IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO.

3, SEPTEMBER 2006 767

Dynamic Modeling, Design, and Simulation of a
Combined PEM Fuel Cell and Ultracapacitor System
for Stand-Alone Residential Applications
M. Uzunoglu, Member, IEEE, and M. S. Alam, Senior Member, IEEE
AbstractThe available power generated from a fuel cell (FC)
power plant may not be sufcient to meet sustained load demands,
especially during peak demand or transient events encountered
in stationary power plant applications. An ultracapacitor (UC)
bank can supply a large burst of power, but it cannot store a
signicant amount of energy. The combined use of FC and UC
has the potential for better energy efciency, reducing the cost
of FC technology, and improved fuel usage. In this paper, we
present an FC that operates in parallel with a UC bank. A new
dynamic model and design methodology for an FC- and UC-
based energy source for stand-alone residential applications has
been developed. Simulation results are presented using MATLAB,
Simulink, and SimPowerSystems environments based on the math-
ematical and dynamic electrical models developed for the proposed
system.
Index TermsCombined system, dynamic modeling, fuel cell
(FC), proton exchange membrane fuel cell (PEMFC), ultracapaci-
tor (UC).
I. INTRODUCTION
F
UEL CELL (FC) power plants are electrochemical devices
that convert the chemical energy of a reaction directly into
the electrical energy. Among the various next-generation power
plants, FCs, especially the proton exchange membrane fuel cells
(PEMFCs), are considered to be one of the promising energy
sources due to high efciency and are environment friendly [1].
However, for stationary and vehicular applications, an FC
power plant may not be sufcient to satisfy the load demands,
especially during peak demand periods or transient events. By
operating the FC and UC in parallel, both steady-state and peak
power demands can be satised. Without the ultracapacitor
(UC) bank, the FC power plant would have to supply all power
demand, thus increasing the size and cost of the FCpower plant.
The UC bank is designed to provide the difference between the
load demand and the output power generated by the FC power
plant.
UCs are electrical energy storage devices, which offer sig-
nicantly better energy densities than conventional capacitors,
Manuscript received August 18, 2005; revised January 4, 2006. This work
was supported in part by the U.S. Department of Energy under Grant DE-FG02-
02ER63376. Paper no. TEC-00297-2005.
M. Uzunoglu is with the Department of Electrical and Computer Engi-
neering, University of South Alabama, Mobile, AL 36688 USA, He is also
with the Yildiz Technical University, Besiktas, Istanbul 34349, Turkey (e-mail:
muzunoglu@usouthal.edu).
M. S. Alam is with the Department of Electrical and Computer En-
gineering, University of South Alabama, Mobile, AL 36688 USA (e-mail:
malam@usouthal.edu).
Digital Object Identier 10.1109/TEC.2006.875468
better power densities than conventional batteries, and can be
constructed in modular and/or stackable format. The charge and
discharge times of a UCvaries fromfractions of a second to sev-
eral minutes, while providing maintenance-free operation. UCs
provide lowest cost per Farad, extremely high cycling capabil-
ity, and are environmentally safe [2] [4]. The capacitance of
UCs may vary from a few Farads to several thousand Farads per
cell [5]. Because of the aforementioned unique characteristics,
UCs are utilized for a wide range of applications. Therefore,
a UC bank can effectively serve as a cost effective alternative
to batteries for residential applications, especially during short
peak demand periods.
In this paper, a new model consisting of two energy sources
(FC system and UC bank) is proposed for residential applica-
tions. Therefore, the dynamic behavior of each component must
be carefully considered and modeled in order to ensure efcient
power ow. Recently, a number of UCmodels has been reported
in the literature, which includes the classical equivalent model,
lumped-parameter or distributed parameter electrical models,
ladder circuit model, and Debye polarization cell model [6]
[8]. The classical equivalent circuit model of the UC is suitable
for slow discharge and pulse load applications. The capacitor
voltage of this model is very similar to the waveform generally
measured in the laboratory [6]. To model the dynamic char-
acteristics of the PEMFC system, we utilized the load prole
data obtained for the FC powered smart energy management
and control (SEMaC) project funded by the DoE [9]. This load
prole data was also utilized in [10].
A 5-kW PEMFC system can essentially satisfy the general
power requirements of a typical residential home. However, the
power requirement varies signicantly during different periods
of the day, and from one home to another home. For example,
the load prole obtained from a 2500-ft
2
house with all elec-
trical devices occupied by two adults and four children, depicts
that during the peak demand periods, the power requirement
varies from 7 to 7.4 kW exceeding the 5-kW maximum power
available from FC system. To overcome the above mentioned
limitations, we present a combined FC and UC system, where
the FC systems supplies the base load (up to 5 kW) and the
UC bank supplies the additional power required during the peak
power demand and load switching.
This paper focuses on designing and dynamic modeling of a
combined PEMFCand UCbank system, as well as developing of
power ow control strategies. Simulation results obtained using
MATLAB, Simulink, and SimPowerSystems are presented to
verify the effectiveness of the proposed technique.
0885-8969/$20.00 2006 IEEE
768 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006
II. DYNAMIC MODELING OF A PEMFC
The FC model used in this paper is realized in MATLAB
and Simulink. Then this model is embedded into the Sim-
PowerSystems of MATLAB as a controlled voltage source.
The FC system model parameters used in this model are as
follows:
B, C constants to simulate the activation over voltage in
PEMFC system and [V];
CV conversion factor [kmol of Hydrogen per kmol of
methane];
E Nernst instantaneous voltage [V];
E
0
standard no load voltage [V];
F Faradays constant [C/kmol];
I

