8-Lump kinetic model for fluid catalytic cracking with olefin detailed distribution study

• Published in: Fuel (Guildford) Vol. 225; pp. 322 - 335
• Main Authors: Sani, A. Golrokh, Ebrahim, H. Ale, Azarhoosh, M.J.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.08.2018; Elsevier BV
• Subjects: Algorithms; Alkenes; Catalytic activity; Catalytic cracking; Classification; Deactivation; Failure detection; Fluid catalytic cracking; Gas oil; Gasoline; Genetic algorithms; Liquefied petroleum gas; Lumped kinetics model; Mathematical models; Olefins; Parameter estimation; Particle swarm optimization; Petroleum; Polyolefins; Propylene; Reactors; Riser reactor; Sorting algorithms; Vacuum; Vacuum technology; VGO catalytic cracking; VGO catalytic cracking; Riser reactor; Lumped kinetics model; Olefins
• ISSN: 0016-2361; 1873-7153
• DOI: 10.1016/j.fuel.2018.03.087

•8-lump kinetics was proposed for fluid catalytic cracking (FCC) of vacuum gas oil (VGO).•Olefin distribution of FCC products was described by the reaction network.•Non-dominated sorting genetic algorithm II (NSGA-II) and Chaotic particle swarm optimization (C-PSO) were employed.•Kinetic parameters were estimated using the superior method.•Effect of FCC severity on the concentration profiles of products was investigated. Modeling fluid catalytic cracking (FCC) riser reactors is of significant importance in FCC unit control, optimization and failure detection, as well as development and design of new riser reactors. In this study, kinetic behavior of vacuum gas oil (VGO) catalytic cracking is studied by developing an 8-lump kinetic model to describe the product distribution. The feedstock and products are divided into eight lumps by reasonably simplifying reaction network, including VGO feed, diesel oil and gasoline, LPG, butylenes, propylene, ethylene, light gases, and coke. A time-on-stream non-selective catalytic activity equation is also assumed to model deactivation mechanism. Twenty-seven pairs of model kinetic parameters are estimated using two different optimization methods, namely: non-dominated sorting genetic algorithm II (NSGA-II), and chaotic particle swarm optimization (C-PSO) algorithm. Performances of both optimization methods are compared and C-PSO algorithm is selected as the superior method in terms of computation time and finding the global optimum. In the current research, based on validated estimated parameters of the preferred C-PSO method, the effects of some operating parameters on product yields distribution are investigated and discussed. This model can be used to predict the riser key products and their compositions with high degree of accuracy which may be especially useful for the conventional FCC processes with olefins production streams.