Dehydrogenation of propane in a combined membrane reactor with hydrogen-permeable palladium module

Article abstract of DOI:10.1134/S0965544113010039

The basic features of the propane dehydrogenation reaction in a combined membrane reactor with a hydrogen-permeable palladium module and chromia-alumina catalyst (9.0 wt % Cr) in the temperature range of 520-580 C have been studied. Under optimum conditions (T, 550 C; propane space velocity and a stripping gas flow rate of 600-900 h^-1 and 100-250 cm³/min, respectively), the feedstock conversion to propylene increases by a factor of 1.6-2.0. It has been suggested that compliance between the rate of H² formation and its withdrawal through the membrane is of considerable importance.

Full text of DOI:10.1134/S0965544113010039

ISSN 0965ꢀ5441, Petroleum Chemistry, 2013, Vol. 53, No. 1, pp. 27–32. © Pleiades Publishing, Ltd., 2013.
Original Russian Text © L.P. Didenko, V.I. Savchenko, L.A. Sementsova, P.E. Chizhov, L.A. Bykov, 2013, published in Neftekhimiya, 2013, Vol. 53, No. 1, pp. 30–36.
Dehydrogenation of Propane in a Combined Membrane Reactor
with HydrogenꢀPermeable Palladium Module
L. P. Didenko, V. I. Savchenko, L. A. Sementsova, P. E. Chizhov, and L. A. Bykov
Institute of Problems of Chemical Physics, Russian Academy of Sciences,
pr. Akademika Semenova 1, Chernogolovka, Moscow oblast, 142432 Russia
eꢀmail: ludi@icp.ac.ru
Received July 3, 2012
Abstract—The basic features of the propane dehydrogenation reaction in a combined membrane reactor
with a hydrogenꢀpermeable palladium module and chromia–alumina catalyst (9.0 wt % Cr) in the temperaꢀ
ture range of 520–580°C have been studied. Under optimum conditions (T, 550°C; propane space velocity
and a stripping gas flow rate of 600–900 h–1 and 100–250 cm3/min, respectively), the feedstock conversion
to propylene increases by a factor of 1.6–2.0. It has been suggested that compliance between the rate of H2
formation and its withdrawal through the membrane is of considerable importance.
Keywords: palladium membrane module, propane, catalytic dehydrogenation
DOI: 10.1134/S0965544113010039
An alternative to the relevant petrochemical proꢀ 130
cess for manufacturing olefins is catalytic dehydrogeꢀ mesh size of 2
nation of light alkanes using natural gas, associated
µ
m made of finely weaved stainless steel with a
µ
m (Fig. 1).
Investigation of the technical characteristics of the
resulting membrane module (MM) in the temperature
range of 25–550 and a transmembrane pressure of
gas, and refinery gas as inexpensive and available feedꢀ
stock, whose reserves are in great abundance, but the
use is extremely inefficient. Because of endothermicꢀ
ity and thermodynamic limitations of the reaction,
high temperatures are required for achieving acceptꢀ
able yields. However, there are side reactions of feedꢀ
stock cracking and hydrocarbon deposit formation
occurring under these conditions, which reduce the
selectivity for the desired products and lead to catalyst
deactivation. For these reasons, the currently pracꢀ
ticed processes for the manufacturing of olefins from
light hydrocarbons are energyꢀ and capitalꢀintensive
and require new engineering solutions to improve their
performance.
°С
up to 176.5 kPa showed that it has good thermal stabilꢀ
ity and mechanical strength parameter and is close to
composite palladium membranes in properties (perꢀ
meability, Н2 adsorption limitations to the perforꢀ
mance, activation energy of permeability) it [1]. At the
same time, its advantage is 100% hydrogen selectivity.
In this work, a palladium MM was used to remove
Н2 from the products of the reaction of catalytic dehyꢀ
drogenation of propane. The purpose of the work was
to study the main features of this reaction by varying
the temperature and the feedstock and strippingꢀgas
flow rates.
