Control of continuous emulsion polymerization

The phenomenon of continuous oscillation has been attributed to the intermittent particle generation coupled with a slow washout, as described previously. The application of advanced control techniques to stabilize these oscillations has met with limited success [17,41,42] and it has been shown that the use of a plug flow reactor (PFR) as the first reactor will stabilize the system. All particle nucleation takes place in the PFR. Subsequent growth of the particles takes place in the downstream CSTRs. The segregation of particle nucleation and growth prevents the onset of oscillation. 

The multivariable adaptive control algorithm was applied in simulation for the PFR/CSTR reactor system described above. On-line measurement of monomer conversion (via density) and particle size (via light scattering) were assumed. White noise was added to the model inputs and outputs simulating actuator and sensor errors, respectively. 

The ability of the controller to handle process disturbances was examined by simulating abrupt variations in feed quality. This was simulated by increasing the concentration of inhibitor in the monomer from zero to 10 ppm. The variation in feed quality is a common problem in industrial practice and results from the deliberate addition of inhibitor to monomer to prevent premature polymerization. If monomer purification to remove inhibitor is not done (and it often is not in commercial operations), the switching of monomer feed tanks can produce undesirable and unexpected effects on the process outputs. In the open loop, the increase in inhibitor concentration from zero to 10 ppm causes a drop in the monomer conversion from 0.247 to 0.169. The particle size output also experiences a decline from 0.762 to 0.737. The reduction in polymerization rate is a direct result of the decreased initiator flux into the polymer particles, and the drop in particle size reflects the diminished 

---

A Simulation Study on the Use of a Dead-Time Compensation Algorithm for Closed-Loop Conversion Control of Continuous Emulsion Polymerization Reactors... 

The objective of this paper is to illustrate, by simulation of the vinyl acetate system, the utility of the analytical predictor algorithm for dead-time compensation to regulatory control of continuous emulsion polymerization in a series of CSTR s utilizing initiator flow rate as the manipulated variable. 

R. F. Dickinson, Dynamic behavior and minimum norm control of continuous emulsion polymerization reactors, PhD thesis, Univ. Waterloo, Ontario, 1976... 

Leffew, K. W. and Deshpande, P. B. (1981) A simulation study on the use of a dead-time compensation algorithm for closed-loop conversion control of continuous emulsion polymerization reactors, in Emulsion Polymers and Emulsion Polymerization (eds D.R. Basset and A.E. Hamielec), ACS, Washington, pp. 533-66. 

Several control techniques have been developed to compensate for large dead-times in processes and have recently been reviewed by Gopalratnam, et al. (4). Among the most effective of these techniques and the one which appears to be most readily applicable to continuous emulsion polymerization is the analytical predictor method of dead-time compensation (DTC) originally proposed by Moore ( 5). The analytical predictor has been demonstrated by Doss and Moore (6) for a stirred tank heating system and by Meyer, et al. (7) for distillation column control in the only experimental applications presently in the literature. Implementation of the analytical predictor method to monomer conversion control in a train of continuous emulsion polymerization reactors is the subject of this paper. 

The utility of the analytical predictor method of dead-time compensation to control of conversion in a train of continuous emulsion polymerizers has been demonstrated by simulation of the vinyl acetate system. The simulated results clearly show the extreme difficulty of controlling the conversion in systems which are operated at Msoap-starvedM conditions. The analytical predictor was shown, however, to provide significantly improved control of conversion, in presence of either setpoint or load changes, as compared to standard feedback systems in operating regions that promote continuous particle formation. These simulations suggest the analytical predictor technique to be the preferred method of control when it is desired that only one variable (preferably initiator feed rate) be manipulated. 

One unique but normally undesirable feature of continuous emulsion polymerization carried out in a stirred tank reactor is reactor dynamics. For example, sustained oscillations (limit cycles) in the number of latex particles per unit volume of water, monomer conversion, and concentration of free surfactant have been observed in continuous emulsion polymerization systems operated at isothermal conditions [52-55], as illustrated in Figure 7.4a. Particle nucleation phenomena and gel effect are primarily responsible for the observed reactor instabilities. Several mathematical models that quantitatively predict the reaction kinetics (including the reactor dynamics) involved in continuous emulsion polymerization can be found in references 56-58. Tauer and Muller [59] developed a kinetic model for the emulsion polymerization of vinyl chloride in a continuous stirred tank reactor. The results show that the sustained oscillations depend on the rates of particle growth and coalescence. Furthermore, multiple steady states have been experienced in continuous emulsion polymerization carried out in a stirred tank reactor, and this phenomenon is attributed to the gel effect [60,61]. All these factors inevitably result in severe problems of process control and product quality. 

