Development of Next Generation

micro-CHP System
Based on High Temperature Proton Exchange
Membrane Fuel Cell Technology

Alexandros Arsalis
Dissertation submitted to the Faculty of
Engineering and Science, Aalborg University
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
in
Mechanical Engineering

Department of Energy Technology

Aalborg University
2011

Development of Next Generation micro-CHP System: Based on High

Temperature Proton Exchange Membrane Fuel Cell Technology
Alexandros Arsalis 2011
All rights reserved. No part of the material protected by this copyright notice
may be reproduced or utilized in any form or by any means, electronic or
mechanical, including photocopying, recording or by any information storage
and retrieval system, without the prior permission of the author.
Printed in Denmark by UniPrint, 2012
First print, January 2012
ISBN 978-87-92846-03-7

Development of Next Generation micro-CHP System

Abstract
Novel proposals for the modeling and operation of a micro-CHP (combined-heat-andpower) residential system based on HT-PEMFC (High Temperature-Proton Exchange
Membrane Fuel Cell) technology are described and analyzed to investigate the
technical feasibility of such systems. The proposed systems must provide electricity,
hot water, and space heating for an average single-family household in Denmark. A
complete fuel processing subsystem, with all necessary BOP (balance-of-plant)
components, is modeled and coupled to the fuel cell stack subsystem. The research
project is divided into five main study topics: (a) Modeling, simulation and validation
of the system in LabVIEWTM environment to provide the ability of Data Acquisition of
actual components, and thereby more realistic design in the future; (b) Modeling,
parametric study, and sensitivity analysis of the system in EES (Engineering Equation
Solver). The parametric study is conducted to determine the most viable
system/component design based on maximizing total system efficiency; (c) An
improved operational strategy is formulated and applied in an attempt to minimize
operational implications, experienced when using conventional operational strategies;
(d) Application of a GA (Genetic Algorithm) optimization strategy. The objective
function of the single-objective optimization strategy is the net electrical efficiency of
the micro-CHP system. The implemented optimization procedure attempts to maximize
the objective function by variation of nine decision variables; (e) The micro-CHP
system is optimized by formulating and applying a process integration methodology.
The methodology involves system optimization targeting in net electrical efficiency
maximization. Subsequently a MINLP (Mixed Integer Non-Linear Programming)
problem optimization strategy is applied to minimize the annual cost of the HEN (Heat
Exchanger Network). The results obtained throughout this research work indicate the
high potential of the proposed micro-CHP system, since net electrical efficiencies of up
to 44% were reached, which are far and away higher than heat engine-based systems.
Another interesting aspect is the simplicity of the systems fuel processing subsystem,
which makes it more competitive, in terms of commercialization prospects, than other
fuel cell-based micro-CHP systems.

Keywords: Fuel cell, PEMFC, PBI, Micro-CHP system, Combined-heat-and-power,

Optimization, Pinch analysis, Process integration, Fuel processing, Operational strategy
Authors address: Alexandros Arsalis, Aalborg University, Department of Energy
Technology, Pontoppidanstrde 101, DK-9220 Aalborg ., Denmark
E-mail: aar@et.aau.dk
Supervisors: Associate Professor Mads Pagh Nielsen, Department of Energy
Technology, Aalborg University, Aalborg ., Denmark (primary)
Professor Sren Knudsen Kr, Department of Energy Technology,
Aalborg University, Aalborg ., Denmark (secondary)

Thesis
assessment
committee:

Associate Professor Thomas Condra, Department of Energy Technology,

Aalborg University, Aalborg ., Denmark (chairman)
Professor Per Alvfors, Department of Chemical Engineering and
Technology, Division of Energy, KTH-Royal Institute of Technology,
Stockholm, Sweden
Dr. Jeppe Grue, Vattenfall A/S Heat Nordic, Aalborg, Denmark

Research
program:

Mechanical Engineering

Contents
List of Publications
Papers included in the thesis
Relevant papers not included in the thesis

9
9
10

Acknowledgements

11

Abbreviations

13

1
1.1

1.2

1.3
1.4
1.5

Introduction
15
Literature Review
15
1.1.1 Fuel Cell Fundamentals
15
1.1.2 Proton Exchange Membrane Fuel Cells
16
1.1.3 High Temperature PEMFC
18
1.1.4 Comparison of Fuel Cell Technologies
19
1.1.5 Combined Heat and Power
20
1.1.6 Micro-CHP Technology
21
1.1.7 Micro-CHP Classification based on the Conversion Process
23
Comparison Analysis Between the Proposed micro-CHP System and a
Centralized CCGT Power Plant/Heater Combination
27
1.2.1 Efficiency
28
1.2.2 Environmental Considerations
29
1.2.3 Cost
30
Background and Motivation
30
Definition of the Research Question
31
Objectives of the Research Project
31

2
2.1
2.2

Overview of Research Methodologies

Design of Stationary Fuel Cell Systems
Modeling of Stationary Fuel Cell-based CHP Systems
2.2.1 Fuel Processing Subsystem
2.2.2 Fuel Cell Subsystem
2.2.3 Thermal Management Subsystem

35
35
36
37
40
41

3
3.1
3.2
3.3
3.4

Description of Publications
Paper I
Paper II
Paper III
Paper IV

43
43
43
44
45

3.5
3.6

Paper V
Overall Evolution of the Configuration Topology

46
48

Summary of Principal Results and Discussion

51

5
5.1
5.2

Conclusions and Future Work

Conclusions
Future Work

61
61
63

References

65

List of Publications
Papers included in the thesis
This thesis is based on the work contained in the following five journal
articles1, referred to by Roman numerals in the text:
I

Arsalis, A., Nielsen, M.P. & Kr, S.K. (2011). Modeling and off-design
performance of a 1 kWe HT-PEMFC (high temperature-proton exchange
membrane fuel cell)-based residential micro-CHP (combined-heat-andpower) system for Danish single-family households. Energy 36(2), 9931002.

II

Arsalis, A., Nielsen, M.P. & Kr, S.K. (2011). Modeling and parametric
study of a 1 kWe HT-PEMFC-based residential micro-CHP system.
International Journal of Hydrogen Energy 36(8), 5010-5020.

III Arsalis, A., Nielsen, M.P. & Kr, S.K. (2012). Application of an improved
operational strategy for a high temperature-proton exchange membrane
fuel cell-based micro-combined heat and power system for Danish singlefamily households. Energy. (submitted)
IV Arsalis, A., Nielsen, M.P. & Kr, S.K. (2012). Modeling and optimization
of a 1 kWe HT-PEMFC-based micro-CHP residential system.
International Journal of Hydrogen Energy. (in press)

1. Every research study topic matches exactly with every corresponding journal article
publication. Therefore the terms paper/article/study topic are used interchangeably throughout this
thesis work.
9

V Arsalis, A., Nielsen, M.P. & Kr, S.K. (2012). Optimization of a high
temperature PEMFC-based micro-CHP system by formulation and
application of a process integration methodology. Energy Conversion and
Management. (submitted)
Articles I and II are reproduced with the permission of the publishers.

