Downloaded from SAE International by Norges Teknisk-Naturvitenskapelige Universitet, Tuesday, September 07, 2021

SAE TECHNICAL
PAPER SERIES 2007-01-3491

Control of Automotive PEM Fuel Cell Systems

Seo-Ho Choi, Yo-In Song, Kyung-Won Suh, Nam-Woo Lee, Dae-Jong Kim,
Sang-Uk Kwon and Tae-Won Lim
Hyundai Motor Company

14th Asia Pacific Automotive

Engineering Conference
Hollywood, California, USA
August 5-8, 2007

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-0790 Web: www.sae.org
 Downloaded from SAE International by Norges Teknisk-Naturvitenskapelige Universitet, Tuesday, September 07, 2021

The selection for publication of this manuscript is based on standard practices and guidelines of the
originating country’s engineering and scientific society in the Asia Pacific region.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of SAE.

For permission and licensing requests contact:

SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: permissions@sae.org
Fax: 724-776-3036
Tel: 724-772-4028

For multiple print copies contact:

SAE Customer Service

Tel: 877-606-7323 (inside USA and Canada)
Tel: 724-776-4970 (outside USA)
Fax: 724-776-0790
Email: CustomerService@sae.org

ISSN 0148-7191
Copyright © 2007 SAE International
Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE.
The author is solely responsible for the content of the paper. A process is available by which discussions
will be printed with the paper if it is published in SAE Transactions.

Persons wishing to submit papers to be considered for presentation or publication by SAE should send the
manuscript or a 300 word abstract of a proposed manuscript to: Secretary, Engineering Meetings Board, SAE.

Printed in USA
 Downloaded from SAE International by Norges Teknisk-Naturvitenskapelige Universitet, Tuesday, September 07, 2021

2007-01-3491

Control of Automotive PEM Fuel Cell Systems

Seo-Ho Choi, Yo-In Song, Kyung-Won Suh, Nam-Woo Lee
Dae-Jong Kim, Sang-Uk Kwon and Tae-Won Lim
Hyundai Motor Company

Copyright © 2007 SAE International

ABSTRACT Code Generation) environment. Thus, the fuel cell

system are modeled by using the Simulink´, and the
In order to understand the automotive PEM fuel cell controllers are designed also using Simulink´ and
system, mathematical system modeling is conducted Stateflow´, and the C code for the controller is
and the model is implemented and simulated by using generated with the developed models by using ACG.
the Matlab´/Simulink´. The components such as fuel cell
stack, air supplier, and radiator are modeled individually MAIN SECTION
and integrated into a system level. The PEM fuel cell
system operation control includes thermal management, FUEL CELL SYSTEM - An example of fuel cell system is
air supply control, hydrogen supply control, fuel cell shown in Figure 1. The fuel cell system consists of air
stack protection control, and load following control. In supply system, water and thermal management system,
the thermal management, the inlet and outlet and hydrogen supply system.
temperature of coolant are controlled to operate the fuel
cell stack in desired temperature range and to prevent
flooding inside the fuel cell stack. In air supply control
and hydrogen supply control, the flow rates of air and
hydrogen are controlled not to starve the fuel cell stack
according to the output current. A control structure for
the system is developed and confirmed by using the
developed simulation model. Besides the system
operation control, start up and shutdown control are also
important issues because they are closely related to the
fuel cell stack life time. The degradation mechanisms of
electrodes and membrane during start up and shutdown
are well understood but they are difficult to model, thus
the start up and shutdown control strategies are
designed based on the experimental results.
Figure 1. Fuel cell system configuration
INTRODUCTION The air supply system, which supplies air to the fuel cell
stack, consists of an air filter, an air blower or
Fuel cells are electrochemical devices that convert the compressor, and a humidifier. The water and thermal
chemical energy of fuels directly into electricity and heat. management system controls the temperature of the fuel
Among many types of fuel cells, the PEM(Polymer cell stack and humidifier, and collects water from the
Electrolyte Membrane or Proton Exchange Membrane) cathode exhaust by condensation. The humidifier
fuel cell is one of the most promising alternative power consumes water, thus we need to collect water from the
sources for automotive applications because it has a cathode exhaust to meet the water balance. Recently,
potential to be produced with low cost, and provides long many companies are using hollow fiber gas to gas
life time, high power density, and high efficiency. The humidifiers that use humid and hot cathode exhaust gas
fuel cell system has many components and is very to humidify the dry air. By using the gas to gas humidifier,
complicated, thus it needs multidisciplinary approach in the condenser and de-ionized water tank can be
designing plant model. If we design the controller based eliminated since de-mineralized water is not used for
on the model, we can reduce the development time. humidification. In order to increase the utilization of
Nowadays, some simulation tools, such as Simulink´ hydrogen, it is needed to re-circulate the hydrogen. This
and Stateflow´ from Mathworks, support ACG (Auto is done by hydrogen recycle blower. The pressure drop
 Downloaded from SAE International by Norges Teknisk-Naturvitenskapelige Universitet, Tuesday, September 07, 2021

