570

Energy & Fuels 2008, 22, 570575

Desulfurization of Liquid-Phase Butane by Zeolite Molecular Sieve

13X in a Fixed Bed: Modeling, Simulation, and Comparison with
Commercial-Scale Plant Data
A. Shams, A. Molaei Dehkordi,* and I. Goodarznia
Department of Chemical and Petroleum Engineering, Sharif UniVersity of Technology,
Tehran 11155-9465, Iran
ReceiVed August 19, 2007. ReVised Manuscript ReceiVed October 16, 2007

This paper deals with the modeling and simulation of binary liquid-phase adsorption of methyl mercaptan
and hydrogen sulfide from a liquid butane stream by zeolite molecular sieve 13X in a fixed bed. The model
equations account for the effect of axial dispersion and the inter- and intraparticle mass-transfer resistances at
isothermal operating conditions. Orthogonal collocation and Gills fourth-order RungeKutta methods were
used to solve the dimensionless general forms of the 4N-coupled ordinary differential equations for simultaneous
adsorption of the solutes by the adsorbent in a fixed bed. The model predictions were compared to the
commercial-scale plant data of an Iranian petrochemical company (Bandar Imam, Iran), and a fair agreement
was observed for the fixed-bed absorber of 1.7 m in diameter and 5.06 m in height. Moreover, the influences
of the bed Reynolds number, Peclet number, total inlet sulfur concentration, feed temperature, and diameter
of the spherical adsorbent on the breakthrough curve were investigated.

1. Introduction
Adsorption is widely used in the chemical industry for the
separation of solutes from a fluid stream onto a solid surface,
i.e., adsorbent.1 The single-component adsorption from a stream
has been modeled successfully, but such modeling development
for multicomponent systems that have to be applied for industrial
process is scarce. Moreover, the behavior of breakthrough curves
of a multicomponent system is totally different from that of
single-component system. This is because that the adsorption
behavior of various kinds of solutes is entirely different.
The removal of low-concentration sulfur compounds from
hydrocarbon or fuel streams by adsorption onto an adsorbent
in fixed beds is an important commercial treatment process. The
traditional process for the desulfurization of natural gas or
liquefied petroleum gas (LPG) may not reduce the sulfur
contents as low of concentrations.2 Moreover, the literature data
concerning the adsorption of the mercaptans are scarce.3 On
the other hand, the applicability of activated carbon for the
desulfurization of mercaptans strongly depends upon the surface
characteristics and the porous texture.4 Nevertheless, there is a
lot of data regarding the desulfurization of light hydrocarbons
by means of activated carbon,2 but such literature data are scarce
for zeolite molecular sieves; e.g., the adsorption of ethyl
mercaptan and n-butyl mercaptan on MFI zeolite was studied
by Garcia et al.5 and Ruthven et al.6 that presented an
* To whom correspondence should be addressed. Telephone: +98-2166165412. Fax: +98-21-66022853. E-mail: amolaeid@sharif.edu.
(1) Yang, R. T. Adsorbents: Fundamentals and Applications; John Wiley
and Sons: New York, 2003; pp 17.
(2) Weber, G.; Bellat, J. P.; Benoit, F.; Paulin, C. Adsorption 2005, 11,
183189.
(3) Bashkova, S.; Bagreev, A.; Bandosz, T. J. EnViron. Sci. Technol.
2002, 36, 27772782.
(4) Bashkova, S.; Bagreev, A.; Bandosz, T. J. Ind. Eng. Chem. Res.
2002, 41, 43464352.
(5) Garcia, C. L.; Lercher, J. A. J. Phys. Chem. 1991, 95, 1072910736.