FC
FC system feedback current [A];
k
1
proportional-integral (PI) gain;
K
an
anode valve constant [

kmol Kg(atms)
1
];
K
H
2
hydrogen valve molar constant [kmol/(atms)];
K
H
2
O
water valve molar constant [kmol/(atms)];
K
O
2
oxygen valve molar constant [kmol/(atms)];
K
r
modeling constant [kmol/(sA)];
M
H
2
molar mass of hydrogen [kgkmol
1
];
N
0
number of series fuel cells in the stack;
p
H
2
hydrogen partial pressure [atm];
p
H
2
O
water partial pressure [atm];
p
O
2
oxygen partial pressure [atm];
q
H
2
molar ow of hydrogen [kmol/s];
q
O
2
input molar ow of oxygen [kmol/s];
q
methane
methane ow rate [kmol/s];
q
in
H
2
hydrogen input ow [kmol/s];
q
out
H
2
hydrogen output ow [kmol/s];
q
r
H
2
hydrogen ow that reacts [kmol/s];
q
req
H
2
amount of hydrogen owrequired to meet the load
change [kmol/s];
R universal gas constant [(1 atm)/(kmolK)];
R
int
FC internal resistance[];
T absolute temperature [K];
U utilization rate;
V
an
volume of the anode[m
3
];
V
cell
dc output voltage of FC system [V];

1
,
2
reformer time constants [s];

3
time constant of the PI controller [s];

H
2
hydrogen time constant [s];

O
2
oxygen time constant [s];

H
2
O
water time constant [s];

act
activation over voltage [V];

ohmic
ohmic over voltage [V].
The relationship between the molar ow of any gas (hydro-
gen) through the valve and its partial pressure inside the channel
can be expressed as [11]
q
H
2
p
H
2
=
K
an
_
M
H
2
= K
H
2
. (1)
For hydrogen molar ow, there are three signicant factors:
hydrogen input ow, hydrogen output ow, and hydrogen ow
during the reaction [11]. The relationship among these factors
can be written as
d
dt
p
H
2
=
RT
V
an
_
q
in
H
2
q
out
H
2
q
r
H
2
_
. (2)
According to the basic electrochemical relationship between
the hydrogen ow and the FC system current, the ow rate of
reacted hydrogen is given by [11]
q
r
H
2
=
N
0
I

FC
2F
= 2K
r
I

FC
. (3)
Using (1) and (3), and applying Laplace transform, the hy-
drogen partial pressure can be obtained in the s domain as [11]
p
H
2
=
1/K
H
2
1 +
H
2
s
_
q
in
H
2
2K
r
I

