Chemical Engineering Journal 284 (2016) 224232

Contents lists available at ScienceDirect

Chemical Engineering Journal

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

Preparation of high purity propane from liquefied petroleum gas in a

fixed bed by removal of sulfur and butanes
Huan Xiang, Huiping Zhang, Pengfei Liu, Ying Yan
School of Chemistry and Chemical Engineering, South China University of Technology, Guangzhou 510640, PR China

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

High purity propane was prepared

from LPG in a fixed bed.

Sulfur compounds in LPG were

removed by adsorption over zeolite

13X and NaY efficiently.

Butanes (i-butane and n-butane) in
LPG were removed by adsorption
over activated carbon efficiently.

a r t i c l e

i n f o

Article history:
Received 13 July 2015
Received in revised form 15 August 2015
Accepted 18 August 2015
Available online 8 September 2015
Keywords:
Fixed bed
Adsorption
Sulfur
Butanes
Liquefied petroleum gas

a b s t r a c t
High purity propane was prepared from liquefied petroleum gas (LPG) in a fixed bed by removal of sulfur
and butanes (i-butane and n-butane). First, sulfur compounds were removed from LPG by fixed bed
adsorption over three commercial zeolites (13X, NaY and 5A). Then, adsorption dynamics of propane,
i-butane and n-butane on activated carbon were investigated by evaluating effects of the feed flow rate,
bed length and temperature on breakthrough curves. Breakthrough curves for sulfur, propane and
butanes adsorption were analyzed by YoonNelson model. The experimental results indicated that zeolite 13X and NaY presented better desulfurization performance than zeolite 5A at the temperature of
303 K, feed flow rate of 20 mL/min and bed length of 15 cm. Breakthrough curves for sulfur, propane
and butanes fitted relatively well to YoonNelson model. The YoonNelson model is strongly valid for
breakthrough curves of n-butane, with all correlation coefficient exceeding 0.96. Butanes were more
selectively adsorbed on activated carbon than propane. The adsorption selectivity of activated carbon
for these three adsorbates decreased in the order of n-butane > i-butane > propane. Activated carbon
had been proved to be a desirable adsorbent to remove butanes from LPG to prepare high purity propane
under experimental conditions.
2015 Elsevier B.V. All rights reserved.

1. Introduction
Refrigerators are one of the major energy consuming domestic
appliances in household environment [1]. Chlorofluorocarbons
(CFCs) were used as refrigerants in most of the domestic refrigerators and other small refrigeration units for many decades because
of their excellent cooling characteristics. However, CFCs, such as
R12, were found to cause global warming and damage the strato Corresponding author. Tel./fax: +86 2087111975.
E-mail address: yingyan@scut.edu.cn (Y. Yan).
http://dx.doi.org/10.1016/j.cej.2015.08.160
1385-8947/ 2015 Elsevier B.V. All rights reserved.

spheric ozone layer due to their chlorine chemical effect [2]. In

contrast, hydrocarbons, another kind of refrigerant, have been
demonstrated to be energy efficient and environment-friendly
[36]. Hydrocarbons have no ozone depletion effect and global
warming effect (see Table 1). They are also universally available
in large amounts at low price.
Among all hydrocarbons, propane (R290) is found to be the
most convenient refrigerant in small refrigerators [3]. As a new
environment-friendly refrigerant, R290, with its total sulfur content below 2 ppm, is comprised of at least 99.5% propane. Since
large amounts of propane are contained in LPG (see Table 2), LPG

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

Table 1
Properties of some refrigerants [7,8].
Refrig.

Chemical
formula

Molecular weight
(kg/kmol)

Natural

ODP

GWP
(100 yr)

R12
R134a
R290
R600
R600a

CCl2F2
CH2FCF3
C3H8
n-C4H10
i-C4H10

120.9
102.0
44.0
58.1
58.1

No
No
Yes
Yes
Yes

0.82
0
0
0
0

8100
1300
0
0
0

is considered to be a potential feedstock to manufacture R290 due

to its local availability and relatively low price. In addition, the key
issues in producing R290 from LPG lie in removing sulfur compounds and hydrocarbons other than propane, especially butanes,
which are also seemed as main components of LPG.
Sulfur compounds, such as H2S and mercaptans, coming from
crude oil and odorized gas added for safety regulations, are widely
existed in LPG. As sulfur can cause corrosion to equipments, LPG
must be thoroughly desulfurized in producing R290. For the
removal of sulfur compounds from hydrocarbon fuels, three kinds
of methods are commonly adopted, hydrodesulfurization (HDS),
selective catalytic oxidation (SCO) followed by adsorption and
adsorption desulfurization. HDS is operated at elevated temperature (300400 C) and pressure (4050 bar) and requires high
hydrogen to feedstock ratio [11,12]. Because of the large energy
cost of compressing hydrogen, HDS is categorized as an impractical
way in fuel processors. Moreover, the conversion of olefin to alkane
results in massive losses of olefin during the HDS process. While
SCO does not require additional hydrogen compared to HDS, it
needs an extra unit to separate unconverted oxygen and nitrogen
from hydrocarbon fuels [13]. In contrast, adsorption desulfurization does not need any hydrogen and is usually operated at relative
low temperature and pressure. Hence, it is a promising and applicable method in removing sulfide from LPG. Up to present, activated carbon [14,15] and zeolite [1618] are two of the most
widely studied adsorbents for desulfurization. Takatsu et al. [19]
studied the adsorption of sulfur compounds in LPG on activated
carbon, CuO/ZnO/Al2O3, Ag-exchanged b-zeolite and Ag/CeO2 at
room temperature. Among the tested adsorbents, Ag-exchanged
b-zeolite and Ag/CeO2 presented better adsorption capacity. They
also deduced that the larger the negative charge of the sulfur atom,
the stronger the coordination bonding between Ag+ and the sulfur
atom of the compounds.
The main processes for light hydrocarbons separation are
absorption process, cryogenic process, membrane process, and
adsorption process [20,21]. Adsorption separation process has