Relevant papers not included in the thesis

VI Arsalis, A., Nielsen, M.P. & Kr, S.K. (2009). Modeling and simulation of
a residential micro-CHP system based on HT-PEMFC technology. EFC0917045. In European Fuel Cell Technology & Applications - Piero Lunghi
Conference. Rome, Italy, December 15-18.

10

Acknowledgements
This PhD research project was supported by Danish companies Dantherm
Power and Danfoss. It was also co-funded by the HyFC (Hydrogen Fuel Cell
Academy) and Aalborg University.
Thanks to my academic supervisors Mads Pagh Nielsen and Sren Knudsen
Kr for providing valuable advice, comments and suggestions throughout the
research study.
Special thanks go to Anders Korsgaard for providing simulation models in
the initial stages of the project, and also prompt feedback, when needed,
throughout the project.
I would also like to thank my colleagues at the Fuel Cell Research Program
for exchanging fruitful ideas on the research project.
Finally, I am indebted to my family who has helped me, while I devoted
time and energy to this research work.
Alexandros Arsalis
Aalborg
October 2011

11

(This is an example of an empty page. A white rectangle is drawn on top of the page number.)

12

Abbreviations
AFC
BOP
CHP
EES
GA
GAMS
HEN
HHV
HT
ICE
IRR
LHV
LT
MCFC
MEA
MINLP
PAFC
PBI
PEMFC
RC
RE
RH
SC
SE
SMR
SOFC
VI
WGS

Alkaline Fuel Cell

Balance-of-Plant
Combined Heat and Power
Engineering Equation Solver
Genetic Algorithm
General Algebraic Modeling System
Heat Exchanger Network
Higher Heating Value
High Temperature
Internal Combustion Engine
Internal Rate of Return
Lower Heating Value
Low Temperature
Molten Carbonate Fuel Cell
Membrane Electrode Assembly
Mixed Integer Non-Linear Programming
Phosphoric Acid Fuel Cell
Polybenzimidazole
Polymer Electrolyte Membrane Fuel Cell
Rankine Cycle
Reciprocating Engine
Relative Humidity
Steam-to-Carbon ratio
Stirling Engine
Steam Methane Reformer
Solid Oxide Fuel Cell
Virtual Instrument
Water Gas Shift

13

(This is an example of an empty page. A white rectangle is drawn on top of the page number.)
(This

14

Introduction

This chapter presents an overall literature review on the fuel cell and microCHP technologies. Further on, the background and motivation for the research
study are explained, and the project objectives are outlined. Finally, the
research question is posed and analyzed.

1.1 Literature Review

Starting from the general field of fuel cell technology and combined heat and
power (CHP), the literature review is narrowed down to the PEMFC (Proton
Exchange Membrane Fuel Cell) type, and more specifically to the PBI type. In
addition, CHP technology is analyzed and broken down to small-scale
residential micro-CHP technology. The different types of micro-CHP systems
are described to indicate their advantages and disadvantages. Finally an
overview of the implications and prospective of micro-cogeneration
technology is given.
1.1.1 Fuel Cell Fundamentals

Fuel cell technology is based on the principle of direct electrochemical energy

conversion (Barbir, 2005; Larminie & Dicks, 2003; Mench, 2008; OHayre,
Colella, Cha, & Prinz, 2009). In other words, fuel cells are not heat engines
since no combustion takes place. Also, chemical energy input is converted to
electrical energy, without the intermediate step of mechanical energy
production, as in heat engines. In terms of electrochemistry, fuel cells are also
related to batteries, but the difference is that fuel cells are not depleted as
batteries. Figure 1 illustrates the operating principle of fuel cell technology.
Two electrodes exist in a fuel cell arrangement, namely cathode and anode,
which are separated by the electrolyte. The oxidant and the fuel enter the fuel
cell stack through the cathode and the anode, respectively. The fuel cell

15

reaction typically produces water, which is at a higher temperature than the

reactants. More importantly, this exothermic reaction is also associated with
the production of electricity with an external load between the two electrodes.
Fuel cell performance is assessed with the aid of voltage-current curves
(polarization curves), plotted for different loads. These curves can also
illustrate the overpotentials (losses), which are divided between activation,
ohmic and concentration losses.
The five main fuel cell types are distinguished by their electrolyte material
and they are the following: (a) Polymer Electrolyte Membrane Fuel Cell
(PEMFC) (b) Phosphoric Acid Fuel Cell (PAFC), (c) Alkaline Fuel Cell
(AFC), (d) Molten Carbonate Fuel Cell (MCFC), (e) Solid Oxide Fuel Cell
(SOFC). Every fuel cell type has some distinctive characteristics in terms of
operational temperature, pressure, fuel intake and tolerance, building materials
(electrolyte, catalyst, cell components), and performance. These characteristics
result in advantages or disadvantages, depending on the application, as it will
be shown in the following sections. A summary of these characteristics is given
in Table 1.
Table 1 Description of major fuel cell types.
PEMFC

PAFC

Electrolyte

Polymer
membrane

Mobile ion

MCFC

SOFC

Liquid H3PO4 Liquid KOH

Molten
carbonate

Ceramic

H+

H+

OH-

CO32-

O2-

Operating
temperature

60-200

205

65-220

650

600-1000

Catalyst

Platinum

Platinum

Platinum

Nickel

Nickel

Cell
components

Carbon -based Carbon -based Carbon- based Stainlessbased

Ceramicbased

Fuel intake

H2, CH3OH

H2, CO, CH4

H2

AFC

H2

H2, CH4

1.1.2 Proton Exchange Membrane Fuel Cells2

The following electrochemical half reactions in the PEMFC occur

simultaneously in the two electrically conductive electrodes,
H 2 2H + +2e - (anode)
1
O 2 +2H + +2e - H 2 O (cathode)
2

(1)

The most common material for the polymer membrane of the ionically
conductive material is NafionTM(Barbir, 2005). The membrane must be thin,
2. This section refers primarily to conventional Nafion-based PEMFCs.
16

flexible and coated with a platinum-based catalyst, while the electrode support
material must be a porous carbon. The total assembly is a sandwich structure,
namely the Membrane Electrode Assembly (MEA), connecting the anodecatalyst-membrane-catalyst-cathode components. The fuel cell operating
temperature for NafionTM-based fuel cell stacks cannot exceed 90C, because
the membrane requires continuous hydration to maintain its conductivity
capabilities. Therefore a rigorous water management is required throughout the
operation.

e-

load

product gases &

depleted fuel out

product gases &

depleted oxidant out

H2
AFC

OH-

H 2O

H2O

H2

PAFC
PEMFC

O2

O2
H+

H2O

H2
MCFC

(CO) CO2

CO3=

CO2

H 2O
SOFC

(CO)

H2

(CH4)

H 2O

O2

O=

O2

fuel in

oxidant in

anode

electrolyte

cathode

Figure 1 Schematic of the operating principle for different fuel cell types [based on the
description found in (Barbir, 2005)].

