Chemical Engineering Science 60 (2005) 3199 3206

www.elsevier.com/locate/ces

Analysis of the CSTR with two-phase ow under

phase equilibrium conditions
A. Yermakova, V.I. Anikeev
Boreskov Institute of Catalysis, SB RAS, Pr. Akademika Lavrentieva, 5 Novosibirsk 630090, Russian Federation

Received 28 February 2004; received in revised form 23 December 2004; accepted 11 January 2005
Available online 4 March 2005

Abstract
A mathematical model of the ow gasliquid continuous stirred tank reactor (CSTR) is presented under the assumption of thermodynamic
equilibrium of the two-phase ow. Phase equilibrium was calculated using the RedlichKwongSoave (RKS) equation of state. An efcient
algorithm based on homotopy method was used to obtain a numerical solution of the system of nonlinear algebraic equations of the model.
By an example of the liquid-phase hydrogenation of benzene, steady states were analyzed in isothermal and adiabatic regimes and at
partial removal of heat released as a result of exothermic reaction. It was shown that under certain conditions, a multiplicity of stationary
solutions is possible due to nonlinearity of the kinetic reaction, on the one hand, and deviation of the reaction mixture properties from
the ideal state, on the other hand.
2005 Elsevier Ltd. All rights reserved.

Keywords: Mathematical modeling; Multiphase reactions; Phase equilibria; Thermodynamics process

1. Introduction
charts (Yablonsky et al., 1983; Shinnar, 1988; Sophos et al.,
1980; Morachevsky, 1989; Bertucco et al., 1997).
In a continuous stirred tank reactor (CSTR) with two-
For the gasliquid systems where chemical reactions
phase gasliquid ow, a stationary composition of the phases
occur, conditions of phase equilibrium remain the same
involved in the reaction at specied kinetic laws forms as
as for the phase equilibrium without a chemical reaction
a result of chemical transformations and two phases seek-
(Morachevsky, 1989). The main distinction in the descrip-
ing a thermodynamic equilibrium. Phase equilibrium is of-
tion of such systems relates to the equations of balance lim-
ten calculated as a rough approximation, assuming, e.g., that
itations, expressing the laws of conservation of matter. For
the solubility of gas components in the liquid phase obeys
example, if no chemical reaction proceeds in the system,
Henrys law and does not depend on the solution compo-
the laws of conservation provide for a constant total number
sition, and fugacity of liquid components in the reaction is
of moles of each component distributed over the equilib-
negligible or described by Raoults law. In our opinion, these
rium phases. With a chemical reaction, the total number of
phase equilibrium problems of multicomponent mixtures
moles of each component will follow the laws of chemical
must be solved more accurately using the modern methods
kinetics and stoichiometry, depending on the reaction step.
of applied thermodynamics. In numerous publications, ther-
Phase equilibrium between gas and liquid forms stationary
modynamic methods for calculation of phase and chemical
concentrations of the reactants in the liquid phase, and thus
equilibrium successfully supplement the traditional meth-
the rate of chemical reaction in this phase. Phase equilib-
ods of mathematical modeling of chemical reactors and ow
rium also determines the amount and composition of gas
and liquid ows at the reactor outlet. For example, the ow
Corresponding author. Tel.: +7 383 239 7447. of dry gas fed to the reactor is saturated with vapor com-
E-mail address: anik@catalysis.nsk.su (V.I. Anikeev). ponents of the liquid phase in the gasliquid system, while

0009-2509/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2005.01.001
3200 A. Yermakova, V.I. Anikeev / Chemical Engineering Science 60 (2005) 3199 3206