225

been widely applied in recovering hydrocarbons with high purity

due to its convenience and high efficiency. It is worth mentioning
that activated carbon, which has high surface area and large
adsorption capacity, has been proved to be an attractive material
for adsorbing butane. Marbn et al. [22] studied the breakthrough
behavior of activated carbon fiber monolith in n-butane adsorption
and proposed a complete breakthrough model to predict its breakthrough profile. The results showed a good agreement between the
experimental and the simulated data. Kawakami et al. [23,24] provided an industrially advantageous method and system for obtaining high purity propane from low purity propane. Propane with a
purity of 99.99 vol.% or higher was obtained by adsorbing ethane,
propylene and butanes (i-butane and n-butane) over activated carbon molecular sieve and activated carbon, respectively. Besides,
there is no other literature described the removal of sulfur compounds and butanes from LPG for producing high purity propane
by fixed bed adsorption.
To obtain high purity propane (R290) from LPG, sulfur compounds and butanes must be removed. Adsorption is one of the
most feasible methods due to its high efficiency. However, it seems
difficult to find a material that can remove sulfur and butanes from
LPG by fixed bed adsorption simultaneously and efficiently. Therefore, a combined process is a good choice to prepare high purity
propane from LPG, as shown in Fig. 1. During this process, high
purity propane can be obtained from LPG in the fixed bed by
removing sulfur and butanes, respectively.
However, the adsorption process has its own limitations. Since
it is a non-stationary process, some kinds of cycle will be needed,
where adsorbents can be regenerated or substituted. In the industrial adsorption processes, adsorbent regeneration is by far the
most time and energy consuming step. Hence, the method for
adsorbents regeneration, such as pressure swing adsorption (PSA)
[25] and temperature swing adsorption (TSA) [26], is another crucial issue that must be considered. This study is just a preliminary
exploration of preparing high purity propane from LPG by fixed
bed adsorption, so the regeneration of used adsorbents is not
investigated.
The purpose of the present work is to prepare high purity propane by adsorption separation of sulfur and butanes from LPG in a
fixed bed. The removal of sulfur compounds and butanes from LPG
over three different zeolites (13X, NaY and 5A) and activated carbon will be studied, respectively. Breakthrough experiments with
commercial LPG are carried out to select excellent adsorbents for
desulfurization. Furthermore, adsorption dynamics of propane, ibutane and n-butane will be investigated by evaluating effects of
the feed flow rate, bed length and temperature on breakthrough
curves, and the YoonNelson model is used to analyze breakthrough curves for sulfur, propane and butanes.

Table 2
LPG composition by country [9,10].
Country

Australia
Belgium
Canada
France
Hong Kong
Italy
Korea
Mexico
Netherlands
New Zealand
Poland
Singapore
Spain
United Kingdom
United States

LPG composition% (vol/vol)

Ethane

Propane

Propylene

Butanes

010
05

0100
50100
92.5100
50100
2040
40100
1035
60
50100
6070
2060
2040
87.5100
50100
92.5100

045
050
05
050

050
010
02.5
05
6080
050
6590
40
050
3040
4080
6080
05
05
02.5

05
05
05

05

05
05
05

050

050

05
050
05

2. Experimental
2.1. Materials
LPG (Guangzhou Huakai Oil and Gas Co., Ltd) was used as adsorbate. The compositions of sulfur, propane, i-butane, and n-butane
in LPG were 35 ppm, 97 1%, 0.02%, and 0.01%, respectively. Zeolite
13X (Anhui Mingmei Mining Co., Ltd), NaY (Dalian Adsorbent Factory), 5A (Zhengzhou Tianxiang Inorganic Material Co., Ltd) with a
particle diameter of 35 mm, and activated carbon (Ningxia
Huayang Activated Carbon Co., Ltd) with an average particle diameter of 0.9 mm, were used as adsorbents. Zeolites were crushed and
sieved to the required particle size (250425 lm) before being
used in the adsorption experiments.

226

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

Fig. 1. Flow diagram for a LPG processor.

