You are on page 1of 8

© EYEWIRE

E
ENERGY AND ENVIRONMENT CRISES ARE FAST BECOMING THE BIGGEST
problems around the world so, as a consequence, new renewable and clean energy power
sources must be considered. One of the prevalent alternative sources of electric power is the fuel
cell, discovered by Sir William Grove in 1839. One expects that fuel cell power generation sys-
tems will be used in a growing number of areas: in portable applications, in transportation appli-
cations, and in stationary power applications, for which fuel cell systems can provide both
power and heat with cogeneration efficiencies as high as 80%. Numerous recent works have
already highlighted the possibility of using the fuel cell in distributed power generation systems.
The fuel cell utilizes the chemical energy of hydrogen (H2 ) and oxygen (O2 ) to generate
electricity without pollution, as shown in Figure 1. The byproducts are simply pure water and
heat. There are several types of fuel cells, which are characterized by the employed electrolyte.
One of the most promising is the small, lightweight, and relatively easy to build polymer elec-
trolyte membrane fuel cell (PEMFC), first used by NASA in the 1960s as part of the Gemini
space program.
A single cell voltage of the fuel cell is given by Gibb’s free energy ΔG and is equal to 1.23 V.
This theoretical value is never reached even at no-load. For the rated current, the voltage of an
elementary cell is about 0.6–0.7 V. Then a fuel cell is always an assembly of elementary cells
that constitute a stack as shown in Figure 2.

Digital Object Identifier 10.1109/MPE.2007.911814

january/february 2008 1540-7977/08/$25.00©2008 IEEE IEEE power & energy magazine 69


Fuel Cell System
A A fuel cell stack requires fuel, oxidant,
Electron Electron
and coolant in order to operate. The com-
Flow Flow
position, pressure, and flow rate of each of
O2 O2 these streams must be regulated. In addi-
H2 H2
tion, the gases must be humidified and the
coolant temperature must be controlled.
To achieve this, the fuel cell stack must be
surrounded by a fuel system, fuel delivery
system, air system, stack cooling system,
Dilute Acid Dilute Acid and humidification system.
Electrolyte Electrolyte Once operating, the output power gen-
erated by the fuel cells must be condi-
tioned and absorbed by a load. Suitable
Pt Electrodes Pt Electrodes
The Electrolysis of Water
alarms must shut down the process if
The Electrolysis of Water
Is Reversible. unsafe operating conditions occur and a
cell voltage monitoring system must mon-
figure 1. Fuel cell principle discovered by Sir William Grove. itor fuel cell stack performance. These
functions are performed by electrical con-
trol systems. As an example, Figure 3 shows a simplified dia-
gram of the PEM fuel cell system.
When a fuel cell operates, its fuel (hydrogen and oxygen)
flows are controlled by a “fuel cell controller,” which receives
current demand. This current demand is the fuel cell current ref-
erence iFCREF (see Figure 3) coming from the energy manage-
ment controller. The fuel flows must be adjusted to match the
reactant delivery rate to the usage rate by the fuel cell controller.

Fuel Cell Dynamic Limitation


It is widely accepted that one of the main weak points of the
fuel cell is its time constants dominated by temperature and
fuel delivery system (pumps, valves, and in some cases, a
figure 2. PEMFC stack (500 W, 40 A) composed of 23 hydrogen reformer). As a result, fast load demand will cause
cells of 100 cm2 developed by ZSW Company (Germany). a high voltage drop in a short
time, recognized as “fuel
starvation phenomena.”
− VFC (t ) + For clarity, Figure 4
iFCREF (t )
presents the 0.5-kW PEM
Hydrogen PEMFC Stack Fuel Cell fuel cell voltage response to
iFC (t )
from Bottle 500 W, 23 Cells Controller a current profile. The tests
operate in two different
Flux Controller ways: current step and cur-
rent slope. It shows the drop
of voltage curve in Figure
Electric 4(a), compared with Figure
Heater Air from
Humidifier 4(b), which implies that
Compressor
Pressure fuel supply and delivered
Controller electrical current do not
Condenser Heat Exchanger
Excess coincide. Fuel flows (partic-
ularly the delay of air flow)
Excess have difficulties following
the current step. This condi-
tion of operation is evident-
figure 3. Simplified diagram of the PEM fuel cell system. vFC and iFC are the fuel cell ly hazardous for the fuel
voltage and current. cell stack.