One of these solutions is membrane technology. As
a result of combining the reaction process and hydroꢀ
gen removal from the reaction zone in a one memꢀ
brane reactor, it is possible to increase the proportion
of the desired product in the equilibrium product
composition. Regarding the membrane material, palꢀ
ladium is of great interest since it possesses the highest
Н2 selectivity. Note that monolithic palladium memꢀ
branes (foil) have 100% hydrogen selectivity. However,
their low mechanical strength is a disadvantage. To
ensure the strength, a Pd foil thickness must be at least
EXPERIMENTAL
The reaction was carried out in a combined memꢀ
brane reactor (Fig. 2) comprising a hydrogenꢀpermeꢀ
able module based on Pd/Ag foil (15.0 wt % Ag) of
30
at the Moscow special alloy processing plant The
hydrogenꢀpermeable module was placed between a
feed–retentate supply chamber ( ) and Н2 permeate
withdrawal chamber ( ). In compartment
, 1 cm3
µ
m thickness produced by the technology designed
3
1
3
50 m, but the permeability of the foil is significantly
µ
(0.61 g) of a chromia–alumina catalyst (9.0 wt % Cr,
grain fraction 0.2–0.4 mm) prepared by a modified
coprecipitation technique was placed.
reduced with such a thickness.
In this study, a Pd/Ag foil with a thickness of 30 m
µ
was used as a hydrogen permeable material. To prevent
its destruction during operation, a disk from the foil
The catalyst used in the study was from one batch,
was fixed between two metal grids with a thickness of and it was not subjected to regeneration.
27

28
DIDENKO et al.
(а)
Pd–Ag alloy foil of 30
thickness
µ
m
Stainless steel grid
20 µm
а – cross section
b – top view
(b)
100 µm
Fig. 1. General view of the membrane module.
Propane (100%) was fed to the catalyst through flow rate of the products at the reactor outlet,
inlets located on the periphery of reactor compartꢀ
ment , and the products were withdrawn through the
cm3/min;
uct in the mixture leaving the reactor, %; and
number of carbon atoms in the product.
The selectivity of the formation of hydrocarbon
deposit (SС, %) was calculated as:
X
prod is the volume concentration of a prodꢀ
3
n
is the
central aperture and sent to a chromatograph. As a
driving force for Н2 withdrawal through the memꢀ
brane, a stripping gas, nitrogen, entered counterꢀcurꢀ
rently to propane into compartment 1 of the reactor.
SC = 100 − S
+ SC H + SC H ,
2 6 3 6
(
)
CH4
The gas streams were controlled by RRGꢀ12 gas
flow meters (Elektropribor, Zelenograd). The reactor
was heated an electric oven. The temperature at the
membrane and in the oven was monitored with
where
are the selectivities for methꢀ
S
, S , S
C2H6 C3H6
CH4
ane, ethane, and propylene, respectively, %.
The main characteristics of the reaction were studꢀ
chromel–alumel thermocouples. The product comꢀ ied at
Т
= 520, 550 and 580 C and at various flow rates
°
position was analyzed in the onꢀline mode using a Kriꢀ of feedstock and stripping gas. The effect of Н2
stallꢀ5000 chromatograph with FID and thermal conꢀ removal through the membrane on the propane conꢀ
ductivity detectors. The Н2 content in the products version and the product composition was revealed in
was determined on a column with 13X molecular comparative experiments under the same conditions
sieves (2 mm
hydrocarbon composition of the products was deterꢀ installed instead of the MM for this purpose.
mined using a HPꢀAl/KCl column (0.5 mm 30 m,
80 , helium carrier gas). The concentration of the
products was calculated using the method of absolute
calibration. The accuracy of the analysis was 99.2%.
The propane conversion ( , %) and the selectivity for
products ( , %) were calculated by the formulae:
×
2 m, 50°С, argon carrier gas). The without Н2 removal, a gasꢀtight stainless steel plug was
×
°С
RESULTS AND DISCUSSIONS
The principal propane dehydrogenation reaction :
С3Н8
С3Н6 + Н2,
(1)
α
S
occurs in the test system, as well as side reactions of
propane cracking and the formation of hydrocarbon
deposit (HD).