2.2. Characterization
The N2 adsorptiondesorption isotherms of the adsorbents
were tested using an ASAP-2020 apparatus (Micromeritics Instrument Co., USA) in static measurement mode at 77 K. The samples
were degassed under high vacuum at 473 K for 2 h before measurements. The obtained adsorption isotherms were used to analyze porosity parameter of the adsorbents. The specific surface
areas (SBET) were calculated by the BET equation. The micropore
surface areas (Smicro) and micropore volume (Vmicro) were determined by t-plot method.
2.3. Adsorption experiments
Adsorption measurements were carried out in a continuous
flow system at atmospheric pressure. A schematic diagram of the
apparatus was shown in Fig. 2. The fixed bed was a stainless steel
pipe with a diameter of 2 cm. A water bath was employed to provide a constant temperature environment for the fixed bed.
Zeolite 13X, NaY and 5A were used to adsorb sulfur compounds
from LPG in a fixed bed at the temperature of 303 K, feed flow rate
of 20 mL/min and bed length of 15 cm. Prior to desulfurization
experiments, the adsorbents were activated in muffle in air at
573 K for 4 h. Sulfur content of the effluent was sampled every 3
or 5 min until the adsorbent was saturated, and the total sulfur
concentration C(t) of every sample was measured by microcoulometer (WK-2D, Jiangsu). In addition, the breakthrough concentration of sulfur was selected as 2 ppm in this work.
Adsorption breakthrough experiments of propane, i-butane and
n-butane were performed at different conditions (feed flow rate,
bed length and temperature) using LPG as a test gas. To minimize
axial dispersion effects, the fixed bed length to particle diameter
ratio was larger than 20 for all experiments [27]. Activated carbon
was dried in oven at 383 K overnight in order to remove the physically adsorbed water prior to the experiments. The inlet and outlet
concentration were monitored by gas chromatography (Aglient
7890A, USA) equipped with a flame ionization detector (FID) and
a GS-GasPro capillary column. The adsorption data were also taken
at 5 min sampling interval until activated carbon was saturated,
and analyses were made at least three times and the results were
finally averaged. After that, the breakthrough curves with the form
of variations of C/C0 versus time were obtained.
3. Theory
The performance of fixed bed is described through the concept
of breakthrough curve. The breakthrough time and the shape of the
breakthrough curve are significant characteristics for determining
the operation and dynamic response of an adsorption column.
Hence, prediction of breakthrough curves is one of the essential
factors for the successful design of a column adsorption process.
A large variety of mathematical models have been proposed for

predicting adsorption breakthrough curves, such as Thomas, Wolborska, BohartAdams, Bed Depth Service Time (BDST) and Yoon
Nelson models [2830].
YoonNelson model is based on the assumption that the rate of
decrease in the probability of adsorption for each molecule is proportional to the probability for adsorption and breakthrough on the
adsorbent [31]. It is not only less complicated than other models,
but also requires no detailed data concerning the characteristics
of adsorbate, the type of adsorbent and the physical properties of
adsorption bed [32]. YoonNelson model is widely used to describe
breakthrough curves.
The YoonNelson equation used in this study is expressed as:

t s

1
C
0 ln
C0  C
k

where k0 is the rate constant (min1), s is the time required for 50%
adsorbate breakthrough (min), t is the breakthrough time (min), C is
the breakthrough concentration (ppm), C0 is the initial concentration (ppm) of adsorbate.
4. Results and discussion
4.1. N2 adsorptiondesorption characterization
The N2 adsorptiondesorption isotherms of commercial zeolites
(13X, NaY and 5A) and activated carbon are displayed in Fig. 2. The
BET surface areas as well as pore volume of the samples are summarized in Table 3 based on the N2 adsorptiondesorption isotherms. As can be seen in Fig. 3, the N2 adsorptiondesorption
isotherms of the adsorbents all belong to type I adsorption isotherm. At the beginning, the volume adsorbed increases obviously
with increasing relative pressure, which could be ascribed to the
volume filling of micropores. As relative pressure keeps increasing,
the volume adsorbed increases continually, which should be
caused by the multilayer adsorption. In addition, the hysteresis
loops observed in N2 adsorptiondesorption isotherms at a high
relative pressure P/P0 of 0.80.97 could be caused by the existence
of mesopores in these adsorbents.
4.2. Adsorption desulfurization
The breakthrough curves for the total sulfur in LPG over zeolite
13X, NaY and 5A at 20 mL/min and 303 K are shown in Fig. 4. The
experimental data shown in this figure are scattered, which is
mainly resulted from the fact that the outlet concentration of sulfur compounds is fluctuating with time. While the sulfur concentration increased rapidly and become equal to that at the inlet
over 5A zeolite fixed bed immediately, 13X and NaY zeolite adsorbents showed low sulfur values at 5 min after the start of the
adsorption runs. Under experimental conditions, breakthrough
times for zeolite 13X and NaY, with the breakthrough point of
2 ppm, were 32 min and 5 min, respectively. From comparison of

227

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

Fig. 2. Schematic diagram of adsorption apparatus. 1. LPG, 2. Drying column, 3. Fixed bed, 4. Water bath, 5. Microcoulometer, 6. Gas chromatography, V. Manual valve, PG.
Pressure gauge, R. Rotameter.