equilibrium model for multicomponent sorption into zeolite

molecular sieve. Generally, in most commercial-scale processes,
the intraparticle diffusion phenomenon is the rate-limiting step
that controls the overall process and the sharpness of the
breakthrough curve.7
The adsorption of a single sulfur compound from an aqueous
solution onto an absorbent has been successfully modeled,8 but
such a modeling strategy for multicomponent adsorption of
mercaptans is scarce. The breakthrough curves of multicomponent systems are different relative to those of singlecomponent systems. For example, if the feed contains a weakly
adsorbed component 1 and strongly adsorbed component 2 in
a fixed bed of adsorbent, the component 1 can be replaced by
the stronger one and, hence, the component 1 goes faster than
the component 2 within the column.9 The various models
concerning the multicomponent liquid-phase adsorption in fixed
beds were presented, but none of them described the behavior
of our chemical system.10-12
As far as we searched, there is no published work on the
binary desulphurization of liquid-phase butane stream by means
of zeolite molecular sieve 13X in a fixed bed. Thus, the main
objectives of the present investigation were to (1) model the
adsorption of methyl mercaptan and hydrogen sulfide from a
liquid-phase butane stream in a fixed bed of zeolite molecular
sieve 13X, (2) study the influences of various parameters on
(6) Ruthven, D. M.; Loughin, K. F.; Holborow, K. A. Chem. Eng. Sci.
1973, 28, 701709.
(7) Ma, Z.; Whitley, R. D.; Wang, N. H. L. AIChE J. 1996, 42, 1244
1262.
(8) Webi, T. W.; Charavort, R. K. AIChE J. 1974, 20, 228238.
(9) Yang, R. T. Gas Separation by Adsorption Processes; Imperial
College Press: London, U.K., 1998; pp 173175.
(10) Cooney, D. O.; Strusi, F. P. Ind. Eng. Chem. Fund. 1972, 11, 123
126.
(11) Olafadehan, O. A.; Susu, A. A. Ind. Eng. Chem. Res. 2004, 43,
81078116.
(12) Kaczmarski, K.; Mozotti, M.; Storti, G.; Morbidell, M. Comput.
Chem. Eng. 1997, 21, 641660.

10.1021/ef700500x CCC: $40.75 2008 American Chemical Society

Published on Web 12/06/2007

Desulfurization of Liquid-Phase Butane

Energy & Fuels, Vol. 22, No. 1, 2008 571

the breakthrough curve, and (3) compare the model predictions

with commercial-scale plant data.

The model describes the adsorption of solutes from an upward

liquid stream in a fixed bed of zeolite particles and takes into
account the axial diffusion, external mass-transfer resistance, and
intraparticle mass transfer of solutes with the following essential
assumptions: (a) isothermal conditions, (b) constant coefficients of
the dispersion and diffusion phenomena, (c) constant porosities of
the particle and bed, (d) neglecting radial distribution of solutes in
the fixed bed, (e) neglecting radial variations of fluid velocity in
the bed, i.e., dispersed-plug flow, (f) taking dilute solution, and (g)
taking the uniform size and homogeneous structure of the spherical
adsorbent particles.
Then, the mass balance applied to each solute in the flowing
liquid phase in the column (the macrosystem) leads to the following
equations:

( )

Ci
Ci
2Ci
1 - b qi
+ U
- DLi 2 +
) 0
t
z
b
t
z
(i ) CH3SH (1) and H2S (2)) (1)
where Ci, U, z, DLi, b, and qi are the concentration of solutes,
superficial velocity of the flowing fluid in the fixed bed, z direction,
axial diffusivity of solute i, porosity of the fixed bed, and mean
adsorbed amount of solute i, respectively. The following initial and
boundary conditions complete the macrosystem mass balance:

DLi

(i ) 1, 2); t e 0, 0 e z e L

Ci
) U (Ci - Cfi)
z
Ci
) 0
z

(i ) 1, 2); t > 0, z ) L

(1 + a)
kfi
R

(i ) 1, 2)

(5)

(i ) 1, 2)

(6)

q/i
) 0
t

(i ) 1, 2)

(7)

pDpi

(i ) 1, 2); t e 0, r, 0 e z e L

Yi
) kfi(Ci - Yi)
r

(11)

Cts )

(12)

i)1

To solve the governing equations, it is recommended to transform

these equations into dimensionless forms. Thus, the model equations
(eqs 1 and 7) can be transformed into the following dimensionless
equations:
2
Ci
1 Ci Ci (1 - b) 3 kfiL
)
(C - Yi)
Pei 2

b
R U i

()

(i ) 1, 2)
(13)

where
Ci ) Ci / Cfi , ) tU / L, ) z / L, Pei ) UL / DLi ,
Yi ) Yi / Cfi (14)

