See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228413866

Mathematical Modeling of an Industrial Naphtha Reformer with Three

Adiabatic Reactors in Series

Article  in  IRANIAN JOURNAL OF CHEMISTRY & CHEMICAL ENGINEERING-INTERNATIONAL ENGLISH EDITION · September 2009

CITATIONS READS

7 561

5 authors, including:

Ali Fazeli Shohreh Fatemi

University of Tehran University of tehran
18 PUBLICATIONS   196 CITATIONS    187 PUBLICATIONS   1,278 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Design of a Hybrid Control System for Pressure Swing Adsorption Processes View project

layer by layer deposition- membrane separation View project

All content following this page was uploaded by Shohreh Fatemi on 01 June 2014.

The user has requested enhancement of the downloaded file.

Iran. J. Chem. Chem. Eng. Research Note Vol. 28, No. 3, 2009

Mathematical Modeling of an Industrial Naphtha

Reformer with Three Adiabatic Reactors in Series

Fazeli, Ali; Fatemi, Shohreh*+; Mahdavian, Mohammad; Ghaee, Azadeh

School of Chemical Engineering, University College of Engineering, University of Tehran, I.R. IRAN

ABSTRACT: A mathematical model for commercial naphtha catalytic reformer of Tehran refinery
was developed. This model includes three sequencing fixed beds of Pt/Al2O3 catalyst at the steady
state condition using detailed kinetic scheme involving 26 pseudo-components connected by a
network of 47 reactions, in the range of C6 to C9 hydrocarbons. The reaction network consisted
dehydrogenation, hydrogenation, ring expansion, paraffin and iso-paraffin cracking, naphthene
cracking, paraffin isomerization and hydrodealkylation of aromatics. The kinetic model was fine
tuned against industrial plant data using a feed characterized by PIONA (Paraffin, Iso-paraffin,
Oleffin, Naphthene and Aromatics) analysis. The final outlet results of the reformer such as RON
(Research Octane Number), yield and outlet reformate compositions have shown good agreement
with actual conditions of Tehran Refinery reforming unit.

KEY WORDS: Kinetic model, Modeling, Naphtha reforming, Fixed bed reactor.

INTRODUCTION
The catalytic reforming process is one of the most lumps of paraffin, naphtane and aromatic. However these
critical operations in petroleum refineries to produce kinetic models could not be a proper tool for modeling
gasoline with high octane number [1-6]. and simulation of a commercial process, in the case of
To design new plants and optimize the existing ones, Tehran refinery naphtha reforming. In this study, an
an appropriate mathematical model for simulation the appropriate kinetic model was developed and fine tuned
industrial catalytic reforming process is needed. However for Tehran refinery catalytic reformer to process the
using simple lump kinetic models could not show straight run naphtha. The feed was characterized by
reasonable and appropriate results in scaled up reforming naphthenes, isoparaffins, paraffins and aromatics lumps
processes, therefore introduction of new kinetic lumps of each carbon number from C6-C9 consisting 26
and reaction networks are still being the interested pseudo-components. The major reforming reactions such
research and development subjects in the field of as dehydrogenation, dehydrocyclization, isomerization,
modeling and simulation of reforming processes. hydrocracking and hydrodealkylation were considered in
The applied reaction and their kinetics reported in the the model. The kinetic equations and their initial
literature are different and usually simplified to the three parameter estimations were taken from the literature [3,5,7].
* To whom correspondence should be addressed.
+ E-mail: shfatemi@ut.ac.ir
1021-9986/09/3/97 6/$/2.60

97
Iran. J. Chem. Chem. Eng. Fazeli, A., et al. Vol. 28, No. 3, 2009

The kinetic parameters were fine tuned by direct Table 1:Catalyst characterization.
search (Nealder-Mead simplex) method. Suitable coupling Kind of Catalyst Pt/γ-Al2O3
of optimization method with a fast ODEs solver, choosing
Size and shape Spherical-1.8 mm diameter
proper reactions, kinetics, objective function and weight
Particle density 690 kg/m3
factors for a special commercial process are critical
stages of the present kinetic modeling. Particle porosity 0.6