the liquid phase dissolves components of the gas phase; as phases, gas and liquid. The total change of molar ows in
a result, the liquid-phase components are removed from the the reactor is described by the material balance equations
reactor. In some cases, this removal may be so signicant
that it prevents the occurrence of the stationary regime in Fi n0i ni + qi (x, T ) mcat = 0. (6)
a two-phase ow reactor in the region of actual process Here n0i = n0iG + n0iL is the molar ow of the ith component
parameters. fed to the reactor (g mol/s); ni = niG + niL is the same ow
The present work is aimed at analyzing the stationary at the reactor outlet; i = 1, 2, 3 relates to benzene, cyclohex-
states in an ideal backmixed reactor with liquid and gas ane and hydrogen. Indexes G and L indicate, respectively,
ows under the assumption that the stationary regime pro- the gas and liquid ows. Weight of the catalyst loaded to the
vides thermodynamic equilibrium of the phases at the reac- reactor is denoted as mcat . qi (x, T ) is the rate of the ith com-
tor outlet. ponent consumption/formation. It is apparent for reaction
The subject of our study is the reaction of liquid-phase (1) that q1 = r, q2 = r and q3 = 3r; x stands for the vec-
hydrogenation of benzene, which is currently one of the tor of molar fractions of the liquid-phase components. Con-
most thoroughly investigated reactions, its kinetics being centrations of benzene (cA ) and hydrogen (cH ), presented
presented as complex nonlinear equations (Temkin and in kinetic expression (2) in units of mol/l, are converted to
Murzin, 1989; Van Meerten et al., 1977; Van Meerten and corresponding molar fractions x1 and x3 using relations
Coenen, 1977; Toppinen et al., 1996). These kinetic equa-
tions were written out using detailed experimental data on x1 = cA vmL and x3 = cH vmL , (7)
the mechanism of elementary reactions in the liquid phase
proceeding on the slurry catalyst surface. where vmL is the molar volume of the liquid phase.
Normalization of Eqs. (6) to the total molar ow at the
reactor inlet
2. Kinetic model NS 
 
S0 = n0iG + n0iL
The equation of benzene hydrogenation in the liquid phase i=1
can be presented as follows:
converts Eqs. (6) to the dimensionless form
R, cat
C6 H6 + 3H2 C6 H12 . (1) Fi zi0 zi + qi (x, T )
= 0. (8)
According to Toppinen et al. (1996), the reaction rate of ben- 0 + z0 , z = n /S 0 = z + z ,
=
Here zi0 = n0i /S 0 = ziL iG i i iL iG
zene hydrogenation in the liquid phase on nickel catalyst in mcat /S (g s/mol).
0
the temperature range of 368398 K at pressure 2.04.0 MPa 0 , z0 and
are specied as the given parameters of the
ziG iL
is described by the kinetic equation problem, while normalized molar ows ziG and ziL form
k 1 K A K H cA cH due to chemical and phase transformations in the reactor.
r= (mol/s/gcat ), (2) In the case of phase equilibrium between gas and liquid,
[3KA cA + (KH cH )1/2 + 1]3
system (8) is supplemented by the equations characterizing

 
E 1 1 the equality of each component fugacities in both the phases:
k1 = 0.0013 exp (mol/s/gcat ), (3)
R T T0 FNS +i ln(ziL /SL ) + ln iL (x, T , P ) ln(ziG /SG )
T0 = 373 K, KA = 1.83 l/mol, KH = 7074.5 l/mol, ln iG (y, T , P ) = 0, i = 1, 2, 3. (9)
(4) Here iL (x, T , P ) and iG (y, T , P ) stand for the fugacity
E = 53 900 J/mol. (5) coefcients of the ith component in liquid and gas phases as
functions of molar composition of phase x, y, temperature
Here, cA is the benzene concentration and cH is the concen- and pressure. Fugacity coefcients are calculated using the
tration of hydrogen dissolved in the liquid phase, mol/l. equation of state. SL and SG in Eqs. (9) denote the sums
NS
 NS
1 
SL = ziL = niL ,
3. Mathematical model of the reactor S0
i=1 i=1
NS NS

3.1. Isothermal regime 1
SG = ziG = niG , (10)
S0
i=1 i=1
Consider the isothermal regime in the benzene hydrogena-
tion reactor at a constant pressure. Flows of liquid benzene while x and y are the NS -dimensional vectors of the molar
and gaseous hydrogen are fed to the reactor. Changes in the fractions of components in liquid and gas phases, respec-
total ow of components at the reactor outlet are caused by tively: xi = ziL /SL , yi = ziG /SG , NS = 3 is the number of
the chemical reaction. This ow is split into two equilibrium components.
 A. Yermakova, V.I. Anikeev / Chemical Engineering Science 60 (2005) 3199 3206 3201