in
C3H8
out
C3H8
X
Vin − X Vout
α =
×100;
XCinH Vin
3
8
Since both Pd/Ag foil and stainless steel grids can
exhibit catalytic activity in the dehydrogenation of
propane, experiments in the absence of the catalyst
were run prior to the study. Measurements were made
nXпродVout
S =
×100,
in
out
3(X Vin − X Vout)
C3H8
C3H8
where Vin is the propane volumetric flow rate at the
in the reactor with the membrane module at
Т =
in
C3H8
550 and closed exit from the permeate compartꢀ
°
С
reactor inlet, cm3/min;
is the volume concentraꢀ
X
ment. It was found that in this case at a propane space
velocity of 600 h–1, the conversion was 0.9% and the
selectivity for СН4, С2Н4, С2Н6, С3Н6, or HD was
11.5, 6.1, 3.1, 34.6, or 44.7%, respectively. Thus, the
tion of propane in the gas stream at the reactor inlet,
out
C3H8
%;
is the volume concentration of propane in the
products at the reactor outlet, %; Vout is the volumetric
X
PETROLEUM CHEMISTRY Vol. 53
No. 1
2013

DEHYDROGENATION OF PROPANE IN A COMBINED MEMBRANE REACTOR
Section 1 (direction of main gas streams)
29
1
2
3
Stripping gas Gas supply to the Supply of stripping
and hydrogen reaction chamber
gas to the
permeate chamber
Section 2 (direction of gas streams
in the permeate chamber)
4
Gas supply Gas withdrawal
) membrane module, ( ) retentate
Fig. 2. Construction arrangement of the membrane reactor: (
compartment, and ( ) supply and exhaust pipes.
1
) permeate compartment, (
2
3
4
dehydrogenation of propane on the foil and grid
occurs to a small extent and almost does not affect the
performance of the main reaction. Under the above
specified conditions, the yield of propylene was about
30%.
С3Н8 + Н2
→
СН4 + С2Н6.
(2)
An increase in the СН4/С2Н6 ratio indicates that
there is an additional methane formation route. This
route can be the interaction of intermediate, HD preꢀ
cursor forms with Н2 present in the system. It is known
[3, 4] that because of the occurrence of this reaction,
the yield of HD deactivating the catalyst decreases to a
certain extent. The experimental data presented in the
table indicate an increase in the feedstock conversion
to HD in the membrane reactions, especially at the
initial stage. But the contribution of the hydrogenolyꢀ
sis reaction is apparently insufficient for reducing the
catalyst deactivation rate.
Catalytic studies at
Т = 520°С, feeds flow rates of
5–15 cm3/min (from 300 h–1 to 900 h–1), and various
flow rates of the stripping gas showed that the propane
conversion is lower than in the “nonmembrane” reacꢀ
tion in all cases (Fig. 3). In this case, theСН4 content
in the product composition increases. The table shows
the change in the СН4/С2Н6 ratio in the products at
feed space velocity of 300 h–1. One can see that this
ratio increases upon the withdrawal of Н2 through the
membrane, as well as with an increase in the strippingꢀ
gas flow rate.
Data for the reaction run at a temperature of 550
and a propane flow rate of 10 cm3/min (600 h–1) are
presented by curves in Figs. 4a and 4b). One can
°С
1, 2, 3
Methane and ethane are formed by the propane see that in the experiments without hydrogen withꢀ
cracking reaction:
drawal (curves 1), the yield of propylene in the beginꢀ
PETROLEUM CHEMISTRY Vol. 53
No. 1
2013

30
DIDENKO et al.
this case, the Н2 content in the products is reduced to
15% (curves ), which corresponds to removal through
60
50
40
30
20
10
(а)
2
the membrane of ca. 30% of the Н2 produced. The
propylene yield decreases with time and becomes
nearly equal to “nonmembrane” value by the
120th min. The С3Н6 selectivity at the beginning of
the reaction is 81%, the remainder being СН4, С2Н6
and HD. A decrease in the yield of propylene can be
due to deactivation of the catalyst by HD formed. The
increase in their output when removing Н2 through the
membrane is typical to almost all membraneꢀcatalytic
reactions. One of the main reasons for this is the
decrease in the content of Н2 in the reaction mixture.
In addition, in the test membrane–catalyst system
a decrease in the MM permeability can occur due to
the formation of HD on the MM. To verify this
assumption, the MM permeability was studied before
and after the membraneꢀcatalytic reaction. It was
found that the permeability of the MM after the reacꢀ
tion decreases by 20–50% depending on the reaction
1
2
3
0
30
(b)
25
20
15
10
5
1
2
3
conditions. After annealing of the grids in air at
Т =
600 , the original membrane permeability is
°С
restored. Thus, the decrease in the MM permeability
is caused by the buildup of HD on the grids, whereas
the Pd/Ag foil is resistant toward the deposition. Pubꢀ
lished data also show that the formation of HD does
not occur on the surface of bulk Pd and Pd–Ag memꢀ
branes [3].