FC
_
(4)
where

H
2
=
V
an
K
H
2
RT
. (5)
Similarly, water partial pressure and oxygen partial pressure can
be obtained. The polarization curve for the PEMFC is obtained
fromthe sumof the Nernsts voltage, the activation over voltage,
and the ohmic over voltage. Assuming constant temperature and
oxygen concentration, the FC output voltage may be expressed
as [10], [12], [13]
V
cell
= E +
act
+
ohmic
(6)
where

act
= Bln(CI

FC
) (7)
and

ohmic
= R
int
I

FC
. (8)
Now, the Nernsts instantaneous voltage may be expressed as
[10]
E = N
0
_
E
0
+
RT
2F
log
_
p
H
2
_
P
O
2
p
H
2
O
__
. (9)
The fuel cell system consumes hydrogen, according to power
demand and the reformer continuously generates hydrogen for
stack operation. The mathematical form of the reformer model
can be expressed as [10], [14]
q
H
2
q
methanol
=
CV

2
s
2
+ (
1
+
2
)s + 1
. (10)
During operational conditions, to control the hydrogen ow
rate according to the output power of the FCsystem, a PI control
system is used. To achieve this feedback control, FC current
from the output is taken back to the input while converting the
hydrogen into molar form [10], [14], given by
q
req
H
2
=
N
0
I

FC
2FU
. (11)
The amount of hydrogen available from the reformer can be
used to control the methane ow rate by using a PI controller
UZUNOGLU AND ALAM: DYNAMIC MODELING, DESIGN, AND SIMULATION OF A COMBINED PEMFC 769
Fig. 1. Dynamic model of the FC system.
[10], [14], expressed as
q
methanol
=
_
k
1
+
k
1

3
s
__
N
0
I

FC
2FU
q
in
H
2
_
. (12)
The hydrogenoxygen ow ratio r
HO
in the reformer deter-
mines the oxygen ow rate [10]. The MATLAB and Simulink-
based FC system model developed in this paper using the above
equations is shown in Fig. 1.
III. UC SYSTEM MODEL
In this section, the mathematical modeling of the UC bank
used in the simulation program is introduced. The parameters
used in the mathematical modeling of the UC are as follows:
C capacitance [F];
C
UCtotal
the total UC system capacitance [F];
EPR equivalent parallel resistance [];
ESR, R equivalent series internal resistance [];
E
UC
the amount of energy drawn from the UC bank
[Ws];
n
s
the number of capacitors connected in series;
n
p
the number of series strings in parallel;
R
UCtotal
the total UC system resistance [];
t time [s];
V (t) the voltage after time t [V];
V
i
the initial voltage before discharging starts [V];
V
f
the nal voltage after discharging starts [V].
Fig. 2 shows the classical equivalent circuit of the UC unit.
The model consists of a capacitance (C), an equivalent series
resistance (ESR, R) representing the charging and discharging
resistance, and an equivalent parallel resistance (EPR) repre-
senting the self-discharging losses [6], [14]. The EPR models
leakage effects and only impacts long-term energy storage per-
formance of the UC [15], [16].
Fig. 2. Classical equivalent model for the UC unit.
The voltage state of a UC (i.e., RC circuit) with a capacitance
C draining charge into a resistance R may be described as [7]
V (t) = V
i
exp
_