The use of a precision digital density meter as supplied by Mettler Instruments (Anton Paar, Ag.) appeared attractive. Few references on using density measurements to follow polymerization or other reactions appear in the literature. Poehlein and Dougherty (2) mentioned, without elaboration, the occasional use of y-ray density meters to measure conversion for control purposes in continuous emulsion polymerization. Braun and Disselhoff (3) utilized an instrument by Anton Paar, Ag. but only in a very limited fashion. More recently Rentsch and Schultz(4) also utilized an instrument by Anton Paar, Ag. for the continuous density measurement of the cationic polymerization of 1,3,6,9-tetraoxacycloundecane. Ray(5) has used a newer model Paar digital density meter to monitor emulsion polymerization in a continuous stirred tank reactor train. Trathnigg(6, 7) quite recently considered the solution polymerization of styrene in tetrahydrofuran and discusses the effect of mixing on the reliability of the conversion data calculated. Two other references by Russian authors(8,9) are known citing kinetic measurements by the density method but their procedures do not fulfill the above stated requirements. 

The most common continuous emulsion polymerization systems require isothermal reaction conditions and provide for conversion control through manipulation of initiator feed rates. Typically, as shown in Figure 1, flow rates of monomer, water, and emulsifier solutions into the first reactor of the series are controlled at levels prescribed by the particular recipe being made and reaction temperature is controlled by changing the temperature of the coolant in the reactor jacket. Manipulation of the initiator feed rate to the reactor is then used to control reaction rate and, subsequently, exit conversion. An aspect of this control strategy which has not been considered in the literature is the complication presented by the apparent dead-time which exists between the point of addition of initiator and the point where conversion is measured. In many systems this dead-time is of the order of several hours, presenting a problem which conventional control systems are incapable of solving. This apparent dead-time often encountered in initiation of polymerization. 

MacGregor and Tidwell (1979) illustrate some of the steps involved in running plant experimentation, building these process and disturbance models, and implementing simple optimal controllers on some continuous condensation polymerization processes. A number of similar applications to continuous emulsion polymerization processes have also been made. 

Temeng, K. O. and Schork, F. J. (1989) Closed-loop control of a seeded continuous emulsion polymerization reactor system, Chem. Eng. Commun. 85,193-19. 

Althongh continuous emulsion polymerization technology is more than 60 years old it is still attracting researchers investigating the dynamic behavior in order to achieve control (164) and also developing procedures that allow transfer of products from batch or semibatch processes to continuous reactors in order to increase prodnctivity (165). 

The batch emulsion polymerization is commonly used in the laboratory to study the reaction mechanisms, to develop new latex products and to obtain kinetic data for the process development and the reactor scale-up. Most of the commercial latex products are manufactured by semibatch or continuous reaction systems due to the very exothermic nature of the free radical polymerization and the rather limited heat transfer capacity in large-scale reactors. One major difference among the above reported polymerization processes is the residence time distribution of the growing particles within the reactor. The broadness of the residence time distribution in decreasing order is continuous>semibatch>batch. As a consequence, the broadness of the resultant particle size distribution in decreasing order is continuous>semibatch>batch, and the rate of polymerization generally follows the trend batch>semibatch>continuous. Furthermore, the versatile semibatch and continuous emulsion polymerization processes offer the operational flexibility to produce latex products with controlled polymer composition and particle morphology. This may have an important influence on the application properties of latex products [270]. 

Until recently all data other than temperature and pressure on emulsion polymerization systems were obtained almost exclusively through manual, off-line analysis of monomer conversion, emulsifier concentration, particle size, molecular weight, etc. Recent interest in closed-loop control of both batch and continuous emulsion polymerizations has resulted in the development of a number of sensors capable of continuously monitoring the states of an emulsion polymerization system. These sensors provide a continuous (or frequently-updated discrete) electronic signal which is compatible with state-of-the-art digital data acquisition and control systems. 

Thus it is rapidly becoming possible to continuously monitor the critical states of an emulsion polymerization. The challenge, then, for the polymerization reaction engineer is to make full use of this data in designing reactor trains and open and closed-loop control policies to tailor polymer properties and end-use needs. 

Semicontinuous and Continuous Emulsion Polymerization In senticontinuous reactors, monomers, surfactant, initiator, and water are continuously fed into the reactor. Monomer droplets form if the rate at which the monomer is fed into the reactor exceeds the polymerization rate. This is not a desirable situation because the presence of free monomer in the system lowers the capability for controlling the polymer characteristics [8]. 

---