Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Comparison of multi-stack fuel cell system

architectures for residential power generation


applications including electrical vehicle charging
John Cardozo ∗ , Neigel Marx∗ † , Loı̈c Boulon∗ ‡ and Daniel Hissel†
∗ Institut de Recherche sur l’Hydrogène, Université du Québec à Trois-Rivières, Trois-Rivières (QC), Canada.
† FCLAB, FR CNRS 3539, FEMTO-ST/Energy UMR CNRS 6174, University of Franche-Comte, Belfort, France.
‡ Corresponding author, Email: loic.boulon@uqtr.ca

Abstract—Multistack fuel cell systems can be more efficient, architecture, degraded mode operation capabilities can be more
more reliable and possess a longer lifetime than single fuel cell or less expensive to implement [5].
systems. However, these features depend on the architecture However, most studies investigating the benefits of using
used to link the fuel cells. Some architectures enable the use of
distribution methods which impact the behavior of the system. fuel cells for residential power generation consider single stack
The aim of this paper is to evaluate the impact of multiple fuel cell systems [6]. Although their use in a co-generation of
architectures and distribution methods for multi-stack fuel cell heat and power environment has been proven to be highly
systems on the ageing and hydrogen consumption in the case of a efficient, the advantages of multi-stack systems versus single
residential application including the charge of an electric vehicle. stack systems has not been studied.
The ageing of the system depends on the number of start/stop
cycles and the duration of use. The cost of use of the fuel cell The aim of this paper is to evaluate the impact of multiple
system depends on the hydrogen consumption, system cost and architectures and distribution methods for multi-stack fuel cell
degradation. The results show that the parallel architecture is the systems on the ageing and hydrogen consumption in the case
less expensive and minimizes ageing if the ratio of the degradation of a residential application including the charge of an electric
per start/stop cycle to the degradation per hour of use is low. vehicle.
The series architecture is the most effective when that ratio is
high. This first section presents the different MFC architectures
Index Terms—Multistack fuel cell systems, ageing, Fuel con- that will be compared. The power distribution method used
sumption, Residential power generation, PEM Fuel Cell in this study is presented in the second section. The third
section is dedicated to the fuel cell system model used in
I. I NTRODUCTION simulations. The fourth section presents the simulation results
The environmental crisis has partly contributed to the recent for the chosen scenarios and an analysis of the results of each
advance in emission-free transportation. Plug-in electric vehi- system.
cles offer an energy efficient alternative to more conventional
fossile fuel based transportation for small trips. II. MFC ARCHITECTURES
However, the charge of electric vehicles (EVs) leads to a MFC systems link multiple fuel cells together through
significant increase of the residential power consumption and architectures. The electric, fluidic and thermal architectures
thus stress on the grid. Indeed, charging a single EV at home can be set up using series, cascaded, parallel or series-parallel
can lead to a doubling of the average load for average house- configurations [3]. Like single fuel cell systems, MFC systems
holds [1]. Currently, residential fuel cell generation systems include a power converter, a reactant supply subsystem and a
are considered as an alternative to the grid for residential thermal management subsystem. This study focuses on the
power production. Their efficiency in this type of scenario is electrical architecture. The thermal and fluidic architectures
unknown. are set to use the parallel configuration. The different MFC
There are multiple technical barriers preventing the pen- architecture are presented in the following subsections.
etration of fuel cells in the market: their limited reliability,
durability [2], and cost. Multi-stack fuel cell (MFC) systems A. Series architecture
offer many advantages that could push fuel cell technology The fuel cells of the system are linked in a series con-
above those barriers [3]. The MFC architecture offers high figuration and are coupled to a single power converter to
efficiency ratings on a larger power range than the single fuel adapt the output voltage to the DC-bus voltage. The output
cell architecture. It also allows for degraded mode operation voltage is high. This leads to low gain values which in turn
providing higher reliability when part of the system fails [4]. leads to low stress on the converter. There is however no
Also, as a modular architecture, it can be adapted to achieve possibility to control the individual power provided by each
various power ratings by increasing or decreasing the number fuel cell. Nevertheless, degraded mode capabilities can be
of modules. This ultimately results in a simpler design process achieved through the use of by-pass systems. The architecture
for fuel cell system manufacturers. Depending on the chosen is presented on figure 1a.

