- Email: [email protected]
Energy efficiency of membrane-based fuel processors – PEM fuel cell systems
international journal of hydrogen energy 35 (2010) 3712–3720
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
Energy efficiency of membrane-based fuel processors – PEM fuel cell systems L. Salemme, L. Menna, M. Simeone*, G. Volpicelli Dipartimento di Ingegneria chimica, Universita` degli Studi di Napoli Federico II, P.le V. Tecchio, 80 – 80125 Napoli, Italy
article info
abstract
Article history:
This work presents a simulative energy efficiency analysis performed on fuel processor –
Received 13 October 2009
PEMFC systems, considering methane as fuel and steam reforming or autothermal
Received in revised form
reforming as processes to produce hydrogen. Computation of energy efficiency takes into
20 January 2010
account the power required by the auxiliary units, coupling of the fuel processor with the
Accepted 22 January 2010
fuel cell as well as heat recovery and integration.
Available online 25 February 2010
Two system configurations were simulated and compared: an innovative configuration, based on an integrated membrane reactor directly coupled with the fuel cell, and
Keywords:
a conventional configuration, based on a classic reforming reactor followed by a conven-
Fuel processor
tional CO clean-up section, constituted by water gas shift and preferential CO oxidation
PEM fuel cell
reactors.
Hydrogen
Reforming temperature, plant pressure, steam to methane and oxygen to methane inlet
Membrane reactor
ratios were considered as process parameters. The effect of the addition of steam as sweep
Steam reforming
gas is also presented and discussed.
Autothermal reforming
ª 2010 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved.
Sweep gas
1.
Introduction
Proton Exchange Membrane Fuel Cells (PEMFC) systems have received growing attention in the literature for stationary and portable power generation. The main advantage of PEMFC is the capability to generate power with high efficiency, reduced on-site emissions and fast response to load changes. The ideal fuel for PEMFC is hydrogen with low carbon monoxide content to avoid poisoning of the fuel cell. Despite the great potential of PEMFC, implementation of a hydrogen supply chain on a relatively large scale is quite complex and technical, economical and safety issues still need to be addressed before hydrogen can be distributed from large production sites to end users to generate power with PEMFC. A possible way to overcome these difficulties is to generate hydrogen on site, reforming an easily transportable fuel in a compact fuel processor.
Conventional fuel processors are constituted by a syngas production unit (steam reforming (SR) or autothermal reforming (ATR) reactor) followed by a CO clean-up section, constituted by two water gas shift (WGS) reactors and one preferential CO oxidation (PROX) reactor [1–11]. This wellknown and mature CO clean-up technology, however accounts for most of the fuel processor volume [12] and hinders the possibility to realize a compact system. Enhanced compactness can be achieved by replacing the conventional CO clean-up section with a highly selective hydrogen membrane. Besides compactness, this solution, providing a pure hydrogen stream, allows to operate the PEMFC without a purge stream (generally referred to as anode off-gas). This peculiarity can reduce the energy losses related to the presence of the anode off-gas, that limit the global energy efficiency of the conventional ATR-based fuel processor – PEMFC systems [4].
* Corresponding author. Tel.: þ39 817682269; fax: þ39 812391800. E-mail address: [email protected] (M. Simeone). 0360-3199/$ – see front matter ª 2010 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2010.01.096
international journal of hydrogen energy 35 (2010) 3712–3720
Although several authors have investigated membrane based fuel processors [13–26], only few contributions are available which address the behavior of the entire system, including not only the membrane based fuel processor, but also the fuel cell, the auxiliary units and the heat exchangers [20–26]. Most of these studies refer to liquid fuels [20–23] and only few contributions are available when methane is employed [24–26]. In particular, Campanari et al. [24] analyzed an integrated membrane-SR reactor coupled with a PEMFC, showing that a higher global energy efficiency can be achieved, with respect to conventional fuel processors, if a membrane reactor is employed. Lyubovsky et al. [25] analyzed a methane ATR-based fuel processor – PEMFC system, with a membrane unit placed downstream the WGS unit and operating at high pressure, concluding that high global energy efficiency can be obtained if a turbine is introduced in the system to generate additional power from the expansion of the hot gases produced by the combustion of the membrane retentate stream. In 2009, Salemme et al. [26] analyzed the energy efficiency of a methane ATR-based fuel processor – PEMFC system, with a membrane unit integrated in the WGS reactor and showed that a high global energy efficiency is obtained when a steam sweep gas is sent to the membrane reactor in counter-current mode. This work presents a comprehensive analysis of fuel processor – PEMFC systems based on integrated membrane reactor technology for pure hydrogen production, considering methane as a fuel and SR and ATR as reforming processes. The manuscript analyzes four fuel processor configurations, each coupled with a PEMFC: FP.A) SR reactor, followed by two WGS reactors and a PROX reactor. FP.B) ATR reactor, followed by two WGS reactors and a PROX reactor. FP.C) Integrated membrane-SR reactor. FP.D) Integrated membrane-ATR reactor. Operating parameters, such as steam to methane and oxygen to methane inlet ratios, reforming temperature, as well as pressure, are screened to identify the conditions that maximize global energy efficiency. The effect of the addition of steam as sweep gas is also presented and discussed.
2.
3713
Methodology
The simulations were performed in stationary conditions, by using the commercial package Aspen Plus [27]. The selected property method was Peng–Robinson and the component list was restricted to CH4, O2, N2, H2O, CO, H2 and CO2. Methane was considered as fuel, fed at 25 C and 1 atm, with a constant flow rate of 1 kmol/h. Feed to the system was completed with a liquid water stream and an air stream (for the ATR Fuel Processors only), fed at 25 C and 1 atm. Aspen Plus was used to calculate product composition throughout the plant as well as energy requirements of each unit. The configurations simulated (flowsheets) are presented in the Fuel processors and Model libraries section, where the assumptions and the model libraries used to simulate the process are presented. The quantities employed to calculate energy efficiency are defined in the System efficiency section.
2.1.
Fuel processors description and model libraries
Fig. 1 reports the flowsheet of a conventional SR-based fuel processor (FP.A) coupled with a PEM fuel cell. The fuel processor consists of a reforming and a CO clean-up section. The reforming section is an isothermal reactor (SR), modeled by using model library RGIBBS. The CO clean-up section consists of a high (HTS) and low (LTS) temperature water gas shift reactor followed by a PROX reactor. HTS and LTS were modeled by using model library RGIBBS; the reactors were considered as adiabatic and methane was considered as an inert in order to eliminate the undesired methanation reaction, kinetically suppressed on a real catalytic system. The PROX reactor was modeled as an isothermal stoichiometric reactor, RSTOIC, in which two reactions take place: oxidation of CO to CO2 and oxidation of H2 to H2O; the air fed to the PROX reactor (AIR-PROX) was calculated in order to achieve a 50% oxygen excess with respect to the stoichiometric amount required to convert all the CO to CO2. The RSTOIC specifics were completed with the assignment of total conversion of CO and O2. The PEMFC was simulated as the sequence of the anode, modeled as an ideal separator, SEP, and the cathode, modeled as an isothermal stoichiometric reactor, RSTOIC. The presence
Fig. 1 – Flowsheet of fuel processor FP.A coupled with a PEMFC.