t
RC
_
. (13)
The RC time constant determines the effective period of the
charging and discharging processes for some initial voltage on
the capacitor [7].
The amount of energy drawn from the UC bank is directly
proportional to the capacitance and the change in terminal volt-
age [5], [15], expressed as
E
UC
=
1
2
C
_
V
2
i
V
2
f
_
. (14)
The effective specic energy for a prescribed load can be
supplied by various UC bank congurations. In practical appli-
cations, the required amount of terminal voltage and energy or
the capacitance of UC storage system can be built using multi-
ple UCs in series and parallel. The terminal voltage determines
the number of capacitors that must be connected in series to
form a bank and the total capacitance determines the number of
capacitors which must be connected in parallel in the bank. The
total UC system resistance and the total UC system capacitance
770 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006
Fig. 3. Arrangement of capacitors in a UC bank.
of the UC bank are calculated as [14], [15]
R
UCtotal
= n
s
ESR
n
p
(15)
C
UCtotal
= n
p
C
n
s
. (16)
Fig. 3 shows, how a number of UC units can be arranged
to build a UC bank that is capable of providing the peak load
demand [14]. The UC bank model has been implemented in
MATLAB and SimPowerSystems for this study.
IV. COMBINED FC AND UC CONTROL SYSTEM
In this section, we present the integration of an FC system in
parallel with a UC bank, which is realized as an energy source
for stand-alone residential applications. Various topologies may
be used for integrating the UC with the FC power system [14],
[15], [17] [20]. For example, the UC system integration can
be done with or without a power electronic converter via series
and parallel connection. Because of the high specic power
and power density of UC bank, it may be possible to eliminate
the dc/dc converter for voltage regulation so that it can deliver
higher output [14], [17], [18], [21]. Also, the direct integration
of the FCwith the UCbank is possible for relatively low-voltage
applications with a dc bus of less than 50 V [2].
In this paper, we integrate a UCbank with the FCpower plant
using a power diode, as shown in Fig. 4. This diode prevents the
ow of reverse current from the UC bank into the FC system.
This FC vehicle system topology, proposed by Honda [17], is
utilized for stationary applications. The hybrid power source
corresponds to the direct integration of the FC and UC in paral-
lel, as illustrated in Fig. 4 [14], [17], [18], [21].
The direct integration of the UC bank with the FC system
is attractive, because it does not require a high power dc/dc
converter. Moreover, the complexity, cost, weight, and volume
of the system are signicantly reduced. Since the amount of
Fig. 4. Combination of FC system and UC bank.
losses in the diode is less important than that in a dc/dc converter,
the efciency of the overall system also increases [14], [17].
The UCbank and the load terminal voltages depend on the FC
system terminal voltage, which prevents the power capability of
the UC bank from being fully utilized. Therefore, the UC bank
size is restricted by the terminal voltage of the FC system, as in
the battery/UC hybrid system of [18].
The power sharing between the FC system and UC bank is
determined by the total resistance between these two systems.
The main control strategy for the combined system reported
in [1], [14], [15], [17], and [19], is modied as shown in Fig. 4
and can be summarized as follows.
1) During low power demand periods (<5 kW), the FC sys-
tem generates up to its load limit, and the excess power is
used to charge the UC. The charging or discharging of the
UC bank occurs according to the terminal voltage of the
overall load requirements.
2) During high power demand periods (5 kW), the FC sys-
tem generates the rated power and the UC is discharged to
meet the extra power requirements that cannot be supplied
by the FC system.
3) Short-time power interruptions in the FC system can only
be supplied by the UC bank.
4) The UC bank is designed to avoid overcharge or under-
charge conditions.
5) About 75% of the initial energy stored in the UC bank
can be utilized if the terminal load voltage is allowed to
decrease to 50% of its initial value.
To realize the control system of the above mentioned
combined system, PI controllers, ideal switching elements,
and current and voltage sensors are used in the simulation
model.
V. PCU
Because the combined model produces dc electrical energy,
it must be converted to ac electrical energy for domestic elec-
trical loads. The relationship between the ac and dc quantities
at the inverter terminals provides the interconnecting equations
between the combined system and the load [20], [22]. The pro-
posed model for domestic loads requires a dc/ac inverter, which
also acts as an energy buffer to match the different dynamics
UZUNOGLU AND ALAM: DYNAMIC MODELING, DESIGN, AND SIMULATION OF A COMBINED PEMFC 771
of the FC system, UC bank and various loads. The proper re-
sponse of the combined system to the overall load dynamics
can be achieved by generating appropriate switching signals to
the dc/ac inverter while modulating for both active and reactive
powers.
In real-world applications, the power conditioning unit (PCU)
consists of a dc/dc converter and a dc/ac inverter. However, in
this study, only a simple model of a dc/ac inverter is considered.
Adc/dc converter has a very small time constant (milliseconds or
microseconds) when compared to the time constants (seconds)
for the FC system. Therefore, the dc/dc converter is omitted
because it will have negligible effect on the dynamic response
of the system [10], [11]. Also, it complicates the system and
may cause numerical instabilities as reported in [11].
The parameters used in the mathematical modeling of the
PCU in the simulation program are as follows:
I
dc
dc current [A];
I
FC
FC system current transferred to dc bus [A];
I
L
load current [A];
I
UC
UC bank current [A];
k
2
, k
3
, k
4
, k
5
constants of the PI controller;
m inverter modulation index;
P
ac
ac output power [W];
P
L
load power [W];
V
ac
ac output voltage of the inverter [V];
V
dc
combined system dc terminal voltage [V];
V
r
reference voltage signal [V];
V
s
load terminal voltage [V];
X reactance of the line connecting the combined
system and the load [];
phase angle of the ac voltage [rad];
load phase angle [rad];
The ac side of the inverter serves as a source with adjustable
magnitude and phase angle of the voltage. The dc side of the
inverter operates as a current source [22]. The ac voltage and
active power at the inverter output can be expressed as a function
of the modulation index and phase angle as [10], [23]
V
ac
= mV
dc