To maintain operation within this low temperature range only platinum-based

catalysts and polymer membranes can be used. This means that the building

17

cost of this fuel cell technology will remain high. On the other hand, power
densities are higher than other fuel cell types, ranging from 300 to
1000*mW/cm2 (OHayre et al., 2009). Further on, on-off operations and startups are also faster than other fuel cell technologies (e.g., SOFC). For these
reasons, PEMFC is the primary fuel cell technology for vehicular and portable
applications (<1 kWe) (Mench, 2008). In addition, due to their high sensitivity
and low tolerance to sulfur and carbon monoxide, hydrogen is usually
preferred as the system fuel, although effort has been made to use other more
convenient fuelling options, such as methanol. Usage of fossil fuels, such as
natural gas, can add a further cost burden to the fuel cell systems capital cost,
because more complicated fuel processing will be required.
1.1.3 High Temperature PEMFC

The previous section discussed the implications encountered when operating

NafionTM-based PEMFCs. The disadvantage of operating near the two-phase
region can be alleviated with the utilization of a phosphoric acid doped
polybenzimidazole (PBI) material (see Figure 2 and Figure 3). This material is
doped with a strong acid, such as H3PO4, to create an ionic conductivity. This
material allows operation between 120 to 200C (Bchi, Inaba, & Schmidt,
2009; OHayre et al., 2009; Jianlu Zhang et al., 2006), and therefore no
membrane hydration is required, leading to a relaxation of the water
management requirements. More importantly the performance of the PBI
membrane, in terms of conductivity, remains comparable to the one for
NafionTM membranes.

Figure 2 High temperature PEMFC stacks from Serenergy (Serenergy, 2011).

18

The high operational temperature contributes in higher tolerances to sulfur and

carbon monoxide, and also yields a higher quality exhaust mixture out of the
fuel cell stack. Moreover PBI membranes offer increased mechanical strength,
lower building costs, and thermal stability (Jianlu Zhang et al., 2006).
On the other hand, several disadvantages are associated with PBI
membranes. The most significant ones include slow kinetics of the oxidant
reduction reaction, membrane oxidative degeneration, challenging
electrocatalyst ink catalyst blending with the PBI material, and durability
issues in relation to acid-leaching (OHayre et al., 2009).

Figure 3 HT-PEMFC MEA performance polarization curve (Stolten, 2010).

1.1.4 Comparison of Fuel Cell Technologies

PAFC- and MCFC-based systems are generally applicable for larger-scale

applications (e.g. 200 kWe systems) and therefore they are not suitable for
micro-CHP system applications (Barbir, 2005; Barclay, 2006; Larminie &
Dicks, 2003; OHayre et al., 2009). In particular PAFC technology is a mature
technology, with proven reliability/long-term performance and low-cost
electrolytes (OHayre et al., 2009). On the other hand, it requires expensive
platinum catalysts, is susceptible to carbon dioxide and sulfur poisoning, and
the electrolyte is a corrosive liquid that must be replenished during operation
(OHayre et al., 2009). In addition, PAFC technology has reached a certain
level of limitation, in regards to research improvement as indicated in the
literature (Larminie & Dicks, 2003). MCFC technology requires very careful
operation and complex BOP components (OHayre et al., 2009), including
CO2 recycling, corrosive molten electrolytes, and relatively expensive

19

materials. Also, degradation and lifetime issues have been reported in the
literature (OHayre et al., 2009).
SOFC is a promising fuel cell technology (Zink et al., 2007) and therefore
still under development. Nevertheless, current SOFC-based systems have
several disadvantages, as compared to HT-PEMFC technology. These include
slower start-up times (due to their higher operational temperature), need for air
and fuel preheaters and more complex cooling systems (Larminie & Dicks,
2003; OHayre et al., 2009). Also the need for higher temperature operation
suggests greater heat losses and thereby more expensive insulation is required.
Finally, it requires pressurization of fuel and air, which suggests greater BOP
component (e.g. air compressor) power losses. On the other hand, SOFC-based
systems allow the use of carbon monoxide content in the reformate gas as fuel,
with the use of an internal reforming process (without the need of a separate
unit) (Arsalis, 2007; Calise, Dentice dAccadia, Palombo, et al., 2006), which
suggests a significant advantage over other fuel cell technologies. Also the
high reaction rates achievable by the SOFC technology allow the use of
cheaper catalysts and thereby reducing their capital cost (Larminie & Dicks,
2003).
1.1.5 Combined Heat and Power

The conventional method of covering electrical, heating (e.g. hot water) and
cooling (e.g. space cooling) load demands is by purchasing electricity from the
electricity network grid and with a fossil fuel-fired boiler. A different method
of covering these loads is combined-heat-and-power (CHP) (or cogeneration),
which can aid in the reduction of running (fuel) costs and in effect increase the
total efficiency. Therefore, CHP is defined as the combined generation of
electrical power and heat from a single chemical energy source.
The useful recovery of the biggest portion of waste heat can increase the
system efficiency from 30-40%, up to 75-90%, depending on the application
and the size of the system. In addition to the efficiency increase, CHP can also
lead in the reduction of emissions, since a smaller amount of fuel is required.
Therefore CHP is the preferred choice of generating electricity and heat,
provided the capital cost is within viable limits in terms of lifetime and
payback time.
Cogeneration systems can be distinguished into two main configurations.
The first configuration is the production of a high temperature fluid product,
which can be used to generate electricity (e.g. gas turbine), while the exhausted
heat is at a low temperature and can be used throughout several thermal
processes for district heating purposes, or in some cases for additional
production of electricity (Arsalis, 2008). Alternatively, the high temperature

20

fluid product (e.g. from a waste-to-energy power plant) can be used in a heat
recovery steam generator to produce superheated steam for a Rankine cycle
(steam turbine) process. In some cases it is also possible to use the hot gases
directly into a Brayton cycle (gas turbine), without the need of a heat recovery
boiler.
1.1.6 Micro-CHP Technology

For the purposes of this research study, micro-CHP (Dentice dAccadia, Sasso,
Sibilio, & Vanoli, 2003; Ferguson, 2004; Gunes & Ellis, 2003; Hawkes &
Leach, 2005a, 2005b, 2007, 2008, 2009; Hawkes et al., 2007, 2006; Hubert,
Achard, & Metkemeijer, 2006; Pehnt et al., 2006) is defined as the
simultaneous production of electricity and heat in a residential application for
systems up to 5 kWe. Micro-CHP systems typically have (or expected to have)
a lifetime of ten to twenty years, which is comparatively lower than large-scale
CHP systems. They are designed to exhibit minimum total efficiencies of 75%.
In order for these systems to make a breakthrough to the power and heat
market, many requirements should be satisfied. The main requirements are (a)
low cost, (b) compact volume and size, (c) easy installation and (d) automated
operation without demanding routine maintenance checks.
Depending on the load profile, an appropriate heat-to-power ratio must be
selected based on the demand. For example, a system with a high heat-topower ratio will be inappropriate for a household requiring a high electrical
load. Nowadays, most newly-built households have efficient insulation and are
more electricity-demanding than heat-demanding, as compared to the past.
Therefore, in general, even lower heat-to-power ratios will be required in the
future. Also the production of additional heat, if needed, can be provided by
condensing gas-fired boilers, which have become very efficient in recent years,
with efficiencies near or above 90% (Hawkes & Leach, 2005a). Therefore,
focus is primarily given on systems exhibiting high electrical efficiencies,
rather than thermal efficiencies.
For the better utilization and distribution of power and heat, micro-CHP
systems must be grid-interconnected, although micro-CHP systems can also
operate in a stand-alone (island) mode. The reason for grid-interconnection is
that a stand-alone micro-CHP system will be required to operate continuously
within an electricity-led operational strategy. As a consequence of this practice,
large amounts of heat will have to be vented to the atmosphere. Otherwise a
thermal storage tank of massive dimensions will be required, which is
inappropriate and unacceptable for a residential design. Another consequence
is also the rate of heat loss, which increases with the increasing size of the

