Temperature control system for PEM Fuel cell
Fuel cell research technology Group Babol University of Technology

S.M.Rakhtala, T.Fadaiian
R.Ghaderi, A.Ranjbar and M.Shakeri mh_rakhtala@stu.nit.ac.ir Abstract
In This paper, design of the temperature control system of PEM fuel cell is developed with integrated control techniques by Atmega 128 micro-controller along with optimized performance of fuel cell system. The temperature control system will guarantee the correct performance of fuel cell around its optimal operation point. In order to reduce high frequency noises, an averaging filter has been applied in the micro controller software. Different parameters such as temperature and electrical loading are simulated by MATLAB/SIMULINK software and also implemented with experimental unit on single cell. The simulation results of dynamic model as well as the implementation results indicate that the proposed control scheme works very well.
1. Introduction
A large number of researches have been conducted to manage the control system for different types of fuel cell. Alfredo et al. [1] developed a control system for portable PEM fuel cell. The control system will guarantee the correct performance of the stack around is optimal operation point this means that the air flow and the stack temperature are controlled to correct value. Shaw et al. [2] presented a system that provides continues PEMFC electrical terminal monitoring under realistic load profiles. The data collected in this will ultimately establish electrical terminal diagnostics for fuel cell for reliable fuel cell operation.
In this work, a single cell of PEM fuel cell is designed and fabricated using a MEA with 100 cm2 active area (10 cm × 10 cm). According to the modeling and simulation of the fuel cell system conducted in MATLAB/SIMULINK under different temperatures it can be concluded that the fuel cell performance is directly influenced by the temperature. A temperature control system has also been designed and fabricated in order to control temperature during the operation of the fuel cell. So a PID controller is applied in hardware unit of the temperature control system to stabilize the operating temperature of fuel cell. Also using an averaging filter, the high frequency noise in the temperature measurement is reduced.
2. Hardware of the fuel cell control system
Hardware of the control system consists of the following blocks: 1) Central processing unit, 2)
Analog inputs, 3) Digital inputs, 4) Analog outputs, 5) Digital outputs, 6) RS 232 serial port.
The hardware is capable of controlling 32 analog inputs which consist of temperature and pressure inputs. It is also used to process 16 digital inputs (i.e. start, stop and Emergency stop…) and to control 16 digital outputs (i.e. switch valves, alarm signals…) and 6 analog PWM outputs.
Overall block diagram of fuel cell control system is shown in Fig.1.
2.1. Central processing unit and data controller
Control system has been composed of a main board which relies on an ATmega128 microcontroller. This microcontroller has the following practical characteristics [3]:
• Functional frequency of 16MHZ.
• Eight analog inputs.
• Six PWM outputs.
• Two serial ports.
•

1

2.2. Temperature analog input unit
Temperature processing unit consists of anode input temperature, cathode input temperature, stack body temperature and anode output temperature (condenser).A K-type thermocouple is used to control the temperature at fuel side and provide the temperature feedback. This unit has the capability to sense 16 temperatures instantaneously and precisely. To process and control these datas, they are transferred to the analog input unit of main board.

Fig. 1. Overall block diagram of fuel cell control system.

A tape heater adjusts the temperature of air and injected fuel to the stack. This is applied on the fuel way. In order to control the heater’s temperature level, a PWM pulse provided by PID microcontroller is applied to the heater control board. This board consists of power system including TRIAC and command board including TRIAC DRIVER and is used to control the heater’s current and voltage.
In fuel cell system, it is vital to control pump speed and heater’s temperature.PWM outputs are applied to control pump speed in order to adjust the flow rate. Scheme of the heater control board is presented in Fig.2.
HEATER CONTROLLER
AIR SUPPLY

HEATER CONTROLLER H2

Fig. 2. Scheme of designed and constructed temperature board.

The microcontroller is programmed in such a way that is able to regulate all activities and duties in the system and will show proper response in critical situations. In the next stage, according to

2

predetermined command, an industrial computer is placed along with the main system and is derived by the corresponding software.
The following tasks will be done by the software in PC:
1) Controlling of the constant and initial values.
2) Calibration coefficients.
3) Reading sent information.
4) Drawing the desired graphs.
The main analog input and Output parameters in temperature control process are given in the following tables.
Table1.Measured analog parameters and their usage.
Serial
Number

