Chemical Engineering Journal 158 (2010) 241249

Contents lists available at ScienceDirect

Chemical Engineering Journal

journal homepage: www.elsevier.com/locate/cej

Dynamic modeling and process optimization of an industrial sulfuric acid plant

Anton A. Kiss a, , Costin S. Bildea b , Johan Grievink c
a

AkzoNobel Research, Development and Innovation, Process & Product Technology, Velperweg 76, 6824 BM, Arnhem, The Netherlands
University Politehnica Bucharest, Centre for Technology Transfer in Process Industries, Polizu 1-7, RO-011061 Bucharest, Romania
c
Delft University of Technology, Department of Chemical Engineering, Julianalaan 136, 2628 BL, Delft, The Netherlands
b

a r t i c l e

i n f o

Article history:
Received 15 October 2009
Received in revised form 8 January 2010
Accepted 11 January 2010
Keywords:
Sulfuric acid
Adiabatic reactors
Absorption
Optimization
SOx emissions
Energy savings

a b s t r a c t
The current legislation imposes tighter restrictions in order to reduce the impact of chemical process
industry on the environment. In this context, this study presents the dynamic model, simulation and
optimization results for an industrial sulfuric acid plant. The dynamic model, implemented in PSE gPROMS
includes a catalytic reactor (ve pass converter), heat exchangers such as economizers and feed-efuent
heat exchangers, mixers, splitters and reactive absorption columns. The kinetic parameters were tted
to the real plant data, while the remaining model parameters were estimated using classical correlations.
The modeling results agree very well with the real plant data.
The model implemented in gPROMS is useful for evaluating the dynamic behavior of the plant and for
minimization of the total amount of SOx emissions. The SOx emissions could be signicantly reduced
by over 40% by optimizing operating parameters such as air feed ow rates or split fractions. However,
only minor increases in energy production can be achieved due to the plant already operating near full
capacity. The simulations also show that operational problems may occur when the process is disturbed
due to production rate changes or catalyst deactivation, the non-linear response of the plant leading
to sustained oscillations. Besides controllability, operability and optimization studies the gPROMS plant
model is also useful for operator training and various scenario assessments.
2010 Elsevier B.V. All rights reserved.

1. Introduction
As the largest-volume industrial chemical produced in the
world, consumption of sulfuric acid is often used to monitor a countrys degree of industrialization. Sulfuric acid is produced every year
in quantities larger than any other chemical [1,2]. Nowadays, the
worldwide production exceeds 160 million of tonnes, with US and
Asia as the top consumers (Fig. 1).
Remarkably, sulfuric acid is also a particularly corrosive and
dangerous acid, with extreme environmental and health hazards if
not manufactured, used, and regulated properly. Sulfuric acid has
a wide range of uses including: phosphate fertilizer production,
dyes, alcohols, plastics, rubber, ether, glue, lm, explosives, drugs,
paints, food containers, wood preservatives, soaps and detergents,
pharmaceutical products, petroleum products, pulp and paper. The
common lead-acid storage battery is one of the few consumer
products that actually contain H2 SO4 . Sulfuric acid is also used
extensively as a solvent for ores and as catalyst for petroleum
rening and polymer manufacture. Note that agricultural fertilizers
represent the largest single application for sulfuric acid, account-

Corresponding author. Tel.: +31 026 366 1714; fax: +31 026 366 5871.
E-mail addresses: [email protected], [email protected] (A.A. Kiss).
1385-8947/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.cej.2010.01.023

ing for up to 65% of its usage. The most common process used for
making phosphate fertilizers consists of two steps:
(1) Production of phosphoric acid and gypsum, by reacting phosphate rock with H2 SO4 .
Ca3 (PO4 )2 (s) + 3H2 SO4 (lq) + 6H2 O (lq)
2H3 PO4 (lq) + 3CaSO4 2H2 O (s)

(1)

(2) Reaction of phosphoric acid with ammonia to make ammonium

phosphates.
2H3 PO4 + 3NH3 NH4 H2 PO4 + (NH4 )2 HPO4

(2)

Sulfuric acid plants are distributed throughout the industrialized world, as follows: 35% Asia, 24% North America, 7% South
and Central America, 10% Western Europe, 10% Eastern Europe,
11% Africa and 3% Oceania and Australia [1,2]. Most of the sulfuric acid plants are located near their product acids point of
usei.e. near phosphate fertilizer plants, nickel ore leach plants
and petroleum reneries. The reason for this is because elemental sulfur is cheaper to transport than sulfuric acid. Note also that
the volatility of the sulfuric acid price is due to the small imbalances between acid demand and supply, as well as the difculty of
storing large quantities of acid. The recent large increase in price