1/ 2
through the anode channel in the stack should be small, 0 RT PH ˜ PO2
since the hydrogen is the lightest molecules, and it is VOC VOC  ln( 2 )
2F PH 2O
very hard to pressurize by using a blower. The produced
water at the cathode transports to the anode through the
membrane. If the anode gas is re-circulated, the dry where PH 2 , PO2 , and PH 2O are partial pressures of
hydrogen from the tank is mixed with humid re-circulated hydrogen, oxygen, and steam gases in bar respectively.
hydrogen, thus the anode humidifier can be removed. We have to consider irreversibilities since the reaction
The nitrogen in the cathode also transports through the can not be reversible. The ohmic losses come from flow
membrane to the anode, and the impurities in hydrogen resistance of ions in the electrolyte and the electrical
accumulate in anode channel, thus the anode should be resistance of the electrodes. The conductivity of
purged to remove the accumulated nitrogen and hydrogen proton in the membrane changes with the
impurities. temperature and water contents(moles of water per one
mole of sulphonic acid sites) of the membrane. The
MODELING OF FUEL CELL SYSTEM – The models of effect of the proton conductivity in the membrane is
system components are developed to have a strict more dominant than the electrical resistance of the
component level modularity. Because of the modularity, electrodes. The amount of voltage drop can be modeled
the effects of component type and size can be easily to be linear to the fuel cell current.
studied. The modularity also provides the environment of
designing controller by using MBD(Model Based Design). VOhm I ˜ ( Relectrode  Rmembrane (T , O ))
The designed controller model are used to make
production code used in ECU by using ACG(Auto Code
Generation). The activation losses are determined by the activation
energy of the reactions. Tafel reported that the
Fuel cell stack – The input and output of a hydrogen fuel overvoltage at the surface of the electrodes can be
cell are presented in Figure 1. The input energy to the denoted by
fuel cell is the enthalpy difference between the input
i
gases and output gases. VOver A ln( )
io
l“ŒŠ›™Š› 

where i is the cell current, io is the exchange current

that becomes higher if the reaction is faster, and A is
El od
j c tr o

the Tafel constant. For a hydrogen fuel cell, the Tafel

An
e e
ˆ› l y
– te

constant A is given by the equation as follows:

‹Œ

o ‹™–ŽŒ• vŸ ŽŒ•

RT
A
2DF

oŒˆ›
~ˆ›Œ™ When the reaction is very fast, the hydrogen and oxygen
concentrations at the catalyst surface becomes low, and
Figure 2. The input and output of a fuel cell rapid voltage drop coccurs.

If the reaction is reversible, the electrical energy will be

1.2
equal to the Gibbs free energy change between input
and output gases. Thus, the open circuit voltage can be 1.1 Ό΄ΚΞΦΝΒΥΚΠΟΎ͑Տͮͧͦ͟

expressed as: 1
Ό΄ΚΞΦΝΒΥΚΠΟΎ͑Տͮͩͦ͑͟
Ό΄ΚΞΦΝΒΥΚΠΟΎ͑Տͮͦ͑͢͟͡
ΌͶΩ ΡΖΣΚΞΖΟΥΎ͑΃͹͑͑ͲΟ͠ʹΒ͖͖ͫͩ͢͡͠͡
Cell voltage(V)

ΌͶΩ ΡΖΣΚΞΖΟΥΎ͑΃͹͑͑ͲΟ͠ʹΒ͖͖ͫ͢͢͢͡͠͡
0.9
 'g f ΌͶΩ ΡΖΣΚΞΖΟΥΎ͑΃͹͑͑ͲΟ͠ʹΒ͖͖ͫͧͣ͢͟͢͠͡
VOC Temperture=70ఁ
2F 0.8

