The Comparison of Artificial Intelligence and Traditional Approaches In FCCU Modeling

Mithat Zeydan


FCCU (Fluid Catalytic Cracking Unit) is a part of oil refinery production process whereby valuable products such as gasoline, LPG (Liquid Petroluem Gas), diesel are manufactured in a short period of time. The objective of this paper is to find the most robust model by comparing the models of FCCU that are developed using different methodologies. The models of FCCU are developed by using Artificial Neural Network (ANN), Fuzzy Logic, Neuro-Fuzzy, and traditional methodology. In this paper, the criteria used for measuring the performance of different models is root mean squared error (RMSE). The models are applied to the real data obtained from TUPRAS (Turkish Petroleum Refineries Corporation)-FCCU. Kurihara (1967) model is used as the traditional model for comparing with intelligence modeling techniques. Finally, the Fuzzy Neural Network (FNN) model was found as the model with the minimum RMSE. Qwicknet 2.23, MATLAB 6.5, and Neuro-solutions 4.1 softwares have been used for the construction of ANN, fuzzy, and neuro-fuzzy models, respectively.


Fluid catalytic cracking, fuzzy modeling, fuzzy neural network, artificial neural network

Full Text:


Online ISSN 1943-670X

The International Journal of Industrial Engineering is a Non-Profit Organization under Section 501(c)(3).

The journal is registered with the Library of Congress and the Copyright Clearance Center. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopy, recording or otherwise, without the prior written permission of the journal.

The International Journal of Industrial Engineering is abstracted by Thomson Reuters Science Citation Index, Engineering Index/Compendex, EIPAGE1, and CURRENT CONTENTS/ENGINEERING, COMPUTING AND TECHNOLOGY.