242

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

Fig. 1. Sulfuric acid production (left) and consumption (right) in the world.

is due to Chinas increasing demand for fertilizer, hence sulfuric

acid.
The benets of the applications of model-based optimization
and control to industrial plants can be highlighted with some recent
contributions. For example, model-based control was applied to
temperature control in an industrial batch reactor and to product quality control in a batch crystallization [3,4]. Simon et al. [5]
reported the model-based control of a liquid swelling constrained
batch reactor subject to recipe uncertainties. Dynamic modeling
was primarily used to make a comprehensive comparison of control strategies for Dividing-Wall Columns [6]. The dynamics and
control of a new process for fatty acid esterication by dual reactive distillation are discussed by Dimian et al. [7]. Moreover, several
applications of dynamic modeling to control plants of industrial
signicance were reported in the book of Dimian and Bildea [8].
Among other standard simulation environments used in the
chemical industry, gPROMS has a proven track record of being
successfully applied in major research areas, such as: reactive
separation processes [9], pressure-swing absorption [10], polymerization processes [11], dynamic modeling [12], process design and
control [13], optimization of design and operation [14], and several
other application areas. For a comprehensive list of publications
the reader is directed to the website of Process Systems Enterprise
(www.psenterprise.com/academic/publications.html).
2. Problem statement
The literature study shows that during the last two decades the
research in the area of sulfuric acid production captures interesting topics requiring reliable (dynamic) modeling, as for example:
numerical-simulation of a periodic ow reversal reactor for SO2
oxidation [15], modeling of hot and cold start-ups [16], study of
unsteady-state catalytic oxidation of SO2 by periodic ow reversal [17], modeling the oxidation of SO2 in a trickle-bed reactor
[18], modeling of SO2 oxidation in a xed-bed reactor with periodic
ow reversal [19], modeling of SO2 oxidation taking into account
dynamic properties of the catalyst [20], oxidation of SO2 in a tricklebed reactor packed with activated carbon at low liquid ow rates
[21], optimization studies in H2 SO4 production [22], optimization
of an adiabatic multi-bed catalytic reactor for the oxidation of SO2
[23], simultaneous determination of SO3 and SO2 in owing gases
[24], model-based experimental analysis of xed-bed reactors for
SO2 oxidation [25].
The previous studies focused either only on the SO2 oxidation reactor or solely on the SO3 absorption. In contrast, this
study presents a complete dynamic model of an industrial sulfuric
acid plantcurrently operated by Phosphoric Fertilizers Industry
(www.p.gr), as well as the results of the simulation and optimization of the plant. Due to its powerful features, gPROMS was selected

to perform all simulation tasks [26]. The dynamic model developed in this study includes also a graphical user interface built
in Microsoft Excel that allows scenario evaluation and operator
training. The model of the complete plant was successfully used
for dynamic simulations to evaluate the non-steady-state behavior of the plant and detect changes in product quality, as well
as to minimize the total amount of sulfur oxides released in the
atmosphere.
3. Process description
There are two major processes used for the sulfuric acid production: the lead chamber process [27] and the current contact
process [28]. The main steps in the latter process consist of burning
sulfur (S) in air to form sulfur dioxide (SO2 ), converting SO2 to sulfur trioxide (SO3 ) using oxygen (O2 ) from air, and absorbing SO3 in
water (H2 O) or a diluted solution of sulfuric acid (H2 SO4 ) to form a
concentrated solution of acid (>96%).
S (s) + O2 (g)  SO2 (g) H R = 296 810 kJ/kmol

(3)

SO2 (g) + 1/2 O2 (g)  SO3 (g) H R = 96 232 kJ/kmol

(4)

SO3 (g) + H2 O (lq)  H2 SO4 (lq) H R = 132 000 kJ/kmol

(5)