MODEL DEVELOPMENT Reactor

Reactor11 Reactor 22 R
Reactor Reactor
eactor 33
Reactor

The schematic of the reforming process modeled in

this work is shown in Fig. 1. The feed to the reforming
unit is hydrotreated straight run naphtha. It consists of
T -1 T -2
T- 1 T- 2

naphthenes (Alkyl cyclohexanes (ACH) and Alkyl

cyclopentanes (ACP)), paraffins (normal paraffins (nP)
and isoparaffins (iP)) and aromatics (A), containing
lumps of carbon number from C6-C9. The major
reforming reactions are endothermic causing temperature
NetHydrogen
Net Hydrogen
decrease of the reaction stream and catalyst along the
Com pressor
Compressor
bed. To prevent reaction rate reduction caused by
temperature decrease, the catalyst is distributed in three P - 6- -
P -
6 --

FuelGGas
Fuel as

adiabatic reactors in presence of heaters installed between LPG

LPG
Flash
Flash
the reactors. As shown in Fig. 1, the naphtha feed is Stabilizer
S tabilizer

mixed with hydrogen and heated to desired temperature Naphtha

N aphtha Fresh
Naphtha FreshFeed
Feed Reformate
Reform ate

before entering the first reactor. In order to delay coke

formation of the catalysts in the fixed bed reformers, the
Fig. 1: Schematic of naphtha reforming process.
process should work at high pressure of hydrogen.
In reforming, a bifunctional catalyst is used in which
The partial pressure of each component, in
an acidic function provided by a cholorinated alumina
thereaction rate equation, was related to the molar flow
carrier which is combined with a metal function provided
rate by the following equation:
by platinum. [4,7]. Table 1 exhibits the catalyst properties
of Tehran reforming process. The operational conditions
of Tehran refinery reformer and naphtha feed Fi
Pi = 26
× Pt ;i = 1, 2, 3, ..., 26 (2)
characterization are shown in table 2. Fi
The reaction network considered in this study is i =1
introduced by various reactions such as Dehydrogenation, The homogeneous energy balance of each adiabatic
Ring expansion, Isomerization, Hydrocracking, Naphthene reactor was written as:
cracking, Hydrodealkylation, for C6 to C9 hydrocarbons In which the reaction enthalpies and component heat
which were totally 47 reactions. [3, 5, 7]. capacities were replaced from Sinnott [8].
The kinetic models of the above reactions were Ordinary differential equations of mass and heat
incorporated into a non-isothermal plug flow reactor balances were solved simultaneously by variable order
model. The partial mass balances of each component solver based on the numerical differentiation formulas
were written as follows for 26 components consisting (NDFs) method with MATLAB 7.3 software. Inlet
liquid hydrocarbons, LPG and Hydrogen: conditions of each reactor were set to outlet conditions of
prior reactor in except of temperature that is changed
47
dFi because of intermediate heaters. The input to the model
dW
= ( R j × StCo j,i ) ; i = 1, 2, 3, ..., 26 (1)
were feed flow rate and composition of all lumps in the
j=1