0
20
40
60
80
100 120
Time, min
The study of the effect of the strippingꢀgas flow rate
on the membraneꢀcatalytic reaction showed that its
performance remains almost unchanged with an
increase in the flow rate to 130 cm3/min, but the yield
of С3Н6 is nearly identical to the “nonmembrane”
Fig. 3. Time variation of propane conversion and H conꢀ
tent in the reaction mixture at
2
T = 520°C: (1) without H
2
stripping and ( ) with H stripping. C H space velocity =
2
2
3 8
value at a flow rate of 40 cm3/min (Fig. 4, curve
3).
–1
3
300 h (5 cm /min). Stripping gas flow rates: (2) 60 and
3
(3) 250 cm /min.
Such an effect of the strippingꢀgas flow rate can be
explained in terms of the results of the earlier study on
the hydrogen permeability of the MM [1]. It was found
that the Н2 permeability increases with an increase in
the stripping gas flow rate to a certain limit, and the
permeability curve reaches a limit at flow rates above
80 cm3/min and ceases to depend on the flow rate of
the stripping gas, which can be explained by the maxꢀ
imum possible removal of hydrogen entering the perꢀ
meate compartment, other conditions being fixed.
Taking into account these data, it can be assumed that
the rate of Н2 removal from the reaction mixture does
not change as the stripping gas flow rate increases from
100 to 130 cm3/min; so, the change in the flow rate of
the stripping gas in this interval does not affect the
reaction performance. But the rate of Н2 removal from
the reaction mixture decreases when the flow rate of
the stripping gas decreases to 40 cm3/min. As is seen
ning of the reaction is ~30% and decreases with an
increase in the reaction time to 24% as a result of gradꢀ
ual deactivation of the catalyst.
The removal of hydrogen through the membrane at
a strippingꢀgas flow rate of 100 cm3/min results in an
increase in the yield of propylene by a factor of 1.6. In
Time variation of the CH4/C2H6 ratio in the products.
T
= 520° )
C, feed flow rate, 5 cm3/min (300 h–1
CH /C H
2 6
4
with H2 withdrawal
Reaction
time, min
without H2
withdrawal
stripping gas flow rate,
cm3/min
from Figs. 4a and 4b (curves 3), it is accompanied by
an increase in the Н2 content of in the reaction mixꢀ
ture, and the yield of С3Н6 in this case is almost the
same as that obtained in the experiments without
removal of Н2. That is, as the rate of Н2 removal
through the MM decreases, there is no increase in the
conversion of feedstock to propylene relative to the
equilibrium rate.
60
250
10
40
4.0
2.0
2.0
2.1
9.8
2.9
2.7
2.6
14.5
5.6
2.3
2.4
80
120
PETROLEUM CHEMISTRY Vol. 53
No. 1
2013

DEHYDROGENATION OF PROPANE IN A COMBINED MEMBRANE REACTOR
31
To explain this experimental result one can refer to
the published data [5–7]. They showed that the memꢀ
braneꢀcatalyst system is an integral unit, where a conꢀ
sistency between the rate of Н2 formation and its
removal through the membrane is required. If it is
absent, then the Н2 withdrawal rate (if the catalyst is
not effective) or the rate of hydrogen transport across
the membrane (if the permeability of the membrane is
insufficient) is the limiting step of the membrane catꢀ
alytic process. In either case, the excess conversion of
the feedstock into the desired product with respect to
the equilibrium values cannot be reached. Taking
these data into account, we can assume that with a
decrease in the strippingꢀgas flow rate to 40 cm3/min
and, hence, in the rate of Н2 removal through the
membrane, this consistency is violated. For this reaꢀ
son, the removal of Н2 does not increase the converꢀ
sion of the feedstock to the target product.