The simplied owsheet of the industrial sulfuric acid production process consists of a sulfur burner, multi-pass converter, heat
exchangers and absorbers as shown in Fig. 2.
Filtered ambient air is drawn through a high efciency drying
tower by the main compressor to remove moisture. The compressed dry air enters a refractory-lined furnace where molten
sulfur is burned to produce SO2 . The hot SO2 combustion gas is
then cooled in a steam boiler to the proper temperature to promote
conversion to SO3 in the conversion step. A multi-bed catalytic adiabatic reactor is used as the SO2 oxidation reaction is limited by
the chemical equilibrium. Note that O2 does not oxidize SO2 to SO3
without a catalyst, hence this is compulsory. The catalyst used here
is vanadium oxide (V2 O5 ) mixed with an alkali metal sulfate [29].
This mixture is supported on small silica beads, and it is a liquid at
the high temperature inside the reactor [30,31]. Yet other catalysts
were also reported recently [32].
The overall process is designed to give a conversion of sulfur
dioxide to sulfuric acid of over 99.7%. Several conversion steps,
addition of fresh air and inter-stage cooling are necessary as the
reaction is reversible and exothermal [30]. SO2 conversion is further
improved and tail gas emissions are reduced through an intermediate SO3 absorption step (Abs1). This adsorption step takes place
after the fourth bed of catalyst and changes the gas composition,
thus shifting the equilibrium curve to higher conversions, as shown
in Fig. 3 and explained hereafter. The absorption of SO3 is nalized

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

243

Fig. 2. Simplied owsheet of the sulfuric acid production plant.

in the second absorber (Abs2). For heat-integration reasons, two

feed-efuent heat exchangers (FEHE) are used. Remarkably, the
operation of the industrial reactor follows the maximum reaction
rate curve.
During operation, the ow rates of air fed to the sulfur burner
and to the converter passes 3 and 4 can be changed. Moreover, it
is possible to change the amount of energy that is exchanged in
the feed-efuent heat exchangers FEHE1 and FEHE2 through bypassing a fraction of the cold stream (the outlet of the absorption
step performed in Abs1).

4. Dynamic model
Due to its powerful features, gPROMS was selected as modeling
environment. gPROMS provides comprehensive facilities for developing, validating and executing gPROMS models, by performing
activities such as steady-state and dynamic simulation, optimization and parameter estimation. gPROMS modeling language has an
object-oriented character. Thus, the user denes classes of Models which are instantiated by Units. The Units can be aggregated to
form complex Models [26].
According to this paradigm, the model of the sulfuric acid production plant presented in Fig. 2 is build by dening the Models
for sulfur burner, catalytic bed, heat exchanger, feed-efuent heat
exchanger and absorption column, dening the appropriate number of Units of each type and connecting them according to the
topology of the owsheet. To achieve the connection between the
Units, each Model is written such that the molar ow rate, molar
composition, pressure and temperature of its outlet streams are
calculated within the model. Then, these variables are equated to
variables associated to the downstream unit, where the values are
used to compute any variable that is needed for model solving.
Alternatively, gPROMS offers the facility to dene Stream types,
which are instantiated and used to connect units. Fig. 4 shows
graphical schematics of the units used for modeling, as well as the
notation used.
In the following, the main differential and algebraic equations
employed in Models are given per operating unit:

(1) Sulfur burner (complete consumption of sulfur feed)

Mass balance:

Fig. 3. Temperature-conversion diagram.

Fi,out = Fi,in + i ,

 = FSulfur,in

(6)

244

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

Fig. 4. Models used to describe the sulfuric acid plant.

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

(2) Catalytic reactor bed (pseudo-homogeneous, axial dispersion

model, adiabatic operation)
Mass balance:
dc
dci
= u i + b i r
dt
dz

d2 c
+ Dz 2i
dz

(7)

Mass balance gas (A):

G

= f cp,f u

dT
dt

dT
d2 T
+ b r (Hr ) + kz 2
dz
dz

NA
(9)

with the boundary conditions:

at z = 0 :

dc
= u (cin c);
Dz
dz

dc
= 0;
dz

dT
= 0;
dz

(10)

dP
=0
dz

(11)

1/2
2

22.414 (1 + K2 pSO2 + K3 pSO3 )2

[kmol/(kg.cat.s)]

(12)

The parameters of the kinetic model were adjusted such that

the predicted behavior agrees with the real plant data (see
Fig. 3). In this work, the following values were used:

 10.68 + 11300 

[1/atm1/2 ]

k1 = 2.1125 105 exp

 3599.78 
T

(13)

[kmol/(kg.cat.atm2 .s)]
(14)

K2 = 14.641 [atm1 ]
1

K3 = 6.5775 [atm

(15)
(16)

(3) Heat exchangers (constant temperature on the shell side, Tc )