21

thermal storage tank3. In addition, the benefit of importing cheap electricity

(e.g. from wind power) from the network grid will not be utilized. It should be
noted though, that import/export of electricity should be minimized in order for
the application to conserve its on-site power generation/consumption
characteristics. On-site power generation minimizes transmission losses, which
can have an even greater factor if the power transmitted has a high purchase
cost.

Figure 4 Classification of micro-CHP technologies based on the conversion process.

Japan has been the leading market for micro-cogeneration in the last decade,
but recently a rapid growth of the technology has also been observed in
Europe. An increase in the sales of micro-CHP systems by around 25%,
between 2009 and 2010, suggests that the market will grow even at higher
proportions in the near future. 20,000 to 70,000 micro-CHP units are expected
to be sold by 2015 (Brown, 2011).
3. Although theoretically all the heat can be recovered to cover the space heating and hot water
loads, it should be noted that a realistic system will not be able to completely recover this heat.
22

1.1.7 Micro-CHP Classification based on the Conversion Process

The conversion process of a micro-CHP system can be based on fuel-air

combustion or direct electrochemical conversion (Pehnt et al., 2006). The
classification of micro-CHP systems in regards to their conversion process is
shown in Figure 4. Combustion-based systems combust the fuel-air mixture at
a high temperature and eventually mechanical energy is produced. The
mechanical energy is then used to drive an electric generator and produce
electrical energy. Combustion-based micro-CHP systems are based on various
thermodynamic cycles, including reciprocating engines (Otto cycle), steam
engines (Rankine cycle) and Stirling engines (Stirling cycle). Alternatively, the
conversion process can be based on direct electrochemical conversion from
chemical energy to electrical energy (i.e. fuel cell technology). As indicated in
the previous section, in addition to the production of electrical energy, the
conversion process also results in the production of heat, which can be
recovered as needed. The status of the different technologies is discussed in the
following subsections and summarized in Table 2.
Reciprocating Engines
A reciprocating engine (RE) based on a spark-ignited, internal combustion
piston-cylinder engine operates on the Otto thermodynamic cycle, which
includes four consecutive processes: (a) isochoric, (b) isentropic expansion, (c)
isochoric, and (d) isentropic compression.
The performance of current RE-based micro-CHP systems is the most
promising, as compared to other heat engine-based micro-CHP systems. This
technology is also the most mature in the micro-cogeneration market, since a
range of different systems have been commercially available in the last decade
(Pehnt et al., 2006). Vaillant/Honda has reported electrical and total
efficiencies of 26.3% and 92%, respectively, for their ecoPOWER 1 kWe unit
(Energy Efficiency News, 2011; Vaillant, 2011). Japanese company Yanmar
has also developed an ICE unit with an electrical power output of 5 kWe
(Dijkstra, 2010). Finally, BAXI has developed a 5.5 kWe system (SenerTec
Dachs) with an electrical and total efficiencies ranging at 28 and 88-91%
(BAXI, 2011a; Thomas, 2008), respectively.
Rankine Cycle
A Rankine cycle-based (RC) engine is based on a thermodynamic cycle
consisting of four consecutive ideal processes: (a) isobaric, (b) isentropic, (c)
isobaric, and (d) isentropic. The liquid water is heated (typically with a boiler)
until superheated steam is produced. The superheated steam is then used to

23

drive a steam expander, producing mechanical energy. The mechanical energy

is then converted to electrical energy by means of an electric generator.
In 2006, German company OTAG developed a 3 kWe unit (Slowe, 2010),
while in 2010, UK-based company Energetic Group has revealed a 1 kWe unit,
Kingston, which is expected to have an electrical efficiency at around 10%
(Energetix Group, 2010).
Table 2 Competing micro-CHP technologies: Status and performance summary.

Company

Type

el ,net (%)

tot (%)

Market status

Vaillant/Honda ecoPOWER 1.0

RE

26

92

CA

Yanmar

RE

N/A

N/A

CA

BAXI SenerTec Dachs

RE

28

88-91

CA

Energetix Kingston

RC

10

N/A

CA

OTAG

RC

N/A

N/A

CA

BAXI Ecogen

SE

14

90

CA

WhisperGen

SE

10-12

90

CA

Bosch ENATEC

SE

10

N/A

CA

Stirling Denmark SM5A

SE

21

85

Demo

Hexis Galileo 1000 N

SOFC

30

90

Demo

Ceres Power

SOFC

N/A

N/A

Demo

Osaka Gas-Kyocera

SOFC

N/A

N/A

Demo

Ceramic Fuel Cells

SOFC

60

N/A

Planned

ClearEdge Power

LT-PEMFC

N/A

90

Demo

Tokyo Gas/Panasonic Ene-Farm

LT-PEMFC

40

90

Demo

BAXI Innotech GAMMA 1.0

LT-PEMFC

32

83

Demo

Plug Power

HT-PEMFC

32

90

Demo

Dantherm

HT-PEMFC

40

85-90

Demo

CA: Commercially available

Stirling Engines
A Stirling engine (SE) is an external combustion engine operating on the
Stirling thermodynamic cycle, which includes four consecutive processes: (a)
isothermal expansion, (b) isochoric cooling, (c) isothermal compression, and
(d) isochoric heating. The SE typically includes three pistons, two outer and
one displacer. The latter circulates the cylinder products into the chamber,
which is cooled or heated by their respective outer streams. The two outer
pistons can be used to regulate the capacity of the combustion process.
The main characteristics of Stirling engines, as compared to other heat
engines, include adequate part-load performances, low emission rates and low
vibration and noise levels. In addition, the external combustion, closed cycle

