Skip to Main content Skip to Navigation


Abstract : Advanced control is becoming imperative for various processes, and polymerization processes are absolutely no exception. The major objectives of on-line monitoring and control of polymerization processes are maintaining the process safety, manufacturing polymers with well-defined properties, and reducing costs.

Polymers are found in a large variety of products. It is a well accepted fact that a significant quantity of polymers are manufactured in emulsion polymerization, and that the economic importance of these products is beyond question. The reaction medium in emulsion polymerization involves the dispersion of one or more monomers (which are usually partially or totally hydrophobic) in a continuous aqueous phase* with a water soluble initiator and an emulsifier. The reaction therefore takes place under heterogeneous conditions, and the final product, called a latex, consists of suspended polymer particles in water, stabilized by the emulsifier.

Industrial emulsion polymerization processes can be classified into two groups according to the final use of the latex. In some emulsion polymerization processes the produced latex is an intermediate product, and it is coagulated after the fact to give the final bulk polymer (e.g. polyvinyl chloride, styrene butadiene rubber). In other processes, the produced polymer is used directly (or perhaps after some intermediate formulation steps) in the form of a latex. Polymers produced in this way are used as water-based paints (acrylics lattices), adhesives (polyvinyl acetate latex), and finishes for textiles, paper or leather (acrylic polymers).

* Of course not all emulsions use water as the continuous phase. Water-in-oil, or inverse processes also exist. However, we will limit our discussion to the more common oil-in-water processes.

The final latex properties depend on the polymer molecular weight distribution, particle size distribution, morphology, glass transition temperature, and composition (when several monomers are used). These properties are strongly determined by the amount and type of additives, initiator, emulsifier and chain transfer agent, the monomer addition policy, the reaction temperature and the reactor configuration. Hence, emulsion polymerization processes are characterized by a large number of manipulated variables, which means that, in order to describe the relationships between them, we would need highly complex process model.

In order to ensure the economical production of a polymer with the desired properties under safe conditions, important decisions must usually be made during plant operation, by means of automatic on-line control. In the context of control, the process is a dynamic system (described by a set of differential equations) with well-known inputs (that can usually be used as manipulated variables), process 'states' that are the variables representing the evolution of the process, and finally measured process outputs (which are usually a combination of the process states). Control of the process states is based on a suitable process model. It also requires some on-line state measurements. In spite of the importance of emulsion polymerization, process control is still a particularly difficult task. The difficulties in controlling emulsion polymerization are often associated with the lack of on-line measurements of several properties, the high reaction rate, the sensitivity of the reaction to small amounts of additives, and the highly complex reaction networks that result in a complex nonlinear model combined with a number of unknown variables.

One of the main control objectives in an emulsion polymerization process is to track polymer composition. This is of absolute importance in terms of final properties of the produced polymer-e.g. glass transition temperature, particle morphology, mechanical and chemical resistances. Controlling polymer composition is indispensable if several monomers with different reactivity ratios are involved in the reaction at a nonazeotropic monomer composition, which is the case of most industrial systems. If no control action is taken in this case, the polymer composition will vary during the reaction, and this will lead to a heterogeneity in the polymer properties (unless the polymer is made in a CSTR). Polymer composition can however be controlled by employing appropriate monomer feed flow rates. The most efficient monomer addition policy, in terms of composition control and simultaneous minimization of process time consists of applying a variable monomer feed flow rate of the more reactive monomer(s), that should be a function of the residual amount of monomer remaining in the reactor and their reactivity ratios. Therefore, controlling polymer composition requires on-line measurement of the polymer composition and the instantaneous quantity of residual monomers, an adequate mathematical model correlating the monomer feed flow rates with the produced polymer composition (inputs-outputs), and finally the use of a robust control technique.

On-line sensors developed for the direct monitoring of emulsion polymerization can be classified into two broad categories: sensors that require a sampling device or a circulating loop to perform the analysis (e.g. gas chromatography and densimetry), and in situ sensors (e.g. Spectroscopic and ultrasonic probes). The copolymer composition can be monitored by means of on-line gas chromatography (based on the analysis of the residual monomer), or by densimetry (if combined with an additional sensor). However, the measurements in these sensors present some difficulties related to the latex sampling and treatment, and sometimes, time delays caused by long analysis times. This is not to say that we cannot use these sensors, by rather we must be aware of their limitations. In the second category, the analysis is performed in the reactor, and therefore problems related to time delays and to plugging of the sampling loop are minimized. However, some sensors such as the infrared spectroscopy, while very promising, are still in the stage of development, and we are still faced with the problem of coagulation on the sensor. When these techniques become available, they will provide valuable information on the state of the reactor. Nevertheless, given the complicated nature of emulsion systems, we will still need to apply advanced control strategies such as those outlined in what follows.

Due to the difficulty of performing on-line measurements of all the states of the process, combined with a lack of accurate phenomenological models, estimation techniques have been developed to infer estimates of certain variables that are not available on-line, from auxiliary measurements. State observers, or software sensors, are designed based on the process model and on the process outputs. If the observability conditions are satisfied, the observer reconstructs some of the unmeasurable states of the model from the available measurements. Good state observers can overcome modeling uncertainties and measurement noise, and are therefore useful for process monitoring, control, fault detection, and are also used as filters of random effects associated to the measurements.

Despite the highly nonlinear behavior of the polymerization process, mainly due to the reaction rate, linear estimation (e.g. Kalman filter) techniques have usually been used to infer information about the evolution of several unmeasurable states of the process. Similarly, linear control techniques (PID, adaptive) have usually been applied to track several variables of the linearized nonlinear system. This is due to the difficulty of dealing with nonlinear systems and to the fact that nonlinear estimators are usually restricted to a class of nonlinear systems. However, linear estimation and control techniques are inadequate for highly nonlinear processes. Nowadays, recent developments in nonlinear theory allow us to implement nonlinear estimation (e.g. high gain observer) and control (e.g. geometric control) techniques to several classes of systems without extensive calculations.

The main objective of this research is to apply such nonlinear techniques to estimate and control the polymer composition, and to maximize the process productivity under safe reactor operation. The fundamental on-line information about the evolution of the process is obtained by calorimetry. Calorimetry is often useful in vinyl polymerization reactions, since they tend to be very exothermic. On an industrial level, the main use of calorimetry is currently reactor safety (by reactor temperature control), and the estimation of the rate of heat produced by the polymerization, which is proportional to the overall reaction rate.

In Chapter 2 we will review the theoretical nonlinear estimation and control techniques that will be used throughout this research. Calorimetry will be discussed in Chapter 3. The observer development to estimate the polymer composition in co- and terpolymerization processes is presented in Chapters 4 and 5. In Chapter 6, we develop a control strategy for polymer composition for both co- and terpolymerization processes. Finally, in Chapter 7, we treat the topic "Maximizing productivity under safe conditions", where the polymer composition is controlled and the reaction rate is simultaneously maintained at a predefined maximum limit.
Complete list of metadata
Contributor : Nida Sheibat-Othman Connect in order to contact the contributor
Submitted on : Monday, March 9, 2009 - 2:42:40 PM
Last modification on : Thursday, August 4, 2022 - 5:13:55 PM
Long-term archiving on: : Saturday, November 26, 2016 - 6:16:18 AM


  • HAL Id : tel-00366694, version 1



Nida Sheibat-Othman. ADVANCED STRATEGIES FOR COMPOSITION CONTROL IN SEMI-CONTINUOUS EMULSION POLYMERIZATION. Chemical and Process Engineering. Université Claude Bernard - Lyon I, 2000. English. ⟨tel-00366694⟩



Record views


Files downloads