70 IEEE power & energy magazine january/february 2008


Reliability and lifetime are the most essential considera- formances of the whole system. Moreover, one can take
tions in such power sources. The hydrogen and oxygen star- advantage of this fast auxiliary power source to achieve an
vations cause severe and permanent damage to the actual hybrid source in order to disassociate mean power siz-
electro-catalyst of the fuel cell. The fuel starvation must be ing from peak transient power sizing; the aim being a reduc-
absolutely avoided even if the operation under fuel starvation tion in volume and weight, and in the case of fuel cells used as
is momentary just within one second. main energy source, the possibility of regenerative braking.
It is therefore recommended, when utilizing a fuel cell, to
employ a power loop or a current loop in order to prevent Supercapacitors
overloads and fault conditions and to associate it with, at least, Recent progress in supercapacitor technology has principally
a fast auxiliary power source to improve the dynamic per- been applied in computer memory backup systems, but with the

Tek Stopped Single Seq 1 Acqs 28 Oct 05 16:56:06 Tek Stopped 1 Acqs 28 Oct 05 17:01:48
Ch1: FC Voltage [4 V/div]
Ch1: FC Voltage
Fuel Starvation Phenomena

40 A 40 A
Ch2: FC Current [10 A/div]
Ch3: Hydrogen Flow 4 A/s
[2 Liter/Min/Div]

Ch4: Air Flow Ch2: FC Current


1 [20 Liter/Min/Div] 1
5A 5A
2 2 Ch3: Hydrogen
Flow
3 3
Ch4: Air Flow
Time: 1 s/Div Time: 4 s/Div
4 4
(a) (b)

figure 4. Fuel cell dynamic characteristics to (a) current step and (b) current slope: 4 A/s

Activated Carbon Electrode


−Ions +Ions

εo εr A
C=
d
εo the Permittivity of Free Space
εr the Relative Permittivity of the Dielectric
Organic A the Plate Surface Area
Electrolyte d the Plate Separation
Power Supply Load

Charging Discharging

figure 5. Principle of operation of a supercapacitor.

january/february 2008 IEEE power & energy magazine 71


latest increases in capacitor energy storage levels, higher power mathematically formulated the first main theorem of thermo-
applications [especially uninterruptible power supply (UPS) and dynamics, in the 1800s. To maximize the capacitance C, the
hybrid vehicle] have become practicable. Electrochemical capac- area A must be maximized and d minimized. Nowadays, the
itors are presently called by a number of names: supercapacitor, equivalent plate separation distance at the double layer con-
ultracapacitor, or electrochemical double-layer capacitor. These sists of a few electrolyte molecule diameters of about 10−10
terms are used interchangeably, and they refer to a capacitor that m. This plate separation distance is impossible in a conven-
stores electrical energy in the interface that lies between solid tional capacitor due to traditional dielectric breakdown, con-
electrodes and an electrolyte, as delineated in Figure 5. sisting of ionization followed by spark discharge. Terminal
The double-layer capacitor phenomenon was discovered voltage of the supercapacitor is limited though, due to disso-
by Helmholtz, one of the greatest natural scientists, who ciation of the electrolyte. This limits the maximum voltage

SAFT Supercapacitor Module:


Six Cells in Series

A SAFT Supercapacitor
(3,500 F, 400 A, 2.5 V)
SAFT
SuperC Bank:
33 F, 270 V

figure 6. SAFT supercapacitor: a single cell, a module in six-series, and a bank in 108 cells in series.

Fuel Cell PFC


Fuel Cell iFC Converter + 42 V dc Bus
+
PLoad
vBus
− − +
Traction
Motor

Power
SuperC SuperC Conver ter
Bank i Converter
SuperC

− +
PSuperC

vSuperC vBus
Control Control
Signal Signal

Energy Management Controller

figure 7. Concept of the actively controlled hybrid source for a distributed generation system. pLoad , pFC , and pSuperC
are the load, fuel cell, and supercapacitor powers, respectively; vBus and vSuperC are the dc bus and supercapacitor volt-
ages; iFC and iSuperC are the fuel cell and supercapacitor currents.