4.3. Adsorption dynamics of propane, i-butane and n-butane

Table 3
Surface areas and pore porosities of zeolites and activated carbon.
Sample

5A
13X
NaY
Activated carbon

Surface area (m2/g)

Volume (cm3/g)

SBET

Smicro

Vtotal

Vmicro

375
410
542
1214

326
395
463
1165

0.2181
0.3179
0.3365
0.6979

0.1528
0.1993
0.2172
0.6232

the adsorption desulfurization on these three adsorbents, it was

found that zeolite 13X and NaY presented better desulfurization
performance than zeolite 5A. The possible reason could be that
both of zeolite 13X and NaY are faujasite, and therefore have larger
pore diameter than zeolite 5A for sulfur species to access and be
adsorbed. Thus, zeolite 13X and NaY can be employed in adsorbing
sulfur compounds from LPG. According to Wakita et al. [33], Na+
sites which lie in large cavity can affect the adsorption performance of sulfur compounds, and sulfur compounds can be
adsorbed coordinately on Na+ through the lone electron pair of
the sulfur atom. These observations are in agreement with the
results of Salem [34] who showed that zeolite 13X had higher
adsorption capacity for sulfur at low concentration ranges than
zeolite 5A which was unfavorable for sulfur adsorption from naphtha. In addition, the introduction of transition metals to zeolite 13X
or NaY, such as Cu+ and Ni2+, can improve its sulfur removal performance [3537].

Adsorption dynamics in multi-component systems has been a

subject of intensive research because it is of both theoretical and
practical importance [38]. Such information is useful in understanding the way that different adsorbates interact during their
diffusion and adsorption inside an adsorbent particle. Figs. 57
depict the breakthrough curves for the adsorption of propane, ibutane and n-butane in LPG on activated carbon at different operating conditions. In these figures, propane, which moved more
quickly than i-butane and n-butane, was adsorbed on the fresh
activated carbon downstream of the bed and appeared in the effluent stream first. Propane was replaced or purged out from the
adsorbent while i-butane and n-butane were being adsorbed in
its place. Furthermore, i-butane was also replaced by n-butane
which appeared at the outlet of the adsorption column last.
At the beginning, all of the components were adsorbed on activated carbon according to their different diffusion rate and concentration in the gas phase. Meanwhile, no equilibrium was achieved
due to the large adsorption capacity of activated carbon. Thereafter, as the filling of pores in activated carbon approached completion, the process proceeded in such a way that the adsorption
equilibrium was approached by replacement. The more selectively
adsorbed component would replace the less one. This raised the
fluid phase composition of the less selectively adsorbed component to such an extent that it rose to a peak composition above that

50
500

(d)

300

(c)
(b)
(a)

200

40

Outlet sulfur content (ppm)

Volume adsorbed (cm3/g)

400

(c)
(b)

30

(a)
20

10

100

2 ppm level
0
0
0.0

0
0.2

0.4

0.6

0.8

1.0

Relative pressure (P/P0)

Fig. 3. N2 adsorptiondesorption isotherms of the samples at 77 K: (a) 5A, (b) 13X,
(c) NaY and (d) activated carbon.

40

80

120

160

Time (min)
Fig. 4. Breakthrough curves for total sulfur adsorption on three different adsorbents: (a) 5A, (b) 13X, (c) NaY, and sulfur breakthrough concentration of 2 ppm as a
reference. Experiments were carried out at 20 mL/min feed flow rate and 303 K.

228

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

1.8

1.8

(a)

(b)

1.6

1.4

1.4

1.2

1.2

1.0

1.0

C/C0

C/C0

1.6

0.8
0.6

0.8
0.6

propane
i-butane
n-butane

0.4
0.2

0.4

propane
i-butane
n-butane

0.2

0.0

0.0
0

40

80

120

160

200

240

280

320

30

60

90

Time (min)

120

150

180

210

Time (min)

1.8

(c)

1.6
1.4
1.2

C/C0

1.0
0.8
0.6

propane
i-butane
n-butane

0.4
0.2
0.0
0

30

60

90

Time (min)
Fig. 5. Breakthrough curves for LPG adsorption on activated carbon at different feed flow rates: (a) 20 mL/min, (b) 40 mL/min and (c) 60 mL/min. Experiments were carried
out with 15 cm bed length and at room temperature.

of the inlet composition, leading to a roll-up or overshoot phenomenon. The phenomenon had also been mentioned by Ahmed
et al. [39] for the adsorption of methane, ethane and propane mixture on zeolite 4A. According to the breakthrough order, n-butane
was more selectively adsorbed than i-butane, and propane was less
selectively adsorbed than i-butane, i-butane replaced propane, and
n-butane replaced i-butane, and thus the adsorption selectivity of
activated carbon for the three adsorbates decreased in the order
of n-butane > i-butane > propane. This phenomenon can be useful
for obtaining high purity propane without containing any
i-butane or n-butane from LPG.

4.3.1. Effect of the feed flow rate

The effect of feed flow rate on the breakthrough curve of propane, i-butane and n-butane in LPG is shown in Fig. 5. It can be
seen that all breakthrough curves of these three components shift
towards the origin as the feed flow rate increases. When the flow
rate increased from 20 mL/min to 60 mL/min, the time required
for complete column saturation decreased from 210 min to
55 min for propane, and from 265 min to 70 min, 285 min to
80 min for i-butane and n-butane, respectively. At the highest flow
rate tested, a less sharp breakthrough curve is obtained, which is
resulted from poor utilization of the column capacity. The use of
high flow rate reduces the time that adsorbates are in contact with