978-1-4673-7637-2/15/$31.00 ©2015 IEEE


(a) (b) (c)
Fig. 1: MFC architectures : (a) series, (b) parallel, (c) series-parallel

B. Parallel architecture architecture. The optimal power distribution leads to behavior


The fuel cells are connected to the DC-bus in a parallel that is close to that of the daisy chain when it is used on an
configuration through individual power converters. This leads unbalanced system. Therefore, only the daisy chain will be
to a high number of power converters. Since each converter used with the series-parallel and parallel architectures.
adapts the voltage of a single fuel cell, the voltage gains The daisy chain distribution method is a sequential distri-
will be higher than that of the series configurations which bution method. The power is provided by the first fuel cell in
in turn will lead to higher stress on the converters. This the chain until the requested power exceeds the rated power of
architectures allows for individual current control through the that fuel cell. The second fuel cell then provides the remaining
dedicated converters. Degraded mode operation is available to power. And so on until the requested power is achieved:
this configuration without any addition to the architecture. The
architecture is presented on figure 1b. PF C1 = PLOAD (1)
C. Series-Parallel Architecture
This architecture blends the series and parallel architectures PF C2 = PLOAD − PF C1 (2)
together. Multiple series-connected MFC systems are linked in
parallel to the DC-bus. Compared to the parallel architecture, PF C3 = PLOAD − PF C2 − PF C1 (3)
this architecture provides reduced capabilities for individual
power control but the number of power converters are reduced
If the system is unbalanced, it is more efficient to use the
and the stress they endure is reduced as well. The architecture
most efficient fuel cells first because the first fuel cells in the
is presented on figure 1c.
sequence are used more often than the others. The fuel cells
are ranked according to their average efficiency:
The parallel and series-parallel architectures both enable
some degree of individual current control. These capabilities  Pmax
are leveraged through the use of distribution methods. 1
ηaverage = η(P )dP (4)
Pmax P =0
III. P OWER DISTRIBUTION METHODS
Power distribution methods can be used when individual where η(P ) is the efficiency of the system when proving
current control is made available through the chosen electrical power P . In order to compare the different systems, a model
architecture. They are used to split the power reference be- of the fuel cell system is required.
tween the different fuel cells composing the multi-stack system
[7]. The following power sharing methods have been studied: IV. F UEL CELL MODEL
• Equal Distribution: As its name implies, it shares the load
equally between the fuel cells of the system. The model presented in this section is adapted from [8].
• Daisy-Chain Distribution: It is a sequential method, is It is adapted to air breathing technology. Air-breathing fuel
uses the lowest amount of fuel cells required to supply cells are mostly used in low power applications like mobile
the load. devices but have also been used in higher power application
• Optimal Distribution: it is the result of an optimization like automotive applications [9]. The fuel cell system model
process which maximizes the steady-state efficiency in includes the stack, the power converter and the blowers. The
the system. stack model include the electrochemical phenomena and the
The equal distribution will not be used in this study be- thermal behavior. The equations of the model are presented in
cause its use leads to a behavior close to that of the series table I.
A. Electrochemical subsystem TABLE I: Fuel cell model equations
The electrochemical reactions within the cells define the Electrochemistry:
output voltage, the heat generation and reactant consumption.
The voltage calculation is based on the assumption that all Vcell,s = En − ΔVact − ΔVconc − ΔVohm
 
cells are working at the same operation point. Therefore, the ΔH − T ΔS R TF C PH2 ∗ PO2 1/2
En = − + ∗ ln
stack voltage can be calculated by multiplying the cell voltage 2F 2F P0 3/2
 
by the number of cell. The cell voltage calculations include the R TF C j + jloss
ΔVact = K1 ln
static activation, ohmic and concentration over-voltages, and F j0
the dynamic effect of membrane hydration on the voltage. j
ΔVconc = K3 TF C ln (1 − )
jL
B. Thermal behavior
ΔVohm = Rm IF C
The thermal behavior of the stack depends on the heat   
PO2 Ec TF C
exchange sources and the thermal capacity of the stack. The j0 = K2 exp − 1−
P0 R TF C  T0  
free convection, forced convection and internal heat generation tmb 1 1
Rm = K4 exp −K5 ∗ −
have been modeled. This model considers that all cells operate b11 ∗ Λmb − b12 T0 TF C
at the same temperature and possess the same behavior. dVcell,d dI
Vcell,d + τ =K
dt dt
C. Ancillaries
VF C = Ncell (Vcell,s + Vcell,d )
The ancillaries use part of the power that is generated by
Ncell IF C
the stack in order to control its operation. As such they require qH2 ,r =
2F
precise modeling because they have an important impact on
Ncell IF C
energy calculations. qO2 ,r =
4F
1) Fan: Fans play an important role in the operation of a
fuel cell as they carry out two distinct functions: supplying qint = Ncell (ΔH/2F − Vcell ) IF C
the the electrochemical reaction with oxygen and cooling the
system. They are composed of a DC motor linked to fan blades Thermal behavior:
that expel air out of the stack. The DC-motor is modeled
as an electromotive force, an internal resistance and internal qconv,f ree = hf ree A (TF C − TAir )
inductance. The rotation of the fan blades induce a resistant qconv,f orced = h̄ At ηtherm θb,lm
torque on the motor’s shaft.
2) Power converter: Fuel cells are energy sources with qconv = qconv,f ree + qconv,f orced