72 IEEE power & energy magazine january/february 2008


(2.5–3 V) of a supercapacitor cell used in this experiment to peak power is usually for a 95% efficient discharge in which
2.5 V. Electrode area in the supercapacitor is maximized by only 5% of the energy from the device is dissipated as heat in
use of activated carbon with an effective surface area up to the equivalent series resistance (ESR). For a corresponding
3,000 m2 /g of material. The large surface area combined with high-efficiency discharge, batteries would have a much lower
the high capacitance per unit area yields the very large capac- power capability.
itance seen in the supercapacitor. Additionally, the main disadvantage of the batteries is a
The first high-power supercapacitors were developed by slow charging time, limited by a charging current [known as
the Pinnacle Research Institute (PRI) for
U.S. military applications such as laser
weaponry and missile guidance systems.
However, only in the 20th century did PLoad = PFC + PSuperC
supercapacitors become well known in the
context of hybrid electric vehicles promoted

Power
Load
by the Department of Energy (DOE) under
a supercapacitor development program.
The supercapacitors are true capacitors t5
in that energy is stored via electrostatic 0 t 1 t 2 t 3 t 4 t
charges on opposing surfaces, and they
can withstand a very large number (thou- Maximum Power
Fuel Cell

sands to millions) of charge/discharge


Power

Average Power
cycles without degradation. They are also
similar to batteries in many respects,
including the use of liquid electrolytes and 0 t
the practice of configuring various size Minimum Power
cells into arrays to meet the power, energy,
SuperC
Power

and voltage requirements of a wide range


of applications.
Supercapacitors with ratings from a few
0 t
farads to a thousand farads have been man-
ufactured by Panasonic, Montena, Ness,
ELNA, Maxwell Technologies, EPCOS,
and SAFT, Inc. As an example, Figure 6
illustrates the photographs of the recent figure 8. Power profile of a hybrid power source.
supercapacitor prototype by
SAFT Company (France):
✔ a cell (3,500 F, 2.5 V, 400 A)
✔ a module (583 F, 15 V,
400 A)
✔ a bank (33 F, 270 V, 400 A).
These capacitors are not yet Fuel Cell
available commercially; nonethe- Converter
SAFT Supercapacitor Bank: 292 F
less, prototype specifications have
Control Panel
been made available. Motor and
Generator Load
Supercapacitor
Versus Battery
Only one-half the energy at the
peak power from the battery is in
the form of electrical energy to
the load, and the other one-half is
dissipated within the battery as
heat in the equivalent series
resistance. This is to say that the dSPACE
efficiency of batteries is around Interfacing Card
50%. For supercapacitors, the figure 9. Hybrid system test bench.

january/february 2008 IEEE power & energy magazine 73


The fuel cell utilizes the chemical energy
of hydrogen and oxygen to generate electricity
without pollution.

the battery state-of-charge (SOC)]; in contrast, the superca- cell generator) to produce the compatibility and performance
pacitor may be charged in a short time depending on a high characteristics needed to be load-compatible. The superca-
charging current (power) available from the main source. For pacitors will be used to improve the load-following character-
example, a SAFT supercapacitor module (583 F, 15 V, 400 A) istics of a power source by providing a more robust power
may be charged from zero voltage (zero-of-charge) to the response to changes in system loading.
maximum voltage (maximum-of-charge) within 22 s at a Figure 7 illustrates the functional block diagram of the
constant current of 400 A. hybrid control structure showing all of the major compo-
nents, which include fuel cell source, fuel cell
converter, supercapacitor storage device, superca-
Tek Run Hi Res 1 Acqs 15 Sep 05 12:09:12 pacitor converter, and energy management con-
troller. The energy management controller
50 A operates to handle the energy balance (voltage
conversion) between main source, auxiliary
Ch2: iSuperCREF source, and load (as presented in Figure 8) by
controlling both the main and auxiliary convert-
ers. The load is capable of providing regenerative
braking that can be used to charge the auxiliary
source. The system operates primarily on main
Ch4: iSuperC source power and draws over- or under-energy
from the supplementary source only for peak or
transient energy requirements, such as vehicle
4 0A acceleration (a high load step). This structure is
called a series hybrid architecture that is com-
monly employed for all-electric hybrids.
As the fuel cell is not current-reversible, a
Ch2↓ 10.0 A M 100 μs 500 kS/s 2.0 μs/pt step-up (boost) converter is selected to adapt the
Ch4 10.0 A Ω A Ch2 \ −9.6 Y
low dc voltage delivered by the fuel cell to the 42
figure 10. Supercapacitor current response to a step 0 A to 50 A. V dc bus (new standard voltage for automotive
system, “PowerNet”). Supercapacitors are con-
How Fast Can Supercapacitor Energy nected to the 42-V dc bus by means of a two-quadrant dc/dc
and Battery Energy Be Used? converter (bidirectional converter). Supercapacitor current,
Supercapacitors fit between traditional capacitors and batteries which flows across the storage device, can be positive or
in terms of time constant, specific energy, and specific power. negative, allowing energy to be transferred in both directions.
Even though it is true that a battery has the largest specific To manage energy flows, one may then define three oper-
energy (meaning more energy is stored per weight than other ating modes (or states):
technologies), it is important to consider the availability of 1) charge mode, in which the main source supplies energy
that energy. For example, how fast can it be used? This is the to the auxiliary source and/or to the load (refer to Fig-
traditional advantage of capacitors. With a time constant of ure 8: t2 − t4 ),
less than 0.1 s, energy can be taken from a capacitor at a very 2) discharge mode, in which both the main source and the
high rate. On the contrary, the same size battery will not be auxiliary source supply energy to the load (refer to Fig-
able to supply the necessary energy in the same time period. ure 8: t1 − t2 ),
3) recovery mode, in which the load supplies energy to the
Concept of Fuel Cell/Supercapacitor storage device (regenerative braking) (refer to Figure 8:
Hybrid Power Source t4 − t5 ).
It is expected that the very fast power response and high spe- The main objective of the control is to regulate the dc bus
cific power of supercapacitors can be used to complement the voltage (energy balance in the dc bus). Taking into account
slower power output of the main source (particularly the fuel the fuel cell dynamics, the fuel cell is only operating in nearly