Ci ) 0
Ci ) 1 +

(i ) 1, 2), e 0, 0 e e 1

( )

1 Ci
Pei

Ci
)0

(i ) 1, 2), > 0, ) 0

(i ) 1, 2), > 0, ) 1

(15)
(16)

(17)

The dimensionless form of the mass balance applied to the adsorbate

in the pore fluid for spherical pellets becomes

[(

) (

Yi
2Y1 2 Y1
2Y2 2 Y2
)
Mi
+
+
S
+
i
i
2
2

)]

(i ) 1, 2)
(18)

where
i )

4Lkfi
Bii
U

gi ) p +

(i ) 1, 2)

/
q/i
q0i
Cfi Y

(i ) 1, 2)

(19)
(20)

g3 )

/
q2/
q02
Cf2 Y

(21)

/
q1/
q01
Cf1 Y

(22)

where p, r, Dpi, and q/i are the porosity of the adsorbent, radial
distance in the adsorbent particles, diffusivity coefficient of solute
i within the adsorbent particles, and instant adsorbed amount of
solute i onto the adsorbent, respectively.
The initial and boundary conditions of eq 7 are as follows:
Yi ) 0

(i ) 1, 2)

When the appropriate isotherm equations (q/i ) are taken for each
adsorbate and eqs 111 are solved, the concentration profiles of
solutes, i.e., methyl mercaptan and hydrogen sulfide, can be
obtained. Then, the total sulfur concentration (Cts) can be determined
by

(4)

where kfi and a are the film mass-transfer coefficient and form factor,
respectively.
The mass balance applied to the adsorbates in the pore fluid gives

(3)

where ki, Yi, and R are the mass-transfer coefficient, concentration

of solute i within the zeolite particles, and radius of particles,
respectively. The mass-transfer coefficient, ki, may be given by13

p 2 Yi
Yi
- 2
r Dpi
t
r
r r

(10)

Subject to the transformed initial and boundary conditions

(i ) 1, 2); t > 0, z ) 0

qi
) ki(Ci - Yi)r ) R
t

p

2
q/i
q/i Yj
)
t
Yj t
j)1

(2)

where Cfi and L denote the concentration of solute i in the flowing

fluid and height of the fixed bed of zeolite molecular sieve,
respectively.
The last term of eq 1 can be replaced by a linear driving force
(LDF) model as13

ki )

(i ) 1, 2); t > 0, r ) R, 0 e z e L

On the other hand, we have

2. Mathematical Model

Ci ) 0

Yi
) 0
r

(8)

(i ) 1, 2); t > 0, r ) R, 0 e z e L (9)

(13) Tien, C. Adsorption Calculation and Modeling; Butterworth Heinemann: Boston, MA, 1994.

g4 )

1 ) g3(g1 - g4)(g2 - g3)

(23)

2 ) g4(g3 - g2)

(24)

M1 ) g2g31

(25)

M1
g (g - g4)
1 3 1

(26)

S1 ) 2(g3g4 - g4g2 - g1g3)

(27)

M2 ) g41 -

572 Energy & Fuels, Vol. 22, No. 1, 2008

Shams et al.

S1
g (g - g4) - 2g4
1 3 1

S2 )

)
Bii )

r
R

(28)
(29)

(i ) 1, 2)

(30)

A restatement of the initial and boundary conditions in terms of

the dimensionless variables yields
Yi ) 0

(i ) 1, 2); e 0, 0 e e 1, 0 e e 1

( )

4 Yi
) (Ci - Yi)
Shi

2Aj,1
(MiY1,1 + SiY2,1)
j

(i ) 1, 2); (j ) 1, 2, ..., NR)

(39)

()

kfiR
pDpi

i,j ) Bj,1 +

(31)

where NR is the number of collocation points in the radial direction

inside the zeolite particle. The initial and boundary conditions
become

4
Shi

Yi,j ) 0

NR+1

e 0, (i ) 1, 2); (j ) 1, 2, ..., NR + 1)

AN+1,kYi,k ) (Ci,j - Yi,NR+1)

k)1

(40)

> 0, (i ) 1, 2);
(j ) 2, 3, ..., NR + 1) (41)

where Shi ) 2kfiR/Dpi.