3714
international journal of hydrogen energy 35 (2010) 3712–3720
of the SEP unit allows to model a purge gas (anode off-gas, AOG) required for mass balance reasons, if the hydrogen stream sent to the PEMFC is not 100% pure. In agreement with the literature, the hydrogen split fraction in the stream H2 at the outlet of the SEP was fixed at 0.75 [6,11], whereas the split fractions of all the other components were taken as 0. The RSTOIC unit models the hydrogen oxidation reaction occurring in the fuel cell. The reactor specifics were completed by considering an operating temperature of 80 C and pressure of 1 atm; the inlet air at the cathode (AIR-FC), fed at 25 C and 1 atm, guarantees a 50% excess of oxygen in the RSTOIC reactor. In agreement with the literature [28], these conditions were considered as sufficient to assign total hydrogen conversion. The anode off-gas is sent to a burner, modeled as an adiabatic RSTOIC, working at atmospheric pressure with 50% excess air (AIR-B). The heat required by the SR reactor working at temperature TSR is supplied by the heat exchanger H-B, where the stream coming from the burner is cooled to TSR þ 10 C. Model library HEATER was used for this purpose. An additional methane stream (CH4-B) is sent to the burner to satisfy the global heat demand of the system, when needed. Fig. 2 reports the flowsheet of a conventional ATR-based fuel processor (FP.B) coupled with a PEM fuel cell. For the sake of simplicity, the description of the flowsheet will be carried out by indicating the differences with respect to the flowsheet of Fig. 1, which are concentrated only in the reforming section. Indeed, in FP.B, the reforming section is constituted by an adiabatic reactor (ATR), modeled by using model library RGIBBS. The integrated membrane reactors were simulated by discretizing the membrane reactor with N reactor-separator units. With this approximation, reactors are assumed to reach equilibrium and the separators are modeled as ideal separators, SEP, whose output is given by a stream of pure hydrogen (permeate) and a stream (retentate) containing the unseparated hydrogen and all the balance. The amount of hydrogen separated (niH2 ;P ) is calculated assuming equilibrium between the partial pressure in the retentate and permeate side, according to Eq. (1): PR $
niH2 ;R niH2 ;P niR niH2 ;P
hydrogen in the retentate stream; niR is the total mole flow of the retentate stream; PiH2 ;P is hydrogen partial pressure in the permeate side of the membrane, calculated as:
PiH2 ;P ¼
niH2 ;P $PP i nH2 ;P þ nSG
(2)
where PP is the pressure in the permeate side of the membrane, taken as 1 atm in all the simulations, and nSG represents the molar flow rate of steam sweep gas (SG), which can be introduced to increase the separation driving force in the membrane. When present, the sweep gas is produced by liquid water, fed at 25 C and 1 atm to a heat exchanger and sent to the membrane reactor in counter-current mode. The high hydrogen purity of the stream sent to the PEMFC allows to take as zero the anode off-gas, simplifying the model of the PEMFC to the cathode side (RSTOIC) only. Figs. 3 and 4 report the flowsheet used to simulate a membrane-SR reactor (FP.C) and a membrane-ATR reactor (FP.D) coupled with a PEMFC, respectively. The membrane-SR reactor was discretized with 30 units (Fig. 3) while the membrane-ATR reactor was discretized with 20 units (Fig. 4); the number of units required to model each membrane reactor was assessed by repeating the simulations with an increasing number of reactor-separator units and was chosen as the minimum value above which global efficiency remained constant within 0.1%. All the reactors were considered as operating at the same pressure. Auxiliary power units for compression of the reactants were considered in the configurations where pressure was explored as an operation variable, i.e. FP.C and FP.D. 100 C was chosen as the minimum exhaust gas temperature (Tex), when compatible with the constraint of a positive driving force in the heat exchangers present in the plant. Finally, it is worth mentioning that the assumptions made to model the system are the same for all the configurations investigated and do not affect the conclusions drawn in this comparative analysis.
! ¼ PiH2 ;P
(1)
where PR is the pressure in the retentate side of the membrane, equals to reactor pressure; niH2 ;R is the mole flow of
2.2.
System efficiencies
Energy efficiency, h, was defined according to the following Eq. (3):
Fig. 2 – Flowsheet of fuel processor FP.B coupled with a PEMFC.
international journal of hydrogen energy 35 (2010) 3712–3720
3715
Fig. 3 – Flowsheet of fuel processor FP.C coupled with a PEMFC.
h¼
Pe Pa ðnCH4 þ nCH4 ;B Þ$LHVCH4
(3)
where Pa is the electric power required by the auxiliary units for compression of methane, air and water, nCH4 is the inlet molar flow rate of methane to the reactor, nCH4 ;B is the molar flow rate of methane fed to the burner, LHVCH4 is the lower heating value of methane and Pe is the electric power generated by the fuel cell, calculated as: Pe ¼ nH2 $LHVH2 $hFC
PN
HR ¼
niH ;P PN 2 i nH2 ;R þ i¼1 nH2 ;P i¼1
(5)
where niH2 ;P is the molar flow rate of hydrogen separated by the i-th membrane unit, nH2 ;R is the molar flow rate of hydrogen in the retentate stream (RETENT) at the exit of the last separator and N is the number of separators. According to the definitions given above, h can be expressed as it follows:
(4)
where nH2 is the molar flow rate of hydrogen that reacts in the fuel cell, LHVH2 is the lower heating value of hydrogen and hFC is the electrochemical efficiency of the fuel cell, taken as 0.6 [29]. In the membrane-based fuel cell systems (configurations FP.C and FP.D), an important parameter is the global hydrogen recovery (HR), defined as:
h ¼ hR $HR$hFC fa
(6)
where fa is the fraction of inlet methane required to run the auxiliary units, defined by Eq. (7): fa ¼
Pa $ð1 aÞ ðnCH4 $LHVCH4 Þ
Fig. 4 – Flowsheet of fuel processor FP.D coupled with a PEMFC.
(7)
3716
international journal of hydrogen energy 35 (2010) 3712–3720
Table 1 – Simulation results and best operating conditions for FP.A and FP.B.
3.
Simulation results
FP.A (SR) FP.B (ATR)
xCH4
a
h
TEx ( C)
91.0 98.8
0.0 0.0
48.0 38.5
226 444
Operating conditions
FP.A (SR) FP.B (ATR)
P (atm)
H2O/CH4
O2/CH4
TSR ( C)
1 1
2.5 3.5
– 0.56
670 –
where a is the ratio between methane flow rate fed to the burner and total methane flow rate fed to the system, defined by Eq. (8): a¼
nCH4 ;B nCH4 þ nCH4 ;B
(8)
hR is the hydrogen production efficiency in the fuel processor, defined by Eq. (8): P nH2 ;R þ Ni¼1 niH2 ;P $LHVH2 hR ¼
2.3.
(9)
$ð1 aÞ
nCH4 $LHVCH4
Model analysis tools
The Calculator Tool was used to calculate the amount of hydrogen separated by the membrane unit, by introducing the relation reported in Eq. (1) and to calculate the air flows required by the PROX reactor, by the PEM fuel cell and by the burner, respecting the 50% oxygen excess specification. The Sensitivity Tool was used to evaluate energy efficiency and thermal efficiency, varying H2O/CH4 inlet ratio, O2/CH4 inlet ratio, temperature of SR reactor, pressure and steam sweep gas flow rate. The Design Specification Tool was used to determine the amount of methane fed to the burner (CH4-B), in order to satisfy the heat duty of the SR reactor, or to pre-heat the inlet streams of the ATR reactor to 350 C.
Simulations on the conventional systems were performed at 1 atm exploring as operative parameters the molar ratio between water and methane at reactor inlet (H2O/CH4) and reactor temperature, for the SR-based fuel processor (FP.A), and water to methane and oxygen to methane (O2/CH4) inlet ratio, for the ATR-based fuel processor (FP.B). Table 1 reports the simulation results and the value of the operative parameters given as simulation input that maximize the energy efficiency, h, for FP.A and for FP.B, respectively. In agreement with Ersoz et al. [4], FP.A shows the highest global efficiency (48.0%) at TSR ¼ 670 C and H2O/CH4 ¼ 2.5. It should be noticed that, in the optimal conditions, methane conversion (xCH4 ) is lower than unity; however, the non converted methane is not energetically wasted, since it contributes to the energy content of the AOG, used to sustain the endothermicity of the SR reactor. In this conditions, no addition of methane to the burner is needed (a ¼ 0). For FP.A the minimum exhaust gas temperature achievable is 226 C, indeed, further heat recovery is hindered by temperature cross over in the heat exchangers. As reported in [26], FP.B shows the highest global efficiency (38.5%) at O2/CH4 ¼ 0.6 and H2O/CH4 ¼ 3.5; the value of h is significantly lower than what achieved with FP.A, mainly due to the autothermal nature of the ATR process, that limits the possibility to recover the energy content of the AOG. This reflects into a higher exhaust gas temperature in FP.B (444 C) than in FP.A (226 C). The simulations on the systems with integrated membrane reactors (FP.C and FP.D) were performed considering also pressure and sweep gas to methane inlet ratio (SG/CH4) as operative parameters to be optimized in the range 3–15 atm and 0–2, respectively. Fig. 5a reports the energy efficiency of FP.C as a function of pressure, at SG/CH4 equals to zero (continuous line). Energy efficiency rapidly increases with pressure in the range 3–5, where no methane addition to the burner is required to sustain the endothermic steam reforming reaction. As pressure increases above 5 atm ca., energy efficiency continues to grows with pressure, but at a lower rate, because methane addition to the burner becomes necessary. The dotted line,
b
70
70
60
60
50
50
(%)
(%)
a
Results
40 30
40 30
SR without CH4-B SR
20
20
10
ATR
10
0
0 3
5
7
9
P (atm)
11
13
15
3
5
7
9
11
13
P (atm)
Fig. 5 – Energy efficiency h as a function of pressure for FP.C (a) and FP.D (b).
15
3717
international journal of hydrogen energy 35 (2010) 3712–3720
Table 2 – Simulation results for FP.C. Operating conditions: SG/CH4 [ 0; H2O/CH4 [ 2.5; T [ 600 8C.