2
50
40
30
20
10
(а)
3
1
0
20
40
60
80
100
120
30
(b)
1
25
20
15
10
5
As illustrated by curve 2 in Fig. 4a, the “membrane
effect” caused by Н2 withdrawal through the memꢀ
brane is reduced over time because of an increased HD
yield. A longer “membrane effect” can be achieved by
increasing the feedstock space velocity and, thus,
reducing the time of its contact with the catalyst. This
3
2
is demonstrated by the experiments at 550°С and a
propane flow rate of 15 cm3/min (900 h–1) (Fig. 5),
which show that the deactivating effect of HD
decreases. In the experiments without the removal of
hydrogen, the yield of С3Н6 is 13% at the beginning of
0
20
40
60
80
100
Time, min
120
the reaction and varies slightly with time (curve
The removal of hydrogen through the membrane
increases the yield of С3Н6 in the beginning of the
reaction by more than twofold (curve ). Moreover,
2).
Fig. 4. Effect of H removal through the membrane on
propane conversion.
(10 cm /min). (1) without H stripping and (2, 3) with H
stripping. Strippingꢀgas flow rates: (
2
–1
T = 550°C, C H flow rate 600 h
1
3 8
3
2
2
the “membrane effect” persists for a long time and
exceeds the equilibrium value by a factor of 1.5 at a
time of 120 min.
2) 100 and (
3
)
3
40 cm /min.
At a propane flow rate of 5 cm3/min (300 h–1) and,
hence, an increased contact time of the feedstock with
the catalyst, the yield of HD increases, which leads to
rapid deactivation of both the catalyst and the MM.
When Н2 is withdrawn through the membrane, the
deactivating effect of HD further increases. The result
is a decrease in the yield of propylene relative to the
equilibrium value upon Н2 removal.
HD, maintaining a sufficiently high yield of olefins at
an optimum temperature.
Catalyst deactivation by the HD is a serious probꢀ
lem in the dehydrogenation of alkanes, and upon the
removal of Н2 through membrane this problem
becomes more heavy due to increasing formation of
HD. In this regard, the membrane–catalytic process
requires frequent catalyst regeneration, and the retenꢀ
tion of the catalytic activity after the regeneration is
important. It has been shown [8] that the activity of
chromia–alumina catalyst used in this work insignifiꢀ
An increase in temperature has a similar effect. At
= 580°С, rapid deactivation of the membrane–catꢀ
Т
alyst system occurs in the beginning of the reaction,
and the variation in feedstock and strippingꢀgas flow
rates over a wide range does not increase the yield of
С3Н6 relative to the equilibrium value.
cantly changes after thermal treatment in air at 550°С
and subsequent reduction with hydrogen at the same
temperature.
In general, the results of the study show that the
optimization of reaction conditions is very important
The results of the present study also indicate an
for the enhancement of feedstock conversion into the important role of optimization of the reaction with
desired product. First of all, it is necessary to optimize respect to the flow rate of the stripping gas. At fixed
the temperature for a maximum yield of the desired values of temperature and feed flow rate, the strippingꢀ
product to be achieved with minimal formation of gas flow rate should provide the rate of Н2 transport
hydrocarbon deposit. A difference in the activation across the membrane corresponding to the rate of its
energies of the principal reaction and the HD formaꢀ formation, which, by our assumption, is a necessary
tion reaction makes it possible to reduce the buildup of condition for increasing the yield of the desired prodꢀ
PETROLEUM CHEMISTRY Vol. 53
No. 1
2013

32
DIDENKO et al.
of withdrawal of Н2 becomes lower than the rate of its
formation.
A major problem with composite palladium memꢀ
branes is their rapid deactivation by coke at temperaꢀ
tures above 500 . It was found that during coking,
carbon is incorporated in the palladium membrane
layer, thereby increasing the number of defects and, as
a result, reducing both the membrane permeability
and Н2 selectivity [10].
In general, analysis of the available results shows
that further development of monolithic palladium
membranes is promising. Their properties, such as
high thermal stability, 100% hydrogen selectivity, and
resistance to deactivation by hydrocarbon deposit, are
important in the processes of highꢀtemperature dehyꢀ
drogenation of light hydrocarbon feedstock compliꢀ
cated by deposition of hydrocarbons.
40
30
20
10
1
2
°
С
0
20
40
60
80
100
120
Time, min
Fig. 5. Time variation of propylene yield at a feed space
velocity of 900 h and
ping and ( ) with H stripping at a stripping gas flow rate of
250 cm /min.