Measured

1

Description

Parameters

Temperature

Analog input
1c ~ 1mv

Temperature
2

2 sample cell

Air input

Type K

×

1c ~ 1mv

Type K

×

Type K

×

Type K

×

Analog input

Hydrogen input

1c ~ 1mv

Temperature
4

Processed

Analog input

Temperature
3

Type

Analog input

Hydrogen output

1c~ 1mv

Table2.Output analog parameters.
Serial
Number

control
Parameter

Tape Heater for
5

Description

0~220V AC

Hydrogen
Tape Heater for

6

Type
Analog output
(PWM output)

0~220 V AC

air supply

Processed

×

Analog output
(PWM output)

×

3. Temperature control method
Every fuel cell stack presents an optimal working temperature at which the reaction occurs with an optimum efficiency. In addition, the reaction itself generates heat changing the stack temperature and thus, forcing the control system to bring it back to the optimum temperature point. Scheme of the heating system is presented in Fig .3.
In this paper in order to control the temperature of fuel cell a conventional PID controller with output temperature feedback has been applied. Temperature data has been sampled as n sample per second and then saved in microcontroller RAM. An averaging filter has been applied in the microcontroller software as Eq. (1):
N −1

Tn ( K ) n n =0

T (K) = ∑

(1)

This method can reduce the measuring noise of temperature. The filtered temperature (Eq.1) is analogous to the actual one. Actual temperature is compared with set point temperature and the error signal is generated. An additional serial communication line enables the system to be monitored using a PC platform. Set point temperature is transferred from PC to microcontroller and then according to PID controller, this process is controlled. Controller outputs are two PWM signals which are generated by a carrier wave with 5 volts amplitude and 20 KHz frequency.
These PWM signals are applied to the heater driver of anode and cathode sides to trigger the

3

TRIAC switches. Fig.4 shows schematic of the temperature process. The control law will calculate the duty-cycle of PWM signal From Eq. (2):
Dpwm = (k p +

kI
+ kD S)et
S

(2)

Where
D pwm : The duty cycle of PWM pulse kp : Proportional gain kD :

Derivative gain kI : Integral gain et : Error signal
TEMPERATURE CONTROL
FEED BACK SIGNAL’S

ATMEGA 128
ANALOG INPUTS
DIGITAL OUT

TEMPERATURE
DRIVER

PWM PULSE

AC
DRIVER
HEATER
TT2
STACK of
PEM
SV Air

SV H2 H2 Heater
TT1

PT

TT4

Air
Heater

TT3

Fig .3. Scheme of the heating system of the PEM fuel cell

Fig .4. Schematic diagram of the temperature process

4. Dynamic Modeling
Steady state operation of fuel cell stack and voltage variations versus load current under determined operating conditions are illustrated in Fig .5.Fuel cell output voltage is a function of temperature, load current, anode pressure, cathode pressure, cathode flow rate and anode flow rate. Fuel cell polarization curve consists of three regions, as shown in Fig.5.
The output voltage of a single cell is expressed as:

VFc = VCell

reversible

− Vact − Vohmic − Vconc

(3)

VCell reversible: Thermodynamic voltage
V act : Activation over voltage
Vohmic : Ohmic over voltage
Conc. : Concentration over voltage
VFc: Fuel cell output voltage

4

4.1. Key parameters in fuel cell dynamic modeling
Key parameters in PEM fuel cell modeling and analysis, to realize the optimum operating conditions are as follow:
1. Fuel cell operating temperature
2. Anode pressure
3. Cathode pressure
4. Load current or fuel cell output current ( I Fc )
Overall block diagram of fuel cell simulation in Matlab/Simulink is presented in Fig .6.

Fig. 5. Fuel cell polarization curve

4.2. PEM fuel cell dynamic equations
The maximum voltage each cell can provide under standard temperature and pressure
( PFuel = Po 2 = 1atm ,25 o c ) is known as cell reversible voltage. Voltage difference between cell temperature and standard temperature ( 25 o c ) is given by Eq(4)[4,5]:
VCell
VCell

reversible

reversible

=

= V Nernst + VTdifference

− ∆G f
2F

−

1

∆ S ( 298 .15 k )
RT  P ∗ ∗ ( P ∗ O 2 ) 2
(T − Tref ) + ln  Fuel *
2F
2F
( P H 2O ) 2



(4)






Where
T: Operating temperature in Kelvin
Tr ef : Reference temperature in Kelvin
F: Faraday constant
∆ S f : Molar anthropy variation under standard condition

Fig. 6. Over all block diagram of fuel cell simulation in Matlab/Simulink

5

(5)

4.2.1. Ohmic over voltage Modeling
Ohmic over voltage is described by the following equation [4, 5]:
Rohm = Rohm 0 + Rohm1 + Rohm 2

(6)