0.7

where 'g f molar Gibbs free energy change, and VOC 0.6

is the fuel cell open circuit voltage, F is Faraday 0.5

constant that is the charge on one mole of electrons. 0.4
The change of operating pressure of the fuel cell affects 0 200 400 600 800 1000 1200

the value of 'g f , and the output voltage. If the open Current density(mA/cm )
2

0
circuit voltage at standard condition is known as VOC , Figure 3. Comparison of simulation and experimental
then the voltage in different pressure conditions can be results(cell voltages vs. current)
calculated by Nernst equation, and it is given by
 Downloaded from SAE International by Norges Teknisk-Naturvitenskapelige Universitet, Tuesday, September 07, 2021

But we do not consider the concentration losses in the By using above equations, the modular model of the
fuel cell stack model, since we will not operate the blower, which is implemented in Simulink´, is presented
in Figure 5.
system at this high current density region. According to
above equations, we can calculate the cell voltage in
different conditions. Figure 3 presents the voltage versus
current curve according to change of the water contents
and relative humidity. It shows that the simulation results
well agree with experimental results.

Air supply system – Blower type air suppliers can be

used for ambient type fuel cell system where the
maximum operating pressure is 0.5 barg.

P1, T1

P2, T2

Figure 4 Input and output of compressor

When the compression is isentropic, the output

temperature can be calculated by

J 1
Figure 5. Blower model implemented in Simulink´
§P · J
T2 _ isen T1 ¨¨ 2 ¸¸
© P1 ¹
The flow rate controller controls the output air mass flow
rate by controlling the blower motor speed. The PI
Since the compression cannot be isentropic, if we define controller is used to control the air feed into the fuel cell
stack. Figure 6 compares the experimental and
the isentropic efficiency as simulation results of the blower and controller by using
the model implemented in Figure 5.
Pisentropic c p (T2 _ isen  T1 )m (T2 _ isen  T1 )
Kisen
Preal c p (T2  T1 )m (T2  T1 )

Then, the output temperature can be denoted by

J 1
§ ·
T1 ¨ § P2 · J ¸
T2 T1  ¨ ¸  1
Kisen ¨¨ ¨© P1 ¸¹ ¸
¸
© ¹

The work done is the change in enthalpy of air, thus the

power can be expressed as

J 1
§ § ··
¨ T ¨§ P · 1 ¸¸
Pshaft c p ¨ T1  1 ¨ ¨¨ 2 ¸¸  1¸ ¸m
¨ Kisen ¨ © P1 ¹ ¸¸
© © ¹¹
Figure 6. Comparison of simulation results of air flow
control with experimental ones
Since the blower is operated by electric motor, if we
consider efficiency of motor, then we get the electric
Radiator- The effectiveness-NTU method is used to
power used by the blower :
model the radiator because we know the inlet coolant
temperature and the air inlet temperature to the radiator.
Pshaft
Pelectric The effectiveness is defined as below:
K motor
 Downloaded from SAE International by Norges Teknisk-Naturvitenskapelige Universitet, Tuesday, September 07, 2021

actual heat transfer rate q

H
maximum possible heat transfer rate qmax

When the heat capacity ratio of cold fluid Cc is smaller

than that of the hot fluid Ch , we define the smaller one
between Cc and Ch as Cmin , and larger one as Cmax .
Then the qmax is expressed as:

qmax Cmin ˜ (Th,i  Tc ,i )

where Th ,i is the inlet temperature of hot fluid, and Tc ,i

is the inlet temperature of the cold fluid to the heat Figure 8 Integrated fuel cell system model
exchanger. For radiator, Th ,i is the coolant inlet
Developed component models are integrated into a
temperature and Tc ,i is the air inlet temperature, system level as shown in Figure 8. Inputs and outputs of
respectively. the component model are decided to have modularity
For a cross-flow heat exchanger with both flows are and forward facing characteristics, so that the system
unmixed, the effectiveness is given by model can be used in designed controller by using
MBD(Model Based Design).
­ NTU 0.22 ½
H 1  exp® *
> 
exp  C * NTU 0, 78  1 ¾  @ CONTROL OF FUEL CELL SYSTEM – Developed fuel
¯ C ¿
cell system model is used in designing controller. Figure
9. shows the off line simulation model that include plant
where NTU is the number of transfer unit defined as model and controller model. The controller model has
UA / Cmin , and C * is the heat capacity ratio defined as the temperature controller, air flow rate controller,
Cmin Cmax . hydrogen controller, and safety algorithms.
The amount of heat transfer rate is expressed as:

q H ˜ qmax H ˜ Cmin Th,i  Tc ,i  Ch Th,i  Th ,o 

By rearranging above equation, the coolant outlet

temperature of the radiator is expressed as:
q
Th , o Th ,i 
Ch

Figure 7. shows the comparison of heat radiation

between simulation and experiment by changing the flow
rate and air speed that flows through radiator fins. As
you can see in the figure, the simulation results agree
well with experimental results.
Figure 9. Off line controller and plant simulation model
45.0
[Experiment] Coolant flow rate 20L/min
[Experiment] Coolant flow rate 60L/min
The designed controller model is used in generating
40.0 [Experiment] Coolant flow rate 120L/min
[Simulation] Coolant flow rate 20L/min
production code with I/O blocks by using ACG(Auto
[Simulation] Coolant flow rate 60L/min Code Generation). The I/O blocks are used to generate
Heat radiation(Kw)

35.0 [Simulation] Coolant flow rate 120L/min

basic software. Since the controller has several

30.0 sampling time, the multitasking is used with rate
monotonic scheduler. For ACG, we used the real time
25.0
workshop embedded coder from Mathworks.
20.0

15.0

10.0
1 2 3 4 5 6
Speed of air(m/s)

Figure 7 Comparison of heat radiation between

simulation and experiment
 Downloaded from SAE International by Norges Teknisk-Naturvitenskapelige Universitet, Tuesday, September 07, 2021

Figure 12. shows the cell voltage variations when the

load is relatively high. The cell voltage is quite even, and
it means the air flow and hydrogen flow is well controlled.

CONCLUSION

A simulation model was developed to have component

level modularity and forward facing characteristics. The
component models, such as a fuel cell stack, a radiator,
and an air blower, are verified with experimental results.
In designing controller, MBD(Model Based Design) is
used by using Simulink´ and Stateflow´ from Mathworks.
The designed controllers are also used in generating C
code by using ACG(Auto Code Generation). The
development time and calibration time was drastically
reduced by using MBD and ACG.

REFERENCES
Figure 10. Controller model with I/O blocksets and 1. Sang-Kwon Kim, Seo-Ho Choi, “Development of
scheduler Fuel Cell Hybrid Vehicle by Using Ultra-Capacitors
as a Secondary Power Source,” SAE 2005 world
congress
EXPERIMENTAL RESULTS – Figure 11. shows the 2. Seo-Ho Choi, Soon-Woo Kwon, et. al, “Control of
experimental results when the code, which generated Power Distribution in Fuel Cell Hybrid Vehicle,” The
using ACG, is compiled and downloaded into ECU. It
20th Electric Vehicle Symposium, 2003
shows that the inlet temperature is controlled, even
3. Yongjin Sung, Yong Sun Park, et. al, “Thermal and
though the power is changing rapidly.
Water Management of Automotive PEM Fuel Cell
System,” SAE 2004 world congress
4. Larminie, J. and Andrew Dicks. “Fuel Cell Systems
Explained,” Ontario, John Wiley & Sons Ltd, 2003
5. Springer, T. E., T. A. Zawodzinski and S. Gottesfeld.
“Polymer electrolyte fuel cell model,” Journal of The
Electrochemical Society 138, 2334-2342, 1991
6. Fuller, T., and J. Newman. “Water and Thermal
Management in Solid-Polymer-Electrolyte Fuel
Cells,” Journal of the Electrochemical Society 140,
1218-1225, 1993
7. Adam Z. Weber, John Newman, “Transport in
Polymer-Electrolyte Membranes,” “III. Model
Validation in a Simple Fuel-Cell Model,” Journal of
The Electrochemical Society, 151 (2) A326-A339,
2004
Figure 11. The experimental results with FC system and
8. M.M. Mench, Q.L. Dong, C.Y. Wang, “In situ water
designed controller.
distribution measurements in a polymer electrolyte
fuel cell,” Journal of Power Sources 124, 90-98,
2003

CONTACT

Seo-Ho Choi is the senior research engineer for Hyundai

Motor Company and holds a Doctor of Philosophy
degree from KAIST(Korea Advanced Institute of Science
and Technology). He has worked in fuel cell vehicle
development division for 7 years.

e-mail : seoho@hyundai-motor.com
Figure 12. The individual cell voltages of the fuel cell
stack