24

operational nature of SE restricts the exposure of the moving parts of the

engine to the products of combustion. Thereby component degradation is
minimized. In addition, SEs allow greater fuel flexibility, which means that
liquid or gaseous fuels, biofuels, etc. are compatible (Pehnt et al., 2006).
SE is not a mature technology in the micro-cogeneration marketplace, when
compared to RE-based systems. Nevertheless, because of the advantages
outlined above, the technology has attracted a significant interest in recent
years. BAXI develops a 1 kWe/6 kWth unit, so-called Ecogen, which is capable
of achieving electrical and total efficiencies of 14% and 90%, respectively
(Gummert, 2008). WhisperGen develops a 1.2 kWe/8 kWth unit, which is
capable of achieving electrical and total efficiencies of 10-12% and 90%,
respectively (Pehnt et al., 2006). Bosch develops a 1 kWe/6 kWth unit, socalled ENATEC, which is capable of achieving an electrical efficiency of 10%
(Beckers, 2006). Finally, Stirling Denmark has developed a larger-scale
demonstration unit (9 kWe/25 kWth), which has been tested to perform
electrical and total efficiencies at 21% and 85%, respectively (Thomas, 2008).
Fuel Cells Systems
Fuel cell-based micro-CHP systems (see Figure 5) are either based on the
PEMFC or SOFC types (Inui, Yanagisawa, & Ishida, 2003). Fuel cells are
currently the least mature technology in the area of micro-CHP systems, when
compared to heat engine-based applications. Nevertheless, because of their
promising features, primarily low emissions and high net electrical efficiencies,
these systems have attracted a lot of interest for research and development. In
addition, some justification also comes from synergies with other renewable
energy sources (e.g. wind power). In this sense, a PEMFC-based system is
capable of providing fast regulation services to the grid responsible.
BAXI has developed a 1 kWe/1.7 kWth LT-PEMFC-based unit (Innotech
GAMMA 1.0), which is capable of achieving electrical and total efficiencies of
32% and 83%, respectively (BAXI, 2011b). Korean company ClearEdge
Power has been developing a 5 kWe LT-PEMFC-based unit with a total system
efficiency reaching 90% (ClearEdge Power, 2011). Tokyo Gas, in association
with Panasonic, has been developing a 0.75 kWe LT-PEMFC-based unit
capable of achieving an electrical efficiency of 40%, while the total system
efficiency is expected to reach 90% (Tokyo Gas Ltd., 2011).
Hexis has developed a 1 kWe/2 kWth SOFC-based unit (Galileo 1000 N),
which is capable of achieving electrical and total efficiencies of 30% and 90%,
respectively (Callux, 2009). Some other companies, namely Osaka GasKyocera, Ceres Power and Ceramic Fuel Cells are also in the process of
developing SOFC-based micro-CHP units (Delta Energy & Environment Ltd,

25

2010). In particular, Ceramic Fuel Cells is planning to develop an SOFC-based

unit (1.5 kWe/0.5 kWth), with a remarkable electrical efficiency reaching up to
60% (Ceramic Fuel Cells Ltd., 2010). Plug Power has developed a 5 kWe HTPEMFC-based unit, which is capable of achieving electrical and total
efficiencies close to 32% and 90%, respectively (Vogel & Tsou, 2007).
exhaust

Electricity
inverter
PEMFC micro-CHP system

heat exchanger

hot water

space heating

natural gas
electricity import/export

Figure 5 Arrangement of a residential PEMFC-based micro-CHP system.

In Germany, a residential fuel cell program, named Callux, has been

initiated in 2008, and the first phase is expected to be completed in 2012. The
consortium combines the expertise of several industrial bodies and companies.
The fuel cell manufacturers involved in this project include companies such as
BAXI, Hexis and Vaillant (Callux, 2009). Several units have already been
installed and currently, they are being tested in German households.
In Denmark, a micro-CHP project has been initiated in 2006, managed by a
national consortium and consisting of nine companies and governmental
bodies, such as the Danish Energy Agency. The consortium combines all the
expertise necessary to develop, test, and demonstrate micro-CHP systems. The
Danish Ministry of Climate and Energy finances the project by 40% of the total
cost. The project is scheduled to be completed by 2012, and will test units
based on various fuel cell technologies, such as NafionTM-based PEMFC, PBI-

26

based PEMFC and SOFC. It is divided into three phases, and two fuel types are
being tested: hydrogen and natural gas. The choice of fuel depends on the fuel
cell technology and its availability at the installation site. A preliminary
assumption concerning the operational control of the system, indicated that the
electrical load must be fulfilled, but also the design should secure the right
amount of heat will be available on demand for the space heating and hot water
loads. Further on, an auxiliary burner for peak production of thermal energy is
under consideration to be integrated to the end-user system (Danish MicroCHP, 2009; Korsgaard, Nielsen, & Kr, 2008).
The HT-PEMFC-based micro-CHP units are developed by Dantherm Power
and operate on natural gas. Experimental tests, with pure hydrogen fuel,
showed start-up times of 30 to 60 minutes, with an electrical efficiency of 40%
(hydrogen fuel, LHV-based). From the operating point of the finalized design,
an efficiency of 50% is expected (hydrogen fuel, LHV-based). Further
experimental tests and calculations showed a potential system efficiency of 85
to 90% (LHV-based) (Danish Micro-CHP, 2009). A fast adaption to load
variations was also observed. In the first phase, Danish micro-CHP developed
unit prototypes with PEMFC and SOFC technologies. In the second phase, ten
micro-CHP units are scheduled to be installed and tested at selected consumer
households in the municipalities of Lolland and Snderborg. In the third and
last phase, micro-CHP systems will be installed and demonstrated at around
100 households in the two aforementioned municipalities. This will allow the
Danish micro-CHP system project to gather and analyze realistic experiences
related to installation, operational procedures, maintenance needs and also
consumer satisfaction (Danish Micro-CHP, 2009). Currently, HT-PEMFC
activities are on standby, pending durability issues to be resolved.

1.2 Comparison Analysis Between the Proposed micro-CHP

System and a Centralized CCGT Power Plant/Heater
Combination
In this section the proposed HT-PEMFC-based micro-CHP system is compared
to a more conventional combination of producing electricity and heat for a
single-family household in Denmark. The best available technology today is
analyzed and set as a suitable baseline for comparison with the proposed
micro-CHP system.
It should be noted though, that combined cycle gas turbine (CCGT) and
micro-CHP systems are not necessarily in competition with each other, because
they can coexist peacefully in the future. If both technologies are used in
parallel, more options will be available and therefore the better utilization of