Equations 13 and 18 represent a set of four coupled, stiff,
nonlinear partial differential equations that can not be solved
analytically but must be integrated numerically by any known
method that is stiffy stable.

Notice that the boundary points ) 0 and 1 are the external

collocation points in eq 13, whereas the boundary point ) 1 is
only taken as an external collocation point in eq 18. Further details
of the solution method were found elsewhere.16 Therefore, the
problem has now been reduced to solving simultaneously the set
of nonlinear differential equations with the initial conditions and
using appropriate equilibrium isotherms. The Aj,k and Bj,k parameters
were evaluated on the basis of the weighting function, i.e., R ) 0
and ) 0.17 Moreover, the approximation order of each orthogonal
polynomial concerning the equations of the flowing fluid and
micropore diffusion was 4.

3. Solution Methodology

4. Results and Discussion

The governing partial differential equations, eqs 13 and 18, could

be solved by the orthogonal collocation method.14 According to
this method, the partial differential equations could be reduced to
4N-coupled ordinary differential equations that were solved by
Gills fourth-order RungeKutta method.15 According to the
orthogonal collocation method, the gradient and laplacian operators
for the variables Ci and Yi of eqs 13 and 18 are given by
polynomial functions. Applying this method to variables and ,
eq 13 yields

4.1. Equilibrium Isotherms. The equilibrium isotherms

determined for the binary liquid system, i.e., methyl mercaptan
and n-butane and hydrogen sulfide and n-butane on zeolite 13X
were given by18

Yi
)0

dCi,j
1
)
d
Pei

(i ) 1, 2); > 0, ) 1, 0 e e 1 (32)

(i ) 1, 2); > 0, ) 1, 0 e e 1

N+1

j,kCi,k

k)2

(33)

j,kCi,k + (Bj,1 - Aj,1)Ci,1 + (Bj,N+2 -

k)2

(i ) 1, 2);
(j ) 2, 3, ..., N + 1) (34)

where i ) ((1 - b)/b)(3/R)(kfiL/U) and N is the number of interior

collocation points in the axial direction. The initial and boundary
conditions are transformed as follows:
e 0, 0 e e 1, (i ) 1, 2); j ) 2, 3, ..., N + 2) (35)
1
Pei

Ci,1 ) 1 +

( )
N+2

(i ) 1, 2); > 0, ) 0

A1,kCi,k

(36)

k)1

N+2

(i ) 1, 2); > 0, ) 1

N+2,kCi,k ) 0

(37)

k)1

Also, applying orthogonal collocation to the system model equations

for the radial direction of spherical particle yields

[ (

dYi,j j
)
M
d
i i

2
j

NR+1

k)2
NR+1

2
Bj,kY1,k +
j

NR+1

) ]

j,kY2,k

k)2

0.165e
1 + 0.0139e

(2293 /

k)2

+ i,j

) (

j,kY1,k

NR+1

+ Si

Bj,kY2,k +

k)2

(i ) 1, 2); (j ) 1, 2, ..., NR) (38)

(14) Richard, G. R.; Do., D. D. Applied Mathematics and Modeling

for Chemical Engineers; John Wiley and Sons: New York, 1995; pp
290296.
(15) Gerald, C. F.; Wheatley, P. O. Applied Numerical Analysis;
Addison-Wesley: New York, 1999; pp 477486.

q/2 )

1 + 0.0139e

Y1

Y1 + 0.062e

0.176e
(2293 /

T)

T)

(1845 /

N+1

Aj,N+2)Ci,N+2 - i(Ci,j - Yi,N+1)

Ci,j ) 0

(2293 /

q/1 )

(1845 /

Y2

(42)

T)

Y2

Y1 + 0.062e

T)

T)

(1845 /

T)

Y2

(43)