Where
Rohm0 =Constant resistance
Rohm1 =Variable resistance, Dependant on temperature
Rohm1 =Variable resistance, Dependant on current
Ohmic over voltage can be expressed as a function of temperature and current. Replacing numeric values for above parameters, Eq (6) can be expressed as follows [6,5]:

R int ernal = ξ 1 + ξ 2T + ξ 3 I Fc

(7)

ξ1 = Rohm0 = 0.01605 ; Constant resistance coefficient ξ 2 = K RT = −3.5 × 10 −5 ; Constant temperature coefficient

ξ 3 = K RI = 8 × 10 −6 ; Constant current coefficient

4.2.2. Activation over voltage modeling
Activation over voltage is due to the slowness of chemical reaction taking place on electrode surface [2, 3].
I 
Vact = Aln Fc 
I 
 o
A=

(8)
(9)

RT nα F

Combining the above equations, Tofel equation is calculated as:
Vact = ηo + (T − 298)a + T.b. ln(I Fc )
To simulate activation voltage drop in simulink the following equation is applied [6, 5]:
V act = ξ 1 + ξ 2 T + ξ 3 .T . ln( Co

*
2

) + ξ 4 .T . ln( I Fc )

(10)

(11)

Where ξi : Parametric coefficients in modeling of each cell.
Constant values used for

ξi are as follows:

ξ3 = 7.4 × 10−5
Applied equation in simulink: ξ 4 = −0.000187

ξ 2 = 0.00312

ξ1 = −0.9514

*
Vact = − 0 .9514 + 0 .00312 .T + 7 .4 × 10 −5. T . ln( Co 2 ) − 0 .000187 .T . ln( I Fc )

(12)

Where
T: Temperature in Kelvin.
*
: Oxygen concentration at gas/liquid interface.
Co 2 ( mol / cm 3 )
*

In the above block diagram Co2 is unknown and is considered as the input to the block. It can be said that water amount plays an important role in cathode. The thick layer of water produced in cathode is due to the combination of oxygen and the flowing hydrogen ion in electrolyte layer.
Effective concentration of oxygen on cathode catalyst is estimated by oxygen concentration in cathode gas/water interface and can be expressed by Eq (13) [6]:
*
Co 2 =

Po 2

*

 − 498 
5.08 × 10 6 exp 

T

(13)

6

Oxygen partial pressure in gas/water interface is a function of water concentration on cathode channel and is given by Eq (14):




1
Sat
*
P02 = PH 2 o 
− 1

 4.192 J Fc  Cathode 



 exp
1.334
.xH 2o
T




(14)

Fuel partial pressure is also described by Eq (15):




1
*
Sat
PFuel = 0.5 PH 2o 
− 1
  1.653 J Fc  anode 
.xH 2o
 exp

1.334



 T


(15)

Where
JF c ;

( I Fc A0 = A.cm −2 )

Cathode xH 2 o =

anode
H 2o

x

anode
H 2o

=
=

sat
PH 2 o

Pa ir sat PH 2o

PF uel

: Fuel cell current density

: Water molar coefficient on cathode

: Water molar coefficient on anode

sat
PH 2o

: Water molar coefficient on anode
PF uel
According to above mentioned equations, oxygen and fuel partial pressure can be summarized as
[6]:

x

*
PO2 =

*
PFuel =

Pair
Sat
− PH 2o
 4.192 J Fc 

 exp 1.334

T


(16)

0.5 PFuel
Sat
− PH 2o
 1.653 J Fc  exp  T 1.334 




(17)

In these equations, water saturation pressure is expressed as [6]:
Sat
PH 2 o = 10 ( −2.179 + 0.02953 .T − 9.1837 ×10

−5

(18)

.T 2 +1 .4454 ×10 −7.T 3 )

4.2.3. Concentration over voltage
This over voltage represents the mass transport losses resulting from the change in concentration of reactants as the fuel is used [7].
V=
Conc

I
RT P2
RT
ln(
) → V = ln( 1− Fc )
Conc
nF
P
nF iL 1

(19)

P1 : Pressure in zero current
P2 : Pressure in fuel cell current i L : Maximum electrical current density
5. Tests and simulation Results
In this section, simulation results are investigated and compared with experimental tests. The effect of different parameters such as current and temperature on polarization and P-I curves is presented. Polarization and P-I curves for a PEM fuel cell at 60 o c have been obtained via dynamic

7

equations simulation in Matlab/Simulink and are given in Fig.7. A step current (1A) has been applied to the system and the output voltage is sampled.
The values of testing parameters are p air = 1atm , p fuel = 1atm , A = 100cm2 , N0 = 1 ; Where p air and p fuel are input pressures of anode and cathode sides respectively, N0 is the number of cells, A is the
Active area. According to the simulation results it can be seen that, the maximum power occurs at
26A and 8.64W. For the values higher than 26A the more current increases, the power decreases dramatically. This region is known as concentration over voltage.
Pair =1atm; P fuel=1 ; N0=1 ; T=60 ; E0=1.21; iL=30 ; na=2 ; A=100;
0.8

10

c

T=60 °

9

0.7

8
0.6
7

V ( VOLT )

V

0.5

6

P ( WATT )

P

0.4

5

????