50
xCH4
a
TEx ( C)
HR
fa
hR
h
3.0 5.0 7.0 9.0 12.0 15.0
70.6 86.3 91.8 94.5 96.6 97.6
0.0 1.8 12.8 17.2 20.4 22.0
803.8 100.0 100.0 100.0 100.0 100.0
58.0 85.9 91.9 94.4 96.2 97.1
0.5 0.7 0.8 0.9 1.0 1.1
80.4 98.7 94.2 92.6 91.5 91.0
27.5 50.2 51.2 51.5 51.8 51.9
45
(%)
P (atm)
SR ATR
40 35
superimposed to Fig. 5 as an aid to this discussion, represents the value of energy efficiency that would be calculated if the methane sent to the burner was not factored in the computation. The trend of h vs P is the combined effect of hydrogen recovery (HR), reforming efficiency (hR), the power of the auxiliary units (related to fa), whose values are reported in Table 2 together with the value of methane conversion (xCH4 ) and fraction of methane sent to the burner (a). In particular, HR increases with pressure due to the increase of hydrogen separation driving force through the membrane; hR is positively influenced by the trend of HR with pressure, due to the positive effect on reaction equilibrium of increasing hydrogen separation (see xCH4 ) and it is negatively influenced by the addition of methane to the burner, since only part of the energy content of the methane sent to the burner is actually transferred to the reacting mixture. fa increases with pressure, due to increasing compression ratios. To complete the picture, it should be kept in mind that the heating value of the retentate decreases with pressure, as a consequence of higher xCH4 and HR. This, in turn, influences the quantity of methane sent to the burner to sustain the endothermic steam reforming reaction. In the low pressure range, the positive effect of HR and xCH4 on energy efficiency overrules the negative effect of fa and a. The plateau value reached at higher pressure indicates that the drawback of fa and a compensates the positive effect of HR and xCH4 . As reported by [24] the energy efficiency of an SR-based system can be increased if a membrane reactor is used (FP.C), in place of a conventional reactor. This is due to the possibility to recover a higher amount of heat in FP.C than in FP.A. Indeed the heat exchanger network needed in FP.A has to satisfy the temperature requirements of the Shift and PROX reactors resulting in a higher exhaust gas temperature (226 C), while in FP.C the heat exchanger network allows to
30 0.0
0.5
1.0
1.5
2.0
SG/CH4 Fig. 6 – Energy efficiency h as a function of SG/CH4 ratio for FP.C (continuous line) and FP.D (dashed line).
cool the exhaust gas to 100 C (as chosen in the methodology), without any temperature cross over. Fig. 5b reports the energy efficiency of the system with a membrane ATR-reactor (FP.D) as a function of pressure. As for the case of FP.C, energy efficiency shows a continuous increase with pressure, but the values are significantly lower, due to limited recovery of the energy contained in the retentate stream and to the negative contribution of the compressor (see Tex and fa in Table 3). It should be noted that, in FP.D, the maximum value of h (37.2%) is even lower than what is obtained with the conventional ATR reactor [h of FP.B (38.5%)]. This should be attributed to the fact that, notwithstanding the absence of the AOG stream, dilution of the reacting mixture with nitrogen reduces HR (affecting, in turn, also xCH4 ) leading to a retentate with relatively high amount of methane and hydrogen, whose heating value cannot be totally recovered. It should be pointed out that, due to the exothermic nature of the reactions, no additional methane to the burner is required, i.e. a ¼ 0, and the exhaust gas stream leaves the plant at quite high temperatures. Data are reported in Table 3. Fig. 6 reports the energy efficiency of FP.C and FP.D as a function of SG/CH4 at constant pressure (10 atm). Simulation details are reported in Tables 4 and 5. The behavior of h as a function of SG/CH4 is due to the combined effect of HR and hR. Hydrogen recovery is enhanced by the presence of the sweep gas, as a consequence of reduced
Table 4 – Simulation results for FP.C. Operating conditions: H2O/CH4 [ 2.5; P [ 10 atm; T [ 600 8C.
Table 3 – Simulation results for FP.D. Operating conditions: SG/CH4 [ 0; H2O/CH4 [ 1.2; O2/CH4 [ 0.5. P (atm)
xCH4
a
TEx ( C)
HR
fa
hR
h
3.0 5.0 7.0 9.0 12.0 15.0
85.2 88.4 90.0 90.9 91.8 92.4
0.0 0.0 0.0 0.0 0.0 0.0
1369.1 1248.1 1178.1 1132.7 1089.8 1063.6
5.8 60.2 75.6 82.7 88.0 90.9
1.6 2.6 3.3 3.9 4.6 5.2
60.1 67.6 71.4 73.8 76.2 77.8
0.5 21.8 29.1 32.7 35.6 37.2
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
0.0 0.1 0.5 1.0 1.5 2.0
95.4 99.8 100.0 100.0 100.0 100.0
18.6 25.7 28.5 30.0 31.5 33.0
100.0 100.0 100.0 100.0 100.0 100.0
95.1 99.1 100.0 100.0 100.0 100.0
0.9 0.8 0.8 0.8 0.8 0.8
92.1 89.0 86.8 84.9 83.1 81.4
51.6 52.1 51.3 50.2 49.1 48.1
3718
international journal of hydrogen energy 35 (2010) 3712–3720
Table 5 – Simulation results for FP.D. Operating conditions: H2O/CH4 [ 1.2; P [ 10 atm; O2/CH4 [ 0.5.
Table 6 – Simulation results for FP.C. Operating conditions: H2O/CH4 [ 2.5; T [ 600 8C.
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
0.0 0.1 0.5 1.0 1.5 2.0
91.3 95.1 99.1 100.0 100.0 100.0
0.0 0.0 0.0 0.4 4.0 7.0
1115.9 894.7 463.8 100.0 100.0 100.0
84.9 95.0 99.1 99.9 100.0 100.0
4.1 4.1 4.1 4.1 3.9 3.8
74.8 81.5 88.6 90.8 88.1 85.5
34.0 42.4 48.6 50.3 48.9 47.5
hydrogen partial pressure in the permeate side, and this holds true, both for FP.C and FP.D; on the other hand, hR decreases continuously with SG/CH4, for the case of FP.C, while it shows a non monotone trend for the case of FP.D. The continuous decrease of hR is due to the fact that, in FP.C, production of sweep gas is always coupled with addition of methane to the burner, that overrules the increment of HR. For this reason, being h combination of hR and HR, after an initial small increment, it decreases with addition of sweep gas. The non monotone trend of hR with SG/CH4 present in FP.D is due to the fact that in an autothermal reforming membrane system, up to a certain amount, sweep gas can be produced by employing the excess energy which cannot be recovered in other ways, i.e. a ¼ 0. Therefore, as long as production of sweep gas does not require addition of methane to the burner, hR, increases with SG/CH4 while a further increase of SG/CH4 above this value causes an increase of a, but only marginal increments of HR, thus leading to a decrease of hR and h. As shown in Fig. 6, energy efficiency of FP.D can be highly improved by adding sweep gas, increasing from 34.0% (SG/CH4 ¼ 0) to 50.3% (SG/ CH4 ¼ 1.0). Fig. 7 reports energy efficiency of FP.C and FP.D as a function of pressure, calculated (for each pressure) in correspondence to the best SG/CH4 values. Data are reported in Tables 6 and 7. In the pressure range investigated, h of FP.C has a monotone increasing trend and the corresponding optimal value of SG/CH4 decreases from 0.5 to 0.1, thus indicating
(%)
52
P (atm) 3.0 5.0 7.0 9.0 12.0 15.0
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
0.5 0.2 0.1 0.1 0.1 0.1
99.2 99.4 99.6 99.7 99.1 99.8
24.8 24.9 25.1 25.3 25.6 25.6
100.0 100.0 100.0 100.0 100.0 100.0
97.4 98.1 98.5 98.8 99.1 99.2
0.4 0.5 0.7 0.8 0.9 1.0
88.0 88.8 89.1 89.2 89.2 89.3
51.1 51.8 52.0 52.1 52.1 52.1
that the best way to operate a membrane steam reforming system is to increase the pressure without addition of sweep gas. On the other hand, h of FP.D shows a nonmonotone trend with a maximum of 50.6% at 7 atm. In the pressure range investigated, the optimal value of SG/CH4 decreases from 1.5 to 1.0. Column 7 of Tables 6 and 7 reports the power required by the auxiliary units ( fa), showing that power consumption is much higher in FP.D than in FP.C. This factor is the main responsible for the drop of h of FP.D above 7 atm. Fig. 7 shows that the best way to operate an autothermal reforming membrane system is to moderately increase pressure and to add sweep gas to improve HR. Finally, it should be noted that the addition of sweep gas in FP.D allows to reach energy efficiency values significantly higher than 38.5% (FP.B) and similar to the energy efficiency of steam reforming based systems.
3.1.
Analysis of thermal stress on the membrane
In analyzing possible membrane reactor configurations, it should be kept in mind that, due to limited thermal stability of the highly selective membranes, membrane units should not be exposed to temperatures higher than 600 C. While FP.C always meets this constraint (since reactor temperature is fixed at 600 C), FP. D does not. Indeed, in the optimal conditions, the first reactors reach temperatures as high as 720 C. It is also well known that ATR reactors are characterized by a non-uniform temperature profile, with a hot spot in the initial part of the catalyst bed [30]. Therefore, the actual realization of an integrated membrane reactor would require significant improvements of membrane compatibility with high temperatures. A more realistic configuration of an ATR-based membrane reactor should consider a first ATR reactor, where most of the methane oxidation takes place,
SR ATR
51
Table 7 – Simulation results for FP.D. Operating conditions: H2O/CH4 [ 1.2; O2/CH4 [ 0.5.