In the present work, because the feed was a dilute solution and
the amount of adsorbed solvent may be constant, thus, the
Langmuir equilibrium isotherms were used.
4.2. Parameters of the Model. To solve the model equations,
we need the thermophysical properties of the feed. The density
of the feed was determined by the PengRobinson equation of
state.19 The axial dispersion coefficient, DLi, was calculated using
the Ruthven correlation,20 whereas the diffusivities in the bulk
of fluid were estimated by the Wilke-Chang equation.21
Furthermore, the pore diffusivities, Dpi, were taken to be constant
in the present work and determined using the Suzuki correlation.22 The film mass transfer coefficients, kfi, were determined
using the correlations developed by Ruthven et al.20 All of these
estimated parameters are summarized in Table 1.
4.3. Commercial-Scale Plant Data. The specifications of
the feed introduced into the fixed bed of zeolite 13X at the
(16) Liapis, A. I.; Rippin, D. W. Chem. Eng. Sci. 1977, 32, 619627.
(17) Villadsen, J. V.; Stewart, W. E. Chem. Eng. Sci. 1967, 22, 1483
1501.
(18) Ruthven, D. M. Principle of Adsorption and Adsorption Processes;
John Wiley and Sons: New York, 1995; pp 106122.
(19) Smith, J. M.; Vanness, H. C.; Abbott, M. M. Introduction to
Chemical Engineering Thermodynamics; McGraw-Hill: Boston, MA, 2001;
pp 98100.
(20) Ruthven, D. M. Encyclopedia of Separation Technology; John Wiley
and Sons: New York, 1997.
(21) Treybal, R. E. Mass Transfer Operations; McGraw-Hill: New York,
1990; pp 3337.
(22) Suzuki, M. Adsorption Engineering; Kodansha Ltd., Tokyo, Japan,
1989; pp 6467.

Desulfurization of Liquid-Phase Butane

Energy & Fuels, Vol. 22, No. 1, 2008 573

Table 1. Physical Properties of the Sulfur Components at 311 K

physical quantity

methyl mercaptan (CH3SH)

hydrogen sulfide (H2S)

Dpi 107 (m2/s)

DLi 105 (m2/s)
kfi 104 (m/s)

0.92
2.28
3.02

1.25
2.23
3.15

Table 2. Specification of the Liquid Butane Stream

concentration of CH3SH (kg/m3) 103
concentration of H2S (kg/m3) 103
temperature (K)
pressure (bar)
density (kg/m3)
viscosity (kg m-1 s-1) 104

7.895
1.491
311
15.5
553.3
1.47

Table 3. Specification of the Fixed Bed

packed bed height, L (m)
bed diameter, dB (m)
feed temperature, T (K)
operating pressure (bar)
superficial velocity, U (m/s)
Reynolds number of bed
porosity of bed, b
porosity of the zeolite, p
radius of the zeolite particle (m)

5.06
1.7
311
15.5
0.0139
89 000
0.36
0.9
0.0016

Figure 1. Comparison of model predictions with commercial-scale plant

data. Cts ) 0.009 386 kg/m3; Re ) 89 000; T ) 311 K; Pe ) 3070.

Bandar Imam Petrochemical Company (Iran) are summarized

in Table 2, whereas the dimensions of the packed bed and the
specifications of the zeolite molecular sieve 13X are presented
in Table 3. It should also be noted that, because the adsorption
cycle of the butane sweetening columns of the Bandar Imam
Petrochemical Company is 12 h, thus, the simulation results
were presented for this period of adsorption time.
4.4. Model Validation. The Biot number (kfR/bDpi) representing the rate of the solid-phase mass-transfer resistance to
the fluid-phase mass-transfer resistance was evaluated to be
154.39 and 118.48 for the methyl mercaptan and hydrogen
sulfide, respectively. These values of the Biot number clearly
show that the solid-phase mass-transfer resistance is much
greater than that of the fluid-phase mass-transfer resistance for
both solutes.
Figure 1 demonstrates the variations of the total sulfur
concentration of outlet stream of the commercial-scale packed
bed with time (breakthrough curve), as well as the predicted
breakthrough curves for methyl mercaptan, hydrogen sulfide,
and total sulfur. As may be observed from this figure, there is
a fair agreement between the plant data and those predicted by
the model. In addition, this figure shows the maximum deviation
15% between the plant data and model predictions.
4.5. Influence of Bed Reynolds Number. Figure 2 demonstrates the effect of the bed Reynolds number (volumetric flow
rate of feed) on the breakthrough profiles for the total inlet sulfur

Figure 2. Effect of the bed Reynolds number on the breakthrough curve.

