Advanced search
Start date

Control and real-time optimization of a three-phase reactor

Santos, Marcela Geni Mota dos
Total Authors: 1
Document type: Doctoral Thesis
Press: Campinas. [2006]. 158 f., gráficos, ilustrações, tabelas.
Institution: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Química
Defense date:
Examining board members:
Maciel Filho, Rubens; Nascimento, Claudio Augusto Oller do; Pasotto, Marlei Barbosa; Boss, Edinara Adelaide; Melo, Delba Nisi Cosme
Advisor: Maciel Filho, Rubens; Toledo, Eduardo Coselli Vasco de
Field of knowledge: Engineering - Chemical Engineering
Indexed in: Base Acervus-UNICAMP; Biblioteca Digital da UNICAMP
Location: Universidade Estadual de Campinas. Biblioteca Central Cesar Lattes; T/UNICAMP; Sa59c

The principal objective of this work was the evaluation of classic and hierarchical control strategies in real time. It was used as case study the o-cresol hydrogenation in a three-phase slurry reactor with solid catalyst to validate the real time optimization approaches, due to the inherent importance of this process, since these reactors are present in many processes of the chemical and petrochemicals industries, with growing preference in comparison with other types of more conventional reactors. It was investigated the performance of some versions of the advanced control algorithm DMC (Dynamic Matrix Control) in a quadratic (QDMC - Quadratic Dynamic Matrix Control) and adaptive (STQDMC - Self Tuning Quadratic Dynamic Matrix Control) form under different control strategies (feedback, feedforward and mixed). The process was represented through deterministic and statistical models. An alternative methodology was proposed for the definition of the optimal model parameters for the valid operational range based on principles of response surface analysis. It was evaluated two-Iayers structure optimization, using optimization algorithms based on Levenberg-Marquardt and SQP (Sequential Quadratic Programming) procedures. Traditional two-layers optimization approaches were proposed. In this approach the optimizer generates the set-points for the controller. Besides, an alternative approach was proposed where the optimizer, besides generating the set-points for the controller, also sends the optimal values for the manipulated variables, becoming this procedure, in most of the cases, faster and efficient. In relation to real time optimization, the mathematical bases are presented for the one-Iayer optimization approach implementation. Here the control and optimization problems are simultaneously resolved. According to this work, the two-Iayers optimization approach presents suitable results. Besides, it should always be considered the control and optimization objectives, in order to weight the feedback or feedforward control strategies and even a mixed one. (AU)