The system of Eqs. (8) and (9) in isothermal isobaric All quantities in Eq. (12), due to the normalization made
regime at specied inlet ows ziG 0 and z0 and parameter
above, are related to 1 mol of the initial mixture fed to
iL
dene the problem completely. Unknown are ows ziL and the reactor. The value QCooler is calculated as a fraction of
ziG . Solution of the problem allows determination of each QReact + QEvap , that is
component ow and the equilibrium composition of phases
at the reactor outlet, x, y; the fractions of gas and liquid in the QCooler = (QReact + QEvap ), (16)
total outlet ow are WG = SG /(SL + SG ), WL = 1 WG ; the with the coefcient of proportionality 0

1. Apparently,
coefcient of variation in the total number of moles due to in isothermal regime,  =1, while in adiabatic regime,  =0.
chemical reaction is  =(SL +SG ). Conversions for benzene In adiabatic regime, Eq. (12) is transformed to equality
and hydrogen are determined as follows: of enthalpies of the inlet ow at temperature T 0 and of the
z10 z1 z10 outlet ow at temperature T ad :
XC6 H6 = , XH2 = 3 XC6 H6 . (11)
z10 z30 F2NS +1

NS
 NS

3.2. Nonisothermal regime = ziG H i (T ad , P , y) + ziL H i (T ad , P , x)
i=1 i=1

In the case of the nonisothermal regime of the reactor NS NS
 
0 0
operation, the system of Eqs. (8) and (9) is supplemented ziG Hi (T 0 , P , y0 ) + ziL Hi (T 0 , P , x0 )
by the heat balance equation: i=1 i=1
F2NS +1 = QHeat (QReact QCooler + QEvap ) = 0. (17)
= 0 (J/molinitial mix ), (12) Partial molar enthalpies of components in Eqs. (13)(17) are
where QHeat is the heat necessary for heating the initial calculated using the thermodynamic model presented below.
mixture from temperature T 0 to the reaction temperature T ;
QReact is the heat evolved as a result of chemical transfor-
4. Thermodynamic model
mations; QCooler is the average integral heat removed from
the reaction volume by a cooler; QEvap is the heat of the
Fugacity coefcients in Eq. (9) were calculated using
liquid gas phase transitions, a negative value for evapora-
the RedlichKwongSoave (RKS) equation of state (Soave,
tion of liquid components, and positive value for condensa-
1972):
tion. These heats can be calculated from the following rela-
tions: P = RT /(vm bm ) am (T )/(vm (vm + bm )), (18)
NS
 where am , bm are the mixture coefcients, and vm is its molar
QHeat = ziL
0
[H iL (T , x0 ) H iL (T 0 , x0 )]
volume.
i=1
Fugacity function fi based on this model has the form
NS
+ ziG
0
[H iG (T , y0 ) H iG (T 0 , y0 )], (13) ln fi (
) = ln[mi RT /(vm bm )] + i /(vm bm )

  
i=1 am i vm + bm bm
+ ln
NS
 RT b2m vm (vm + bm )
 
QReact =
HReact r
=
qi H iL (T , x), i vm + bm
ln , (19)
i=1 RT bm vm
(14)
where
where
HReact is the change in enthalpy of reaction (1)  
occurring in the liquid phase, i = 2 j aij , i = 2 j bij bm .
j
NS

QEvap = (ziG ziG
0
)
H i,Evap , (15) Coefcients am and bm for a mixture with specied molar
i=1 composition
are determined as follows:
where
H i,Evap is the change in enthalpy at the gasliquid Ns 
 Ns Ns 
 Ns
interface (J/mol). am = i j aij , bm = i j bij , (20)
Here H iL and H iG denote partial molar enthalpies of the j =1 i=1 j =1 i=1
ith component of liquid and gas, respectively, calculated at
appropriate temperature and pressure with regard to their aii = i (T )(0.42748R2 Tci2 /Pci ), (21)
dependence on the mixture composition. By denition of bii = 0.08664RTci /Pci , (22)
partial molar values, they have dimensionality of J/mol of
the ith component. aij = (1 kij ) aii ajj , (23)
3202 A. Yermakova, V.I. Anikeev / Chemical Engineering Science 60 (2005) 3199 3206

bij = (1 cij )(bii + bjj )/2, (24) 354

  2
i (T ) = 1 + di 1 T /Tci , (25)
353
di = 0.480 + 1.574i 0.1762i . (26)