Cts ) 0.009 386 kg/m3; T ) 311 K.

concentration of Cts ) 0.009 386 kg/m3. As may be noticed

from this figure, an increase in the bed Reynolds number from
58 000 to 142 000 causes a decrease in the breakthrough time
from 11.76 to 6.86 h. This can be explained by a decrease in
the mean residence time of the fluid within the column, which
is not large enough to reach the equilibrium adsorption. On the
other hand, the breakthrough curve becomes steeper with an
increase in the volumetric flow rate of feed, but for Reynolds
numbers larger than 984 000, the slope of the breakthrough
curve remains almost constant. The latter may be attributed to
the different effects of increasing the volumetric flow rate of
feed such that (1) at higher flow velocities the film surrounding
the particles becomes thinner, thereby reducing the external
mass-transfer resistance and causing the breakthrough curve to
become steeper, and (2) axial dispersion of the fluid flow is
enhanced at higher Reynolds numbers, which would decrease
the slope of the breakthrough curve. The breakthrough curve is
more sensitive to mass-transfer resistance relative to axial
dispersion at lower fluid velocities, but the effect of axial
dispersion increases for a much larger volumetric flow rate of
feed.
4.6. Influence of the Peclet Number. The effect of the bed
height on the breakthrough curve of a fixed-bed absorber was
studied by varying the Peclet number, while the volumetric flow
rate of feed was kept constant. Figure 3 shows the variation of
the breakthrough curve with the Peclet number. As may be
observed, as the value of the Peclet number decreases from 3640
to 1820, the breakthrough time decreases from 11.64 to 5.88 h
and the breakthrough curve becomes steeper. This shows that
the outlet concentration of total sulfur will quickly reach the
inlet concentration at smaller bed heights. This behavior also
shows that the steepness of the breakthrough curve is strongly
dependent upon the bed height.
4.7. Influence of the Inlet Sulfur Concentration. The effect
of the inlet sulfur concentration on the breakthrough curve was
shown in Figure 4. Three inlet concentrations considered for
the total sulfur were 0.007 306, 0.009 386, and 0.014 05 kg/
m3. Notice that, during these simulations, other parameters,
including the Peclet number and Reynolds number, were kept
constant. As may be observed from this figure, as the inlet sulfur
concentration increases from 0.009 386 to 0.014 05 kg/m3, the
break point time decreases from 9.84 to 6.48 h. On the other
hand, the steeper breakthrough curves were found for larger
feed concentrations. This is because the adsorbent sites are
quickly filled at higher initial concentrations, which results in
a decrease in the breakthrough time. Moreover, at higher

574 Energy & Fuels, Vol. 22, No. 1, 2008

Figure 3. Effect of the Peclet number on the breakthrough curve. Cts

) 0.009 386 kg/m3; Re ) 89 000; T ) 311 K.

Figure 4. Effect of the total inlet sulfur concentration on the

breakthrough curve. Re ) 89 000; T ) 311 K; Pe ) 3070.

Shams et al.

Figure 6. Effect of the feed temperature on the breakthrough curve.

Cts ) 0.009 386 kg/m3; Re ) 89 000; Pe ) 3070.

the particle radius increases from 0.8 to 1.6 mm, the steepness
of the breakthrough curve decreases. Also, the break point
increases from 8.52 to 9.84 h. The smaller particle size provides
a faster kinetic equilibrium. This allows for faster cycle times,
with no reduction in the adsorption capacity. On the other hand,
as the diameter of particle increases, the thickness of the stagnant
film around the adsorbent particles increases and, also, the total
length of the path inside the pores increases. Under these
conditions, the overall kinetics of the process is slow, because
the time for a molecule of adsorbate to reach the adsorption
site is longer and the diffusion path along the pores is larger.
4.9. Influence of the Feed Temperature. The model predictions for different operating temperatures including 311 and 330
K were given in Figure 6. As may be noticed, an increase in
the operating temperature from 311 to 330 K causes a decrease
in the break point time from 9.84 to 4.56 h. This behavior is
expected, because the adsorption is an exothermic process and,
hence, as the operating temperature increases, the equilibrium
concentration on the solid phase decreases. This means that the
equilibrium is attained quickly for higher operating temperatures.
5. Conclusions

