Kinetic model reduction for control of phenol-formaldehyde reactive systems

.This study aims at improving the performance of a model-based control scheme for chemical processes characterized by very complex kinetics. The highly exothermic phenol-formaldehyde reactive process for the production of phenolic resins is accurately modeled by taking into account a large number of reactions and compounds. Then, two simplified reaction networks, involving only a limited number of reactions and species, are proposed. The accuracy of the simplified models in terms of concentrations and heat release estimates, is assessed by comparing their behavior in a simulation environment. The analysis of the results suggests that the best match, both in terms of concentration estimation and heat estimation accuracy, is obtained with second-order kinetics and involves 4 reactions. The best performing reduced-order model devised in Section 3 has been adopted for the design of a control scheme for a jacketed batch reactor in which the phenol-formaldehyde reaction takes place. A reliable and accurate simulation model of the reactor has been built in a Matlab-Simulink environment, involving the mass balances of the 13 species of the high-order complete model and the energy balances written for the reactor and the cooling jacket, under the assumption of perfect mixing. The tracking error remains very low (below 0.05 [K]) over the entire reaction time and this confirms that the reduced-order model can be successfully used for control purposes.