low voltage – high current behavior, thus requiring voltage 1
TF C = (qint − qconv )
converters that raise levels of connection with the load. The CF C
adaptation of voltage is that of a boost converter and it occurs
with losses which have to be accounted for. Power converter:

1
Iconv = (Vbus − Vconv ) dt − Iloss
D. Simulation results of the fuel cell system model Lconv
1
The model has been fitted to experimental data from four Vconv = VF C
1−α
different fuel cell systems. The resulting efficiency curves for
1
the four systems are presented on figure 2. The power unit IF C =
1−α
Iconv
used for the following is 2.44 kW. It is the maximum power
required by the load. The fuel cell system models are scaled Blowers:
to a 2.5kW global maximum rated output. dIF AN
The differences between the efficiency of the system’s LF AN + RF AN IF AN = VF AN
dt
fuel cells can be explained by the uneven manufacturing
process and the different ageing of the fuel cells due to their E = kΩ
unbalanced use. Tm = k IF AN − Tloss

V. C OMPARISON OF THE ARCHITECTURES 1
Ω= (TM − Taero ) dt
JF AN
A. Problem formulation
qair = ηaero kaero Ω
The following architectures will be compared in this section:  
1
Taero = Nb ρair Cd eb (Rb 4 − R0 4 )(Ω cos θ)2
• Series architecture - no power distribution 8
• Parallel architecture - daisy chain power distribution
• Series-parallel architecture - daisy chain power distribu-
tion
Fig. 2: Efficiency of the 4 fuel cell systems, power unit : Fig. 4: Residential power profile
2,44kW

the degradation of the system is evaluated using the number


These different combinations lead to systems that behaves of start and stop cycles and the operation time:
very differently. This impacts the efficiency of the three
systems. The steady-state efficiency curves of the 3 systems
are presented on figure 3. δtot = δstart/stop + δtime
= Nstart/stop Dstart/stop + ton Dtime (5)

where δtot is the total degradation per cycle, δstart/stop is


the part of degradation due to the start/ cycles, δtime is the
part of degradation due to the operation time, Nstart/stop is
the number of start and stop cycle per FC system, Dstart/stop
is the degradation per start and stop cycle, ton is the operation
time per FC system and Dtime is the degradation per hour of
operation.
The chosen values for these two types of degradations are:
−1
• Dtime = 5μV h [12].
• Dstart/stop = 25μV per cycle [13].

B. Simulation results
Using different power distribution and architectures leads
to different power profiles for each fuel cell of the series,
Fig. 3: Efficiency curves of the different systems, power unit series-parallel and parallel systems. These power curves are
: 2,44kW presented on figure 5 .
The hydrogen consumption, number of on/off cycles and
The parallel and series-parallel system have a clear advan- duration of use are summed up in the table II. Degradation
tage over the series system at low power outputs. However, the results are presented in table III. The end of life is caracterized
series system becomes the most efficient of the three systems by a 0.1 V loss per cell.
over the second half of the power range.
The three systems are evaluated on hydrogen consumption C. Impact on degradation, fuel consumption and system cost
and degradation resulting of their use in a residential appli- Using the aforementioned degradation model the degrada-
cation. The residential load profile used in simulation (figure tion of the system per day can be calculated:
4) is that of a 200 m2 , 4 people US residence [10] coupled These results show the parallel architecture is the best
to an electric vehicle charge profile. The charging profile is architecture in terms of degradation when considering
set between 4pm and 5am and is responsible for doubling the the chosen degradation values. However, ageing tests
average load [11]. have provided very different results depending on the test
The ageing of fuel cells depends on multiple phenomena conditions and fuel cell technology. They can range from
like the duration of use, the condition of operation, the number 0.5μV.h−1 to 22500μV.h−1 for Doperation [14] and from
of start/stop cycles and the shutdown procedure. In this study, 25μV per cycle to 7500μV per cycle for Dstart/stop [15],
Dstart/stop
H= (6)
Dtime
This factor represent the number of hours of operation
that leads to the same degradation an on/off cycle provokes.
Equation (5) becomes

δtot = Dtime (Nstart/stop H + ton ) (7)


(b)
(a) The dependency of the degradation results to the factor
H are presented on figure 6. A value of 5μV.h−1 has been
considered for Dtime .

