow, oxidized white liquor flow, consistency, Kappa input number, Kappa number of output and temperature, totalling 9 process variables. The equation that relates these variables contains parameters to be estimated by minimizing the deviation function through a routine optimization. We used 46080 data (from each process variable) from the industry to estimate the parameters, with the exception of the Kappa number that were performed the total of 1295 measurements. With 9 process variables available it is possible to propose 512 different models. In this work, there is a proposal of a methodology to reduce the number of analyzed models from 512 to 46. The determination of the most suitable model among the 46 models analyzed was based on the value of the coefficient of determination of validation presented by the models. The most appropriate model presented 5 variables, 6 parameters and a regression coefficient of 0.8765. To validate the model, 4021 data of each process variable were used, with the exception of the Kappa number that 115 measurements were performed. The coefficient of determination of the validation was equal to 0.8610. The results show that the model is adequate to predict the Kappa number at the exit of the delignification reactor.