Hale Hapoglu

Generalized predictive control of optimal temperature profiles in a polystyrene polymerization reactor

The performance of generalized predictive control (GPC) was examined when it was applied to the temperature of a free radical solution polymerization of styrene in a jacketed batch reactor. The optimal temperature policies were obtained at different initiator concentrations by applying the optimal control theory to the relevant polymerization reactor. The use of polynomial ARIMAX model related with reactor temperature and heat input was emphasised. Model parameters were determined by using Kalman, Bierman and Levenberg Marquardt algorithms. A pseudo random binary sequence (PRBS) signal was employed to operate the system. The GPC control method was based on ARIMAX model. The performance of the GPC method was compared with that of the PID controller.

Nonlinear control of polymerization reactor

Abstract In this work, nonlinear model based control was applied to the free radical solution polymerization of styrene in a jacketted batch reactor and its performance was examined to reach the required monomer conversion and molecular weight. Optimal temperature profiles for the properties of polymer quality were evaluated using the Hamiltonian optimization method. Total simulation program having mass and energy balances of the jacketed polymerization reactor was used to calculate the optimal trajectories. For control purposes, several experimental and theoretical dynamic studies have been made to observe the validity of simulation program. Experimental and theoretical nonlinear model based control have been investigated to track the temperature at the optimal trajectory Two types of parametric and nonparametric models were evaluated to achieve the temperature control. For this purpose, reaction curve was obtained to calculate the system dynamic matrix as a nonparametric model. In all control work, heat input to the reactor was chosen as a manipulated variable. Nonlinear auto regressive moving average exogenous (NARMAX) giving a relation between heat input and reactor temperature was chosen to represent the system dynamic and this model was used to describe the related control system as a parametric model. NARMAX model parameters were determined by using Levenberg Marquard algorithm. A pseudo random binary sequence (P.R.B.S.) signal was employed to disturb the system. Total simulation program was used to calculate the system and control parameters. Several types and orders were used to construct the NARMAX models. The efficiency and the performance of the nonlinear model based control with the NARMAX model and dynamic matrix were tested to calculate the best model. Nonlinear model based control system was used to control the reactor temperature at desired temperature trajectory experimentally and theoretically. Theoretical simulation results were compared with experimental control data. It was concluded that the control simulation program represents the behavior of the controlled reactor temperature well. In addition, nonlinear model based control keeps the reactor temperature of optimal trajectory satisfactorily.

Control of a polymerization reactor by fuzzy control method with genetic algorithm

The optimal temperature control of a batch jacketed free radical polymerization reactor by fuzzy control method with genetic algorithm (GA) is considered. The manipulated variable was chosen as the heat given by the immersed heater. A key issue in this study is to generate sets of fuzzy control membership function and relation matrix using GA, which can be easily implemented and an efficient method for optimization problems. The fitness function for GA is chosen as the integral of the absolute value of error (IAE). By using the fuzzy parameters obtained for three different optimal temperature profiles, the efficiency of the fuzzy controller with GA was examined by simulation and experimentally. It was seen that GA is able to tune the fuzzy controller efficiently for different situations and therefore to control the temperature of the polymerization reactor.

Adaptive heuristic temperature control of a batch polymerisation reactor

The purpose of this research is to improve and apply the temperature control of a free radical solution polymerisation of styrene and to examine its performance on the basis of adaptive heuristic criticism (AHC) control. This algorithm consists of a three-layer feed forward artificial neural network (ANN) which uses supervised learning with reinforcement in a unique topology. This algorithm has two neurone-like adaptive elements and a difficult learning control problem which can be solved by means of a learning system with a single associative search element (ASE) and a single adaptive critic element (ACE). AHC uses a type of control system whose output value is either maximum or minimum. The controller will take in process data on-line and update the weights to proper ones in the control of the process. The performance results of the AHC controller are compared with the results obtained by using conventional Deadbeat algorithm. AHC control system shows satisfactory behaviour to track the reactor temperature.