Energy balance:
f cp,f

1
kL

1
kG HA

pA
DB
+
CB,L and NB = NA
HA
DA

dT1
dT1
4
Hw Av
FG
(T1 T2 )
=

G cp,G
dt
dz
D2

4
dT
dT
= f cp,f u
Hw (T Tc )
D
dt
dz

(17)

(4) Feed-efuent heat exchangers

Energy balance (tube side):
4
Hw
dT1
dT1
= u1

(T1 T2 )
D 1 Cp,1
dt
dz

(18)

Energy balance (shell side):

dT2
Hw
dT2
+ Av
= u2
(T1 T2 )
2 Cp,2
dz
dt

(22)

(23)

dT2
4
Hw Av
dT2
=
+
FL
(T1 T2 )
L cp,L
dt
dz
D2
+

NA Av
(HR )
L cp,L

(19)

(5) Absorbers
The absorption model assumes one-dimensional mass and
heat transport normal to the interface, thermodynamic equilibrium at the interface, and instantaneous reaction. Component
A is SO3 in gas phase, while B is H2 O in liquid phase.

(24)

Pressure drop:
u2
dP
= ff f
Dp
dz

k1 pO2 pSO2 (1 (pSO3 /Kp pSO2 pO ))

Kp = exp

(21)

Energy balance (liquid):

A kinetic model similar to the one proposed by Froment and

Bischoff [30] was used:
r =

4 d
dcB
=
(FL CB ) NB Av
dt
D2 dz

Energy balance (gas):

G

dT
= f cp,f u (Tin T )
kz
dz
at z = L :

(20)

Molar ux of components:
(8)

Pressure drop:
u2
dP
= ff f
D
dz
p

4 d
dcA
=
(FG CA ) NA Av
dt
D2 dz

Mass balance liquid (B):

Energy balance:
( f cp,f + b cp,cat )

245

(25)

All the model parameters were either measured or estimated using standard correlations available in the open literature
[28,3338]. The results of the estimation t very well in the range
of general characteristics of gasliquid reactors previously reported
in literature [39]. The reference temperature and viscosity for the
gas components used in the model are available in the CRC Handbook of Chemistry and Physics [40]. Moreover, UNIQUAC equation
can be used in the whole concentration range for the calculation of
vaporliquid equilibrium in aqueous sulfuric acid solutions [41].
The plant model was solved in gPROMS. The steady-state solution is found by equating to zero the time derivatives of the model
Eqs (6)(25). The spatial coordinate is automatically discretized by
the solver (rst order, backwards nite differences method was
used) and the resulting system of algebraic equations is solved by
a non-linear solver using block decomposition (BDNLSOL). Convergence is achieved even for very rough initial approximation of the
model unknowns, for example at temperature and concentration
proles along the reactor. Finding the steady-state solution takes
only a few seconds. gPROMS offers facilities to save the solution of
one simulation in order to be used as the initial guess for the next
run. This further reduces the solution time.
The solution of the dynamic model is found by discretizing the
spatial coordinate and integrating the resulting set of differential and algebraic equations. SRADAU, a fully implicit variable-step
Runge-Kutta method was used. The method is efcient for solution
of differential equations arising from discretization of PDE with
strongly advective terms and handles well discontinuities. About
1 min. of computing time is necessary to simulate 1 h of operation. The model is robust, convergence being easily achieved when
step changes of reasonable magnitude were imposed on the control
variables. The sequential quadratic programming method included
in CVP MS (Control Vector ParameterisationMultiple-Shooting)
dynamic optimization solver was used for the optimization problems described in the next section, a solution being obtained in
about 1 h of computer time.

246

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

Fig. 5. Temperature and composition proles along the oxidation reactor.

Fig. 6. Temperature and composition proles along the two absorption columns.

5. Results and discussion

The path of the reaction can be conveniently shown in the XT
diagram, as illustrated by Fig. 3. One equilibrium curve corresponds
to each composition of the reaction mixture at reactor inlet. The
operating lines represent the simultaneous increase of conversion
and rise of temperature which take place along each catalytic bed.
The slope of the operating lines is proportional to the reciprocal
value of the adiabatic temperature rise, 1/Tad . The horizontal lines
represent cooling of the reaction mixture which takes place in the
external heat exchangers or by injection of fresh cold air. The evolution of temperature and conversion in the industrial reactor is
represented in Fig. 3 by a continuous line. The dashed line rep-

resents the same variables as predicted by the model. The model

practically coincides with the industrial reactor for the rst two
catalytic beds, while the error is very small for the next three beds.
For a given inlet composition, the operating lines can approach the
equilibrium curve (dotted line in Fig. 3) but can never cross it. The
intermediate SO3 absorption shifts upwards the equilibrium curve
and therefore allows higher conversions. Fig. 5 presents the steadystate temperature and composition proles along the SO2 oxidation
reactor. The highest temperatures are reached at the end of the rst
and second catalytic bed, the following steps having less impact on
conversion and temperature. Along with the increased conversion,
the SO2 concentration is reduced from about 11% at the reactor inlet
down to less than 0.02%.

