Modelagem e desenvolvimento de estrategia de controle para o reator de sintese de dextrana

Abstract

This work is focused on modeling and development of a control strategy for a dextran reactor. Dextran is a biopolymer, with great potential of utilization in cosmetic industry and medicai area. For each specific purpose, it must have suitable molecular weight. Dextran can be obtained by enzymatic path, with the reaction occurring "in vivo" or "in vitro". The latter has been chosen because it is easier to control, if compared with "in vivo" technique. The dextran reactor is constituted by two stages, two vessels filled with a two-phase mixture. Enzymatic reaction -desorption of the enzyme carries out the first stage; meanwhile, the adsorption of the enzyme takes place in the second stage. Enzyme dextran-sucrase and non-consumed substrate-sucrose -are continuously recycled. From these considerations, it was developed a dynamic model for the process, whose main characteristics were considered to make the representation as real as possible. Mathematical models were based on mass balances involving enzyme and substrate distributions between solid and liquid phases; energy balances were not considered because the thermal effects are negligible. The results obtained from the simulation of mathematical model, using LSODAR algorithm for numerical integration, revealed the natural dynamic behavior-also called pseudo stationary behavior -of the system. The variables did not reach a stationary value, so it was necessary to develop controllers which were capable to maintain the variable of interest in a suitable levei. This variable was the substract concentration on the second stage, which is used to calculate the productivity and the yield of the process. Two types of control algorithms were performed and compared: a conventional feedback PID and the DMC-SISO, utilizing the feed flow on the first reactor as manipulated variable.The results showed that DMC algorithm presents a better robustness if compared to the PID, for the most part of results. It was concluded that the conventional DMC algorithm can be used successfully in control of processes with a natural dynamic behavior as long as the parameters are chosen carefully.