Multiobjective Optimization of Industrial FCC Units Using Elitist Nondominated Sorting Genetic Algorithm

Abstract

This study provides insights into the optimal operation of the fluidized-bed catalytic cracking unit (FCCU). A five-lump model is used to characterize the feed and the products. The model is tuned using industrial data. The elitist nondominated sorting genetic algorithm (NSGA-II) is used to solve a three-objective function optimization problem. The objective functions used are maximization of the gasoline yield, minimization of the air flow rate, and minimization of the percent CO in the flue gas using a fixed feed (gas oil) flow rate. The decision variables and several important state variables corresponding to the optimal conditions of operation are obtained. The optimal solutions correspond to the unstable, saddle-kind, middle steady states. The procedure used is quite general and can be applied to other industrial FCCUs. The optimal results obtained here provide physical insights that can help one in obtaining and interpreting such solutions.