(c) (d)
Fig. 5: Power profiles for the three system. blue: series, green:
parallel, red: series-parallel a) FC 1, b) FC2, c) FC3, d) FC4

TABLE II: Simulation results


Hydrogen Number of duration
System Fuel cell
consumption (kg) on/off cycles of use (h)
FC 1 0 24
FC 2 0 24
Series 1.594 Fig. 6: Degradation per day
FC 3 0 24
FC 4 0 24 Given the chosen degradation model, the parallel system is
FC 1 0 24 always more attractive than the series-parallel system and is
FC 2 3 10.3 the best system when H is under 8.75 hour per cycle. The
Parallel 1.465
FC 3 1 14 degradation is part of the cost of operation. The other part
FC 4 1 4 depends on the hydrogen consumption. As the different system
require different amounts of hydrogen to provide the requested
FC 1 0 24
power a cost evaluation is provided in the following.
Series FC 2 3 10.3
1.507 The cost per day will be calculated as:
Parallel FC 3 3 10.3
FC 4 0 24 Ctot = δdeg Csystem + mH2 CH2 (8)
where Ctot is the total cost per day, δdeg is the degradation
TABLE III: Average degradation per day per day, Csystem is the cost of the system, mH2 is the mass of
Degradation (% per day) hydrogen consumed per day and CH2 is the cost of hydrogen.
System The cost of the system should be dependent on the architecture
100% means the system reached end of life
however for a lack of data they are considered equal for both
Series 0.12
architectures.
Series-parallel 0.123
The different costs are based on the US DOE’s projections
Parallel 0.0966 [16]:
−1
• CH2 = 3.8 $ · kg
−1
• Csystem = 1200 $ · kW × 2.44 kW = 2928 $
[13]. The cost depends on the value of H. Figure 7 presents the
cost of system versus the factor H for all three systems.
In order to investigate the impact of those values, Once again, the parallel system provides the best results at
Dstart/stop will be considered variable and a variable factor low H numbers. Above 11.4 the series system is best out of all
H is defined: three. The advantage in hydrogen consumption of the parallel
[5] D. C. Toquica Cardenas, N. Marx, L. Boulon, F. Gustin,
and D. Hissel, “Degraded Mode Operation of Multi-Stack
Fuel Cell Systems,” in 2014 IEEE Veh. Power Propuls.
Conf. IEEE, Oct. 2014, pp. 1–6. [Online]. Available:
http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=7007041
[6] A. Adam, E. S. Fraga, and D. J. Brett, “Options for
residential building services design using fuel cell based
micro-CHP and the potential for heat integration,” Appl.
Energy, vol. 138, pp. 685–694, Jan. 2015. [Online]. Available:
http://linkinghub.elsevier.com/retrieve/pii/S0306261914011532
[7] J. E. Garcia, D. F. Herrera, L. Boulon, P. Sicard, and
A. Hernandez, “Power sharing for efficiency optimisation into
a multi fuel cell system,” in 2014 IEEE 23rd Int. Symp. Ind.
Electron. Istanbul: IEEE, Jun. 2014, pp. 218–223. [Online]. Available:
http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6864614
[8] L. Boulon, D. Hissel, A. Bouscayrol, and M.-C. Pera,
“From Modeling to Control of a PEM Fuel Cell Using
Energetic Macroscopic Representation,” IEEE Trans. Ind. Electron.,
vol. 57, no. 6, pp. 1882–1891, Jun. 2010. [Online]. Available:
http://ieeexplore.ieee.org/xpls/abs all.jsp?arnumber=5166485
Fig. 7: Cost of operation per day [9] Y. Tang, W. Yuan, M. Pan, and Z. Wan, “Experimental investigation
on the dynamic performance of a hybrid PEM fuel cell/battery
system for lightweight electric vehicle application,” Appl. Energy,
vol. 88, no. 1, pp. 68–76, Jan. 2011. [Online]. Available:
system over the series system enables it to further increase the http://linkinghub.elsevier.com/retrieve/pii/S0306261910003028
[10] M. B. Gunes, “Investigation of a Fuel Cell Based Total Energy System
gap between the two at low H numbers. for Residential Applications Investigation of a Fuel Cell Based Total
Energy,” Ph.D. dissertation, Virginia Polytechnic Institute and State
VI. C ONCLUSION University, 2001.
[11] K. Mets, T. Verschueren, F. De Turck, and C. Develder, “Exploiting
