Estudo do processo de cristalização de açúcar : modelagem e estratégias de controle.

• Autor(a) principal: Torres, Rodrigo Coutinho
• Data de Publicação: 2007
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFSCAR
• Texto Completo: https://repositorio.ufscar.br/handle/ufscar/3985
• topic: Cristalização; Automação; Simulação dinâmica; Controle de processo; Fuzzy logic; ENGENHARIAS::ENGENHARIA QUIMICA

Sugar cane can be considered one of the main crop in the state of São Paulo, both for its contribution to the gross profits of the entire state harvest and the extent of the agricultural land it occupies. Recent figures show that the Brazilian sugar and alcohol industry park has 304 active industries, in addition to 40 new ones in implementation phase. The enlargement of that sector and its growing interest in new technologies represent a favorable context in developing and studying processes related with sugar and alcohol production among universities. This study focuses on the main stage in the sugar production process, the crystallization stage, in an attempt to better understand the deficiencies associated to the equipment where such stage is processed, known as crystallizer (or batch vacuum pan). In addition, more effective control strategies are also sought. A dynamic model of an industrial batch crystallizer was designed using the mass balance classical approach. The open-loop dynamic model simulation results were compared to real data taken from the supervisory system and laboratorial analysis, both information got in an industry located in the middle west of São Paulo state, the Usina Barra Grande. The model comprised a non-linear system of differential-algebraic equations. The mass balance that was adopted, in spite using severe hypotheses, generated good predictions for the equipment s state variables. The analysis of the data from the mentioned industry has shown that the operational stability of the equipment is affected by the crystallizer s automatic control routines. Besides the model, a control software was also build, aiming initially closedloop dynamic model simulation and the firsts adjusts in the new control strategies. After that, the control software had its behavior assessed in real operation time, replacing the more problematic controls in the equipment s programmable logical controller. The control software was encoded in VISUAL BASIC and comprised two control modes: a proportional-integral (PI) mode and a fuzzy mode. The crystallization stage was followed up by means of a graphical interface, where the main process informations were visually available. The informations were stored during the crystallization stage and, in the end, were saved in a text file, to guarantee future analysis. The fuzzy algorithm was created on MATLAB and it was compiled in a dynamic linked library format, which allowed its access by the control software. The tests with the control software were developed in two periods of time. The first period tests were made with the proportional-integral control mode and the second period tests were made with the fuzzy control mode. Both control modes generated extremely satisfying results. There were savings in the batch times, a reduction in the oscillation of the manipulated variables and a decrease in the error between the main controlled variable and its setpoint, when the tests results were compared with data from the actual control.