Tb, K
Here Pci and Tci are the critical pressure and temperature of 352
the ith component of the mixture; i is the acentric factor.
In the above equations,
x for the liquid phase, and

y for the gas phase. Molar volumes vm for gas and 351
liquid phases are calculated directly from Eq. (18) as roots of
cubic polynomial. Fugacity coefcients of the components
of gas and liquid phases are determined as 350
0.0 0.2 0.4 0.6 0.8 1.0
iL (x, T , P ) = fi (x, T , P )/xi /P and Mole fraction
iG (y, T , P ) = fi (y, T , P )/yi /P . (27)
Fig. 1. Boiling temperature of benzenecyclohexane binary mixtures
Coefcients ki,j and cij in relations (23) and (24), allow- vs. benzene content. Solid linethe calculation by RKS EOS, dots
ing for the binary interactions of molecules, are determined experimental data (Sangster et al., 1988).
from the experimental data by the equilibrium in binary sys-
tems C6 H6 C6 H12 , H2 C6 H6 and H2 C6 H12 (Anikeev and
Yermakova, 1998). Because of this, the RKS model ade- known point on or near a homotopy path, a path tracking
quately describes the PVT properties of the mixture under algorithm follows the path by determining new points near
study. This is illustrated in Fig. 1, which shows the calcu- the path (Choi et al., 1996; Allgower and Georg, 1980). The
lated curve of boiling temperature for benzenecyclohexane algorithm presented below is based on the method of arc
binary mixtures at ki,j =0.2854 and ci,j =0.3059 vs. benzene length continuation.
was used as a homotopy parameter.
content in comparison with the experimental data (Sangster The method employs the two-step predictorcorrector al-
et al., 1988). One may see from Fig. 1 that the RKS equation gorithm with linear step predictor according to Euler and
describes the experimental data with high accuracy. step correction by Newtons method (Choi et al., 1996).
Partial molar enthalpies of individual components of liq- Determine the N -dimensional vector of unknown
uid and gas mixtures are determined with the use of the quantities with dimensionality N = 2NS + 1: u
known thermodynamic relations (Sandler, 1999) as [z1L , z2L , z3L , z1G , z2G , z3G , T ]T , the composite vector
with dimensionality N + 1: w = [u,
] and the vector-
j ln iL (T , P , x)
H iL (T , P , x) = Hi0 (T , P0 ) RT 2 , function F [F1 , F2 , . . . , F2NS +1 ]T . Then the system of
jT Eqs. (8), (9) and (12) is presented as
(28)
1 j ln iG (T , P , y) F(u,
) = 0. (30)
H iG (T , P , y) = Hi0 (T , P0 ) .
RT 2 jT The essence of the applied method consists in parametric
(29) representation of solution (30): u = u(s),
=
(s), where
Hi0 (T , P0 ) denotes molar enthalpies of the pure ith com- s is the arc length of a smooth spatial curve plotted from
ponent in the ideal gas state at temperature T and pressure the starting point (s = 0) to the current point. As a starting
P0 = 0.1 MPa. solution we use point T = T 0 ,
= 0 (at mcat = 0, S 0 = 0).
Under these conditions, the contacting gas and liquid ows
come to phase equilibrium without chemical transformation.
5. Numerical analysis of the model The algorithm for solving this problem at such a statement
(system of equations (9)) is well known (Michelsen, 1982).
The system of Eqs. (8), (9) and (12) was calculated with Consecutive steps of the predictor in iterations are found
respect to unknown values ziL , ziG and T at specied pres- by solving the system of linear equations:
sure and values of inlet ows. Solutions were searched in  jF 
parametric form depending on parameter
. (s k )
jw vk = eN+1 (31)
T
(ej )
5.1. The homotopy algorithm
and
It is known that the homotopy continuation methods nd
jw k vk
roots of systems of nonlinear equations by tracking the ho- (s ) = k k , (32)
motopy path from one or multiple starting points. Given a js v 2
 A. Yermakova, V.I. Anikeev / Chemical Engineering Science 60 (2005) 3199 3206 3203

dw k
k+1 = wk +
s k
w (s ), (33) 100
4 3 2
ds 1
90
where k is the iteration number; ej (1
j
N + 1) is the 80
(N + 1)-dimensional unit vector, its j th element being equal 2
70
to unity and all other elements equal to zero. Parameter
k = 1 denes the path direction, and parameter
s > 0