98
Iran. J. Chem. Chem. Eng. Mathematical Modeling of… Vol. 28, No. 3, 2009

Table 2: Naphtha feed and operating conditions of Tehran feed, catalyst loading, recycle gas composition, inlet
refinery reforming plant. temperatures and pressures. The output of the model
Operating Condition Values included temperatures and concentration profiles of
T1 493
reformate as a function of catalyst distribution for each
Inlet temperature of the reactors reactor, gas composition and temperature drop. In
T2 493
(°C) addition the octane number of the feed and liquid
T3 493 reformate obtained from a data base of several pure
P1 3075.6 components of naphtha by a subroutine program.
Pressure (kPa) P2 2920.5 The initial guess values for the frequency factors and
activation energies were obtained from the literature [3]
P3 2799.9
and optimized using the commercial plant data of first
No. of Reactors 3 few days data, by a direct search algorithm (Nealder-
Total catalyst weight (kg) 45693.5 Mead simplex method) due to the minimization of an
R1 20 objective function. This function is introduced as the sum
of absolute relative deviations of the reactors temperature
Catalyst loading (wt%) R2 30
drop, outlet compositions and liquid reformate yield as
R3 50 observed in Eq. 4. The weight factors were introduced for
Feed flow rate (kg/hr) 65341.7 each term of objective function (OF) in order to approach
H2/HC molar ratio 5.6 to desired convergence of model and experiment.
ACH6 2.261 Objective Function (OF)
ACH7 5.952
24
ZCalculated
i − ZiExprimental
= 100 * w i
ACH8 3.878 i =1 ZiExprimental
ACH9 2.852 20
x iCalculated − x iExprimental
= 100 * w i +
ACP6 5.866 i =1 XiExprimental
ACP7 0.821 LY%Calculated − LY% Exprimental
100 * w 21 +
ACP8 4.156 LY% Exprimental
ACP9 0.724 3
∆TrCalculated − ∆TrExprimental
100* w 21+ r
nP6 4.925
r =1 ∆TrExprimental
Fresh feed composition nP7 9.315
(mole %)
nP8 8.582
RESULTS AND DISCUSSION
nP9 5.263 The pre-exponential Arrhenius constants and
iP6 2.063 activation energies were determined after fine tuning with
iP7 7.880 experimental data. Table 3 gives the predicted and the
plant output results of Tehran refinery reforming plant.
iP8 10.930
The predicted temperature drop of the reactors and the
iP9 10.674 reformate yield, compositions and RON from the last
A6 0.944 reactor are compared with actual results of the plant, in
A7 4.915 this table.
As can be seen, temperature differences in all reactors
A8 5.217
are in good consistency with reality. Reformate
A9 2.782 composition, yield and research octane number are also
Feed RON 69.44 close to actual values. The maximum deviation of the
tuned model from operational values is less than 2 %.

99
Iran. J. Chem. Chem. Eng. Fazeli, A., et al. Vol. 28, No. 3, 2009

Table 3: Comparison of predicted and experimental results of 520

Experimental Point
the reforming unit.
Experimental Point
Tehran Refinery 500 Experimental Point
Parameters
Absolute
Experimental Predicted 480
Relative Experimental Point

T (°C)
Deviation Reactor 3
∆T1 54 53.889 0.206
Temperature 460 Reactor 2 Experimental Point
drop in the ∆T2 19 18.951 0.258
reactors (°C) Experimental Point
440
∆T3 5 4.986 0.280
Reactor 1
ACH6 0.058 0.058 0.059 420
0 20 40 60 80 100
ACH7 0.098 0.098 0.119 Catalyst (%)

ACH8 0.081 0.081 0.030 Fig. 2: Temperature profiles in three reforming reactors.

ACH9 0.010 0.010 0.167

250
ACP6 0.912 0.913 0.093 Molar flow rate (kmole/hr)

200 A
ACP7 0.468 0.468 0.047

150 nP iP
ACP8 1.160 1.161 0.034

ACP9 0.010 0.010 0.000 100

ACP
ACH
nP6 4.681 4.666 0.333 50
C5-
Composition nP7 2.660 2.659 0.023
0
(mole %) of 0 20 40 60 80 100
reformate at nP8 0.010 0.010 0.040
the outlet of Catalyst (%)
reactor no.3 nP9 5.398 5.426 0.522 Fig. 3: Lumped component molar flow rate versus catalyst
percent.
iP6 10.863 10.750 1.040

iP7 10.033 10.029 0.040 35

Molar flow rate (kmole/hr)

iP8 2.640 2.639 0.042 30

25
iP9 0.370 0.370 0.016
ACH7
20 ACH9
A6 5.675 5.675 0.004 ACH8
15 ACH6
A7 21.723 21.734 0.051
10
A8 19.705 19.795 0.457
5

A9 13.445 13.448 0.022 0

0 20 40 60 80 100
LY (wt %) 80.24 78.699 1.920 Catalyst (%)

RON 96.8 96.824 0.0248 Fig. 4: Alkylcyclohexanes (ACH) molar flow rates versus
catalyst percent.

100
Iran. J. Chem. Chem. Eng. Mathematical Modeling of… Vol. 28, No. 3, 2009

100
3750
Molar flow rate (kmole/hr)

Molar flow rate (kmole/hr)

80 A8 3700
A7
3650
60
3600
A9
40 Reactor 3 3550
Reactor 2 Reactor 3
3500 Reactor Reactor 2
20 Reactor 1
1
A6 3450
0
0 20 40 60 80 100 0 20 40 60 80 100
Catalyst (%) Catalyst (%)
Fig. 5: Aromatic (A) molar flow rates versus catalyst percent. Fig. 6: Hydrogen molar flow rates versus catalyst percent.