APPLICATION OF ADAPTIVE PID CONTROL WITH GENETIC ALGORITHM TO A POLYMERIZATION REACTOR

This article describes the application of adaptive PID control with genetic algorithm (GA) to a jacketed batch polymerization reactor. This method was used to keep the polymerization reactor temperature at the desired optimal path, which was determined by the Hamiltonian maximum principle method. The reactor was simulated and the model equations of this jacketed polymerization reactor were solved by means of Runge-Kutta-Felthberg methods. A genetic algorithm can be a good solution for finding the optimum PID parameters because unlike other techniques it does not impose many limitations and it is simple. In this research, suitability of these parameters was checked by the integral absolute error (IAE) criterion. The control parameters in the PID algorithm were changed with time during the control of a polymerization reactor. It was seen that the genetic algorithm was able to tune the PID controller used in this system in terms of higher robustness and reliability by changing the parameters continuously.

Optimal temperature control in a batch polymerization reactor using fuzzy-relational models-dynamics matrix control

In this work, fuzzy-relational models-dynamics matrix control (Fuzzy-DMC) was applied to the free radical solution polymerization of styrene in a jacketted batch reactor and its performance was examined to reach the required monomer conversion, molecular weight and chain length in a minimum time. The reactor temperature was controlled by manipulating the heat-input to the reactor. The performance of the Fuzzy-DMC controller was compared with that of the nonlinear generalized predictive control (NLGPC).

Generalized minimum variance control of optimal temperature profiles in a polystyrene polymerization reactor

In this paper, the dynamic, optimization and on-line temperature control of the free radical polymerization of styrene in a cooling jacketed batch polymerization reactor were investigated theoretically and experimentally. Time varying optimal temperature trajectories to obtain a product with desired quality in a minimum time are calculated using optimal control theories. The polynomial CARMA model for reactor temperature and heat input to the reactor was used to design the relevant control system. The CARMA model parameters were calculated by applying the Bierman algorithm. A Pseudo Random Binary Sequence (PRBS) signal was given to disturb the system. Generalized minimum variance (GMV) control based on the CARMA model was applied to the polymer reactor. Theoretical and experimental results were in agreement and it was shown that the GMV method controlled the system very well at optimal trajectory.

New Tuning Method for Generalized Predictive Control of the Production of S. cerevisiae

In this study, S. cerevisiae was produced in a batch bioreactor in aerobic conditions and the growth medium temperature was controlled at its optimal value. In order to control the growth medium temperature, the Generalized Predictive Control (GPC) method was used. The process was described using an Auto Regressive Integrated Moving Average eXogenous (ARIMAX) model. Model parameters were determined by applying Pseudo Random Binary Sequence (PRBS) signals to the process and using Biermans U-D Factorization Algorithm. By using statistical experimental design and Box-Wilson optimization methods, optimal values of the sampling time and weighting factor were determined. A two-level factorial experimental design technique was used to identify a statistical model. The predicted maximum, minimum and the base levels of sampling time and weighting factor were determined on the basis of the previous knowledge about the plant. It was proposed to determine the values of sampling time and weighting factor giving the best control performance. Integral Square Error (ISE) was selected as the optimization criterion. Growth medium temperature was controlled with very small levels of oscillation around the set point by using optimal controller parameters with GPC.

The steady-state and dynamic analysis of packed distillation column based on partial differential approach

The steady-state and dynamic behavior of a binary packed distillation column has been simulated using the two film back-mixing model. The model solution has been obtained employing orthogonal collocation on finite elements. The approach using the Legendre or Jacobi polynomial has been tested on the solution of related models. A pilot plant scaled packed distillation column distilling methanol-water mixture was used for experimental work. The variation of overhead and bottom temperatures have been monitored by an on-line computer control system. A number of comparisons have been made between the results predicted in this work from back mixing model based on the two film theory and those obtained experimentally. In most cases the prediction of this work gave results which were closer to experiment than other numerical solutions, showing that Legendre polynomials for orthogonal collocation on finite element can be applied effectively to simulate the packed distillation column by using a partial differential approach.

Generalized Minimum Variance Control of Growth Medium Temperature of Baker's Yeast Production

The control of optimum growth medium temperature of a batch bioreactor in which S.cerevisiae was grown under aerobic conditions has been studied. The generalized minimum variance (GMV) algorithm was applied for on-line computer control. A controlled autoregressive moving average model relating the bioreactor temperature and heat input was used to show the dynamic behaviour of the system. The heat input to the bioreactor was chosen as a manipulated variable. A pseudo-random binary sequence signal was applied to the system and the model parameters were determined using the Bierman algorithm. More suitable values of the GMV controller parameters were determined using a total simulation program. These control parameters were used in experimental and theoretical work under several disturbances.