Fig. 7. Reactor temperatures prole, for 10% changes in air feed ow rate (nal case).

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

247

Fig. 8. SOx composition after absorption, for 10% disturbance in the air feed ow rate.

Similarly, the steady-state temperature and composition proles along the two absorption columns are shown in Fig. 6.
Counter-current operation is used in both absorbers. The SO3 concentration is reduced from 9.81% to 0.01% in the intermediate
absorber, and from 0.44% to ppm levels in the nal absorber. Note
that the dimensionless length is in fact the ratio between any given
length and the total absorber length.
The dynamic simulation shows that the nominal operating point
of the plant is stable. When the plant is disturbed, a new steadystate is reached in less than 1 h. Fig. 7 presents simulation results
for the temperature at the exits of the catalytic beds, when the ow
rate of air fed to the sulfur burner is increased or decreased by 10%.
The molar fractions of SO2 and SO3 at the outlet of the nal
absorption column are presented, for similar disturbances, in Fig. 8.
Note the non-symmetrical response (50% +200% for SO2 ) to
these 10% disturbances. However, the composition reaches a new
steady-state in a relatively short settling time of less than 1 h.
Multi-variable optimization was performed for several production rates, corresponding to the amount of sulfur fed into the plant
(nominal value and 5-10% changes). Five key variables were identied and manipulated accordingly to carry out the optimization:
the amount of air fed into the sulfur burner,
the ow rates of air fed into converter pass 3 and 4,
the split fractions (by-pass) for cold streams entering the gas-gas
heat exchangers (FEHE1 and FEHE2).
It should be remarked that the plant produces a signicant
amount of energy in the sulfur burner and the heat exchangers

HX1HX4. Therefore, the rst optimization run aimed at maximizing the amount of energy produced. Increasing energy production
is equivalent to maximizing the amount of SO2 converted into
products. A at optimum is expected because practically the entire
amount of heat generated in the reaction is recovered due to tight
energy integration. As the conversion of the process is already very
high, almost 99.85%, any further increase is insignicant. Therefore, not surprisingly, the energy production can be increased only
by almost one percent.
The second optimization target sought to minimize the total
amount of SOx released in atmosphere (i.e. not absorbed in the nal
absorption column). The optimization results reveal that for the
nominal operating point the ow rate of air fed to sulfur burner
should be increased by 30%. Moreover, the ow rates of air entering the catalytic beds 3 and 4 must increase by 80% and decrease
by 10%, respectively. Finally, the cold streams should not be split to
by-pass the heat exchangers. The conversion proles in the reactor for changes of 10% in the feed ow rate are given in Fig. 9
(left). Remarkably, the SOx emissions can be drastically reduced
by 40% in the nominal case or even more, depending on the
operating region (Fig. 9). Therefore, the industrial sulfuric acid plant
can be now fully exploited even at larger production rates, while
respecting the current ecological restrictions.
6. Discussion of the dynamic behavior
It should be noted that the tight heat-integration can lead to
operational difculties when the plant is not well designed. For
example, we considered the kinetics presented in Froment and

Fig. 9. Conversion proles in the reactor, for 10% changes in the air feed ow rate (left). SOx emissions (lower is better) before and after optimization at various feed ow
rates (right).

248

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

Fig. 10. Evolution of bed-outlet temperature 5% change in air feed ow rate (kinetics taken from [30]).