4

0.3

3
0.2
2
0.1

1

0

0
0

3

6

9

12

15

18

21

24

27

30

33

I ( AMP )

Fig. 7. Simulation results of polarization and P-I curves at 60c

5.1. The effect of temperature on fuel cell performance
The performance of the proposed controller system has been simulated at different temperatures using MATLAB.The polarization and P-I curves at different temperatures, obtained via simulation, is presented in Fig .8.Considering the polarization curves, as the temperature increases the polarization curves rise up. So the more temperature increases the cell voltage and current increase.
0.8

10
9

0.7

8
0.6
7
V(VOLT)
‫وﻟﺘﺎ ژ‬

0.5

6

0.4

5 P(WATT)
4

0.3

‫ﺗﻮان‬

V(T=30 C)

3

V(T=70 C)

0.2

V(T=50 C)

2

P(T=30 C)

0.1

P(T=50 C)

1

P(T=70 C)

0

0
0

3

6

9

12

15

18

21

24

27

30

33

I(AMP) ‫ﺟﺮﯾﺎن‬

Fig .8. Polarization and P-I curves obtained from simulation at different temperature

The above mentioned results are completely compatible with dynamic equations. To verify the results from the simulation, the system is also implemented by the actual tests with experimental setup and is presented in Fig.9.The results of computer simulation are compared with the actual test and are in good agreement with each other. Finally it can be concluded at higher temperature, chemical reactions take place with optimal efficiency.

8

6. Conclusion
This paper has presented the methodology that is to be followed in order to achieve the temperature control of a PEM fuel cell system. The control takes care of the system temperature values using a PID controller. In order to reduce high frequency noises in the measured temperature, an averaging filter has been applied. The control system has been developed by
Atmega128 micro- controller. A single cell has been tested under different DC load by Chroma model 63210 .Since the simulation results show that the proposed controllers work very well, we decided to implement the controllers using an experimental setup to show the effectiveness and applicability of the proposed controllers.

0.9

8

0.85

7.5

0.8

7

0.75

6.5

0.7

6
P40

5.5

P48

0.6

5

0.55

4.5

Power

Voltage,V

0.65

P55
P58
P64

0.5
4

P69

3.5

V40

0.45
0.4

V48
3

0.3

V55

2.5

0.35

V58
V64

2

0.25

1.5

0.2

V69
Poly. (P48)
Poly. (P40)

1

0.15

0.5

0.1

0

0.05
0

-0.5
0

4

8

12

16

20

24

current,A

Fig .9. Polarization and P-I curves obtained from experimental test at different temperature
7. REFERRENCES:
[1] AlfredoP.Vega-Leal, F. Rogelio Palomo, Felipe Barragan, Covadonga Garcia, J. Javier Brey ,"Design of control systems for portable PEM fuel cells" Journal of Power Sources 169(2007)194–197.
[2] steven R.shaw "Instrumentation for PEM Fuel Cell Transient Degradation Monitoring"
[3] M. Shakeri ,J. Imen, M. R. Rostami ,November 2008, "A Full Scale Microcontroller Based DMFC Test
Station"Journal of Fuel Cell Science and Technology, ASME,in press.
[4] Jeferson Marian Correa, Felix Alberto Farret , Luciane Neves Canaba,2001 , "An Analysis of the
Dynamic Performance of Proton Exchange Membrane Fuel Cells Using an Electrochemical Model" IEEEIndustrial Electronics conference-IECON'01,PP.141-146.
[5] J.M .Correa, F.A.Farret,V.B.Popov, J.B.Parizzi,June 2003,"Influence of the Modeling parameters on the
Simulation Accuracy of Proton Exchange Membrane Fuel Cells"IEEE Bologna PowerTech Conference,
Italy.
[6] D.Xue , Z.Dong,August1998, "Optimal Fuel Cell System Design Considering Functional Performance and Production Costs"J.Power Source 76(1998)69-80.
[7] James Larminie , Andrew Dicks,2000, "Fuel Cell System Explaind " New York:John Wiley &
Sons,Ltd.

9