the adsorbent, and allows less time for adsorption to occur, thus
leads to an early breakthrough of adsorbates. However, the use
of low flow rates may not be desirable in practice, because it will
result in long overall processing time, especially when large volume feed has to be processed.
4.3.2. Effect of the bed length
The effect of fixed bed length on the breakthrough curve of propane, i-butane and n-butane in LPG is shown in Fig. 6. These figures
showed that the breakthrough curves shifted away from the origin
with increasing bed length. When the fixed bed length was
increased from 8 cm to 15 cm, the time required for complete
column saturation increased from 80 min to 210 min for propane,
and from 130 min to 265 min, 135 min to 285 min for i-butane and
n-butane, respectively. The amount of the adsorbent (activated
carbon) packed in the column rises with increasing bed length
(815 cm) and consequently, adsorption capacity is enhanced,
which finally leads to increasing breakthrough times of all these
adsorbates.
4.3.3. Effect of the temperature
Adsorption temperature is another important operating condition. Fig. 7 shows the breakthrough curves obtained at 293 K,
313 K and 323 K for propane, i-butane and n-butane using 15 cm

229

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

1.8

(a)

(b)

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

C/C0

C/C0

1.8

0.8
0.6

0.8
0.6

0.4
0.2

propane
i-butane
n-butane

0.4

propane
i-butane
n-butane

0.2

0.0

0.0
0

20

40

60

80

100

120

140

160

20

40

60

80

Time (min)

100

120

140

160

180

Time (min)

1.8

(c)

1.6
1.4
1.2

C/C0

1.0
0.8
0.6
0.4

propane
i-butane
n-butane

0.2
0.0
0

40

80

120

160

200

240

280

320

Time (min)
Fig. 6. Breakthrough curves for LPG adsorption on activated carbon with different bed lengths: (a) 8 cm, (b) 10 cm and (c) 15 cm. Experiments were carried out at 20 mL/min
feed flow rate and room temperature.

activated carbon fixed bed and at 40 mL/min feed flow rate. These
figures showed that the breakthrough curves shifted towards the
origin as the adsorption temperature increased, especially for
i-butane and n-butane. When the adsorption temperature was
increased from 293 K to 323 K, the time required for complete column saturation decreased from 185 min to 145 min for i-butane,
and from 185 min to 165 min for n-butane. The adsorption capacity is significantly higher at lower temperature, which indicates
that these adsorptions belong to physical adsorption.
4.4. Application of YoonNelson model
In this work, fixed bed adsorption breakthrough experiments
were carried out using LPG as a test gas in order to simulate industrial process. Hence, modeling breakthrough curves by YoonNelson equation is of guiding importance for the the successful
design of a column adsorption process.
4.4.1. Modeling breakthrough curves for sulfur
First, breakthrough curves for sulfur adsorption on both zeolite
13X and NaY, which exhibited good desulfurization performance,
were analyzed by YoonNelson model. Fig. 8 showed a comparison
between the experimental and theoretical breakthrough curves.
Then, a linear plot of ln[C/(C0  C)] vs. breakthrough time (t) with
slope of k0 and intercept of k0 s could obtained according to the

YoonNelson equation. The value of k0 and s were calculated from

the slope and intercept. The values of these parameters along with
the correlation coefficient (R2) were listed in Table 4.
According to the assumption of YoonNelson model, k0 characterizes the shape of the breakthrough curve. As can been seen in
Fig. 8 and Table 4, the steepness of breakthrough curve for sulfur
over zeolite 13X with larger k0 value was more upright than that
over zeolite NaY. The data in Table 4 also indicated that s values
from the calculation were in good agreement with experimental
results. Thus, the agreement between YoonNelson model and
the experimental results is satisfactory.
However, these experimental data in Fig. 8 were scattered
and the correlation coefficients R2 of breakthrough curves for
sulfur on these two zeolites were relatively low. This is mainly
due to the fact that the outlet concentration of sulfur
compounds in LPG is fluctuating with time. In addition, the
YoonNelson model, due to its limited form, is not able to
calculate process variables and to predict adsorption under a
variety of conditions [40,41], which might be another reason
for the poor fitting capacity of YoonNelson model. Regardless
of the fluctuation of sulfur concentration, it is can found that
the overall trend of the data accord with the breakthrough
curves simulated by YoonNelson model. In some sense, this
model is of theoretical importance for designing adsorption
process for sulfur compounds.

230

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

1.6

1.6

(b)

1.4

1.4

1.2

1.2

1.0

1.0

0.8

0.8

C/C0

C/C0

(a)

0.6

0.6

0.4

0.4

propane
i-butane
n-butane

0.2

propane
i-butane
n-butane

0.2

0.0

0.0
0

30

60

90

120

150

180

210

20

40

60

80

Time (min)

100

120

140

160

180

200

Time (min)

1.6

(c)

1.4
1.2

C/C0

1.0
0.8
0.6
0.4

propane
i-butane
n-butane

0.2
0.0
0

20

40

60

80

100

120

140

160

180

Time (min)
Fig. 7. Breakthrough curves for LPG adsorption on activated carbon at different temperatures: (a) 293 K, (b) 313 K and (c) 323 K. Experiments were carried out with 15 cm
bed length and at 40 mL/min feed flow rate.

1.4

Table 4
Parameters of YoonNelson model for sulfur adsorption on zeolite 13X and NaY.