74 IEEE power & energy magazine january/february 2008


The fuel cell stack must be surrounded by a fuel system,
fuel delivery system, air system, stack cooling system,
and humidification system.

steady-state conditions, and the supercapacitors are function- The energy management controller has been implemented
ing during transient energy delivery or transient energy in the real-time card dSPACE DS1104, through the mathe-
recovery with the following constraints: matical environment of Matlab-Simulink, with a sampling
✔ fuel cell current slope must be limited to a maximum frequency of 25 kHz. Moreover, ControlDesk software
absolute value (for example: 4 A/s) in order to guarantee enables change to the parameters of the control. It is also
matching the reactant delivery rate and the usage rate used for driving the load.
✔ fuel cell current must be
kept within an interval (rated
value, minimum value, or
zero) 42 DC Bus
✔ supercapacitive storage 35
Voltage [V]

device voltage must be kept 28


within an interval (minimum 21 Fuel Cell
value, maximum value). 14
Normally, the system 7
attempts to reach the nomi- 0
1.0
nal (operating) voltage, Load (Motor)
Fuel Cell
called full-of-charge.
Power [kW]

0.5

0.0
Performance of Hybrid
Power Sources −0.5
SuperC
To examine the hybrid source per-
−1.0
formance, a small-scale test bench
1,000
Motor Speed [rpm]

is realized in the laboratory, as


800
depicted in Figure 9. The PEM
600
fuel cell system (Figure 2 and Fig-
ure 3) was achieved by the ZSW 400
Company. It is supplied using pure 200
hydrogen from bottles under pres- 0
sure and with clean and dry air
40 Fuel Cell
from a compressor. The superca-
Current [A]

pacitive storage device (292 F) is 20


obtained by means of twelve SAFT 0
supercapacitors (capacitance:
3,500 F, rated voltage: 2.5 V, rated −20
SuperC
current: 400 A, low frequency −40
equivalent series resistance: 0.8 25.5
SuperC Voltage [V]

m) connected in series. Maxi- 25.0


mum supercapacitor voltage is then 24.5
30 V. One chooses the supercapaci-
24.0
tor operating voltage of 25 V and
23.5
minimum voltage of 15 V (classi-
cally 50% of the maximu voltage). 23.0
0 10 20 30 40 50 60 70 80 90 100
Accordingly, the energy storage
Time [s]
without losses (discharging from
the operating voltage to the mini-
mum voltage) is equal to 58.4 kJ. figure 11. Fuel cell/supercapacitor hybrid source response during drive cycle.

january/february 2008 IEEE power & energy magazine 75


It is widely accepted that one of the main weak points
of the fuel cell is its time constants dominated by temperature
and fuel delivery system.