50
P (atm)
49 3
6
9
12
P (atm) Fig. 7 – Energy efficiency h as a function of pressure calculated at the best SG/CH4 value.
15
3.0 5.0 7.0 9.0 12.0 15.0
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
1.5 1.1 1.0 1.0 1.0 1.0
99.4 99.2 99.9 100.0 100.0 100.0
0.0 0.0 0.0 0.0 0.0 0.0
100.0 100.0 100.0 100.0 100.0 100.0
97.9 99.2 99.6 99.8 99.9 99.9
1.6 2.6 3.3 3.9 4.6 5.2
87.0 89.1 90.2 90.8 91.4 91.6
49.5 50.5 50.6 50.5 50.2 49.8
international journal of hydrogen energy 35 (2010) 3712–3720
followed by a membrane reactor, interposing between the two units a heat exchanger to cool down the temperature before entrance into the membrane reactor, so that the membranes are never exposed to temperatures higher than 600 C. With this configuration, energy efficiency becomes 48.5% and the best operating conditions are P ¼ 7 atm; O2/ CH4 ¼ 0.5; H2O/CH4 ¼ 1.7; SG/CH4 ¼ 1.0.
4.
Conclusion
A simulative energy efficiency analysis was performed on fuel processor – PEMFC systems, considering methane as fuel and steam reforming and autothermal reforming as processes to produce hydrogen. Four configurations were investigated, as follows: FP.A) SR reactor, followed by two WGS reactors and a PROX reactor. FP.B) ATR reactor, followed by two WGS reactors and a PROX reactor. FP.C) Integrated membrane-SR reactor. FP.D) Integrated membrane-ATR reactor. The comparison between the steam reforming based systems (FP.C vs FP.A) showed that the employment of a membrane reactor can increase system efficiency from 48% to 52.1%. Such an efficiency increase is obtained by increasing system pressure from 1 to 15 atm and by adding small amounts of sweep gas. The impact of sweep gas addition on energy efficiency as well as the amount of sweep gas needed, is higher when system pressure is low, i.e. at P ¼ 3 atm, energy efficiency increases from 27.5% to 51.1% if a sweep gas to methane ratio equal to 0.5 is employed; while at P ¼ 15 atm the improvement is only from 51.9% to 52.1%, with sweep gas to methane ratio equal to 0.1. The comparison between the autothermal reforming systems (FP.D vs FP.B) showed that energy efficiency can be improved from 38.5% to 50.6%, if a membrane reactor is employed. To obtain such an energy efficiency improvement, sweep gas addition is required. Indeed, without the addition of sweep gas, the employment of a membrane reactor is not energetically convenient in the pressure range investigated. The best energy efficiency of 50.6 was achieved for P ¼ 7 atm and sweep gas to methane ratio equals to 1.0. Finally, the energy efficiency of FP.D is reduced to 48.5% if a constraint of 600 C is imposed on the maximum temperature achieved in the membrane units.
Acknowledgements This work has been carried out within the framework of the project ‘‘Pure hydrogen from natural gas through reforming up to total conversion obtained by integrating chemical reaction and membrane separation’’ financially supported by MUR (FISR DM 17/12/2002 – year 2001).
3719
references
[1] Xu J, Froment GF. Methane steam reforming, methanation and water-gas shift. Intrinsic kinetics. AIChE J 1989;35:88–96. [2] Semelsberger TA, Brown LF, Borup RL, Inbody MA. Equilibrium products from autothermal processes for generating hydrogen-rich fuel-cell feeds. Int J Hydrogen Energy 2004;29:1047–64. [3] Dupont V, Ross AB, Hanley I, Twigg MCV. Unmixed steam reforming of methane and sunflower oil: a single-reactor process for H2-rich gas. Int J Hydrogen Energy 2006;32:67–79. [4] Ersoz A, Olgun H, Ozdogan S. Reforming options for hydrogen production from fossil fuels for PEM fuel cells. J Power Sources 2006;154:67–73. [5] Roychoudhury S, Lyubovsky M, Walsh D, Chu D, Kallio E. Design and development of a diesel and JP-8 logistic fuel processor. J Power Sources 2006;160:510–3. [6] Francesconi JA, Mussati MC, Mato RO, Aguirre PA. Analysis of the Energy efficiency of a integrated ethanol processor for PEM fuel cell systems. J Power Sources 2007;167:151–61. [7] Docter A, Lamm A. Gasoline fuel cell systems. J Power Sources 1999;84:194–200. [8] Ahmed S, Krumpelt M. Hydrogen from hydrocarbon fuels for fuel cells. Int J Hydrogen Energy 2001;26:291–330. [9] Seo YS, Shirley A, Kolaczkowski ST. Evaluation of thermodynamically favourable operating condition for production of hydrogen in three different reforming technologies. J Power Sources 2002;108:213–25. [10] Cutillo A, Specchia S, Antonini M, Saracco G, Specchia V. Diesel fuel processor for PEM fuel cells: two possible alternatives (ATR vs SR). J Power Sources 2006;154:379–85. [11] Doss ED, Kumar R, Ahluwalia RK, Krumpelt M. Fuel processors for automotive fuel cell systems: a parametric analysis. J Power Sources 2001;102:1–15. [12] Qi A, Peppley B, Karan K. Integrated fuel processors for fuel cell application. Fuel Process Technol 2007;88:3–22. [13] Gallucci F, Basile A. Pd–Ag membrane reactor for steam reforming reactions: a comparison between different fuels. Int J Hydrogen Energy 2008;33:1671–87. [14] Gallucci F, Basile A, Tosti S, Iulianelli A, Drioli E. Methanol and ethanol steam reforming in membrane reactors: an experimental study. Int J Hydrogen Energy 2007;32:1201–10. [15] Chiappetta G, Clarizia G, Drioli E. Design o fan integrated membrane system for a high level hydrogen purification. Chem Eng J 2006;124:29–40. [16] Tong J, Matsumuura Y. Effect of catalytic activity on methane steam reforming in hydrogen-permeable membrane reactor. Appl Catal A General 2005;286:226–31. [17] Li A, Lim CJ, Grace JR. Staged-separation membrane reactor for steam methane reforming. Chem Eng J 2008;138:452–9. [18] Wang L, Murata K, Inaba M. Production of pure hydrogen and more valuable hydrocarbons from ethane on a novel highly active catalyst system with a Pd-based membrane reactor. Catal Today 2003;82:99–104. [19] Fernandes FAN, Soares AB. Methane steam reforming modeling in a palladium membrane reactor. Fuel 2006;85: 569–73. [20] Pearlman JB, Bhargav A, Shields EB, Jackson GS, Hearn PL. Modeling efficiency and water balance in PEM fuel cell systems with liquid fuel processing and hydrogen membranes. J Power Sources 2008;185:1056–65. [21] Lattner JR, Harold MP. Comparison of methanol-based fuel processors for PEM fuel cell systems. Appl Catal B Environ 2005;56:149–69. [22] Lattner JR, Harold MP. Comparison of conventional and membrane reactor fuel processors for hydrocarbon-based PEM fuel cell systems. Int J Hydrogen Energy 2004;29:393–417.
3720
international journal of hydrogen energy 35 (2010) 3712–3720
[23] Manzolini G, Tosti S. Hydrogen production from ethanol steam reforming: energy efficiency analysis of traditional and membrane processes. Int J Hydrogen Energy 2008;33: 5571–82. [24] Campanari S, Macchi E, Manzolini G. Innovative membrane reformer for hydrogen production applied to PEM microcogeneration: simulation model and thermodynamic analysis. Int J Hydrogen Energy 2008;33:1361–73. [25] Lyubovsky M, Walsh D. Reforming system for co-generation of hydrogen and mechanical work. J Power Sources 2006;157: 430–7. [26] Salemme L, Menna L, Simeone M. Analysis of the energy efficiency of innovative ATR-based PEM fuel cell system with
[27] [28]
[29]
[30]
hydrogen membrane separation. Int J Hydrogen Energy 2009; 34:6384–92. Aspen Technology Inc., http://www.aspentech.com. Ratnamala GM, Shah N, Mehta V, Rao PV. Integrated fuel cell processor for a 5-kW proton – exchange membrane fuel cell. Ind Eng Chem Res 2005;44:1535–41. Hou Y, Zhuang M, Wan G. The analysis for the efficiency properties of the fuel cell engine. Renew Energy 2007;32: 1175–86. Simeone M, Salemme L, Allouis C. Reactor temperature profile during autothermal methane reforming on Rh/Al2O3 catalyst by IR imaging. Int J Hydrogen Energy 2008;33: 4798–808.
