Min-gu Kang

SIMULATION AND OPTIMIZATION OF A STYRENE MONOMER REACTOR USING A NEURAL NETWORK HYBRID MODEL

Abstract The modeling and optimization of a dehydrogenation reactor in the industrial styrene monomer plant has been proposed in this study. Because this reactor consumes large amount of expensive high-pressure steam to produce the styrene monomer (usually more than two third of total energy costs), the minimization of the operating cost is highly desirable to maximize the profit. However, it is not easy to develop the accurate mathematical model because of the lack of internal or intermediate measurements of the industrial reactor, and also the lack of experimental results of the catalyst deactivation. To overcome these difficulties, we propose an alternative model of the styrene monomer reactor using a hybrid model in which the mathematical model is combined with neural networks. Major reaction mechanism is described in the mathematical model, and the deactivation effect is modelled in neural network using the real operation data. Using this model, the sensitivity analysis and the optimization of the industrial plant have been performed. The proposed optimal operation enlarges the profit of the plant very much.