27

resources (e.g. natural gas) will improve. In other words, energy availability
and reliability will improve, because dependency to the network grid will
decrease, and therefore grid congestions and blackout events will diminish
(Chicco & Mancarella, 2009; Praetorius et al., 2009). In addition, the current
centralized power production system status-quo will shift towards a more
competitive regime, with obvious benefits (Praetorius et al., 2009). Therefore,
it should be strongly emphasized that the purpose of this research work is not
the investigation of a proposed replacement of centralized CCGT power plants,
but rather the replacement of aging natural gas boilers4, as already noted.
Therefore the discussion below is only provided for comparison purposes.
Centralized power generation is considered more conventional and power
plants are available in various configurations and based in different
technologies. The most significant ones are the following: Steam turbines (ST),
gas turbines (GT) and CCGT. The latter type is considered to be the most
advanced technology, exhibiting overall efficiencies close to 60% (Lund, 2008;
Praetorius et al., 2009). It combines the gas turbine (Brayton) and steam
turbine (Rankine) thermodynamic cycles, where the former is the topping cycle
and the latter the bottoming one. In addition, some other, more novel, central
power plants designs have been under investigation in recent years. These
include hybrid systems, which combine high temperature fuel cells with
turbine cycles (i.e. SOFC-GT, SOFC-ST, SOFC-GT-ST and MCFC-GT).
Therefore the following hypothetical question must be answered:
Can the proposed system compete with centralized CCGT power plants and
heat-only boilers which are in widespread use in single-family households
today?
To answer the above question, a number of comparison parameters are
considered in order to distinguish the characteristics, including advantages and
disadvantages, of the two technologies. The following analysis is performed in
terms of efficiency, environmental considerations and cost.
1.2.1 Efficiency

The efficiency of a CCGT power plant is lower than the efficiency of the
proposed system in many ways. At nominal load, modern CCGT systems can
exhibit electrical efficiencies up to 60%5 (Ang, Fraga, Brandon, Samsatli, &
Brett, 2011), which is higher than the expected electrical efficiency of the
4. This is also the target of the Danish micro-CHP project (Danish Micro-CHP, 2009).
5. One of the most modern CCGT power plants in Denmark is the Silkeborg CHP Plant with a
power output of 105 MWe and an electrical efficiency of 50% (DONG Energy, 2011).
28

proposed system. On the other hand, both systems must operate both at
nominal and part-load. At part-load, the electrical efficiency of the CCGT will
decrease6 (Kehlhofer, Bachmann, Nielsen, & Warner, 1999), while the
electrical efficiency of the proposed system will increase. For example, a
CCGT plant with a nominal load efficiency of 60%, will have a decreased
efficiency of 52% at 50% part-load (estimation based on empirical data found
in (Kehlhofer et al., 1999)).
Further on, due to distribution and transmissions losses, around 6 to 10% of
the efficiency is lost in the case of CCGT (Al-Sulaiman, Hamdullahpur, &
Dincer, 2011; Lund, 2008), while in the case of the proposed system, there are
practically no such losses, because the power is produced and consumed onsite (Raven & Verbong, 2007). It should be noted though that this assumption
is only valid provided electricity is not exported by the proposed system to the
grid (electricity-led operation).
In terms of overall efficiencies, the proposed system can reach efficiencies
up to 91%, which is significantly higher than the corresponding efficiencies for
CCGT, especially in the case where the exhausted heat of the power plant
cannot be usefully utilized (Ang et al., 2011; Westner & Madlener, 2011).
Finally innovative control techniques, such as virtual control of many
micro-CHP systems (virtual power plant), can aid in a more efficient operation
of the micro-CHP systems, since neighboring micro-CHP systems can operate
jointly, transmitting heat and power, as needed (Delta Energy & Environment
Ltd, 2010; Praetorius et al., 2009; Raven & Verbong, 2007).
1.2.2 Environmental Considerations

Fuel cell technology features high electrical and total efficiencies, which in
terms of environmental benefits translate to lower fuel consumption and
decreased CO2 emissions, as compared to a CCGT/boiler combination
(Praetorius et al., 2009). The Kyoto agreement has prompted the EU to adopt
energy policies that favor power and heat production with reduced greenhouse
emissions (De Paepe & Mertens, 2007; Praetorius et al., 2009; Ropenus,
Schrder, Costa, & Ob, 2010). Therefore, it is expected that the proposed
system will be offered with several benefits, such as lowered taxation,
subsides, etc. Therefore energy policies can further favor the adoption of
decentralized systems, such as the proposed system, even though their (actual)
total cost will remain higher than CCGT/heater per kWh.

6. In CCGT power plants, without supplementary firing, the total plant efficiency primarily
depends on the degree of heat recovery, the GT efficiency and the size of the plant (Kehlhofer et
al., 1999).
29

The main problem with the proposed system, in terms of emission rates, is
that the emissions will be exhausted in urban areas. Thereby CCGT is more
emissions attractive, at least locally, in the sense that centralized power plants
are typically located in rural areas (Chicco & Mancarella, 2009).
1.2.3 Cost

Currently fuel cell-based micro-CHP systems are not commercialized and

therefore an estimation of their total cost cannot be accurately calculated.
Nevertheless, the current capital cost of a fuel cell system is estimated to be
very high and not competitive in terms of market diffusion (Chicco &
Mancarella, 2009; Woudstra, van der Stelt, & Hemmes, 2006). A threshold
point7 must be reached, to allow the proposed system to compete with
centralized power plants. The total cost includes the operating (fuel) cost,
capital cost, maintenance cost and taxation cost. In addition the lifetime of the
system must be known.
The proposed system can achieve high overall efficiencies and therefore can
lead in fuel consumption reductions. Therefore, compared to the lower overall
efficiencies of CCGT power plants at both full-load, and especially at part-load
operation, the proposed system is more attractive in terms of fuel savings.
Finally in terms of lifetime and reliability, the proposed system has still a
long way to prove itself (Delta Energy & Environment Ltd, 2010). Both the
fuel cell stack and fuel processing degradation and catalyst deactivation issues
have to be resolved, before the proposed system is able to match CCGT or
other heat engine-based technologies.

1.3 Background and Motivation

The main motivations for the development of HT-PEMFC based micro-CHP
systems for Danish single-family households are the following:
The proven ability of fuel cell technology to achieve higher
operational efficiencies (electrical and overall), with lower emissions
as compared to other more conventional technologies (e.g. heat
engines) (Barbir, 2005; Georgopoulos, 2002; Kim, von Spakovsky,
Wang, & Nelson, 2010; Larminie & Dicks, 2003; OHayre et al.,
2009);
The need for the replacement of gas-fired furnaces in Danish singlefamily households with more efficient alternatives (Danish Micro7. The investment cost for a typical modern CCGT-CHP power plant is estimated at around
1200 /kWe, with a depreciation period of 30 years and IRR of 7.4% (Westner & Madlener,
2011).
30

CHP, 2009; Korsgaard et al., 2008). There is an increasing demand

for cleaner and more efficient energy production in conjunction with
the increasing price and depletion of fossil fuels. Moreover, higher
operational efficiencies suggest lower operational costs (Arsalis,
2008);
The HT-PEMFC technology utilizes PBI membranes, operating at
temperatures between 140 to 200 . Therefore, this technology
allows enhanced electrochemical kinetics, simpler water management
and cooling, and also useful waste heat can be recovered (Bchi et
al., 2009; Korsgaard, Nielsen, Bang, & Kr, 2006; Korsgaard,
Refshauge, Nielsen, Bang, & Kr, 2006; Stolten, 2010; Jianlu Zhang
et al., 2006). In addition lower quality reformed hydrogen may be
used as fuel (Jianlu Zhang et al., 2006), suggesting cheaper and
simpler fuel processing subsystems are required.