(17)
P
ac
=
mV
dc
V
s
X
sin(). (18)
The corresponding load current and dc current can be ex-
pressed as
I
L
=
P
L
V
s
cos()
(19)
I
dc
= mI
L
cos( +) (20)
I
dc
= I
FC
+I
UC
. (21)
Assuming a lossless inverter, P
ac
can be expressed as [10],
[23]
P
ac
= P
dc
= V
dc
I
dc
. (22)
The PCU supplies two loops to control the output ac voltage
and power. The modulation index and the phase angle are used
to control the terminal ac output voltage and the active power
Fig. 5. PCU and load connection diagram.
Fig. 6. Block diagram of the proposed model.
ow from the combined system to the load as shown in Fig. 5.
The voltage at the load terminals is kept constant, using the
PI controller. The modulation index and phase angle associated
with the PI controller can be calculated as
m =
_
k
2
+k
3
s
s
_
(V
r
V
ac
) (23)
=
_
k
4
+k
5
s
s
_
(P
L
P
ac
). (24)
The FC system current feedback signal is calculated from the
load, UC bank current, and the ac output voltage.
VI. SIMULATION RESULTS
The topology used in this study for the combined FC and UC
system, PCU, and load is shown in Fig. 6.
The proposed PEMFC system operates with a UC bank con-
nected in parallel with the dc bus via a diode. The voltage drop
in the diode is assumed to be 0.6 V. The UC bank serves as
a short duration power source to meet load demand that can-
not be met by the FC system, particularly during transient or
peak demand periods. In this study, the UC bank is designed
to provide the difference between the load and the FC system
output power. Simulation results are obtained by developing
a detailed MATLAB, Simulink, and SimPowerSystems-based
software packages using the mathematical and electrical mod-
els of the system described earlier.
The 5-kW PEMFC system currently operated in our labo-
ratory feeds a 500-ft
2
house. Energy consumption varies from
772 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006
Fig. 7. Real power of residential load.
TABLE I
MAXWELL BOOSTCAP PC2500 UC CHARACTERISTICS [24]
house to house, but to simulate a real-world application, the
load prole obtained from a 2500-ft
2
house with all electrical
devices occupied by two adults and four children is used. The
load prole was measured from a 120-V circuit of a residential
home. Fromthis load prole, it is evident that the average power
demand is less than 3 kW. However, the load prole during peak
load periods varies from 7 to 7.4 kW, as illustrated in Fig. 7.
The measurement of this load prole is realized with a 15-s
sampling interval. Fig. 7 shows that the maximum period of
the peak demand is 75 s, when the demand exceeds the 5-kW
maximum power available from the FC system.
The proposed combined FC (commercially available 5 kW)
and UC system is capable of sustaining the extra load of 2.4 kW
for 75 s (i.e, 50 Wh of energy) available from the UC bank
during peak demand periods. Thus, the energy capacity of the
UC bank must have a minimum rating of 50/0.75 = 66.67 Wh.
For the simulated model, we selected the Maxwell Boostcap
PC2500 UC unit, whose characteristics are shown in Table I.
Each UC unit has a nominal voltage of 2.5 V corresponding to
2700 F. Assuming a 48-Vdc output fromthe FCsystem, a string
of 20 UCs in series (135 F per string) is used to represent 48 V.
It can also be used to compensate for the internal resistance
(approximately 1 m per UC unit) within the units.
The energy stored in a 135-F capacitance at 48 Vcorresponds
to 43.2 Wh. The required energy for the above mentioned
residential load is (2400/0.75) (75/3600) = 66.67 Wh. To
provide this required amount of energy, two strings of 20 UC
units should be connected in parallel, which provides a total
capacitance of 270 F, and 86.4 Wh of energy. This design can
be changed to larger or smaller power capacities for different
applications.
The PEMFC system parameters in this work are given in
Table II [10]. The initial active output power of the FC system
TABLE II