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
Energy efficiency of membrane-based fuel processors – PEM fuel cell systems L. Salemme, L. Menna, M. Simeone*, G. Volpicelli Dipartimento di Ingegneria chimica, Universita` degli Studi di Napoli Federico II, P.le V. Tecchio, 80 – 80125 Napoli, Italy
article info
abstract
Article history:
This work presents a simulative energy efficiency analysis performed on fuel processor –
Received 13 October 2009
PEMFC systems, considering methane as fuel and steam reforming or autothermal
Received in revised form
reforming as processes to produce hydrogen. Computation of energy efficiency takes into
20 January 2010
account the power required by the auxiliary units, coupling of the fuel processor with the
Accepted 22 January 2010
fuel cell as well as heat recovery and integration.
Available online 25 February 2010
Two system configurations were simulated and compared: an innovative configuration, based on an integrated membrane reactor directly coupled with the fuel cell, and
Keywords:
a conventional configuration, based on a classic reforming reactor followed by a conven-
Fuel processor
tional CO clean-up section, constituted by water gas shift and preferential CO oxidation
PEM fuel cell
reactors.
Hydrogen
Reforming temperature, plant pressure, steam to methane and oxygen to methane inlet
Membrane reactor
ratios were considered as process parameters. The effect of the addition of steam as sweep
Steam reforming
gas is also presented and discussed.
Autothermal reforming
ª 2010 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved.
Sweep gas
1.
Introduction
Proton Exchange Membrane Fuel Cells (PEMFC) systems have received growing attention in the literature for stationary and portable power generation. The main advantage of PEMFC is the capability to generate power with high efficiency, reduced on-site emissions and fast response to load changes. The ideal fuel for PEMFC is hydrogen with low carbon monoxide content to avoid poisoning of the fuel cell. Despite the great potential of PEMFC, implementation of a hydrogen supply chain on a relatively large scale is quite complex and technical, economical and safety issues still need to be addressed before hydrogen can be distributed from large production sites to end users to generate power with PEMFC. A possible way to overcome these difficulties is to generate hydrogen on site, reforming an easily transportable fuel in a compact fuel processor.
Conventional fuel processors are constituted by a syngas production unit (steam reforming (SR) or autothermal reforming (ATR) reactor) followed by a CO clean-up section, constituted by two water gas shift (WGS) reactors and one preferential CO oxidation (PROX) reactor [1–11]. This wellknown and mature CO clean-up technology, however accounts for most of the fuel processor volume [12] and hinders the possibility to realize a compact system. Enhanced compactness can be achieved by replacing the conventional CO clean-up section with a highly selective hydrogen membrane. Besides compactness, this solution, providing a pure hydrogen stream, allows to operate the PEMFC without a purge stream (generally referred to as anode off-gas). This peculiarity can reduce the energy losses related to the presence of the anode off-gas, that limit the global energy efficiency of the conventional ATR-based fuel processor – PEMFC systems [4].
* Corresponding author. Tel.: þ39 817682269; fax: þ39 812391800. E-mail address: [email protected] (M. Simeone). 0360-3199/$ – see front matter ª 2010 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2010.01.096
international journal of hydrogen energy 35 (2010) 3712–3720
Although several authors have investigated membrane based fuel processors [13–26], only few contributions are available which address the behavior of the entire system, including not only the membrane based fuel processor, but also the fuel cell, the auxiliary units and the heat exchangers [20–26]. Most of these studies refer to liquid fuels [20–23] and only few contributions are available when methane is employed [24–26]. In particular, Campanari et al. [24] analyzed an integrated membrane-SR reactor coupled with a PEMFC, showing that a higher global energy efficiency can be achieved, with respect to conventional fuel processors, if a membrane reactor is employed. Lyubovsky et al. [25] analyzed a methane ATR-based fuel processor – PEMFC system, with a membrane unit placed downstream the WGS unit and operating at high pressure, concluding that high global energy efficiency can be obtained if a turbine is introduced in the system to generate additional power from the expansion of the hot gases produced by the combustion of the membrane retentate stream. In 2009, Salemme et al. [26] analyzed the energy efficiency of a methane ATR-based fuel processor – PEMFC system, with a membrane unit integrated in the WGS reactor and showed that a high global energy efficiency is obtained when a steam sweep gas is sent to the membrane reactor in counter-current mode. This work presents a comprehensive analysis of fuel processor – PEMFC systems based on integrated membrane reactor technology for pure hydrogen production, considering methane as a fuel and SR and ATR as reforming processes. The manuscript analyzes four fuel processor configurations, each coupled with a PEMFC: FP.A) SR reactor, followed by two WGS reactors and a PROX reactor. FP.B) ATR reactor, followed by two WGS reactors and a PROX reactor. FP.C) Integrated membrane-SR reactor. FP.D) Integrated membrane-ATR reactor. Operating parameters, such as steam to methane and oxygen to methane inlet ratios, reforming temperature, as well as pressure, are screened to identify the conditions that maximize global energy efficiency. The effect of the addition of steam as sweep gas is also presented and discussed.
2.
3713
Methodology
The simulations were performed in stationary conditions, by using the commercial package Aspen Plus [27]. The selected property method was Peng–Robinson and the component list was restricted to CH4, O2, N2, H2O, CO, H2 and CO2. Methane was considered as fuel, fed at 25 C and 1 atm, with a constant flow rate of 1 kmol/h. Feed to the system was completed with a liquid water stream and an air stream (for the ATR Fuel Processors only), fed at 25 C and 1 atm. Aspen Plus was used to calculate product composition throughout the plant as well as energy requirements of each unit. The configurations simulated (flowsheets) are presented in the Fuel processors and Model libraries section, where the assumptions and the model libraries used to simulate the process are presented. The quantities employed to calculate energy efficiency are defined in the System efficiency section.
2.1.
Fuel processors description and model libraries
Fig. 1 reports the flowsheet of a conventional SR-based fuel processor (FP.A) coupled with a PEM fuel cell. The fuel processor consists of a reforming and a CO clean-up section. The reforming section is an isothermal reactor (SR), modeled by using model library RGIBBS. The CO clean-up section consists of a high (HTS) and low (LTS) temperature water gas shift reactor followed by a PROX reactor. HTS and LTS were modeled by using model library RGIBBS; the reactors were considered as adiabatic and methane was considered as an inert in order to eliminate the undesired methanation reaction, kinetically suppressed on a real catalytic system. The PROX reactor was modeled as an isothermal stoichiometric reactor, RSTOIC, in which two reactions take place: oxidation of CO to CO2 and oxidation of H2 to H2O; the air fed to the PROX reactor (AIR-PROX) was calculated in order to achieve a 50% oxygen excess with respect to the stoichiometric amount required to convert all the CO to CO2. The RSTOIC specifics were completed with the assignment of total conversion of CO and O2. The PEMFC was simulated as the sequence of the anode, modeled as an ideal separator, SEP, and the cathode, modeled as an isothermal stoichiometric reactor, RSTOIC. The presence
Fig. 1 – Flowsheet of fuel processor FP.A coupled with a PEMFC.