Conversion, %
60
controls the length of admissible step. An approximate so- 50 1
lution w k+1 predicted by the predictor is determined more
40
accurately by Newtons method. If it is rather close to curve
F (u(s),
(s)) = 0, convergence of Newtons method to the 30
true solution wk+1 is ensured and obtained by 56 iterations. 20
10
5.2. Conditions of existence of a solution
0
0 1 2 3 4 5 6 7
Within the model under consideration, the stationary mode (a) Residence Time, g*h/mole
of the reactor operation, which provides existence of a so-
lution of set (30), takes place only when two equilibrium 100
phases occur at the reactor outlet. In this case, the equilib- 90
rium value of the fraction of a phase, for example, of the 80
gas phase WG must comply with the condition
70
Conversion, %

0 < WG < 1. (34) 60 1

2
The occurrence of condition WG = 0 in the outlet ow is 50
3
possible when the entire gas fed to the reactor is dissolved 40 4
5
in the liquid phase. In the other extreme case, when WG = 1, 30
the liquid phase is completely removed from the reactor due
20
to its evaporation to dry gas.
Condition (34) should also be met for the set of equations 10
(30) at
= 0. In the example under consideration, the fulll- 0
ment of this condition consists in nding the admissible limit 0 1 2 3 4 5
of the coefcient of hydrogen excess  at the reactor inlet, (b) Residence Time, g*h/mole
taken to be unity at the stoichiometric ratio n03 /n01 = 3. In-
Fig. 2. Benzene conversion vs. residence time (
). (a) P = 2.0 MPa.
vestigation of the starting solution showed that as  changes T /K: 1343; 2353; 3363; 4373.
1 ,
2 turning points of the
from 1 to 3 (T 0 = 368 K and P = 2.0 MPa), the WG value curves. (b) T = 363 K: P /MPa: 11.0; 22.0; 33.0; 44.0. Dotted
at
= 0 varies from 0.818 to 0.983, i.e., complies with con- line 5hydrogen solubility in the liquid phase calculated by a simplied
dition (34). model according to Henrys law, P = 1.0 MPa.

6. Results of numerical analysis
6.1. Isothermal regimes
At the very beginning, it should be noted that dimension-
ality of the system of equations for the problem stated above
for the isothermal case at T = T 0 = const decreases by unity
(N = 2NS ), and the heat balance equation is excluded from
consideration. Parameter
was chosen as the leading pa-
rameter of continuation, so that j = N + 1 in Eq. (31).
Fig. 2(a) displays the dependencies of benzene conversion
on parameter
at  = 1.2, different values of T and P =
2.0 MPa. Similar dependencies at variation of P and xed
T = 363 K are presented in Fig. 2(b).
The gures show that at certain conditions, for exam-
ple at P = 2.0 MPa, T = 343373 K or at T = 363 K,
P = 0.1 0.2 MPa, the X
curves are S-shaped and there
are three stationary solutions in the interval 
1 ,
2  at the
same value of
, where
1 and
2 denote the values of

at the turning points of the curves. The solutions lying on
the curve branches between two turns are unstable. Note
that these solutions cannot be obtained when other methods
are used instead of the homotopy method, for example,
the determination method. The width of the interval of
existence of unstable solutions 
1 ,
2  is dependent on
pressure and temperature. The region of multiple solutions
may gradually disappear with increasing temperature and
pressure.
To explain possible causes of the emergence of multiple
stationary solutions, consider the dependencies of the reac-
tion rate (Fig. 3(a)) and equilibrium concentration of hydro-
gen in the liquid phase (Fig. 3(b)) on benzene conversion at
different temperatures and P = 2.0 MPa. One may see from
Fig. 3(b) that the concentration of hydrogen in the liquid
3204 A. Yermakova, V.I. Anikeev / Chemical Engineering Science 60 (2005) 3199 3206

0.22 4 25
0.20 4
3
0.18
20 2
Reaction Rate, mole/h/g

0.16
0.14 5 3
0.12 15 1

Conversion, %
0.10
0.08 2
10
0.06
1
0.04
5
0.02
0.00
0 20 40 60 80 100 0
(a) Conversion, % 0.00 0.02 0.04 0.06 0.08 0.10 0.12
(a) Residence Time, g*h/mol
1.40
4
1.30 3 440 4
1.20 2
1 3
1.10
420
X hydrogen, %

