VOL. NO.
47 10
O C T O B E R
2
0
1
4
JOURNAL OF CHEMICAL ENGINEERING O F J A PA N
[ VOL. 47, NO. 10, OCTOBER 2014 ]
▶ Separation Engineering Supported Liquid Membrane Extraction of Reactive Dye Using Fabricated Polypropylene Membrane Norlisa Harruddin, Norasikin Othman, Ani Idris, Zing Yi Ooi and Masahiro Goto –––––––––––––––––––––––––––––––––––––––––– 761 Thickness Reduction of the Zeolitic Imidazolate Framework-8 Membrane by Controlling the Reaction Rate during the Membrane Preparation Nobuo Hara, Miki Yoshimune, Hideyuki Negishi, Kenji Haraya, Shigeki Hara and Takeo Yamaguchi ––––––––––––––––––––– 770 Regeneration of SO2-Loaded Sodium Phosphate Solution in Rotating Packed Bed Zhiguo Yuan, Wei Song, Youzhi Liu, Xiaofeng Kang, Bo Peng and Tao Wang ––––––––––––––––––––––––––––––––––––––––––––– 777
▶ Chemical Reaction Engineering Estimating the Chemical Reaction Kinetics of p-Xylene Oxidation Using Artificial Neural Network without Traditional Kinetic Equations Yaming Dong and Xuefeng Yan ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– 782
The Society of Chemical Engineers, Japan
Photocatalytic Decolorization of Methylene Blue in a Glass Channel Microreactor (SC) Bruno Ramos, Shinichi Ookawara, Yoshihisa Matsushita and Shiro Yoshikawa ––––––––––––––––––––––––––––––––––––––––––– 788
JCEJAQ 47(10) 761-792(2014) ISSN 0021-9592
This article appeared in the Journal of Chemical Engineering of Japan. The attached copy is provided to the author for non-commercial research, education use and sharing with colleagues. Other uses listed below are prohibited: - Reproduction, - Commercial use, - Posting to personal, institutional or third party websites.
Research Paper
Journal of Chemical Engineering of Japan, Vol. 47, No. 10, pp. 782–787, 2014
Estimating the Chemical Reaction Kinetics of p-Xylene Oxidation Using Artificial Neural Network without Traditional Kinetic Equations Yaming Dong and Xuefeng Yan Key Laboratory of Advanced Control and Optimization for Chemical Processes of Ministry of Education, East China University of Science and Technology, Shanghai 200237, P. R. China Keywords: Artificial Neural Network, Hybrid Model, p-Xylene Oxidation, Reaction Kinetic, Reaction Rate An alternative method for the estimation of the chemical reaction rate without the use of traditional kinetic model equations is investigated in this study. The proposed method based on artificial neural network (ANN) is an efficient and accurate means to predict the chemical reaction rate because of the complex and unknown reaction kinetics of the p-xylene (PX) oxidation process. A mechanism model of PX oxidation is also integrated in the proposed neural network reaction rate model to predict the concentrations of target materials in the chemical process. The results show that the neural network model can accurately predict the chemical reaction rate of PX oxidation and that the proposed process hybrid model exhibits better performance than the pure mechanism and pure ANN models.
Received on February 27, 2014; accepted on April 18, 2014 DOI: 10.1252/jcej.14we090 Correspondence concerning this article should be addressed to X. F. Yan (E-mail address:
[email protected]).
the currently adopted industrial operation conditions and proposed a fractional kinetic model to estimate the reaction rate. The agreement between model predictions and experimental data was shown to be satisfactory. Considering the concentrations of different peroxy-free radicals, Sun et al. (2008) proposed a simplified free-radical kinetic model based on free-radical chain reaction mechanisms. The abovementioned reaction kinetic models of PX oxidation to TA are based on mechanism knowledge, which is a result of the presence of materials whose reaction rates are described by simple kinetic expressions. Successful modeling of chemical reactors requires accurate knowledge of reaction kinetics. Although efforts have been exerted to develop kinetic models to estimate reaction rates, the reaction kinetics of PX oxidation remains poorly understood because of the complexity of this process. PX oxidation involves the heat and mass transfer of the gas–liquid phase, the catalytic reaction of free radicals in the liquid phase, the crystallization of solid (precipitation of product TA), etc. Thus, these proposed simple kinetic models represent only a part of the mechanism of PX oxidation. Furthermore, most published studies on PX oxidation kinetics employ kinetic equations to estimate reaction rates, where kinetic parameters are determined from data fitting of experimental values and model results. However, selecting the proper kinetic parameters for kinetic models is difficult with the rapidly changing operation conditions. Psichogios and Ungar (1992) proposed a hybrid modeling approach, in which a neural network was utilized to estimate the unknown biomass growth rate of a fed batch bioreactor; a partial first-principle model was introduced as the basis of the hybrid model and combined with the neural network to predict the output values of the fermentation process. Compared with simple empirical or mechanism models, this hybrid model can be extrapolated beyond the
782
Introduction Terephthalic acid (TA) is a commercially available important aromatic compound that is widely used to produce polyethylene terephthalate polymers. Common chemical processes for TA are produced through air or molecular oxygen oxidation of p-xylene (PX) in acetic acid (HAC) at 180 to 195°C, catalyzed by cobalt acetate and manganese acetate [Co(Ac)2 and Mn(Ac)2], and promoted by hydrogen bromide (HBr) (Raghavendrachar and Ramachandran, 1992). The reaction kinetics of PX oxidation that involves radical chain elementary reactions is highly complex; therefore, analysis of the chemical reactors through experiments is difficult. Mechanism models could aid in such analysis. However, the development of mechanism models is difficult because of the complexity of PX oxidation. Reaction kinetics, which are important in estimating the reaction rate of chemical processes, must be understood to gain insights into the reaction mechanism. The reaction rate of PX oxidation is usually approximated by the nth kinetic model. Cao and Cincotti studied PX oxidation kinetics at low temperature (
