2nd WSEAS/IASME International Conference on RENEWABLE ENERGY SOURCES (RES'08)

Corfu, Greece, October 26-28, 2008

Modeling and Simulation of a PEM Fuel Cell System

Under Various Temperature Conditions
A. A. SALAM, A. MOHAMED, M A HANNAN
Department of Electrical, Electronic and Systems Engineering
Universiti Kebangsaan Malaysia
Bangi, Selangor 43600 MALAYSIA

Abstract: - This paper describes the modeling, simulation and analysis of the dynamic model of a 5kW PEM
fuel cell system. The model includes hydrogen fuel reformer and the PEM fuel cell stack. The model is used to
investigate the effect of temperature on the fuel cell voltage and power flow at various temperatures.
Simulation results show that the temperature has a significant impact on fuel cell output voltage and output
power.

Key-Words: - Dynamic Model, PEM fuel cell, Hydrogen fuel reformer, Fuel cell Stack, Temperature, Power.

1 Introduction PEM fuel cell system. The simulation model of the

The application of fuel cells and other fuel cell was developed based on the mathematical
renewable energy sources like photovoltaic and relationship for PEM fuel cell and subsystems using
wind power for both stationary and mobile power the SimPowerSystems toolbox of Matlab-Simulink.
applications is beneficial for the sustainable energy
conversion. In residential and vehicular application, Table 1 Type of electrolyte, operating temperature
a new kind of commercially viable, efficient and and fuel for different fuel cells
reliable fuel cell is to be fully developed. Proton Operating
Type Electrolyte Fuel
Exchange Membrane (PEM) fuel cell has the Temp. 0C
advantage of low operating temperature [1]. Alkaline
KOH H2 50-200
fuel cell
Several models of PEM fuel cells have been
Phosphoric
developed and simulated [2-3]. Models that have acid fuel
Phosphoric
H2 ~220
been reported in the literature for the PEM fuel cell acid
cell
range from stationary to dynamic models [3-7]. In Proton
this paper, the proposed PEM fuel cell model is exchange Solid
based on the model given in [2], which is modified pure H2 50-100
membrane polymer
to account for temperature variations. The PEM fuel cell
fuel cell system must be accurately modeled in Molten
Lithium H2, CO,
order to be able to apply a suitable nonlinear and CH4 other
carbonate ~650
control scheme. potassium hydro-
fuel cell
Fuel cells are widely regarded as one of the carbonate carbons
most promising energy sources [3 due to their high H2, CO,
Solid oxide Solid oxide CH4 other
energy efficiency, extremely low emission of fuel cell electrolyte hydro-
500-1000
oxides of nitrogen and sulfur, very low noise, and carbons
clean energy production. Fuel cells are classified
based on their operating temperature. Details of
fuel cells classification based on the type of
electrolyte, operating temperature and fuel are 2 PEM fuel cell
shown in Table 1. The PEM fuel cell technology is The PEM fuel cell consists of a solid polymer
the best amongst other fuel cell technologies electrolyte sandwiched between two electrodes
because of its low operating temperature, high- (anode and cathode). In the electrolyte, only ions can
power density, and relatively short startup time [6]. exit and electrons are not allowed to pass through.
This paper discusses the modeling, simulation So, the flow of electrons needs a path like an
and investigates the effect of temperature on the external circuit from the anode to the cathode to
produce electricity because of a potential difference

ISSN: 1790-5095 204 ISBN: 978-960-474-015-4

 2nd WSEAS/IASME International Conference on RENEWABLE ENERGY SOURCES (RES'08)
Corfu, Greece, October 26-28, 2008

between the anode and cathode. The overall According to the basic electrochemical
electrochemical reactions for PEM fuel cell fed with relationships, the relationship between the stack
a hydrogen-containing anode gas and an oxygen- current and the molar flow of reacted hydrogen can
containing cathode gas are as follows: be expressed as,
N0 I
At anode raection, qr H 2 = = 2Kr I (8)
+
2F
2 H 2 = 4 H + 4e (1) Substituting equation (8) into (7), the hydrogen
At cathode reaction, partial pressure can be written as,
O2 + 4 H + + 4e = 2 H 2 O (2)
Overall reaction d RT in
PH 2 = (q H 2 q out H 2 2 K r I ) (9)
2 H 2 + O2 = 2 H 2 O + electricity + heat (3) dt Van

The output stack voltage V is defined as a

function of the stack current, reactant partial
pressures, fuel cell temperature, and membrane
humidity. The potential difference between the
anode and cathode is calculated using the Nernsts
equation and Ohms law and can be written as
follows:
RT PH 2 PO2
0.5
3 Modeling Descriptions
The model used in this paper to simulate a
5kW PEM fuel cell is given in Fig. 1. Using the
equations derived for the PEM fuel cell model, a
simulation model is developed using the
SimPowerSystems toolbox of Matlab-Simulink.
The overall model of the PEM fuel cell consists of