1.00
0.90 2
0.80 400
0.70
T, K

0.60 1
380
0.50
0.40
0.30 360
0 20 40 60 80 100
(b) Conversion, %
340
Fig. 3. (a) Reaction rate vs. benzene conversion. P = 2.0 MPa. T /K: 0.00 0.02 0.04 0.06 0.08 0.10 0.12
1343; 2353; 3363; 4373. Dotted line 5hydrogen solubility in
the liquid phase calculated by a simplied model according to Henrys (b) Residence Time, g*h/mol
law, P = 1.0 MPa, T = 363 K. (b) Hydrogen solubility in the liquid phase
vs. benzene conversion. P = 2.0 MPa. T /K: 1343; 2353; 3363; Fig. 4. Adiabatic regimes. Benzene conversion (a) and temperature of
4373. adiabatic heating (b) vs. residence time
at initial temperature T 0 =343 K.
P /MPa: 11.0; 22.0; 33.0; 44.0.

phase passes through a minimum with increasing benzene
conversion due to reaction (1). Such behavior of the hydro-
gen solubility in the benzenecyclohexane mixture is caused
by the presence of an azeotropic point at ca. 50% benzene
content in the liquid (Fig. 1).
A complicated nature of the reaction rate dependence on
benzene conversion may be explained by the nonlinear ki-
netic equation of the reaction rate (2), on the one hand, and
by the extremal dependence of hydrogen solubility in the
benzenecyclohexane mixture. To estimate the contribution
of each of these components to the rate variation, the equa-
tion of hydrogen equilibrium in system (8) was replaced by
the linear equation representing Henrys law. In this case,
the concentration of hydrogen in the liquid phase at constant
T and P is independent of the mixture composition. Dot-
ted lines in Figs. 2(b) and 3(a) show the calculation results
in comparison with the previously obtained data. Compari-
son of the
dependencies of benzene conversion shows that
idealization of hydrogen solubility leads to disappearance of
the stationary states multiplicity.
6.2. Nonisothermal regimes
First consider the adiabatic regime at a zero value of
heat removal parameter  found from Eq. (16). The calcu-
lation results are presented in Figs. 4(a) and (b) as benzene
conversion (Fig. 4(a)) and temperature of adiabatic heating
T ad (Fig. 4(b)) vs. parameter
at the initial temperature
Ti0 = 343 K and several pressure values. Similar to the case
of isothermal regimes, multiplicity of solutions is also ob-
served here. Unstable solutions lie on the branches of curves
following the turn, and terminate when the liquid-phase ow
disappears due to its complete evaporation into the gas phase.
Turning points at low conversions (
=
sing ) are the singular
points of the problem, i.e., when the matrix determinant in
Eq. (31) vanishes. In calculations by the homotopy method,
 A. Yermakova, V.I. Anikeev / Chemical Engineering Science 60 (2005) 3199 3206 3205

3 heat may appear to be the most practical regime. Figs. 5(a)

100 4 5 and (b) show the dependencies of benzene conversion
90 (Fig. 5(a)) and temperature (Fig. 5(b)) on parameter
at
various values of heat removal coefcient . One may see
80
that at insufcient heat removal, when 
0.8, the reactor
70 temperature even at low conversion attains the values at
2
which the liquid-phase ow disappears, i.e., a complete
Conversion, %

60
ow of liquid reactants evaporates into the gas phase. Sta-
50 tionary regimes possibly occur only at  > 0.8, i.e., while
approaching the isothermal regime.
40

30 7. Conclusions
20 1
The above model, assuming the phase equilibrium condi-
10 tion, is one of the approximations to the model of the system
taking into account the interphase heat and mass transfer.
0.01 0.10 1.00 10.00 100.00
Analysis showed that this equilibrium model provides use-
ful information on the process character, amount and com-
(a) Residence Time, g*h/mol
position of gas and liquid ows at the reactor outlet. In the
case of strongly nonlinear dependencies of the reaction rate
460 3 on the reactants concentration, which are typical of some
liquid-phase catalytic reactions, the multiplicity of station-
ary regimes may be expected. The multiplicity of stationary
440
regimes and the presence of unstable solutions region for
2 4 the chosen reaction of benzene hydrogenation are caused by
420 nonlinear dependence of the reaction rate on the reactants
1 concentration in the liquid phase, on the one hand, and by the
behavior of hydrogen solubility in the benzenecyclohexane
T, K

400 mixture, on the other hand.