FC SYSTEM MODEL PARAMETERS
is assumed to be 3.6 kW. Start-up transients from 0 to 17 s are
not considered and this interval is not shown in the plots to
represents the dormant state of the FC system.
The output power of the FC system is limited to 5 kW and
only single 120 V
rms
circuit of the house is considered. The
inverter is assumed to have an input of 48-V dc and output of
120-V
rms
ac.
In the PCU, PI controllers are used to control the ac output
voltage and active power, which adjusts the modulation index
and phase angle, according to load variations. Thus, the ac output
voltage is kept at 120 V
rms
and the total load power demand
is met from FC system and UC bank by the power ow PI
controller. The power ow PI controller regulates the ow of
current depending on the required load power with a short time
delay. In this study, PCU losses are assumed to be negligible.
Simulation results are obtained for the time interval between
18:30:00 and 19:13:15 h of the recorded load prole as depicted
in Fig. 7. This time interval including peak load demands corre-
sponds to the 2595 s of simulation duration. Figs. 816 showthe
simulation results for the FC system diode output voltage, UC
bank terminal voltage, UC bank current, ac output load voltage,
PCU modulation index, ac voltage phase angle, ac output active
power, FC system output power, and hydrogen ow rate under
load switches, respectively, as a function of time.
FromFig. 8, it is seen that the increase in load power decreases
the FC output voltage. The simulation result for the UC bank
voltage is illustrated in Fig. 9. During each pulse on time, the
terminal voltage at the UC bank drops fromapproximately 48 V
to about 34 V (about 14-V variation) due to the switching of UC
bank fromthe charging to discharging mode. During this period,
the FC system voltage decreases slightly (about 2-V variation).
UZUNOGLU AND ALAM: DYNAMIC MODELING, DESIGN, AND SIMULATION OF A COMBINED PEMFC 773
Fig. 8. Variation of FC system output voltage according to load demand.
Fig. 9. Variation of UC bank terminal voltage according to load demand.
Fig. 10. Variation of UC bank charging and discharging current according to
load switching.
Fig. 11. Variation of ac output power.
Fig. 12. Variation of ac load voltage.
Fig. 13. Variation of modulation index corresponding to load demand.
774 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006
Fig. 14. Variation of ac voltage phase angle.
Fig. 15. Variation of FC system dc output power.
Fig. 10 shows that the variation of UC bank current switching
between negative (charging) and positive (discharging) accord-
ing to the terminal voltage requirement of the overall load. From
Fig. 11, it is evident that the FC system and UC bank together
share this load requirement. During peak load demand, the load
power requirement is higher than the power generated by the FC
system. Therefore, the FC system supplies the available power
and the UC bank supplies the remaining extra power. At this
time, the UC bank discharge current is very high and the UC
bank terminal voltage drops signicantly.
Although the FC and UC voltages are affected by the load
conditions as seen in Figs. 8 and Fig. 9, the PI controlled modu-
lation index of the inverter successfully maintains the ac output
voltage stable as illustrated in Fig. 12. For the modulation index
control, the real ac bus voltage is compared to the reference volt-
age and then the error signal is processed in the PI controller.
Fig. 13 shows the corresponding modulation index adjustment
corresponding to load demand variations.
Fig. 16. Variation of hydrogen ow rate.
To meet the load requirements, the ac voltage phase angle
is controlled as discussed earlier. For phase angle control, the
real load power is compared to the output power of the inverter
and then the error signal is processed in the PI controller which
determines the phase angle , as shown in Fig. 14.