Bischoff and designed a plant which achieves the same performance

as the plant discussed in the previous sections. Although the nominal operating point is stable, the response of the plant is non-linear
and, for certain disturbances, sustained oscillations are observed.
Fig. 10 presents the temperature at the exit of the catalytic bed,
for 5% changes of the air feed ow rate. Obviously, these oscillations are not acceptable since they propagate from the reactor to
absorbers, rendering the plant unstable and the product off-spec.
Other disturbances that may have a similar effect are production
rate changes, catalyst deactivation, or variation of air feed ow rates
caused by daynight or summerwinter temperature differences
and constant volumetric ow operation.
Morud and Skogestad [42] reported a similar behavior of a multibed, heat-integrated industrial ammonia reactor, where small
changes of operating conditions led to onset of sustained oscillations. The non-linear behavior was explained by the presence
of an inverse response for the temperature through the reactor
beds, combined with the positive energy feedback induced by the
two feed-efuent heat exchangers. Nevertheless, a complete nonlinear analysis would be required to reveal all steady-states that
are possible [4348].
7. Conclusions
This work presents the complete model of an industrial sulfuric acid plant, as well as the simulation and optimization results.
Process Systems Enterprise gPROMS was successfully used as an
advanced process modeling, simulation and optimization environment for dynamic simulations to evaluate the non-steady-state
behavior of the plant and detect changes in product quality, as
well as for optimization of the total amount of SOx released in the
atmosphere. In addition, a Microsoft Excel interface was developed
to allow a more user-friendly interaction, thus making the model
suitable for operator training.
The dynamic model includes a catalytic reactor (ve pass converter), heat exchangers such as economizers and feed-efuent
heat exchangers, mixers, splitters and reactive absorption columns.
As no suitable kinetics information was available, the kinetic
parameters were tted to the real plant data, while the remaining
parameters required for modeling were estimated using classical
correlations. The results show that an excellent agreement exists
between the real plant data and the results of the simulations, the
relative error being typically below 1%.
Along with minor benets in energy production, the amount
of SOx emissions could be signicantly reduced by 40% just by
optimizing operating parameters such as air feed ow rates or
split fractions. Besides controllability, operability and optimization
studies the plant model coded in gPROMS is also useful for operator
training and scenario assessments.

The robust dynamic model developed in this work may be further used for:
Controllability analysis to determine the controllability of the
plant and the possible improvements of the control structure,
as well as optimal control policy.
Other dynamic simulations, to detect potential sensitivity problems and which model parameters have greater importance.
Operator training, due to the Excel interface that plays the role
of a control panel that simulates the real plant behavior for any
change of model parameters.
Other optimization studies, such as tuning the amount of energy
produced by the plant or optimize the air feed policy accounting
for day/night temperature variations.

Notation
ap
Av
ci
cp
D
Di
Dz
ff
HA
Hw
kG/L
L
Ni
P
pi
r
R
Re
t
T
Tc
u

Hr

b
f
i

specic surface area of the packing (m2 /m3 )

specic area (m2 /m3 )
concentration of component i (kmol/m3 )
specic heat capacity (J/kgK)
diameter (m)
diffusion coefcient (component i)
axial dispersion coefcient (m2 /s)
friction factor
Henry constant for component A
total heat transfer coefcient (W/m2 K)
partial mass transfer coefcient (gas or liquid phase)
length of the operating unit
molar ux of component i
pressure (Pa)
partial pressure component i (Pa)
reaction rate (kmol/kg.cat)
Rydberg constant (8.3145 J/molK)
Reynolds number
time (s)
temperature (K)
temperature coolant (K)
velocity (m/s)
individual heat transfer coefcient (W/m2 K)
thermal conductivity (W/mK)
enthalpy of reaction (J/mol)
void fraction
density of the bulk (bed) catalyst (kg/m3 )
density of the uid (kg/m3 )
stoichiometric coefcient
thermal conductivity (W/mK)
dynamic viscosity (Pas)