3714
international journal of hydrogen energy 35 (2010) 3712–3720
of the SEP unit allows to model a purge gas (anode off-gas, AOG) required for mass balance reasons, if the hydrogen stream sent to the PEMFC is not 100% pure. In agreement with the literature, the hydrogen split fraction in the stream H2 at the outlet of the SEP was fixed at 0.75 [6,11], whereas the split fractions of all the other components were taken as 0. The RSTOIC unit models the hydrogen oxidation reaction occurring in the fuel cell. The reactor specifics were completed by considering an operating temperature of 80 C and pressure of 1 atm; the inlet air at the cathode (AIR-FC), fed at 25 C and 1 atm, guarantees a 50% excess of oxygen in the RSTOIC reactor. In agreement with the literature [28], these conditions were considered as sufficient to assign total hydrogen conversion. The anode off-gas is sent to a burner, modeled as an adiabatic RSTOIC, working at atmospheric pressure with 50% excess air (AIR-B). The heat required by the SR reactor working at temperature TSR is supplied by the heat exchanger H-B, where the stream coming from the burner is cooled to TSR þ 10 C. Model library HEATER was used for this purpose. An additional methane stream (CH4-B) is sent to the burner to satisfy the global heat demand of the system, when needed. Fig. 2 reports the flowsheet of a conventional ATR-based fuel processor (FP.B) coupled with a PEM fuel cell. For the sake of simplicity, the description of the flowsheet will be carried out by indicating the differences with respect to the flowsheet of Fig. 1, which are concentrated only in the reforming section. Indeed, in FP.B, the reforming section is constituted by an adiabatic reactor (ATR), modeled by using model library RGIBBS. The integrated membrane reactors were simulated by discretizing the membrane reactor with N reactor-separator units. With this approximation, reactors are assumed to reach equilibrium and the separators are modeled as ideal separators, SEP, whose output is given by a stream of pure hydrogen (permeate) and a stream (retentate) containing the unseparated hydrogen and all the balance. The amount of hydrogen separated (niH2 ;P ) is calculated assuming equilibrium between the partial pressure in the retentate and permeate side, according to Eq. (1): PR $
niH2 ;R niH2 ;P niR niH2 ;P
hydrogen in the retentate stream; niR is the total mole flow of the retentate stream; PiH2 ;P is hydrogen partial pressure in the permeate side of the membrane, calculated as:
PiH2 ;P ¼
niH2 ;P $PP i nH2 ;P þ nSG
(2)
where PP is the pressure in the permeate side of the membrane, taken as 1 atm in all the simulations, and nSG represents the molar flow rate of steam sweep gas (SG), which can be introduced to increase the separation driving force in the membrane. When present, the sweep gas is produced by liquid water, fed at 25 C and 1 atm to a heat exchanger and sent to the membrane reactor in counter-current mode. The high hydrogen purity of the stream sent to the PEMFC allows to take as zero the anode off-gas, simplifying the model of the PEMFC to the cathode side (RSTOIC) only. Figs. 3 and 4 report the flowsheet used to simulate a membrane-SR reactor (FP.C) and a membrane-ATR reactor (FP.D) coupled with a PEMFC, respectively. The membrane-SR reactor was discretized with 30 units (Fig. 3) while the membrane-ATR reactor was discretized with 20 units (Fig. 4); the number of units required to model each membrane reactor was assessed by repeating the simulations with an increasing number of reactor-separator units and was chosen as the minimum value above which global efficiency remained constant within 0.1%. All the reactors were considered as operating at the same pressure. Auxiliary power units for compression of the reactants were considered in the configurations where pressure was explored as an operation variable, i.e. FP.C and FP.D. 100 C was chosen as the minimum exhaust gas temperature (Tex), when compatible with the constraint of a positive driving force in the heat exchangers present in the plant. Finally, it is worth mentioning that the assumptions made to model the system are the same for all the configurations investigated and do not affect the conclusions drawn in this comparative analysis.
! ¼ PiH2 ;P
(1)
where PR is the pressure in the retentate side of the membrane, equals to reactor pressure; niH2 ;R is the mole flow of
2.2.
System efficiencies
Energy efficiency, h, was defined according to the following Eq. (3):
Fig. 2 – Flowsheet of fuel processor FP.B coupled with a PEMFC.
international journal of hydrogen energy 35 (2010) 3712–3720
3715
Fig. 3 – Flowsheet of fuel processor FP.C coupled with a PEMFC.
h¼
Pe Pa ðnCH4 þ nCH4 ;B Þ$LHVCH4
(3)
where Pa is the electric power required by the auxiliary units for compression of methane, air and water, nCH4 is the inlet molar flow rate of methane to the reactor, nCH4 ;B is the molar flow rate of methane fed to the burner, LHVCH4 is the lower heating value of methane and Pe is the electric power generated by the fuel cell, calculated as: Pe ¼ nH2 $LHVH2 $hFC
PN
HR ¼
niH ;P PN 2 i nH2 ;R þ i¼1 nH2 ;P i¼1
(5)
where niH2 ;P is the molar flow rate of hydrogen separated by the i-th membrane unit, nH2 ;R is the molar flow rate of hydrogen in the retentate stream (RETENT) at the exit of the last separator and N is the number of separators. According to the definitions given above, h can be expressed as it follows:
(4)
where nH2 is the molar flow rate of hydrogen that reacts in the fuel cell, LHVH2 is the lower heating value of hydrogen and hFC is the electrochemical efficiency of the fuel cell, taken as 0.6 [29]. In the membrane-based fuel cell systems (configurations FP.C and FP.D), an important parameter is the global hydrogen recovery (HR), defined as:
h ¼ hR $HR$hFC fa
(6)
where fa is the fraction of inlet methane required to run the auxiliary units, defined by Eq. (7): fa ¼
Pa $ð1 aÞ ðnCH4 $LHVCH4 Þ
Fig. 4 – Flowsheet of fuel processor FP.D coupled with a PEMFC.
(7)
3716
international journal of hydrogen energy 35 (2010) 3712–3720
Table 1 – Simulation results and best operating conditions for FP.A and FP.B.
3.
Simulation results
FP.A (SR) FP.B (ATR)
xCH4
a
h
TEx ( C)
91.0 98.8
0.0 0.0
48.0 38.5
226 444
Operating conditions
FP.A (SR) FP.B (ATR)
P (atm)
H2O/CH4
O2/CH4
TSR ( C)
1 1
2.5 3.5
– 0.56
670 –
where a is the ratio between methane flow rate fed to the burner and total methane flow rate fed to the system, defined by Eq. (8): a¼
nCH4 ;B nCH4 þ nCH4 ;B
(8)
hR is the hydrogen production efficiency in the fuel processor, defined by Eq. (8): P nH2 ;R þ Ni¼1 niH2 ;P $LHVH2 hR ¼
2.3.
(9)
$ð1 aÞ
nCH4 $LHVCH4
Model analysis tools
The Calculator Tool was used to calculate the amount of hydrogen separated by the membrane unit, by introducing the relation reported in Eq. (1) and to calculate the air flows required by the PROX reactor, by the PEM fuel cell and by the burner, respecting the 50% oxygen excess specification. The Sensitivity Tool was used to evaluate energy efficiency and thermal efficiency, varying H2O/CH4 inlet ratio, O2/CH4 inlet ratio, temperature of SR reactor, pressure and steam sweep gas flow rate. The Design Specification Tool was used to determine the amount of methane fed to the burner (CH4-B), in order to satisfy the heat duty of the SR reactor, or to pre-heat the inlet streams of the ATR reactor to 350 C.
Simulations on the conventional systems were performed at 1 atm exploring as operative parameters the molar ratio between water and methane at reactor inlet (H2O/CH4) and reactor temperature, for the SR-based fuel processor (FP.A), and water to methane and oxygen to methane (O2/CH4) inlet ratio, for the ATR-based fuel processor (FP.B). Table 1 reports the simulation results and the value of the operative parameters given as simulation input that maximize the energy efficiency, h, for FP.A and for FP.B, respectively. In agreement with Ersoz et al. [4], FP.A shows the highest global efficiency (48.0%) at TSR ¼ 670 C and H2O/CH4 ¼ 2.5. It should be noticed that, in the optimal conditions, methane conversion (xCH4 ) is lower than unity; however, the non converted methane is not energetically wasted, since it contributes to the energy content of the AOG, used to sustain the endothermicity of the SR reactor. In this conditions, no addition of methane to the burner is needed (a ¼ 0). For FP.A the minimum exhaust gas temperature achievable is 226 C, indeed, further heat recovery is hindered by temperature cross over in the heat exchangers. As reported in [26], FP.B shows the highest global efficiency (38.5%) at O2/CH4 ¼ 0.6 and H2O/CH4 ¼ 3.5; the value of h is significantly lower than what achieved with FP.A, mainly due to the autothermal nature of the ATR process, that limits the possibility to recover the energy content of the AOG. This reflects into a higher exhaust gas temperature in FP.B (444 C) than in FP.A (226 C). The simulations on the systems with integrated membrane reactors (FP.C and FP.D) were performed considering also pressure and sweep gas to methane inlet ratio (SG/CH4) as operative parameters to be optimized in the range 3–15 atm and 0–2, respectively. Fig. 5a reports the energy efficiency of FP.C as a function of pressure, at SG/CH4 equals to zero (continuous line). Energy efficiency rapidly increases with pressure in the range 3–5, where no methane addition to the burner is required to sustain the endothermic steam reforming reaction. As pressure increases above 5 atm ca., energy efficiency continues to grows with pressure, but at a lower rate, because methane addition to the burner becomes necessary. The dotted line,
b
70
70
60
60
50
50
(%)
(%)
a
Results
40 30
40 30
SR without CH4-B SR
20
20
10
ATR
10
0
0 3
5
7
9
P (atm)
11
13
15
3
5
7
9
11
13
P (atm)
Fig. 5 – Energy efficiency h as a function of pressure for FP.C (a) and FP.D (b).
15
3717
international journal of hydrogen energy 35 (2010) 3712–3720
Table 2 – Simulation results for FP.C. Operating conditions: SG/CH4 [ 0; H2O/CH4 [ 2.5; T [ 600 8C.