5
380
360
340
0.01 0.10 1.00 10.00 100.00
(b) Residence Time, g*h/mol
Fig. 5. Regimes with partial heat removal. Benzene conversion (a) and
temperature (b) vs. residence time
at initial temperature T 0 = 343 K,
P = 2.0 MPa. Heat removal parameter : 10; 20.8; 30.85; 40.90;
50.95.
transition through this point is made by the substitution of
temperature T for the leading parameter of continuation

, i.e., by the substitution of a unit vector (ej )T j = N for
the lower row of matrix (31). Since the liquid-phase ow
disappeared, it was impossible to obtain the second point of
the curve turn and the upper branch of a stable stationary
solution in adiabatic regime at high conversion (similar to
isothermal regime, see Fig. 2).
Analysis of the model of benzene hydrogenation reactor
in adiabatic regime shows that in practice it is impossible to
obtain stationary solutions at high conversion. So a regime of
such reactor operation with a partial removal of the reaction
References
Allgower, E., Georg, K., 1980. Simplical and continuation methods for
approximating xed points and solutions to systems of equations.
SIAM Review 22 (1), 2885.
Anikeev, V.I., Yermakova, A., 1998. Coefcients of binary interaction in
the RKS equation of state for mixtures with supercritical components.
Theoretical Foundation of Chemical Technology 32 (5), 508514.
Bertucco, A., Canu, P., Devetta, L., Zwahlen, G., 1997. Catalytic
hydrogenation in supercritical CO2 : kinetic measurements in a
gradientless internal-recycle reactor. Industrial Engineering and
Chemistry Research 36 (7), 26262633.
Choi, S.H., Harney, D.A., Book, N.L., 1996. A robust path tracking
algorithm for homotopy continuation. Computers & Chemical
Enginerring 20 (67), 647655.
Michelsen, M.L., 1982. The isothermal ash problem. Part II. Phase split
calculation. Fluid phase equilibrium 9, 2140.
Morachevsky, A.G., 1989. Thermodynamics of the LiquidVapor
Equilibrium. Chemistry, Leningrad.
Sandler, S.I., 1999. Chemical and Engineering Thermodynamics. Wiley,
New York.
Sangster, P., Talley, K., Bale, C.W., Pelton, A.D., 1988. Coupled
optimisation and evaluation of phase equilibria and thermodynamic
properties of the benzenecyclohexane system. Canadian Journal of
Chemical Engineering 66 (6), 881896.
Shinnar, R., 1988. Thermodynamic analysis in chemical process and
reactor design. Chemical Engineering Science. Special issue ISCRE
10. Chemical Reaction Engineering 43 (8), 23032318.
3206 A. Yermakova, V.I. Anikeev / Chemical Engineering Science 60 (2005) 3199 3206
Soave, G., 1972. Equilibrium constants from a modied R.K. equation
of state. Chemical Engineering Science 27 (6), 11971203.
Sophos, A., Rotstein, E., Stephanopoulos, G., 1980. Thermodynamic
bounds and the selection of technologies in the petrochemical industry.
Chemical Engineering Science 35 (5), 10491065.
Temkin, M.I., Murzin, D.Yu., 1989. Mechanism of the liquid-phase
hydrogenation of benzene ring. Kinetics and Catalysis 30 (3),
637643.
Toppinen, S., Rantakyla, T.K., Salmi, T., Aittamaa, J., 1996. Kinetics of
the liquid-phase hydrogenation of benzene and some monosubstituted
alkylbenzenes over a nickel catalyst. Industrial Engineering and
Chemistry Research 35 (6), 18241833.
Van Meerten, R.Z.C., Coenen, J.W.E., 1977. Gas phase benzene
hydrogenation on a nickel-silica catalyst IV. Rate equations and curve
tting. Journal of Catalysis 46 (1), 1324.
Van Meerten, R.Z.C., de Graaf, T.F.M., Coenen, J.W.E., 1977. Gas
phase benzene hydrogenation on a nickel-silica catalyst III. Low-eld
magnetization measurements on a hydrogen, benzene, cyclohexene
and cyclohexane adsorption, and benzene hydrogenation. Journal of
Catalysis 46 (1), 112.
Yablonsky, G.S., Bykov, V.I., Gorban, A.N., 1983. Kinetic models of
catalytic reactions. Nauka, Novosibirsk.