A.A. Kiss et al. / Chemical Engineering Journal 158 (2010) 241249

Acknowledgement
This project was funded by the European Commission (OPTABSO Project, Contract no. G1RD-CT-2001-00649). We also thank
Phosphoric Fertilizers Industry and Process Systems Enterprise Ltd.
for the technical assistance.
References
[1] W.G. Davenport, M.J. King, Sulfuric Acid Manufacture, Elsevier, 2005.
[2] S.R.I. Consulting, CEH Marketing Research Report Sulphuric Acid, Chemical
Economics Handbook, SRI Consulting, Menlo Park, CA, USA, 2006.
[3] Z.K. Nagy, B. Mahn, R. Franke, F. Allgower, Efcient output feedback nonlinear
model predictive control for temperature control of industrial batch reactors,
Control Engineering Practice 15 (2007) 839859.
[4] Z.K. Nagy, Model based robust control approach for batch crystallization product design, Computers & Chemical Engineering 33 (2009) 16851691.
[5] L.L. Simon, Z.K. Nagy, K. Hungerbuhler, Model based control of a liquid swelling
constrained batch reactor subject to recipe uncertainties, Chemical Engineering
Journal 153 (2009) 151158.
[6] R.C. van Diggelen, A.A. Kiss, A.W. Heemink, Comparison of control strategies for
dividing-wall columns, Industrial & Engineering Chemistry Research 49 (2010)
288307.
[7] A.C. Dimian, C.S. Bildea, F. Omota, A.A. Kiss, Innovative process for fatty acid
esters by dual reactive distillation, Computers & Chemical Engineering 33
(2009) 743750.
[8] A.C. Dimian, C.S. Bildea, Chemical Process DesignComputer-Aided Case Studies, Wiley-VCH, Weinheim, 2008.
[9] C.P. Almeida-Rivera, J. Grievink, Process design approach for reactive distillation using economics, exergy and responsiveness optimization, Industrial &
Engineering Chemistry Research 47 (2008) 5165.
[10] Q. Huang, A. Malekian, M. Eic, Optimization of PSA process for producing
enriched hydrogen from plasma reactor gas, Separation and Purication Technology 62 (2008) 2231.
[11] M. Asteasuain, A. Brandolin, Modeling and optimization of a high-pressure
ethylene polymerization reactor using gPROMS, Computers & Chemical Engineering 32 (2008) 396408.
[12] R.D. Moita, H.A. Matos, C. Fernandes, C.P. Nunes, N.J. Pinho, Dynamic modelling and simulation of a heated brine spray system, Computers & Chemical
Engineering 33 (2009) 13231335.
[13] A. Lawal, M. Wang, P. Stephenson, H. Yeung, Dynamic modelling of CO2 absorption for post combustion capture in coal-red power plants, Fuel 88 (2009)
24552462.
[14] M.S. Tanvir, I.M. Mujtaba, Optimisation of design and operation of MSF desalination process using MINLP technique in gPROMS, Desalination 222 (2008)
419430.
[15] J.D. Snyder, B. Subramaniam, Numerical-simulation of a periodic-ow reversal
reactor for sulfur-dioxide oxidation, Chemical Engineering Science 48 (1993)
40514064.
[16] K. Gosiewski, Dynamic modeling of industrial SO2 oxidation reactors. 1. Model
of hot and cold start-ups of the plant, Chemical Engineering & Processing 32
(1993) 111129.
[17] H.X. Wu, S.Z. Zhang, C.Y. Li, Study of unsteady-state catalytic oxidation of sulfur
dioxide by periodic ow reversal, Canadian Journal of Chemical Engineering 74
(1996) 766771.
[18] P.V. Ravindra, D.P. Rao, M.S. Rao, A model for the oxidation of sulfur dioxide in
a trickle-bed reactor, Industrial & Engineering Chemistry Research 36 (1997)
51255132.
[19] R.Y. Hong, X. Li, H.Z. Li, W.K. Yuan, Modeling and simulation of SO2 oxidation
in a xed-bed reactor with periodic ow reversal, Catalysis Today 38 (1997)
4758.
[20] N.V. Vernikovskaya, A.N. Zagoruiko, A.S. Noskov, SO2 oxidation method. Mathematical modeling taking into account dynamic properties of the catalyst,
Chemical Engineering Science 54 (1999) 44754482.