50
xCH4
a
TEx ( C)
HR
fa
hR
h
3.0 5.0 7.0 9.0 12.0 15.0
70.6 86.3 91.8 94.5 96.6 97.6
0.0 1.8 12.8 17.2 20.4 22.0
803.8 100.0 100.0 100.0 100.0 100.0
58.0 85.9 91.9 94.4 96.2 97.1
0.5 0.7 0.8 0.9 1.0 1.1
80.4 98.7 94.2 92.6 91.5 91.0
27.5 50.2 51.2 51.5 51.8 51.9
45
(%)
P (atm)
SR ATR
40 35
superimposed to Fig. 5 as an aid to this discussion, represents the value of energy efficiency that would be calculated if the methane sent to the burner was not factored in the computation. The trend of h vs P is the combined effect of hydrogen recovery (HR), reforming efficiency (hR), the power of the auxiliary units (related to fa), whose values are reported in Table 2 together with the value of methane conversion (xCH4 ) and fraction of methane sent to the burner (a). In particular, HR increases with pressure due to the increase of hydrogen separation driving force through the membrane; hR is positively influenced by the trend of HR with pressure, due to the positive effect on reaction equilibrium of increasing hydrogen separation (see xCH4 ) and it is negatively influenced by the addition of methane to the burner, since only part of the energy content of the methane sent to the burner is actually transferred to the reacting mixture. fa increases with pressure, due to increasing compression ratios. To complete the picture, it should be kept in mind that the heating value of the retentate decreases with pressure, as a consequence of higher xCH4 and HR. This, in turn, influences the quantity of methane sent to the burner to sustain the endothermic steam reforming reaction. In the low pressure range, the positive effect of HR and xCH4 on energy efficiency overrules the negative effect of fa and a. The plateau value reached at higher pressure indicates that the drawback of fa and a compensates the positive effect of HR and xCH4 . As reported by [24] the energy efficiency of an SR-based system can be increased if a membrane reactor is used (FP.C), in place of a conventional reactor. This is due to the possibility to recover a higher amount of heat in FP.C than in FP.A. Indeed the heat exchanger network needed in FP.A has to satisfy the temperature requirements of the Shift and PROX reactors resulting in a higher exhaust gas temperature (226 C), while in FP.C the heat exchanger network allows to
30 0.0
0.5
1.0
1.5
2.0
SG/CH4 Fig. 6 – Energy efficiency h as a function of SG/CH4 ratio for FP.C (continuous line) and FP.D (dashed line).
cool the exhaust gas to 100 C (as chosen in the methodology), without any temperature cross over. Fig. 5b reports the energy efficiency of the system with a membrane ATR-reactor (FP.D) as a function of pressure. As for the case of FP.C, energy efficiency shows a continuous increase with pressure, but the values are significantly lower, due to limited recovery of the energy contained in the retentate stream and to the negative contribution of the compressor (see Tex and fa in Table 3). It should be noted that, in FP.D, the maximum value of h (37.2%) is even lower than what is obtained with the conventional ATR reactor [h of FP.B (38.5%)]. This should be attributed to the fact that, notwithstanding the absence of the AOG stream, dilution of the reacting mixture with nitrogen reduces HR (affecting, in turn, also xCH4 ) leading to a retentate with relatively high amount of methane and hydrogen, whose heating value cannot be totally recovered. It should be pointed out that, due to the exothermic nature of the reactions, no additional methane to the burner is required, i.e. a ¼ 0, and the exhaust gas stream leaves the plant at quite high temperatures. Data are reported in Table 3. Fig. 6 reports the energy efficiency of FP.C and FP.D as a function of SG/CH4 at constant pressure (10 atm). Simulation details are reported in Tables 4 and 5. The behavior of h as a function of SG/CH4 is due to the combined effect of HR and hR. Hydrogen recovery is enhanced by the presence of the sweep gas, as a consequence of reduced
Table 4 – Simulation results for FP.C. Operating conditions: H2O/CH4 [ 2.5; P [ 10 atm; T [ 600 8C.
Table 3 – Simulation results for FP.D. Operating conditions: SG/CH4 [ 0; H2O/CH4 [ 1.2; O2/CH4 [ 0.5. P (atm)
xCH4
a
TEx ( C)
HR
fa
hR
h
3.0 5.0 7.0 9.0 12.0 15.0
85.2 88.4 90.0 90.9 91.8 92.4
0.0 0.0 0.0 0.0 0.0 0.0
1369.1 1248.1 1178.1 1132.7 1089.8 1063.6
5.8 60.2 75.6 82.7 88.0 90.9
1.6 2.6 3.3 3.9 4.6 5.2
60.1 67.6 71.4 73.8 76.2 77.8
0.5 21.8 29.1 32.7 35.6 37.2
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
0.0 0.1 0.5 1.0 1.5 2.0
95.4 99.8 100.0 100.0 100.0 100.0
18.6 25.7 28.5 30.0 31.5 33.0
100.0 100.0 100.0 100.0 100.0 100.0
95.1 99.1 100.0 100.0 100.0 100.0
0.9 0.8 0.8 0.8 0.8 0.8
92.1 89.0 86.8 84.9 83.1 81.4
51.6 52.1 51.3 50.2 49.1 48.1
3718
international journal of hydrogen energy 35 (2010) 3712–3720
Table 5 – Simulation results for FP.D. Operating conditions: H2O/CH4 [ 1.2; P [ 10 atm; O2/CH4 [ 0.5.
Table 6 – Simulation results for FP.C. Operating conditions: H2O/CH4 [ 2.5; T [ 600 8C.
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
0.0 0.1 0.5 1.0 1.5 2.0
91.3 95.1 99.1 100.0 100.0 100.0
0.0 0.0 0.0 0.4 4.0 7.0
1115.9 894.7 463.8 100.0 100.0 100.0
84.9 95.0 99.1 99.9 100.0 100.0
4.1 4.1 4.1 4.1 3.9 3.8
74.8 81.5 88.6 90.8 88.1 85.5
34.0 42.4 48.6 50.3 48.9 47.5
hydrogen partial pressure in the permeate side, and this holds true, both for FP.C and FP.D; on the other hand, hR decreases continuously with SG/CH4, for the case of FP.C, while it shows a non monotone trend for the case of FP.D. The continuous decrease of hR is due to the fact that, in FP.C, production of sweep gas is always coupled with addition of methane to the burner, that overrules the increment of HR. For this reason, being h combination of hR and HR, after an initial small increment, it decreases with addition of sweep gas. The non monotone trend of hR with SG/CH4 present in FP.D is due to the fact that in an autothermal reforming membrane system, up to a certain amount, sweep gas can be produced by employing the excess energy which cannot be recovered in other ways, i.e. a ¼ 0. Therefore, as long as production of sweep gas does not require addition of methane to the burner, hR, increases with SG/CH4 while a further increase of SG/CH4 above this value causes an increase of a, but only marginal increments of HR, thus leading to a decrease of hR and h. As shown in Fig. 6, energy efficiency of FP.D can be highly improved by adding sweep gas, increasing from 34.0% (SG/CH4 ¼ 0) to 50.3% (SG/ CH4 ¼ 1.0). Fig. 7 reports energy efficiency of FP.C and FP.D as a function of pressure, calculated (for each pressure) in correspondence to the best SG/CH4 values. Data are reported in Tables 6 and 7. In the pressure range investigated, h of FP.C has a monotone increasing trend and the corresponding optimal value of SG/CH4 decreases from 0.5 to 0.1, thus indicating
(%)
52
P (atm) 3.0 5.0 7.0 9.0 12.0 15.0
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
0.5 0.2 0.1 0.1 0.1 0.1
99.2 99.4 99.6 99.7 99.1 99.8
24.8 24.9 25.1 25.3 25.6 25.6
100.0 100.0 100.0 100.0 100.0 100.0
97.4 98.1 98.5 98.8 99.1 99.2
0.4 0.5 0.7 0.8 0.9 1.0
88.0 88.8 89.1 89.2 89.2 89.3
51.1 51.8 52.0 52.1 52.1 52.1
that the best way to operate a membrane steam reforming system is to increase the pressure without addition of sweep gas. On the other hand, h of FP.D shows a nonmonotone trend with a maximum of 50.6% at 7 atm. In the pressure range investigated, the optimal value of SG/CH4 decreases from 1.5 to 1.0. Column 7 of Tables 6 and 7 reports the power required by the auxiliary units ( fa), showing that power consumption is much higher in FP.D than in FP.C. This factor is the main responsible for the drop of h of FP.D above 7 atm. Fig. 7 shows that the best way to operate an autothermal reforming membrane system is to moderately increase pressure and to add sweep gas to improve HR. Finally, it should be noted that the addition of sweep gas in FP.D allows to reach energy efficiency values significantly higher than 38.5% (FP.B) and similar to the energy efficiency of steam reforming based systems.
3.1.
Analysis of thermal stress on the membrane
In analyzing possible membrane reactor configurations, it should be kept in mind that, due to limited thermal stability of the highly selective membranes, membrane units should not be exposed to temperatures higher than 600 C. While FP.C always meets this constraint (since reactor temperature is fixed at 600 C), FP. D does not. Indeed, in the optimal conditions, the first reactors reach temperatures as high as 720 C. It is also well known that ATR reactors are characterized by a non-uniform temperature profile, with a hot spot in the initial part of the catalyst bed [30]. Therefore, the actual realization of an integrated membrane reactor would require significant improvements of membrane compatibility with high temperatures. A more realistic configuration of an ATR-based membrane reactor should consider a first ATR reactor, where most of the methane oxidation takes place,
SR ATR
51
Table 7 – Simulation results for FP.D. Operating conditions: H2O/CH4 [ 1.2; O2/CH4 [ 0.5.