V = N 0 E0 + ln B ln CI Rint I (4) 5 subsystems such as partial pressures, ideal

2 F PH 2 O voltages, activation losses, ohmic losses and

reformer model. Figs. 2-6 show the various
The relationship between molar gas flow subsystems of the dynamic simulation model of the
through the valve is proportional to its partial PEM fuel cell. The parameters of the simulation
pressure and can be expressed as, model are given as in Table 2.
qH K an
= 2
= KH (5)
2
PH 2
MH 2

qH 2 O K an
= = K H 2O (6)
PH 2 O M H 2O
where,
qH2 : Molar flow of hydrogen
qH2O : Molar flow of water
PH2 : Partial pressure of hydrogen
PH2O : Partial pressure of water
PO2 : Partial pressure of oxygen
kH2 : Hydrogen valve molar constant
kH2O : Water valve molar constant
kan : Anode valve molar constant
MH2 : Molar mass of hydrogen
MH2O : Molar mass of water

For the hydrogen molar flows (kmol/s), there

are three major factors such as the output flow, the
input flow and the flow that takes part in the
reaction. These flows of hydrogen molar are
represented as,

d
PH =
RT in out r
(q q q ) (7)
dt 2 Van H 2 H2 H2

Fig.1 Overall model of PEM fuel cell

ISSN: 1790-5095 205 ISBN: 978-960-474-015-4

 2nd WSEAS/IASME International Conference on RENEWABLE ENERGY SOURCES (RES'08)
Corfu, Greece, October 26-28, 2008

Table 2 Parameters used in PEM fuel cell simulation Oxygen partial pressure:
Representation Value 1
Faradays constant (F) C/mol 96484600 K O2
PO2 = (q in O2 2 K r I ) (11)
Universal gas constant (R) J/kmol K -2
8314.47 1 + O2 s
No load voltage, E0 (V) 1.229 V
Number of cells (N) 42 Water partial pressure:
-1 -7 1
Constant (K=N/4F) ,kmol/(s A) 1.0883 x 10
K H 2O
Valve molar constant for hydrogen (KH2), 4.22 x 10-5 PH 2 O = ( 2 K r I ) (12)
-1
1 + H 2O s
kmol/(s atm)
Valve molar constant for 2.11 x 10-5
oxygen (KO2), kmol/(s-1 atm)
Valve molar constant for 7.716 x 10-6
water (KH2O), kmol/(s atm)
Response time for hydrogen flow (TH2) 3.37
sec
Response time for water flow (TH2O) , sec 18.418
Response time for oxygen flow (TO2) , sec 6.74
Utilization factor, U 0.8
Reformer time constant, 1 2 Fig.2 Hydrogen, Water and Oxygen pressures model
Reformer time constant, 2 2
3.2 Nernst Voltages
Conversion Factor, CV 2 The fuel cell voltage can be calculated using
Activation voltage constant, B (A-1) 0.04777 the Nernst voltages (E) and considering the
Activation voltage constant, C (V) 0.0136 activation (Nact) and ohmic losses (Nohmic) in the
system. The fuel cell voltage is given by,
Stack internal resistance, Rint 0.00303
PI gain constants K5 and K6 10
Vcell = E + N act + N ohmic (13)
Ratio of hydrogen to oxygen (rHO) 1.168
Methane reference signal, Qmethref 0.000015 The variation in temperature affects the ideal
voltage at different temperatures. Considering the
variation of temperature, the Nernst voltage from
references [1,3] can be expressed as,
3.1 Partial Pressures
Using equation (5) to replace the output flow
PH PO
and performing Laplace transform of both sides of E=
1
[ ]
( g + s T + Tref + RTop ln( 2

2 )V

(14)
equation (9) the hydrogen partial pressure can be 2F

PH 2O

expressed as: where R is the gases universal constant (8.314
1 J/K.mol), F is the Faraday constant (96485 C), g
KH2
(10) is the variation of Gibbs free energy, s is the
PH 2 = (q in H 2 2 K r I )
Van variation entropy (-163.2 J/K) in the reaction, Tref
1+ s (298K) and Top are the reference and operational
K H 2 RT
temperatures in K, respectively. Applying the
Similar analysis were carried out in reference constants to the open circuit voltage (Nernst
[4], to derive the partial pressure of oxygen ( PO2 ) maximum voltage) and including the number of
series cell in the stack, the equation can be
and water ( PH 2O ) of equations (11) and (12), simplified as shown in references [3,7] as,
respectively. Considering the equations for the
pressures, the model that represents the water, PH PO (15)
hydrogen and oxygen pressure is given in Fig 2.
[ ]
E = N0 (1.229 0.85103 Top 298 4.31103Top ln( 2 2 )V
PH2O