Fig. 15 shows that during periods of low power demand
(<5 kW), the FC system satises the load demand and at the
same time charges the UC bank. The hydrogen ow rate varies
according to the system power requirements, i.e., feedback cur-
rent as illustrated in Fig 16.
VII. CONCLUSION
A UC-based storage system is designed for a PEMFC-
operated grid independent home to supply the extra power re-
quired during peak demand periods. The parallel combination
of the FC system and UC bank exhibits good performance for
the stand-alone residential applications during the steady-state,
load-switching, and peak power demand. Without the UC bank,
the FC system must supply this extra power, thereby increasing
the size and cost of the FC system. The results corresponding
to high peak load demand during short time periods are not
shown in order to simulate more realistic load prole. The load
prole was created by measuring data at 15-s sampling interval.
However, the proposed model can be used for different load
proles consisting of different transients and short-time inter-
ruption. Also, it can be extended for use in many areas such
as portable devices, heavy vehicles, and aerospace applications.
The lifetime of an FC system can be increased if combined FC
system and UC bank is used instead of a stand-alone FC system
or a hybrid FC and standby battery system.
REFERENCES
[1] L. Gao, Z. Jiang, and R. A. Dougal, An actively controlled fuel
cell/battery hybrid to meet pulsed power demands, J. Power Sources,
vol. 130, no. 12, pp. 202207, May 2004.
UZUNOGLU AND ALAM: DYNAMIC MODELING, DESIGN, AND SIMULATION OF A COMBINED PEMFC 775
[2] T. S. Key, H. E. Sitzlar, and T. D. Geist, Fast response, load-matching
hybrid fuel cell, Final Tech. Prog. Rep., EPRI PEAC Corp., Knoxville,
TN NREL/SR-560-32743, Jun. 2003.
[3] S. Buller, E. Karden, D. Kok, and R. W. De Doncker, Modeling the dy-
namic behavior of supercapacitors using impedance spectroscopy, IEEE
Trans. Ind. Appl., vol. 38, no. 6, pp. 16221626, Nov. 2002.
[4] J. L. Duran-Gomez, P. N. Enjeti, and A. Von Jouanne, An approach to
achieve ride-through of an adjustable-speed drive with yback converter
modules powered by super capacitors, IEEE Trans. Ind. Appl., vol. 38,
no. 2, pp. 514522, Mar.Apr. 2002.
[5] A. Burke, Ultracapacitors: Why, how, and where is the technology, J.
Power Sources, vol. 91, no. 1, pp. 3750, Nov. 2000.
[6] R. M. Nelms, D. R. Cahela, and B. J. Tatarchuk, Modeling double-layer
capacitor behavior using ladder circuits, IEEE Trans. Aerosp. Electron.
Syst., vol. 39, no. 2, pp. 430438, Apr. 2003.
[7] B. E. Conway, Electrochemical Supercapacitors-Scientic Fundamen-
tals and Technological Applications. New York: Kluwer/Plenum, 1999,
pp. 497547.
[8] R. M. Nelms, D. R. Cahela, and B. J. Tatarchuk, Using a Debye polar-
ization cell to predict double-layer capacitor performance, IEEE Trans.
Ind. Appl., vol. 37, no. 1, pp. 49, Jan.Feb. 2001.
[9] M. S. Alam, Fuel Cell Powered Smart Energy Management and con-
trol (SEMaC), Project DE-FG02-02ER63376, Funded by the U.S. Dept.
Energy.
[10] M. Y. El-Sharkh, A. Rahman, M. S. Alam, P. C. Byrne, A. A. Sakla, and
T. Thomas, A dynamic model for a stand-alone PEM fuel cell power
plant for residential applications, J. Power Sources, vol. 138, no. 12,
pp. 199204, Nov. 2004.
[11] J. Padulles, G. W. Ault, and J. R. McDonald, An integrated SOFC plant
dynamic model for power systems simulation, J. Power Sources, vol. 86,
no. 12, pp. 495500, Mar. 2000.
[12] J. C. Amphlett, R. F. Mann, B. A. Peppley, P. R. Roberge, and A. Rodrigues,
Amodel predicting transient response of proton exchange membrane fuel
cells, J. Power Sources, vol. 61, no. 12, pp. 183188, Jul.Aug. 1996.
[13] J. Hamelin, K. Agbossou, A. Laperriere, F. Laurencelle, and T. K. Bose,
Dynamic behavior of a PEM fuel cell stack for stationary applications,
Int. J. Hydrogen Energy, vol. 26, no. 6, pp. 625629, Jun. 2001.