High Dynamics Finally, fuel cells are good energy sources to provide
of Supercapacitor Source reliable power at steady state and supercapacitor energy
Figure 10 presents the transient response during supercapaci- storage devices can advance the load following character-
tor discharging. The supercapacitor converter interfaces istics of a fuel cell by providing a stronger power
between the 42 V dc bus and the supercapacitor bank. The response to changes in system loading. During motor
initial voltage of supercapacitor bank is 30 V. The superca- starts/stops or other considerable steps in load, the super-
pacitor current reference iSuperCREF is Ch2 and the measured capacitors provide the balance of energy needed during
current iSuperC is Ch4. One can observe the high dynamic the temporary load transition periods and also absorb
response of the supercapacitor source from 0 to 50 A in 0.4 excess energy from the generator source (motor braking).
ms. Unquestionably, the fast response of supercapacitor Adding supercapacitor energy storage to distributed
power source can function with the fuel cell main generator power systems improves power quality and efficiency and
to improve the slow dynamics of the whole system. reduces capital expenses by allowing the systems to be
sized more closely to the steady-state power requirements
Drive Cycle rather than over-sizing the main generator to meet tran-
The experimental tests shown below were carried out by con- sient loading requirements.
necting the 42-V dc bus to an active load composed of a two-
quadrant converter, loaded by a dc motor coupled with a dc For Further Reading
generator. The motor functions with the cascade current- W. Friede, S. Raël, and B. Davat, “Mathematical model and
speed control method. A hysteresis and P controller are characterization of the transient behavior of a PEM fuel cell,”
selected for the motor current and speed loops, respectively, IEEE Trans. Power Electron., vol. 19, no. 5, pp. 1234–1241,
with a current limitation at ±60 A. Sept. 2004.
The test, as illustrated in Figure 11, presents waveforms P. Thounthong, S. Raël, and B. Davat, “Test of a PEM fuel
obtained at motor start to 1,000 rpm and stop. The motor cell with low voltage static converter,” J. Power Sources, vol.
starts at t = 10 s, to the final speed of 1,000 rpm, and the 153, pp. 145–150, Jan. 2006.
fuel cell power increases to its rated value of around 500 W. R.M. Nelms, D.R. Cahela, and B.J. Tatarchuk, “Modeling
The peak load power is about 1 kW, which is two times the double-layer capacitor behavior using ladder circuits,” IEEE
fuel cell rated power. Thus, the storage device, which sup- Trans. Aerosp. Electron. Syst., vol. 39, pp. 430–438, Apr.
plies most of the power required during motor acceleration, 2003.
remains in discharge state after motor start. In fact, the final P. Thounthong, S. Raël, and B. Davat, “Utilizing fuel cell
supercapacitor current is 8 A because the steady-state load and supercapacitors for automotive hybrid electrical system,”
power (about 600 W) is greater than the fuel cell rated in Proc. 2005 IEEE Applied Power Electronics Conf.
power (500 W). (APEC), Mar. 2005, pp. 90–96.
After that, the motor speed, at t = 40 s, decelerates P. Thounthong, S. Raël, and B. Davat, “Control strategy
from 1,000 rpm to stop; consequently, the storage device of fuel cell/supercapacitors hybrid power sources for electric
is deeply discharged, demonstrating three phases: first, vehicle,” J. Power Sources, vol. 158, pp. 806–814, July
during t = 40 s to 44 s, it recovers the power supplied to 2006.
the dc bus by the fuel cell and the motor regenerative J.W. Dixon and M. Ortúzar, “Ultracapacitors + DC-DC
breaking (the peak power of about 0.5 kW); then it recov- converters in regenerative braking system,” IEEE Aerosp.
ers the power supplied only by the fuel cell. From t = 44 s Electron. Syst. Mag., vol. 17, pp. 16–21, Aug. 2002.
to 48 s, this power is constant, limited to the fuel cell rated
power. After t = 48 s, the fuel cell power decreases due to Biographies
the supercapacitor voltage regulation. During the two first Phatiphat Thounthong is a lecturer at King Mongkut’s
phases, the fuel cell power is at a rated value of 500 W. In Institute of Technology North Bangkok, Thailand.
the third phase, it decreases down to zero. Excellently, Bernard Davat is a professor at the Institut National Poly-
only small perturbations on the dc bus voltage waveform technique de Lorraine, Nancy, France.
can be seen, which is of major importance for the energy Stéphane Raël is an assistant professor at the Institut
management controller. National Polytechnique de Lorraine. p&e

76 IEEE power & energy magazine january/february 2008

You might also like