1.2
1.0

Adsorbents

k0 (min1)

s (min)

s (experimental, min)

R2

13X
NaY

1.8679
0.0552

36
16

35
16

0.8953
0.6017

C/C0

0.8
0.6
0.4

13X
NaY

0.2
0.0
0

40

80

120

160

Time (min)
Fig. 8. Breakthrough curves for total sulfur adsorption on zeolite 13X and NaY.
Symbols: experimental data, solid lines: prediction of the YoonNelson model.
Experiments were carried out at 20 mL/min feed flow rate and 303 K.

4.4.2. Modeling breakthrough curves for propane, i-butane and

n-butane
The YoonNelson model was adopted to study breakthrough
curves for propane, i-butane and n-butane at different conditions

(feed flow rate, bed length and temperature) as well. However,

because the YoonNelson model is not valid for modeling a breakthrough curve with overshoot, only the initial parts before
C0/C = 1 of breakthrough curves for propane and i-butane were
studied, while the complete breakthrough curves for n-butane,
the most adsorbed component, were investigated. In addition,
when the feed flow rate exceeded 20 mL/min, propane appeared
in the effluent stream quickly. It is not valid to model the ascending
part of breakthrough curves for propane, because there are just few
data points. As a result, breakthrough curves for propane only with
a feed flow rate of 20 mL/min were studied by YoonNelson model
in this study. The parameters predicted by the YoonNelson model
for propane, i-butane and n-butane adsorption on different experimental conditions were listed in Tables 57, respectively.
0
As can be seen from Tables 57, the value of k indicated that the
adsorption rate decreased in the order of propane > i-butane >
n-butane on the same experimental conditions. These observations are also demonstrated by Prasetyo et al. [42] who found that

231

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

Table 5
Parameters of YoonNelson model for propane adsorption on activated carbon on different conditions.
Feed flow rate (mL/min)

Bed length (cm)

Temperature (K)

k0 (min1)

s (min)

s (experimental, min)

R2

20
20
20

8
10
15

Room temperature
Room temperature
Room temperature

0.8348
0.6415
0.7874

24
29
36

24
28
37

0.9705
0.9897
0.8652

Table 6
Parameters of YoonNelson model for i-butane adsorption on activated carbon on different conditions.
Feed flow rate (mL/min)

Bed length (cm)

Temperature (K)

k0 (min1)

s (min)

s (experimental, min)

R2

20
20
20
40
60
40
40
40

8
10
15
15
15
15
15
15

Room
Room
Room
Room
Room
293
313
323

0.0749
0.1133
0.0831
0.0942
0.1637
0.0837
0.1120
0.1785

71
101
139
77
25
78
67
54

75
103
145
75
25
76
68
54

0.9829
0.9206
0.9490
0.9461
0.9902
0.9749
0.9780
0.9698

temperature
temperature
temperature
temperature
temperature

Table 7
Parameters of YoonNelson model for n-butane adsorption on activated carbon on different conditions.
Feed flow rate (mL/min)

Bed length (cm)

Temperature (K)

k0 (min1)

s (min)

s (experimental, min)

R2

20
20
20
40
60
40
40
40

8
10
15
15
15
15
15
15

Room
Room
Room
Room
Room
293
313
323

0.0736
0.0845
0.0360
0.0636
0.1041
0.0635
0.0563
0.0492

108
118
212
143
32
142
120
104

111
118
211
142
32
143
120
104

0.9808
0.9757
0.9936
0.9957
0.9698
0.9956
0.9955
0.9936

temperature
temperature
temperature
temperature
temperature

the adsorption rate decreased with increasing molecular weight.

The value of k0 (Tables 6 and 7) increased with increasing feed flow
rate. This is due to the external mass transfer resistance of adsorbates decreased as feed flow rate increased. These observations
are in agreement with the results of Guo et al. [43] who showed
that increasing SO2 superficial velocity could enhance the adsorption rate. However, no significant effects on the value of k0 were
observed with an increasing in bed length. In addition, the value
of k0 of i-butane and n-butane presented an opposite tendency with
temperature increasing. This phenomenon may be relevant to their
competitive adsorption on activated carbon. The results in Tables
57 also indicated that the value of 50% breakthrough time
decreased with increasing in both feed flow rate and temperature
or decreasing in bed length. This may further explain the phenomenon shown in Figs. 57 that shorter breakthrough times were
obtained at higher feed flow rate and temperature and with shorter
bed length.
For all fitting results of propane and butanes adsorption, s values from the calculation were in good agreement with experimental results. Also, the YoonNelson model fitted the experimental
results reasonably well seen from the correlation coefficients R2,
and the correlation coefficients of n-butane on different conditions
even exceeded 0.96. Hence, this model is valid for the adsorption of
propane and butanes in LPG, especially for the breakthrough
curves of n-butane.

tion performance for LPG than zeolite 5A at the temperature of

303 K, feed flow rate of 20 mL/min and bed length of 15 cm.
Efficient and fast adsorptions of butanes (i-butane and
n-butane) from LPG on activated carbon were proved by fixed
bed experiments at a wide range of operating conditions (feed flow
rate, bed length and temperature). An increase in feed flow rate,
adsorption temperature or a decrease in bed length led to an early
breakthrough of adsorbates. The overshoot phenomenon was
observed for the less adsorbed components (propane) due to the
replacement of the more selectively adsorbed components
(i-butane and n-butane). The adsorption selectivity of activated
carbon for these three adsorbates decreased in the order of
n-butane > i-butane > propane. Activated carbon had been demonstrated to be an appropriate adsorbent for removing i-butane and
n-butane from LPG to prepare high purity propane.
The agreement between YoonNelson model and the experimental results of sulfur, propane and butanes is satisfactory.
YoonNelson model is strongly valid for breakthrough curves of
n-butane, with all correlation coefficients R2 exceeding 0.96.
Further researches should be focused on the adsorption dynamics of sulfur in LPG, detailed simulations and the regeneration of
used adsorbents.