Thus, the important role of optimization of condiꢀ
tions for the membrane–catalytic reaction of propane
dehydrogenation on the temperature and the feed and
strippingꢀgas flow rates in order to obtain the maxiꢀ
mum yield of the desired product with minimal formaꢀ
tion of hydrocarbon deposit was shown. It is suggested
that in order to increase the feedstock conversion to
the desired product, the rate of Н2 removal through
the membrane should match the rate of its formation
under the reaction conditions. Under optimal condiꢀ
–1
T = 550°C: (1) without H stripꢀ
2
2
2
3
uct relative to the equilibrium value. The optimum
conditions in this system are = 550 , the feed space
Т
°С
velocity of 600–900 h–1, and the stripping gas flow rate
of 100 cm3or higher. Under these conditions, the
increase in the feedstock to propylene conversion
makes 1.6–2 times the equilibrium value. In other
cases, either a decrease in the Н2 formation rate
because of catalyst deactivation by hydrocarbon
deposit or reduction in the rate of Н2 removal through
the membrane due to a decrease in the stripping gas
flow rate are observed. In both cases, the corresponꢀ
dence between the catalyst and the membrane is broꢀ
ken and the “membrane effect” is not observed.
tions (Т = 550°С, a propane space velocity of 600–
900 h–1, a stripping gas flow rate of >100 cm3/min),
the yield of propylene in the membrane–catalytic
reaction of propane dehydrogenation increases to
1.6–2.0 times the equilibrium value.
ACKNOWLEDGMENTS
This work was supported by the Ministry of Educaꢀ
tion and Science of the Russian Federation under the
Federal Target Program “Researchꢀandꢀdevelopment
works on priority lines in the scientific–technological
complex of Russia for 2007–2013.”
It is also of interest to compare the results with the
published data on the membrane–catalytic dehydroꢀ
genation of propane using palladium membranes. As
in this paper, the results of other studies indicate an
important role of optimization of the feedstock and
strippingꢀgas flow rates [9]. An increase in the yield of
propylene relative to equilibrium values was obtained
only at low feed rates, which is due to a low membrane
permeability. Thus, in the case of using commercial
tubular membrane made of Pd/Ag (25% Ag), the
“membrane effect” was obtained only at propane
space velocities of 0.1–0.15 h–1 [3], which is ten times
below that of feedstock in the commercial dehydrogeꢀ
nation process. Similarly, for the membrane–catalyst
systems with composite membranes, an increase in the
yield of propylene was obtained at low feed space
velocities. For example, the yield of propylene at
REFERENCES
1. L. P. Didenko, M. S. Voronetskii, L. A. Sementsova,
et al., Al’ternat. Energ. Ekol., No. 10, 154 (2010).
2. L. P. Didenko, A. M. Kolesnikova, M. S. Voronetskii,
Katal. Prom., No. 2, 93 (2011).
3. M. Sheintuch and R. M. Dessau, Chem. Eng. Sci. 51
,
535 (1996).
4. J. P. Collins, R. W. Schwartz, R. Sehgal, et al., Ind. Eng.
Chem. Res. 35, 4398 (1996).
5. P. Ciavarella and D. Casanave, S. Moueddeb, et al.,
Catal. Today 67, 177 (2001).
6. S. Miachon and J. A. Dalmon, Top. Catal. 29, 59
(2004).
7. D. Casanave, P. Ciavarella, and K. Fiaty, Chem. Eng.
Sci. 54, 2807 (1999).
500°С
and a feed space velocity of 0.1 h–1 in [9] was
60–80%, much higher than the equilibrium value.
However, the yield of С3Н6 decreases with an increase
in the propane space velocity and becomes comparaꢀ
ble with the equilibrium value at a velocity of 0.8 h–1 or
higher. The cited authors attribute this to the fact that
at a propane space velocity of 0.8 h–1 or higher, the rate
8. L. P. Didenko, A. M. Kolesnikova, V. I. Savchenko,
et al., Al’ternat. Energ. Ekol., No. 2, 97 (2009).
9. H. Weyten, J. Luyten, K. Keizer, et al., Catal. Today 56
,
3 (2000).
10. R. Dittmeyer, V. Hollein, and K. Daub, J. Mol. Catal.
A: Chem. 173, 166 (2001).
PETROLEUM CHEMISTRY Vol. 53
No. 1
2013