MFC systems can be set up using different architectures. V2G to optimize residential energy consumption with electrical
vehicle (dis)charging,” in 2011 IEEE First Int. Work. Smart
Some architectures provide individual current control capabil- Grid Model. Simul. IEEE, Oct. 2011, pp. 7–12. [Online].
ities. For those systems, the daisy-chain power distribution is Available: http://ieeexplore.ieee.org/xpls/abs all.jsp?arnumber=6089203
used. A dynamic fuel cell model taking ancillaries into account http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6089203
[12] S. Cleghorn, D. Mayfield, D. Moore, J. Moore, G. Rusch, T. Sherman,
is used for energy calculations. The ageing and hydrogen N. Sisofo, and U. Beuscher, “A polymer electrolyte fuel cell
consumption resulting from the use of the three presented life test: 3 years of continuous operation,” J. Power Sources,
systems on a residential power profile including the charge vol. 158, no. 1, pp. 446–454, Jul. 2006. [Online]. Available:
http://linkinghub.elsevier.com/retrieve/pii/S0378775305013546
of an EV are evaluated. [13] Y. Yu, G. Wang, Z. Tu, Z. Zhan, and M. Pan, “Effect of gas
The results show that the choice of the best architecture is shutoff sequences on the degradation of proton exchange membrane
highly dependent on the H factor. For high values of H the fuel cells with dummy load during startup and shutdown cycles,”
Electrochim. Acta, vol. 71, pp. 181–193, Jun. 2012. [Online]. Available:
series architectures becomes the less expensive. Otherwise, the http://linkinghub.elsevier.com/retrieve/pii/S0013468612004938
parallel architecture is more interesting. The use of the series- [14] J. Wu, X. Z. Yuan, J. J. Martin, H. Wang, J. Zhang, J. Shen,
parallel architecture always result in higher degradation and S. Wu, and W. Merida, “A review of PEM fuel cell durability:
Degradation mechanisms and mitigation strategies,” J. Power Sources,
cost than the other architectures. However these results are vol. 184, no. 1, pp. 104–119, Sep. 2008. [Online]. Available:
based on the assumption that the series and parallel systems http://linkinghub.elsevier.com/retrieve/pii/S0378775308011968
have the same cost. This issue should be addressed in future [15] Y. Yu, H. Li, H. Wang, X.-Z. Yuan, G. Wang, and M. Pan,
“A review on performance degradation of proton exchange
works. membrane fuel cells during startup and shutdown processes:
Causes, consequences, and mitigation strategies,” J. Power
R EFERENCES Sources, vol. 205, pp. 10–23, May 2012. [Online]. Available:
http://linkinghub.elsevier.com/retrieve/pii/S0378775312001024
[1] A. Ipakchi and F. Albuyeh, “Grid of the future,” IEEE Power Energy [16] United States Department of Energy, “Fuel Cell Technologies Office
Mag., vol. 7, no. 2, pp. 52–62, Mar. 2009. [Online]. Available: Multi-Year Research, Development, and Demonstration Plan,” Tech.
http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=4787536 Rep., 2012.
[2] Y. Hou, D. Hao, C. Shen, and Z. Shao, “Experimental
investigation of the steady-state efficiency of fuel cell stack
under strengthened road vibrating condition,” Int. J. Hydrogen Energy,
vol. 38, no. 9, pp. 3767–3772, Mar. 2013. [Online]. Available:
http://linkinghub.elsevier.com/retrieve/pii/S0360319913000943
[3] N. Marx, L. Boulon, F. Gustin, D. Hissel, and K. Agbossou, “A review
of multi-stack and modular fuel cell systems: Interests, application
areas and on-going research activities,” Int. J. Hydrogen Energy,
vol. 39, no. 23, pp. 12 101–12 111, Aug. 2014. [Online]. Available:
http://linkinghub.elsevier.com/retrieve/pii/S0360319914016036
[4] L. Palma and P. N. Enjeti, “A Modular Fuel Cell,
Modular DC–DC Converter Concept for High Performance
and Enhanced Reliability,” IEEE Trans. Power Electron.,
vol. 24, no. 6, pp. 1437–1443, Jun. 2009. [Online]. Available:
http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=4939358

You might also like