[14] K.-H. Hauer, Analysis tool for fuel cell vehicle hardware and software
(controls) with an application to fuel economy comparisons of alternative
system designs, Ph.D. dissertation, Dept. Transport. Technol. Policy,
Univ. CA, Davis, CA, 2001.
[15] R. L. Spyker and R. M. Nelms, Analysis of double-layer capacitors
supplying constant power loads, IEEE Trans. Aerosp. Electron. Syst.,
vol. 36, no. 4, pp. 14391443, Oct. 2000.
[16] R. L. Spyker and R. M. Nelms, Classical equivalent circuit parameters for
a double-layer capacitor, IEEE Trans. Aerosp. Electron. Syst., vol. 36,
no. 3, pp. 829836, Jul. 2000.
[17] Honda Fuel Cell Power FCX (Dec. 2004). [Online]. Available:
http://world.honda.com/FuelCell/FCX/FCXPK.pdf, Press Information,
2004.12
[18] L. Gao, R. A. Dougal, and S. Liu, Power enhancement of an actively
controlled battery/ultracapacitor hybrid, IEEE Trans. Power Electron.,
vol. 20, no. 1, pp. 236243, Jan. 2005.
[19] W. Vielstich, A. Lamm, and H. A. Gasteiger, Hy. PowerA technology
platformcombining a fuel cell systemand a supercapacitor, in Handbook
of Fuel CellsFundamentals, Technology and Applications. vol. 4(11),
New York: Wiley, 2003, pp. 11841198.
[20] M. Cacciato, F. Caricchi, F. Giuhlii, and E. Santini, A critical evaluation
and design of bi-directional dc/dc converters for super-capacitors inter-
facing in fuel cell applications, in Proc. 39th IAS Annu. Meeting, 2004,
vol. 2, pp. 11271133.
[21] M. W. Verbrugge, P. Liu, and S. Soukiazian, Activated-carbon electric-
double-layer capacitors: Electrochemical characterization and adaptive
algorithm implementation, J. Power Sources, vol. 141, no. 2, pp. 369
385, Mar. 2005.
[22] C. J. Hatziadoniu, A. A. Lobo, F. Pourboghrat, and M. Daneshdoost, A
simplied dynamic model of grid-connected fuel-cell generators, IEEE
Trans. Power Deliv., vol. 17, no. 2, pp. 467473, Apr. 2002.
[23] M. Tanrioven and M.S. Alam, Modeling, control and power quality eval-
uation of a PEM fuel cell based power supply system for residential use,
in Proc. 39th IAS Annu. Meeting, 2004, vol. 4, pp. 28082814.
[24] Electric Double Layer Capacitor: BOOSTCAP Ultracapacitor [Online].
Available: http://www.maxwell.com/pdf/uc/datasheets/PC2500.pdf
M. Uzunoglu (S95M06) received the B.Sc.,
M.Sc., and Ph.D. degrees from Yildiz Technical Uni-
versity, Besiktas, Istanbul, Turkey, in 1991, 1996, and
2000, respectively, all in electrical engineering.
He is a Faculty Member at Yildiz Technical Uni-
versity and a Research Scholar in Electrical and Com-
puter Engineering Department, University of South
Alabama, Mobile. His current research interests in-
clude alternative energy sources, power system mod-
eling, analysis and control, power quality, voltage
stability, and state estimation.
M. S. Alam (S91M93SM95) is a Professor and
Chair of the Electrical and Computer Engineering
Department, University of South Alabama, Mobile.
His research interests include ultrafast computer ar-
chitectures and algorithms, image processing, pat-
tern recognition, ber-optics, infrared systems, digi-
tal system design, and smart energy management and
control. He is the author or coauthor of more than 325
published papers, including 136 articles in refereed
journals, 175 papers in conference proceedings, and
12 book chapters. He has presented over 55 invited
papers, seminars, and tutorials at international conferences and research insti-
tutions in the USA and, abroad.
Mr. Alam is a Fellow of OSA, a Fellow of the SPIE, a Fellow of the In-
stitution of Electrical Engineers (U.K.), a member of ASEE and AIP. He was
the Chairman of the Fort Wayne Section of IEEE during 19951996. He has
received numerous research, teaching, and service awards including the 2005
Outstanding Scholar of the Year Award from the USA Alumni Association. He
served or serves as the PI or CoPI of many research projects totaling nearly
13 million U.S., and supported by NSF, FAA, DoE, ARO, AFOSR, WPAFB,
SMDC, and ITT industry.