ISSN: 1790-5095 206 ISBN: 978-960-474-015-4

 2nd WSEAS/IASME International Conference on RENEWABLE ENERGY SOURCES (RES'08)
Corfu, Greece, October 26-28, 2008

where, Top is the operating temperature of the
stack. The model that is represented by equation (15)
is shown in Fig 3. The number of cells connected in
series with 42 cells which formed the total stack
voltage for the fuel cell output voltage is simulated.
losses voltage with temperature variation and the
internal resistance can be expressed as,
Rint = 0.01605 3.5 10 5 T + 8 10 5 I (17)
where, I is the fuel cell current. Therefore, the
ohmic losses are given by,

N ohmic = Rint I (18)

The model that represents the ohmic losses is shown

in Fig 5.

Fig.3 Model to represent fuel cell voltage

3.3 Activation Losses

The activation losses of the fuel cell relates
Fig.5 Ohmic losses model
to the slowness of the reactions that take place on the
surface of the electrodes. A portion of the voltage
generated is lost in driving the chemical reaction at 3.5 Reformer Model
the electrodes as mentioned previously. In this
paper, the activation over voltage is considered as A reformer model that produces hydrogen
constant and is expressed in references [1, 3 and 7] through a reforming process is introduced in
as, references [1, 3, 5]. This is a second order reformer
function which can be expressed as,
N act = B ln (C I ) (16)
qH 2 CV
= (19)
where, I is the stack current, B = 0.04777 V q menthanol 1 2 s + ( 1 + 2 ) s + 1
2

and C = 0.0136 A-1. The model used to represent the

activation loss is shown in the Fig. 4. Here, U
represents the fuel cell current, B and C are
constants, respectively.
Fig.4 Activation losses model
3.4 Ohmic Losses
The ohmic losses occur due to the polymeric
membrane resistance to ionic flux, the resistance due
to electrodes, resistance of separating plate to flux of
electrons and gas spreaders, respectively. An ohmic
where, CV is a conversion factor (kmol of
hydrogen kmol-1 of methanol), qmethanol is the
methanol flow rate (kmols-1) and 1,2 are time
constants. Stack current feed back [1, 3, 4] and the
relationship between the increase in stack current
and the required hydrogen can be expressed as,
N0I
q req H 2 = (20)
2 FU
where, U is the utilization rate (coefficient) and
req
q H2 is the required amount of hydrogen flow to
meet load demand (kmols1). To control the
methanol flow rate, the amount of hydrogen required
to meet the load demand is,
k N I
q methanol = k 3 + 3 0 q in H 2 (21)
3 s 2 FU

ISSN: 1790-5095 207 ISBN: 978-960-474-015-4

 2nd WSEAS/IASME International Conference on RENEWABLE ENERGY SOURCES (RES'08)
Corfu, Greece, October 26-28, 2008

where, 3 is the time constant and k3 is the gain rated power at about 1.5s. The model is sub
of the Proportional Integral (PI) controller which is sequentially simulated from the temperature of
used to control the flow rate of methanol in the 25C-80C (298K-353K) and the results of the
reformer. The oxygen flow is controlled using the simulation are shown in Table 3.
ratio rho. The reformer model is shown in Fig 6.
PEM Fuel Cell Power Ouput
4898

4896

4894

4892

4890

Power(W)
4888

4886

4884

4882

4880

Fig.6 Reformer model 4878

0 1 2 3 4 5 6 7 8 9 10
Time(Seconds)

Fig.8 Fuel Cell Stack Output Power

4 Results
In the simulation, it was assumed that the From Table 3, it can be seen that the Nernst
hydrogen, oxygen and water pressures are considered voltage decreases with the increase in temperature
as ideal, hence no variation of the pressures are to be and this fact is illustrated graphically as in Fig 9.
expected. The stack current is considered as 95A The decrease in ideal voltage therefore reduces the
based on the power rating of 5kW and voltage of 50V fuel cell voltage. The stack voltage and fuel cell
DC. The PEM fuel cell output voltage and output power decreases as the temperature increases as
power with respect to the simulation run time of 10 shown in Figs. 10 and 11, respectively.
seconds are plotted and shown in Fig. 7 and Fig. 8,
respectively at a temperature of 25C (298K) and Table 3 Effect of varying temperature
constant current of 95A.
Temp Voltages (Volts) Power
PEM Fuel Cell Voltage
K Eo Ohmic Vfc x 42 P
51.6 298 1.229 0.01321 51.63 4904.85
303 1.22475 0.01304 51.46 4888.7
51.55 308 1.222 0.01286 51.28 4871.6
313 1.21625 0.01269 51.1 4854.5
51.5
318 1.212 0.01251 50.92 4837.4
Voltage(Volts)