249

[21] Y. Suyadal, H. Oguz, Oxidation of SO2 in a trickle bed reactor packed with activated carbon at low liquid ow rates, Chemical Engineering & Technology 23
(2000) 619622.
[22] A.A. Kiss, C.S. Bildea, P.J.T. Verheijen, Optimization studies in sulphuric acid production, Computer Aided Chemical Engineering 21A (2006)
736742.
[23] A. Nodehi, M.A. Mousavian, Simulation and optimization of an adiabatic
multi-bed catalytic reactor for the oxidation of SO2 , Chemical Engineering &
Technology 30 (2007) 8490.
[24] J.G. Ibanez, C.F. Batten, W.E. Wentworth, Simultaneous determination of SO3
and SO2 in a owing gas, Industrial & Engineering Chemistry Research 47 (2008)
24492454.
[25] J.C. Schoneberger, H. Arellano-Garcia, G. Wozny, S. Korkel, H. Thielert, Modelbased experimental analysis of a xed-bed reactor for catalytic SO2 oxidation,
Industrial & Engineering Chemistry Research 48 (2009) 51655176.
[26] Ltd. Process Systems Enterprise, gPROMS Advanced User Guide, 2009.
[27] E.M. Jones, Chamber process manufacture of sulfuric acid, Industrial & Engineering Chemistry 42 (1950) 22082210.
[28] Kirk Othmer, Encyclopedia of Chemical Technology, vol. 22, 3rd edition, John
Wiley & Sons, 1984.
[29] G.A. Bunimovich, N.V. Vernikovskaya, V.O. Strots, B.S. Balzhinimaev, Y.S.
Matros, SO2 oxidation in a reverse-ow reactorinuence of a vanadium catalyst dynamic properties, Chemical Engineering Science 50 (1995) 565580.
[30] G.F. Froment, K.B. Bischoff, Chemical Reactor Analysis and Design, John Wiley
& Sons, 1979 (Chapter 11).
[31] A. Christodoulakis, S. Boghosian, Molecular structure of supported molten salt
catalysts for SO2 oxidation, Journal of Catalysis 215 (2003) 139150.
[32] T.L. Jorgensen, H. Livbjerg, P. Glarborg, Homogeneous and heterogeneously catalyzed oxidation of SO2 , Chemical Engineering Science 62 (2007) 44964499.
[33] C.R. Wilke, P. Chang, Correlation of diffusion coefcients in dilute solutions,
AIChE Journal 1 (1955) 264270.
[34] K. Onda, H. Takeuchi, Y. Okumoto, Mass transfer coefcients between gas and
liquid phases in packed columns, Journal of Chemical Engineering Japan 1
(1968) 5662.
[35] R. Taylor, R. Krishna, Multicomponent Mass Transfer, John Willey & Sons, 1993.
[36] J.K. Klassen, Z. Hu, L.R.J. Williams, Diffusion coefcients for HCl and HBr in 30
wt% to 72 wt% sulfuric acid at temperatures between 220 and 300 K, Journal of
Geophysical Research 103 (1998) 1619716202.
[37] R.H. Perry, D.W. Green, J.O. Maloney, Perrys Chemical Engineers Handbook,
7th edition, McGraw-Hill, 1999.
[38] F.P. Incropera, D.P. DeWitt, Fundamentals of Heat and Mass Transfer, 4th edition, John Wiley & Sons, 2000 (Chapter 11).
[39] G. Bozga, O. Muntean, Chemical Reactors, vol. 2, Editura Tehnica, Bucharest,
2001.
[40] Chemical Rubber Company (CRC), in: Weast, C. Robert (Eds.), CRC Handbook of
Chemistry and Physics, 65th edition, CRC Press, Inc., Boca Raton, Florida, 1984.
[41] F.L.P. Pessoa, C.E.P.S. Campos, A.M.C. Uller, Calculation of vaporliquid equilibria in aqueous sulfuric acid solutions using the UNIQUAC equation in the whole
concentration range, Chemical Engineering Science 61 (2006) 51705175.
[42] J.C. Morud, S. Skogestad, Analysis of instability in industrial ammonia reactors,
AIChE Journal 44 (1998) 888895.
[43] C.S. Bildea, A.C. Dimian, Stability and multiplicity approach to the design of
heat-integrated PFR, AIChE Journal 44 (1998) 27032712.
[44] C.S. Bildea, A.C. Dimian, P.D. Iedema, Nonlinear behavior of reactor-separatorrecycle systems, Computers & Chemical Engineering 24 (2000) 209215.
[45] A.A. Kiss, C.S. Bildea, A.C. Dimian, P.D. Iedema, State multiplicity in CSTRseparator-recycle polymerization systems, Chemical Engineering Science 57
(2002) 535546.
[46] A.A. Kiss, C.S. Bildea, A.C. Dimian, P.D. Iedema, State multiplicity in PFRseparator-recycle polymerization systems, Chemical Engineering Science 58
(2003) 29732984.
[47] A.A. Kiss, C.S. Bildea, A.C. Dimian, P.D. Iedema, Design of recycle systems with
parallel and consecutive reactions by non-linear analysis, Industrial & Engineering Chemistry Research 44 (2005) 576587.
[48] A.A. Kiss, C.S. Bildea, A.C. Dimian, Design and control of recycle systems by
non-linear analysis, Computers & Chemical Engineering 31 (2007) 601611.