Hybrid modeling of ethylene to ethylene oxide heterogeneous reactor
G. Zahedi
a,
⁎
, A. Lohi
b
, K.A. Mahdi
c
a
Process Systems Engineering Centre (PROSPECT), Faculty of Chemical Engineering, Universiti Teknologi Malaysia, UTM Skudai, 81310 Johor Bahru, Johor, Malaysia
b
Department of Chemical Engineering, University of Ryerson, Toronto, Ont. M5B 2K3, Canada
c
Department of Chemical engineering, University of Kuwait, Safat 13060, Kuwait
abstractarticle info
Article history:
Received 2 August 2010
Received in revised form 7 April 2011
Accepted 12 April 2011
Available online 4 May 2011
Keywords:
Simulation
Ethylene oxide reactor
Grey box modeling
Neural network
Dynamic modeling
In this research a dynamic grey box model (GBM) of ethylene oxide (EO) fixed bed reactor has been
presented. In the first step of the study, kinetic model of the existing reactions was obtained using artificial
neural network (ANN) approach. In order to build the ANN model industrial data of a typical EO reactor were
employed. Time, C
2
H
4
,C
2
H
4
O, CO
2
,H
2
O and O
2
mole fractions were network inputs and the multiplication of
reaction rate and catalyst deactivation (r *a)was ANN output. From 164 data, 109 data were employed to train
ANN. After employing different training algorithms, it was found that, the radial basis function network
(RBFN) training algorithm provides the best estimations of the data. This best obtained network was tested
against fifty five unseen data. The network estimations were close to unseen data which confirmed
generalization capability of the obtained network.
In the next step of study, (r*a) was estimated with ANN and then the hybrid model of the reactor was solved.
Simulation results were compared with EO mechanistic model and also with plant industrial data. It was
found that GBM is 8.437 times more accurate than the mechanistic model.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
Ethylene oxide (EO) has many applications like in disinfection,
sterilization of surgical equipments, fumigation and cosmetic [1-3].
Several studies have been carried out on EO reactor modeling [4-11].
The EO epoxidation was optimized by Zhu [6]. Baratti et al. proposed
strategies to optimize the reactor performance [7]. Bingchen et al.
modeled the EO synthesis in a slurry reactor with silver catalyst. They
found that a slurry reactor has higher selectivity than a gas–solid fixed
bed catalytic reactor [8]. Modeling of the transient kinetics of ethylene
oxidation by oxygen over a commercial catalyst was performed by
Harmsen et al. [9]. The simulation of EO reactor in order to find the
optimum temperature was performed by Hwang and Smith [10, 11].
Zhou and Yuan studied optimization of EO reactor [5]. Recently Zahedi
et al. investigated the effect of operating conditions on an industrial
EO reactor performance using a mechanistic (white box) model. The
model was validated with plant industrial data [12].
Dynamic modeling and optimization of an industrial EO multi-
tubular reactor in the presence of silver-based catalyst and ethylene
dichloride as a moderator was investigated by Aryana et al. [13]. They
simultaneously solved a set of nonlinear equations; kinetic heat mass
transfer equations and catalyst deactivation. Their model was in
good agreement with plant data. They also investigated the effect
of operating parameters on reactor performance and optimized the
performance of ethylene oxide reactor to maximize EO yield and
selectivity [13]. Hybrid modeling of EO reactor has not been a subject
of any research based on our literature survey.
In order to provide a realistic model of industrial EO reactor, ac-
curate knowledge of reaction kinetics is desirable. In spite of applying
some reactions for many years, their reaction mechanisms are not still
understood. Sometimes many reactions and intermediates play
significant roles in reaction systems and also kinetic modeling is
related to some state variables which are not measurable using
conventional methods [14, 15]. In this case as elementary reaction
kinetics vary with reaction conditions, applying global reaction rates
for a reaction network is impossible. According to these facts and due
to the lack of precise kinetic models, using some alternative models
like artificial neural network (ANN) instead of mathematical reaction
rates is useful [14, 15].
Generally, there are three main procedures for process modeling;
white box, black box and grey box modeling [16]. White box, first
principle model (FPM) or deterministic model is used when complete
information of the process is accessible and the whole governing
equations of the system are solvable using analytical or numerical
techniques. The governing equations are mass, energy and momentum
balances and empirical equations from thermodynamics and kinetics
[17, 18]. The commercial simulators for simulation of oil refining units
use FPMs [19]. If the governing equations are not available, cannot be
solved or need long computational time; the process simulation can be
carried out using pure ANN. This type of modeling which uses only
input and output data is called black box modeling [17, 18].
Fuel Processing Technology 92 (2011) 1725–1732
⁎ Corresponding author. Tel.:+60 07 5535583; fax: +60 07 5581463.
E-mail address: grzahedi@fkkksa.utm.my (G. Zahedi).
0378-3820/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.fuproc.2011.04.022
Contents lists available at ScienceDirect
Fuel Processing Technology
journal homepage: www.elsevier.com/locate/fuproc