1.4 Definition of the Research Question

Based on the literature review on micro-CHP technology, and the background
and motivation, analyzed and explained in the previous sections, the main
research question can be formulated:
What are the prospects of 1 kWe HT-PEMFC-based micro-CHP systems to
replace gas-fired burners in Danish single-family households in the context of
a technical feasibility study?
To answer this question, this research effort establishes the methods and
framework for physical and empirical parametric modeling of the HT-PEMFCbased micro-CHP system, including its subsystems/components at a level
appropriate for conceptual and preliminary design. The developed system must
be validated and throughout a process of computational simulations, including
design and off-design operational modes, parametric studies, sensitivity
analysis, development of improved operational strategies, and application of
optimization strategies (genetic algorithm, pinch analysis and process
integration methods), conclusions are drawn on the proposed system potential.

1.5 Objectives of the Research Project

The overall goal of this research work is to develop a HT-PEMFC-based
micro-CHP system to replace natural gas heating units in Danish households.
A 1 kWe (nominal power rate) grid-connected micro-CHP system can fulfill

31

the needs of such a household (average four person family residence in

Denmark). To model and then analyze the micro-CHP system configurations as
realistically as possible, detailed system and component thermodynamic,
kinetic and geometric models are developed, implemented, and validated. Then
by means of different studies, such as parametric studies, sensitivity analysis,
genetic algorithm optimization (Chong & Zak, 2001; Ravindran, Ragsdell, &
Reklaitis, 2006), and optimization using process integration techniques (pinch
analysis) (Biegler, Grossmann, & Westerberg, 1997; Kemp, 2007; Sieniutycz
& Jezowski, 2009; Smith, 2005), the system is thoroughly analyzed in the
context of a technical feasibility study. Therefore the full advantages of such a
HT-PEMFC-based micro-CHP system over more conventional systems can be
extracted and analyzed.
The research project is split into five main topics/studies (shown in Figure
6), which can be summarized as follows:
Development of the system in LabVIEWTM for future use with data
acquisition hardware and laboratory tests of actual components;
Development of the system in EES. Subsequent analysis using
parametric study and sensitivity analysis methodologies;
Investigation of the simulation model performance at different
operational loads and formulation of an improved operational
strategy;
Optimization of the system by use of a genetic algorithm
optimization strategy;
Further optimization using pinch analysis and process integration
techniques. Objective functions are formulated for the maximization
of the net electrical efficiency and minimization of total HEN cost.
Thus, this research project includes the following tasks:
Perform a literature review on fuel cell- and conventional-based
micro-CHP systems;
Become familiar with the previously developed HT-PEMFC-based
micro-CHP system by A. Korsgaard (Korsgaard et al., 2008);
Modify existing or develop new models/components required for the
micro-CHP system. These models include: HT-PEMFC stack, SMR
reactor, WGS reactor, catalytic combustor, heat exchangers, steam
generator, water pump, air blower, thermal storage tank;
Design, model and optimize the total system to implement the
outlined studies;
Examine the off-design operational behavior of the system in terms
of efficiency and emissions;
Validate the system using previously reported results;

32

Analyze the results and draw conclusions.

Figure 6 Information flow within the research project effort.

To eliminate possible confusion, while reading this thesis work, the following
assumptions are given:
The fuel cell type was predetermined to be of the HT-PEMFC type. It
is therefore assumed to be a fixed, non-decision variable, since no
evaluation process took place to select which fuel cell technology
type is more applicable (e.g. in terms of efficiency) for a fuel cellbased micro-CHP system;
The computational model development varied throughout the
projects evolution, based on decisions taken by the author and the
supervising committee, or due to knowledge and information gained
while developing the models.

33

(This is an example of an empty page. A white rectangle is drawn on top of the page number.)

34

Overview of Research Methodologies

In this chapter an overview of the development of the fuel cell system is

described in detail. The analysis includes a description of the different
components and subsystems needed in the development and operation of a fuel
cell system.

2.1 Design of Stationary Fuel Cell Systems

A complete system based on fuel cell technology apart from the fuel cell stack,
requires the coupling of some additional components (Barbir, 2005). Each
component performs a different function, including heating, cooling, power
conditioning and controlling. A fuel cell stack must be heated at start-up, but
during operation, and especially during transient loads, it must be cooled to
maintain normal operation. Therefore cooling is used to prevent overheating
and thermal gradients within the stack (OHayre et al., 2009). Cooling and
heating is not required only for the stack, but also for other system processes,
such as fuel processing. Therefore a careful design, based on the selected
application, must be utilized to perform the thermal management of the total
system. This will ensure a minimum dumping of heat from the fuel cell stack
and the fuel processing subsystem. In a stationary application, the additional
waste heat may be suitable for recovery and use by means of cogeneration for
different purposes. These may include covering part of the heating load profile
(e.g. space heating) of a household (Korsgaard et al., 2008), or utilization in an
absorption heat pump system (Zink, Lu, & Schaefer, 2007).
In terms of fueling, stationary fuel cell systems are usually less demanding
than portable or vehicular applications. This is because they do not require fuel
storage and system volume restrictions are usually more relaxed. Therefore
pure hydrogen demand can be avoided, since fossil fuels, which are more

35

readily available8, can be instead utilized. A hydrogen-rich gas can be

produced from natural gas9, throughout a series of fuel processing operations.
These typically include desulfurization10, steam methane reforming and carbon
monoxide cleaning (e.g. water gas shift, preferential oxidation). These steps are
necessary in order to ensure different impurities and poisons have been
completely removed, or at least removed at a tolerable limit, before the
hydrogen-rich gas can enter the fuel cell stack anode. In the case of SOFCs,
only pre-reforming is required, since internal reforming of the fuel is feasible
inside the fuel cell stack, taking advantage of the high operating temperature of
this fuel cell type (Calise, Dentice dAccadia, Palombo, & Vanoli, 2006).
The fuel cells power output requires conditioning to ensure a stable,
reliable electrical output (OHayre et al., 2009). The power is conditioned by
means of regulation and inversion. DC/AC inverters are used to transform the
incoming DC power into an AC power, where DC/DC converters are used to
regulate the power by stepping up or down the incoming transient voltage to an
outgoing constant value. Almost all the electricity can be conserved, since both
inverters and converters are very efficient, with typical values around 85-90%
and 90-98%, respectively (Barbir, 2005; Larminie & Dicks, 2003). The fuel
cell control unit11 use feedback loops between sensors and actuators (i.e.
valves, switches) to maintain operation within a desired range (OHayre et al.,
2009). The total system efficiency is defined as the sum of the net system
electrical efficiency and thermal efficiency (available for cogeneration) divided
by the chemical energy input of the system fuel.