50
P (atm)
49 3
6
9
12
P (atm) Fig. 7 – Energy efficiency h as a function of pressure calculated at the best SG/CH4 value.
15
3.0 5.0 7.0 9.0 12.0 15.0
SG/CH4
xCH4
a
TEx ( C)
HR
fa
hR
h
1.5 1.1 1.0 1.0 1.0 1.0
99.4 99.2 99.9 100.0 100.0 100.0
0.0 0.0 0.0 0.0 0.0 0.0
100.0 100.0 100.0 100.0 100.0 100.0
97.9 99.2 99.6 99.8 99.9 99.9
1.6 2.6 3.3 3.9 4.6 5.2
87.0 89.1 90.2 90.8 91.4 91.6
49.5 50.5 50.6 50.5 50.2 49.8
international journal of hydrogen energy 35 (2010) 3712–3720
followed by a membrane reactor, interposing between the two units a heat exchanger to cool down the temperature before entrance into the membrane reactor, so that the membranes are never exposed to temperatures higher than 600 C. With this configuration, energy efficiency becomes 48.5% and the best operating conditions are P ¼ 7 atm; O2/ CH4 ¼ 0.5; H2O/CH4 ¼ 1.7; SG/CH4 ¼ 1.0.
4.
Conclusion
A simulative energy efficiency analysis was performed on fuel processor – PEMFC systems, considering methane as fuel and steam reforming and autothermal reforming as processes to produce hydrogen. Four configurations were investigated, as follows: FP.A) SR reactor, followed by two WGS reactors and a PROX reactor. FP.B) ATR reactor, followed by two WGS reactors and a PROX reactor. FP.C) Integrated membrane-SR reactor. FP.D) Integrated membrane-ATR reactor. The comparison between the steam reforming based systems (FP.C vs FP.A) showed that the employment of a membrane reactor can increase system efficiency from 48% to 52.1%. Such an efficiency increase is obtained by increasing system pressure from 1 to 15 atm and by adding small amounts of sweep gas. The impact of sweep gas addition on energy efficiency as well as the amount of sweep gas needed, is higher when system pressure is low, i.e. at P ¼ 3 atm, energy efficiency increases from 27.5% to 51.1% if a sweep gas to methane ratio equal to 0.5 is employed; while at P ¼ 15 atm the improvement is only from 51.9% to 52.1%, with sweep gas to methane ratio equal to 0.1. The comparison between the autothermal reforming systems (FP.D vs FP.B) showed that energy efficiency can be improved from 38.5% to 50.6%, if a membrane reactor is employed. To obtain such an energy efficiency improvement, sweep gas addition is required. Indeed, without the addition of sweep gas, the employment of a membrane reactor is not energetically convenient in the pressure range investigated. The best energy efficiency of 50.6 was achieved for P ¼ 7 atm and sweep gas to methane ratio equals to 1.0. Finally, the energy efficiency of FP.D is reduced to 48.5% if a constraint of 600 C is imposed on the maximum temperature achieved in the membrane units.
Acknowledgements This work has been carried out within the framework of the project ‘‘Pure hydrogen from natural gas through reforming up to total conversion obtained by integrating chemical reaction and membrane separation’’ financially supported by MUR (FISR DM 17/12/2002 – year 2001).
3719
references
[1] Xu J, Froment GF. Methane steam reforming, methanation and water-gas shift. Intrinsic kinetics. AIChE J 1989;35:88–96. [2] Semelsberger TA, Brown LF, Borup RL, Inbody MA. Equilibrium products from autothermal processes for generating hydrogen-rich fuel-cell feeds. Int J Hydrogen Energy 2004;29:1047–64. [3] Dupont V, Ross AB, Hanley I, Twigg MCV. Unmixed steam reforming of methane and sunflower oil: a single-reactor process for H2-rich gas. Int J Hydrogen Energy 2006;32:67–79. [4] Ersoz A, Olgun H, Ozdogan S. Reforming options for hydrogen production from fossil fuels for PEM fuel cells. J Power Sources 2006;154:67–73. [5] Roychoudhury S, Lyubovsky M, Walsh D, Chu D, Kallio E. Design and development of a diesel and JP-8 logistic fuel processor. J Power Sources 2006;160:510–3. [6] Francesconi JA, Mussati MC, Mato RO, Aguirre PA. Analysis of the Energy efficiency of a integrated ethanol processor for PEM fuel cell systems. J Power Sources 2007;167:151–61. [7] Docter A, Lamm A. Gasoline fuel cell systems. J Power Sources 1999;84:194–200. [8] Ahmed S, Krumpelt M. Hydrogen from hydrocarbon fuels for fuel cells. Int J Hydrogen Energy 2001;26:291–330. [9] Seo YS, Shirley A, Kolaczkowski ST. Evaluation of thermodynamically favourable operating condition for production of hydrogen in three different reforming technologies. J Power Sources 2002;108:213–25. [10] Cutillo A, Specchia S, Antonini M, Saracco G, Specchia V. Diesel fuel processor for PEM fuel cells: two possible alternatives (ATR vs SR). J Power Sources 2006;154:379–85. [11] Doss ED, Kumar R, Ahluwalia RK, Krumpelt M. Fuel processors for automotive fuel cell systems: a parametric analysis. J Power Sources 2001;102:1–15. [12] Qi A, Peppley B, Karan K. Integrated fuel processors for fuel cell application. Fuel Process Technol 2007;88:3–22. [13] Gallucci F, Basile A. Pd–Ag membrane reactor for steam reforming reactions: a comparison between different fuels. Int J Hydrogen Energy 2008;33:1671–87. [14] Gallucci F, Basile A, Tosti S, Iulianelli A, Drioli E. Methanol and ethanol steam reforming in membrane reactors: an experimental study. Int J Hydrogen Energy 2007;32:1201–10. [15] Chiappetta G, Clarizia G, Drioli E. Design o fan integrated membrane system for a high level hydrogen purification. Chem Eng J 2006;124:29–40. [16] Tong J, Matsumuura Y. Effect of catalytic activity on methane steam reforming in hydrogen-permeable membrane reactor. Appl Catal A General 2005;286:226–31. [17] Li A, Lim CJ, Grace JR. Staged-separation membrane reactor for steam methane reforming. Chem Eng J 2008;138:452–9. [18] Wang L, Murata K, Inaba M. Production of pure hydrogen and more valuable hydrocarbons from ethane on a novel highly active catalyst system with a Pd-based membrane reactor. Catal Today 2003;82:99–104. [19] Fernandes FAN, Soares AB. Methane steam reforming modeling in a palladium membrane reactor. Fuel 2006;85: 569–73. [20] Pearlman JB, Bhargav A, Shields EB, Jackson GS, Hearn PL. Modeling efficiency and water balance in PEM fuel cell systems with liquid fuel processing and hydrogen membranes. J Power Sources 2008;185:1056–65. [21] Lattner JR, Harold MP. Comparison of methanol-based fuel processors for PEM fuel cell systems. Appl Catal B Environ 2005;56:149–69. [22] Lattner JR, Harold MP. Comparison of conventional and membrane reactor fuel processors for hydrocarbon-based PEM fuel cell systems. Int J Hydrogen Energy 2004;29:393–417.
3720
international journal of hydrogen energy 35 (2010) 3712–3720
[23] Manzolini G, Tosti S. Hydrogen production from ethanol steam reforming: energy efficiency analysis of traditional and membrane processes. Int J Hydrogen Energy 2008;33: 5571–82. [24] Campanari S, Macchi E, Manzolini G. Innovative membrane reformer for hydrogen production applied to PEM microcogeneration: simulation model and thermodynamic analysis. Int J Hydrogen Energy 2008;33:1361–73. [25] Lyubovsky M, Walsh D. Reforming system for co-generation of hydrogen and mechanical work. J Power Sources 2006;157: 430–7. [26] Salemme L, Menna L, Simeone M. Analysis of the energy efficiency of innovative ATR-based PEM fuel cell system with
[27] [28]
[29]
[30]
hydrogen membrane separation. Int J Hydrogen Energy 2009; 34:6384–92. Aspen Technology Inc., http://www.aspentech.com. Ratnamala GM, Shah N, Mehta V, Rao PV. Integrated fuel cell processor for a 5-kW proton – exchange membrane fuel cell. Ind Eng Chem Res 2005;44:1535–41. Hou Y, Zhuang M, Wan G. The analysis for the efficiency properties of the fuel cell engine. Renew Energy 2007;32: 1175–86. Simeone M, Salemme L, Allouis C. Reactor temperature profile during autothermal methane reforming on Rh/Al2O3 catalyst by IR imaging. Int J Hydrogen Energy 2008;33: 4798–808.