Figure 5. Effect of the particle radius on the breakthrough curve. Cts

) 0.009 386 kg/m3; Re ) 89 000; T ) 311 K; Pe ) 3070.

concentrations, the slope of isotherm curves is smaller and,

hence, the equilibrium is quickly attained.
4.8. Influence of the Particle Diameter. The effect of the
adsorbent particle size on the outlet concentration was shown
in Figure 5. The particle diameters considered were 1.6 and
3.2 mm. This is because the Universal Oil Products (UOP)
company supplies these two particle sizes for zeolite 13X. To
study the effect of the adsorbent size on the breakthrough curve,
other parameters were kept constant. Figure 5 shows that, as

An investigation was conducted on the modeling and simulation of binary liquid-phase adsorption of methyl mercaptan and
hydrogen sulfide from a liquid butane stream in a fixed bed.
The modeling approach takes into account the effects of axial
dispersion, interparticle mass-transfer resistance, and intraparticle
mass-transfer resistance of the adsorbents. The dimensionless
general forms of governing partial differential equations were
solved by an orthogonal collocation method. The mode predictions were compared to the commercial-scale plant data. It was
found that: (1) An increase in the bed Reynolds number from
58 000 to 142 000 causes a decrease in the breakthrough time
from 11.76 to 6.86 h and, also, the breakthrough curve becomes
steeper. On the other hand, for bed Reynolds numbers larger
than 984 000, the slope of the breakthrough curve remains
almost constant. (2) An increase in the Peclet number as a result
of increasing the fixed-bed height over the range of 18203640
causes an increase in the breakthrough time from 5.88 to 11.64
h and a decrease in the slope of the breakthrough curve. (3)
Decreasing the adsorbent radius from 0.0016 to 0.0008 m as
supplied by the UOP company only decreases the breakthrough
time and may affect the final time of the adsorption cycle. (4)
Increasing the feed temperature decreases the breakthrough time
significantly.

Desulfurization of Liquid-Phase Butane

Acknowledgment. The authors gratefully acknowledge the
financial support provided by the Bandar Imam Petrochemical
Company (Iran).

Nomenclature
a ) form factor
Aj,k ) constant generated in the orthogonal collocation method
Bii ) Biot number of component i
Bj,k ) constant generated in the orthogonal collocation method
Cfi ) concentration of solute i in feed
Ci ) concentration of solute i in the fluid phase of the column
Ci ) dimensionless concentration of solute i in the fluid phase (Ci/
Cfi)
Cts ) total sulfur concentration in the fluid phase
dB ) bed diameter
DLi ) axial dispersion coefficient of component i
Dpi ) diffusivity for component i in the pore fluid
kfi ) film mass-transfer coefficient of solute i
ki ) local mass-transfer coefficient of solute i
L ) packed bed height
N ) number of interior collocation points in the axial direction
NR ) number of collocation points in the radial direction
Pei ) Peclet number of component i
/ ) initial concentration of solute i in the solid phase
q0i
qi ) average concentration of solute i in the solid phase
q/i ) equilibrium concentration of solute i in the solid phase
q/i ) dimensionless concentration of solute i in the solid phase
(q/i /q0i)
R ) external radius of the adsorbent

Energy & Fuels, Vol. 22, No. 1, 2008 575

r ) radial distance in the particle
Re ) bed Reynolds number (FUdB/)
Shi ) Sherwood number of component i
T ) temperature
t ) time
U ) superficial velocity
Yi ) concentration of solute i in the pore fluid phase
Yi ) dimensionless concentration of solute i in the pore fluid
phase (Yi/Cfi)
z ) axial distance in the column
Greek Symbols
R ) constant in the weighting function
) constant in the weighting function
b ) void fraction of the bed
p ) void fraction of the zeolite particles
) viscosity of fluid
) dimensionless distance (z/L)
F ) density of fluid
) dimensionless radius (r/R)2
) dimensionless time (tU/L)
Superscript
* ) equilibrium value
Subscripts
i, j, and k ) integer values
f ) feed
p ) pore phase
EF700500X