5. Conclusions

We gratefully acknowledge the financial supports of the

National Natural Science Foundation of China (Grant No.
21176086 and 21376101), Fundamental Research Funds for the
Central Universities and Science and Technology Planning Project
of Guangdong Province, China (No. 2013B010403002) for this
work.

In this study, the process of producing high purity propane from

LPG in a fixed bed by removal of sulfur and butanes (i-butane and
n-butane) has been investigated. For the adsorption separation of
sulfur from LPG, zeolite 13X and NaY exhibited better desulfuriza-

Acknowledgements

232

H. Xiang et al. / Chemical Engineering Journal 284 (2016) 224232

References
[1] A. Langde, J. Ali, M. Shahid, M. Sultan, Hydrocarbon refrigeration system,
International Conference on Advances in Engineering, Technology and
Management (ICAET-2014), Kannan Devan Hills Munnar, Kerala, India, 2014,
pp. 8082.
[2] M. Hammad, M. Alsaad, The use of hydrocarbon mixtures as refrigerants in
domestic refrigerators, Appl. Therm. Eng. 19 (1999) 11811189.
[3] R. James, J. Missenden, The use of propane in domestic refrigerators, Int. J.
Refrig. 15 (1992) 95100.
[4] E. Bodio, M. Chorowski, M. Wilczek, Working parameters of domestic
refrigerators filled with propanebutane mixture, Int. J. Refrig. 16 (1993)
353356.
[5] R. Richardson, J. Butterworth, The performance of propane/isobutane mixtures
in a vapour-compression refrigeration system, Int. J. Refrig. 18 (1995)
5862.
[6] M. Alsaad, M. Hammad, The application of propane/butane mixture for
domestic refrigerators, Appl. Therm. Eng. 18 (1998) 911918.
[7] S. Wongwises, N. Chimres, Experimental study of hydrocarbon mixtures to
replace HFC-134a in a domestic refrigerator, Energy Convers. Manage. 46
(2005) 85100.
[8] M. Fatouh, M. El Kafafy, Assessment of propane/commercial butane mixtures
as possible alternatives to R134a in domestic refrigerators, Energy Convers.
Manage. 47 (2006) 26442658.
[9] K.J. Morganti, T.M. Foong, M.J. Brear, G. da Silva, Y. Yang, F.L. Dryer, The
research and motor octane numbers of liquefied petroleum gas (LPG), Fuel 108
(2013) 797811.
[10] M. El-Morsi, Energy and exergy analysis of LPG (liquefied petroleum gas) as a
drop in replacement for R134a in domestic refrigerators, Energy 86 (2015)
344353.
[11] I.V. Babich, J.A. Moulijn, Science and technology of novel processes for
deep desulfurization of oil refinery streams: a reviewq, Fuel 82 (2003) 607
631.
[12] D.G. Lffler, K. Taylor, D. Mason, A light hydrocarbon fuel processor producing
high-purity hydrogen, J. Power Sources 117 (2003) 8491.
[13] L. Lv, J. Zhang, C. Huang, Z. Lei, B. Chen, Adsorptive separation of dimethyl
disulfide from liquefied petroleum gas by different zeolites and selectivity
study via FT-IR, Sep. Purif. Technol. 125 (2014) 247255.
[14] H. Cui, S.Q. Turn, Adsorption/desorption of dimethylsulfide on activated
carbon modified with iron chloride, Appl. Catal. B 88 (2009) 2531.
[15] J. Chaichanawong, T. Yamamoto, T. Ohmori, A. Endo, Adsorptive
desulfurization of bioethanol using activated carbon loaded with zinc oxide,
Chem. Eng. J. 165 (2010) 218224.
[16] A.J. Hernndez-Maldonado, R.T. Yang, Desulfurization of liquid fuels by
adsorption via p complexation with Cu (I)-Y and Ag-Y zeolites, Ind. Eng.
Chem. Res. 42 (2003) 123129.
[17] J. Lee, H.T. Beum, C.H. Ko, S.Y. Park, J.H. Park, J.N. Kim, B.H. Chun, S.H. Kim,
Adsorptive removal of dimethyl disulfide in olefin rich C4 with ion-exchanged
zeolites, Ind. Eng. Chem. Res. 50 (2011) 63826390.
[18] F. Tian, Q. Shen, Z. Fu, Y. Wu, C. Jia, Enhanced adsorption desulfurization
performance over hierarchically structured zeolite Y, Fuel Process. Technol.
128 (2014) 176182.
[19] K. Takatsu, G. Takegoshi, H. Katsuno, Y. Kawashima, H. Matsumoto, Room
temperature sulfur removal from LPG with adsorbent for fuel cell applicationadsorption behavior of sulfur compounds, J. Jpn. Petrol. Inst. 50 (2007) 200
207.
[20] U. Daiminger, W. Lind, Adsorption processes for natural gas treatment,
Engelhard Corp, USA, 2004, p. 14.
[21] M.J. Ahmed, S.K. Theydan, Modeling of propane separation from light
hydrocarbons by adsorption on 4A molecular sieve zeolite, J. Nat. Gas Sci.
Eng. 18 (2014) 16.

