Computers and Chemical Engineering 35 (2011) 2886–2895 Contents lists available at ScienceDirect Computers and Chemical Engineering j ourna l ho me pag e: w ww.elsevier.com/locate/compchemeng Multicriteria dynamic optimization of an emulsion copolymerization reactor B. Benyahia ∗ , M.A. Latifi, C. Fonteix, F. Pla Laboratoire Réactions et Génie des Procédés, CNRS-ENSIC,1 rue Grandville, BP 20451, 54001 Nancy Cedex, France a r t i c l e i n f o Article history: Received 24 December 2010 Received in revised form 29 March 2011 Accepted 25 May 2011 Available online 24 June 2011 Keywords: Emulsion copolymerization Chain-transfer agent Core–shell morphology Multiobjective optimization Decision aid method Experimental implementation a b s t r a c t A multicriteria optimization approach based on an evolutionary algorithm has been developed to deter- mine the optimal control policy for a fed-batch emulsion copolymerization reactor, particularly for styrene and butyl acrylate in the presence of n-C12 mercaptan as chain transfer agent. The process model was elaborated and validated experimentally in order to predict the global monomer conversion, the number and weight average molecular weights, the particle size distribution and the residual monomers mass fraction. The process objectives were to produce core–shell particles (hard core and smooth shell) with specific end-use properties and high productivity. This has been achieved by the maximization of the monomers overall conversion at the end of the process and the minimization of the error between the glass transition temperature and a designed profile subject to a set of operational constraints. The nondominated Pareto solutions obtained were ranked according to a decision making aid method based on a decision maker preferences and experience using multi-attribute utility theory. Finally, the best solution was implemented experimentally. © 2011 Elsevier Ltd. All rights reserved. 1. Introduction Multiobjective optimization problems are encountered in most real-world applications and more recently in chemical processes (Fonteix, Massebeuf, Pla, & Nandor Kiss 2004; Garg & Gupta, 1999; Mitra, Majundar, & Raha, 2004; Sakar, Rohani, & Jutan, 2007). Since such problems involve several objective functions with conflicting nature, the final optimum is not unique but a set of nondomi- nated solutions (Pareto’s front) which show a trade-off between the whole objectives. Genetic algorithms (GAs) are well adapted tools to solve multiobjective problems. This kind of technique stands for a class of stochastic optimization methods that simulate the process of natural evolution (mainly genetic algorithms, evolu- tionary programming, and evolution strategies). These algorithms have proven themselves as a general, robust and powerful search mechanism. Moreover, evolutionary algorithms (EAs) seem to be especially suited for multiobjective optimization because they are able to find multiple Pareto-optimal solutions in a single simulation run. Emulsion polymerization is an important industrial process used to produce a great variety of polymers for multiple uses (e.g. paints, adhesives, coatings and varnishes). Moreover, it has significant advantages over bulk and solution polymerization pro- ∗ Corresponding author. Current address: Process Systems Engineering Labora- tory, Department of Chemical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge, MA 02139, USA. E-mail address: benyahia@mit.edu (B. Benyahia). cesses. These advantages result mostly from its multiphase and compartmentalized nature which allows producing, with high polymerization rates, macromolecules of high molecular weights, delivering a high versatility to product qualities. However, the com- plexity of emulsion polymerization systems arising from factors such as their multiphase nature, nonlinear behavior and sensitivity to disturbances, induces more difficulties on modeling and makes the development of optimization procedures of emulsion polymer- ization reactions a very challenging task. The production of polymers with specified end-use properties is one of the key issues in polymer industry. These proper- ties are strongly governed by both chemical and morphological characteristics, mainly: molecular weight distribution (MWD), polymer microstructure, glass transition temperature (T g ), parti- cles size distribution (PSD) and particles morphology. For example, PSD is strongly correlated to the rheological, adhesive and film- forming properties of the final products. On the other hand, MWD affects important end-use properties of the film, such as elasticity, strength, toughness, and resistance to solvents. In radical polymerizations, molecular weights are commonly controlled using chain transfer agents (CTAs). On the other hand, the use of multistage emulsion polymerization to produce parti- cles containing multiple polymer phases is widespread throughout coatings, impact plastics and adhesives industries. Such compos- ite particles often improve various end-use properties compared to related single-phase lattices or latex blends. Moreover, the morphology of the resulting particles is generally controlled by a combination of thermodynamic and kinetic factors (Dimonie, Daniels, Schaffer, & El-Aasser, 1997; Sundberg & Durant, 2003). 0098-1354/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.compchemeng.2011.05.014