Multiobjective Optimization of an Industrial LPG Thermal Cracker using a First Principles Model

Abstract

Thermal cracking of hydrocarbons in the presence of steam is the most important process for the production of ethylene and propylene, which are the raw materials for many petrochemicals. In this work, multiobjective optimization (MOO) of an industrial LPG (liquefied petroleum gas containing mainly propane and butane) thermal cracker is studied using the elitist nondominated sorting genetic algorithm adapted with the jumping gene operator, NSGA-II-aJG. A first principles model based on free radical mechanism for LPG thermal cracking is employed for optimization. Several bi- and triobjective optimization problems are solved; these problems involved maximization of annual ethylene/propylene production, selectivity and run length, and minimization of severity and total heat duty per year. Feed flow rate, steam ratio, inlet temperature, coil outlet temperature, and pressure are the decision variables. MOO of the LPG thermal cracker provides a range of optimal operating conditions and objective values, with the triobjective optimization giving more optimal solutions compared to those from the biobjective optimization for the objectives considered. With this detailed knowledge, a suitable operating point can be selected based on specific requirements in the particular plant.