[22] G. Marbn, T. Valds-Sols, A.B. Fuertes, Modeling the breakthrough behavior

of an activated carbon fiber monolith in n-butane adsorption from diluted
streams, Chem. Eng. Sci. 61 (2006) 47624772.
[23] J. Kawakami, S. Tai, S. Morimoto, M. Miyake, Method and Device for Purifying
Propane, JP Patent 162121 A1, 2011.
[24] J. Kawakami, S. Tai, S. Tanaka, Purification Method and Purification System for
Propane, US Patent 0343341 A1, 2014.
[25] T.L. Saleman, G. Li, T.E. Rufford, P.L. Stanwix, K.I. Chan, S.H. Huang, E.F. May,
Capture of low grade methane from nitrogen gas using dual-reflux pressure
swing adsorption, Chem. Eng. J. 281 (2015) 739748.
[26] T. Eriksson, Y. Kiros, Temperature swing adsorption device for oxygenenriched air, J. Clean. Prod. 76 (2014) 174179.
[27] N. Raghavan, D. Ruthven, Numerical simulation of a fixed-bed adsorption
column by the method of orthogonal collocation, AlChE J. 29 (1983) 922925.
[28] B. Salamatinia, A. Kamaruddin, A. Abdullah, Modeling of the continuous copper
and zinc removal by sorption onto sodium hydroxide-modified oil palm frond
in a fixed-bed column, Chem. Eng. J. 145 (2008) 259266.
[29] J. Liu, Y. Yan, H. Zhang, Adsorption dynamics of toluene in composite bed with
microfibrous entrapped activated carbon, Chem. Eng. J. 173 (2011) 456462.
[30] H. Chen, H. Zhang, Y. Yan, Adsorption dynamics of toluene in structured fixed
bed with ZSM-5 membrane/PSSF composites, Chem. Eng. J. 228 (2013) 336
344.
[31] Z. Aksu, F. Gnen, Biosorption of phenol by immobilized activated sludge in a
continuous packed bed: prediction of breakthrough curves, Process Biochem.
39 (2004) 599613.
[32] W.T. Tsai, C.Y. Chang, C.Y. Ho, L.Y. Chen, Adsorption properties and
breakthrough model of 1, 1-dichloro-1-fluoroethane on activated carbons, J.
Hazard. Mater. 69 (1999) 5366.
[33] H. Wakita, Y. Tachibana, M. Hosaka, Removal of dimethyl sulfide and tbutylmercaptan from city gas by adsorption on zeolites, Microporous
Mesoporous Mater. 46 (2001) 237247.
[34] A.B.S. Salem, Naphtha desulfurization by adsorption, Ind. Eng. Chem. Res. 33
(1994) 336340.
[35] A.J. Hernndez-Maldonado, F.H. Yang, G. Qi, R.T. Yang, Desulfurization of
transportation fuels by p-complexation sorbents: Cu(I)-, Ni(II)-, and Zn(II)zeolites, Appl. Catal. B 56 (2005) 111126.
[36] H. Song, X. Wan, M. Dai, J. Zhang, F. Li, H. Song, Deep desulfurization of model
gasoline by selective adsorption over CuCe bimetal ion-exchanged Y zeolite,
Fuel Process. Technol. 116 (2013) 5262.
[37] V.M. Bhandari, C.H. Ko, J.G. Park, S.S. Han, S.H. Cho, J.N. Kim, Desulfurization of
diesel using ion-exchanged zeolites, Chem. Eng. Sci. 61 (2006) 25992608.
[38] M.J. Ahmed, S.K. Theydan, Temperature distribution studies for light
hydrocarbons (propane, n-butane, and iso-butane) adsorption on 5A
molecular sieve zeolite, Heat Mass Transfer 51 (2015) 3338.
[39] M.J. Ahmed, A.H.A.K. Mohammed, A.A.H. Kadhum, Prediction of breakthrough
curves for light hydrocarbons adsorption on 4A molecular sieve zeolite, Korean
J. Chem. Eng. 27 (2010) 752758.
[40] A. Katsigiannis, C. Noutsopoulos, J. Mantziaras, M. Gioldasi, Removal of
emerging pollutants through granular activated carbon, Chem. Eng. J. 280
(2015) 4957.
[41] Z. Xu, J.G. Cai, B.C. Pan, Mathematically modeling fixed-bed adsorption in
aqueous systems, J. Zhejiang Univ. Sci. A 14 (2013) 155176.
[42] I. Prasetyo, D. Do, Adsorption kinetics of light paraffins in AC by a constant
molar flow-rate method, AlChE J. 45 (1999) 18921900.
[43] J. Guo, A.C. Lua, Adsorption of sulphur dioxide onto activated carbon prepared
from oil-palm shells with and without pre-impregnation, Sep. Purif. Technol.
30 (2003) 265273.