The presented model can also predict the trends of the experimental observations suggests that the model is well
liquid products, LPG, H2 and other necessary outputs. founded and is a good platform for additional process
The present model shows production of the reformate of specific refinement.
96.8 RON from a feed around 69 RON. The model shows
around 79 wt % yield of reformate against 21 wt % LPG Nomenclatures
which is considered in the acceptable range for such a Cp Specific heat capacity (KJ/kmol.K))
high pressure process. The predicted temperature and Ft Total molar flow rate (kmol/hr)
component profiles along the catalyst weight are shown Fi Molar flow rate of i th component (kmol/hr)
in Figs. 2 to 6. LY % Reformate liquid mass yield percent
Pt Total pressure kPa)
CONCLUSIONS Pi Partial pressure kPa)
Nowadays production of gasoline with high octane R Gas constant (J/mol.K)
number from naphtha produced from crude oil refining Rj j th reaction rate (kmol/(hr.kgcat))
through catalytic reforming reactions, attracts great StCoi,j Stochiometry coefficient of
attention. In this research naphtha reforming reactor of componet i in j th reaction
Tehran refinery was kinetically modeled, in high T Temperature (K)
pressure fixed bed reactors, using 47 reactions network W Catalyst weight (kg)
with 26 pseudo-components which enables modeling wi i th weight factor
of hydrogen, C1 to C5 production, liquid reformate Xi Mole fraction of the ith component
composition and temperature variations. Hj Reaction enthalpy(kJ/kmol)
The kinetics was fine tuned by direct search (Nelder- Tr Temperature difference across
Mead simplex) method. This study showed that using fast the reactor no. r
ODE solver of variable order based on the numerical
differentiation formulas (NDFs) is suitable for coupling Received : 18th April 2008 ; Accepted : 14th October 2008
to optimization method and can handle large number of
components and reactions. REFERENCES
The final outlet results of the reformer such as RON, [1] Smith, R.B., Kinetic Analysis of Naphtha Reforming
yield and outlet reformate composition have shown good with Platinum Catalyst, Chemical Engineering
agreement with actual conditions of Tehran Refinery Progress, 55, p. 76 (1959).
reforming unit with global absolute deviation of less than 2 %. [2] Marine, G. B. and Froment, G. F., Reforming of C6
The reforming model contains a rich set of model features Hydrocarbons on a Pt-Al2O3 Catalyst, Chemical
and the confrontation of simulation results with Engineering Science, 37, p. 759 (1982).

101
 Iran. J. Chem. Chem. Eng. Fazeli, A., et al. Vol. 28, No. 3, 2009

[3] Padmavathi, G. and Chaudhuri, K. K., Modeling and

Simulation of Commercial Catalytic Naphtha
Reformers, The Canadian Journal of Chemical
Engineering, 75, p. 930 (1997).
[4] Tasker, U. and Riggs, J. B., Modeling and
Optimization of a Semi Regenerative Catalytic
Naphtha Reformer, AIChE, 43, p. 740 (1997).
[5] Ancheyta Juarez, J. and Villafuerte Macias, E.,
Kinetic Modeling of Naphtha Catalytic Reforming
Reactions, Energy & Fuels, 14, p. 1032 (2000).
[6] Hu, S. and Zhu, F. X. X., Molecular Modeling and
Optimization for Catalytic Reforming, Chemical
Engineering Communications, p. 191, 500 (2004).
[7] Ancheyta Juarez, J., Villafuerte Macias, E., Diaz
Garcia, L. and Gonzalez Arredondo, E., Modeling
and Simulation of Four Catalytic Reactors in
Series for Naphtha Reforming, Energy & Fuels,
15, p. 887 (2001).
[8] Sinnott, R. K., (Editor), “Chemical Engineering
Design”, Coulson and Richardson's Chemical
Engineering, Vol. 6, 4th Edition, Elsevier, (2005).

102

View publication stats