323 1.20775 0.01234 50.75 4821.25

328 1.2035 0.01216 50.57 4804.15
333 1.19925 0.01199 50.39 4787.05
51.45

338 1.195 0.01181 50.21 4769.95

51.4 343 1.19075 0.01164 50.03 4752.85
348 1.1865 0.01146 49.86 4736.7
51.35 353 1.18225 0.01129 49.68 4719.6
0 1 2 3 4 5 6 7 8 9 10
Time(Seconds)

The ohmic losses decrease with increasing

Fig.7 Fuel Cell Stack Voltage output temperature as shown in Fig 12. This is in
accordance with the theory in which as the
The simulation results show that the fuel cell temperature increases, the losses due to internal
output voltage increases from about 51.25 to a resistance will reduce. It is noted in Fig. 12 that the
steady state value of about 51.5 V. From Fig. 8, the ohmic losses can be reduced by operating the PEM
output power shows the fuel cell start to produce the

ISSN: 1790-5095 208 ISBN: 978-960-474-015-4

 2nd WSEAS/IASME International Conference on RENEWABLE ENERGY SOURCES (RES'08)
Corfu, Greece, October 26-28, 2008

fuel cell at a higher temperature than the standard voltage losses. However, other factors like overall
temperature of 298K. efficiency of the system would be reduced.
1.24 4 Conclusions
Dynamic behavior of a 5kW PEM fuel cell system is
1.22
studied and evaluated to investigate the effect of
Nernst Voltage

1.2 temperature on the fuel cell output voltage and

output power at various temperatures. Simulation
1.18 results show that the fuel cell stack voltage and its
output power decrease with increase in temperature.
1.16
The developed dynamic model of a 5kW PEM fuel
cell is useful for future studies such as grid
280 300 320 340 360
Stack Temperature connected fuel cell system.
Fig.9 Nernst Voltage Variation with Temperature
References
[1] M.Y. El-Sharkh, A. Rahman, M.S. Alam, P.C.
52
Byrne, A.A. Sakla, T.Thomas, A dynamic model
51.5 for a stand-alone PEM fuel cell power plant for
Stack Voltage

residential application, Journal of Power Source,

51
138, 2004, pp.199-204.
50.5
[2] M.Uzunoglu and M.S.Alam,Dynamic Modelling,
Design and Simulation of a Combined PEM
50 Fuel Cell and Ultra capacitor System for Stand-
Alone Residential Application, IEEE
49.5
280 300 320 340 360 Transactions on energy conversion, Vol. 21, No.
Temperature(K)
3, 2006.
[3] J. Padulles, G. W. Ault, and J. R. McDonald,An
Fig.10 Stack Voltage Variation with Temperature Approach to the Dynamic Modeling of Fuel Cell
Characteristics for Distributed Generation
4950
Operation, IEEE- PES Winter Meeting, Vol. 1,
4900 2000, pp.134-138.
[4] S. Pasricha, and S. R. Shaw, A Dynamic PEM
Fuel Cell Power

4850
Fuel Cell Model, IEEE Trans. Energy
4800 Conversion, Vol. 21, Issue 2, 2006, pp.484-490.
[5] M.Y.El-Sharkh,N.S.Sisworahardjo, Uzunoglu,
4750
O.Onar, M.S. Alam, Dynamic behaviour of
4700 PEM fuel cell and Microturbine power plants,
280 300 320 340 360
Temperature(K) Journal of Power Source,164 , 2007, pp.315-
Fig.11 Fuel Cell Power Variation with Temperature 321.
[6] J. Purkrushpan, A. G. Stefanopoulou, and H.
0.014 Peng, Modeling and control for PEM fuel cell
0.0135 stack system, Proc. Amer. Control Conf.,
Anchorage, AK, 2002, pp.31173122
Ohmic Losses

0.013
[7] C.Wang, M. H. Nehrir, and S. R. Shaw, Dynamic
model and model validation for PEM fuel cells
0.0125

0.012
using electrical circuits, IEEE Trans. Energy
0.0115 Convers., Vol. 20, No. 2, 2005, pp.442451.
0.011
280 300 320 340 360
Stack Temperature

Fig.12 Ohmic losses due to temperature variation

The simulation results show an improvement in

the fuel cell ohmic losses as the stack temperature
increases from 25C to 80C (300K-355K). It seems
that the higher the operating temperature, the lower

ISSN: 1790-5095 209 ISBN: 978-960-474-015-4