2.2 Modeling of Stationary Fuel Cell-based CHP Systems

A fuel cell system is divided into four main subsystems, introduced in the
previous section: (a) the fuel processing subsystem, (b) the fuel cell subsystem,
(c) the power electronics subsystem, and (d) the thermal management
subsystem (OHayre et al., 2009). The fuel processing subsystem consists of
chemical reactors, a burner and pipelines, where the fuel cell subsystem
includes the fuel cell stack, cooling arrangements and an air blower (or air
8. Natural gas supply lines are usually available in most urban areas.
9. For the purpose of calculation simplicity natural gas is assumed to behave as methane. It
should be noted that for a realistic operation with natural gas, the effect of higher carbons, should
not be neglected. In this case more steam will be needed to address the potential problem of
carbon deposition.
10. The purpose of desulfurization is the removal of sulfur compounds from a fossil fuel (e.g.
natural gas) (Barbir, 2005; Kolb, 2008).
11. The control unit model lies outside of the scope of this research project and therefore it is
not analyzed any further.
36

compressor). It should be noted that SOFCs additionally require a fuel

compressor (Calise, Dentice dAccadia, Vanoli, & von Spakovsky, 2006). The
power electronics subsystem includes the electricity cabling (OHayre et al.,
2009). Finally, the thermal management subsystem includes the heat exchange
network, with the associate heat exchangers, pipelines and actuators.
2.2.1 Fuel Processing Subsystem

The purpose of the fuel processing subsystem is the generation of a hydrogenrich reformate gas, with a low CO-content (composition depends on the
tolerance of the fuel cell type). Additionally other poisonous substances, such
as H2S must be removed. To accomplish these tasks various processes must
take place in appropriate components (e.g. reactors) in a series of several steps.
A simplified configuration of a typical fuel processing subsystem arrangement
is shown in Figure 7. The main components include heat exchangers, chemical
reactors, a burner, pipelines and extraction equipment (Jahn & Schroer, 2005;
Jannelli, Minutillo, & Galloni, 2007; Kolb, 2008; OHayre et al., 2009). A
steam generator converts liquid water into superheated steam. Steam
generation is necessary for the fulfillment of the chemical reactions (e.g. steam
methane reforming) in the reactors. The compressed natural gas is also
preheated to accelerate and facilitate the reforming reaction. Steam generation
and fuel preheating requirements are usually easily accomplished by waste heat
generated by either the fuel cell stack or fuel processing. Then the
methane/steam mixture enters the reformer, where it reacts at a high
temperature (600-700 ), in the presence of a catalyst (to accelerate chemical
kinetics), resulting in a hydrogen-rich reformate gas. The reformate gas then
enters the water gas shift reactor, which increases the quantity of hydrogen in
the stream and decreases the CO content.
The catalytic burner serves into the complete combustion of fuel remnants
in the fuel cell exhaust exiting from the fuel cell anode and cathode. Based on
the fuel processing arrangement the generated heat can be used in an
endothermic reforming reaction (e.g. steam methane reforming) and/or provide
heating for other system needs (e.g. steam generation). If heat is still available
it can be used externally (e.g. thermal storage tank). Depending on the
operating fuel utilization of the fuel cell stack, hydrogen-rich gas may be
depleted at the exiting streamline of the fuel cell stacks anode and used in the
burner. In some cases, the steam generated by the combustion of the hydrogenrich gas in the burner can be reused in other parts of the system (OHayre et al.,
2009). Finally after the burner, a condenser can be utilized to convert steam
back to liquid water by cooling this stream and capturing the latent heat of
condensation. In a fuel cell system, a condenser is important for both

37

recapturing heat and recovering liquid water to achieve neutral system water
balance12. Finally it should be noted that very careful insulation and integration
of the fuel processing components is necessary to minimize heat loss rates.

Depleted fuel
Natural gas
BURNER
air

WGS

Hot flue gas

Reformate fuel

SMR

Reformate fuel

Flue gas
FUEL CELL STACK
air

DESULFURIZER

Natural gas

exhaust

Figure 7 Configuration of a simplified fuel processing subsystem.

Steam Reforming
Steam reforming involves the reaction process of a hydrocarbon with steam. It
is highly endothermic and proceeds at high temperature over a nickel catalyst
(Kolb, 2008; Xu & Froment, 1989a, 1989b) as follows,
1

C x H y xH 2 O(g) xCO y x H 2
2

CO, CO2 , H 2 , H 2 O

(2)

In addition, the high reaction temperature combined with high steam-to-carbon

ratios can aid in the minimization of carbon formation. High pressures can also
help in carbon formation minimization, but they result in lower conversion
rates (Kolb, 2008). The steam reforming of methane can yield towards a H2content of 76% on a dry molar basis (OHayre et al., 2009). Steam reforming is

12. A neutral system water balance describes the situation where all the water consumed by
system components is produced by other system components, without any external water addition
(OHayre et al., 2009).
38

the most efficient reforming option13 in terms of H 2 generation, because of the

absence of oxygen (air) in the reaction, and in turn the outlet stream does not
contain nitrogen (Barbir, 2005). H 2 generation can be increased by operating
the reaction with excess steam to help shift the reactions equilibrium in favor
of H 2 production (Xu & Froment, 1989b). Additional increase in H 2
generation can be accomplished by shifting the CO towards H 2 via the water
gas shift reaction (OHayre et al., 2009):
CO H 2 O (g) CO 2 H 2

(3)

All the main chemical reactions associated with steam methane reforming are
summarized in Table 3. The most common steam reforming design
configuration is the tubular type (e.g. shell-and-tube) and includes a furnace
containing tubes filled with catalysts, through which the steam reforming
reactants flow (OHayre et al., 2009). In a typical shell-and-tube arrangement,
the reforming reaction takes place inside the tubes (tube-side), which are
heated by the hot flue gas flowing in the shell-side. Shell-and-tube is not easily
applicable in low-scale fuel cell-based systems, because integration becomes
more difficult in compact design applications.
Table 3 Steam reforming reactions.
0
hrxn
kJ mol

Reaction #

Reaction Type

Stoichiometric Formula

Steam reforming

CH 4 H 2O(g) CO 3H 2

+206.4

Water gas shift reaction

CO H 2O(g) CO 2 H 2

-41.2

Evaporation

H 2O(l) H 2O(g)

+44.1

Water Gas Shift

The aforementioned steam reforming reaction results in the production of CO
in the reformate fuel. Therefore, the second fuel reforming step involves one,
or a series, of CO-cleanup processes. The initial and main CO-cleanup process
involves passing of the reformate gas through a WGS reactor (Keiski, Salmi,
Niemisto, Ainassaari, & Pohjola, 1996; Kolb, 2008). The CO-content yCO is the
mole fraction of CO in the reformate stream,
yCO

nCO
n

(4)

13. The other main reforming options are partial oxidation and autothermal reforming. Partial
oxidation is less efficient than steam reforming in terms of hydrogen production. In particular for
methane, it is only able to generate two moles of hydrogen per mole of methane, while steam
reforming can generate three moles (Barbir, 2005).

39

where nCO is the number of moles of CO in the reformate stream. The WGS
reaction can reduce the CO-content up to around 0.2%, with the presence of a
shift catalyst (Kolb, 2008).
The reversible WGS reaction is moderately exothermic. At high
temperatures the balance is shifted towards the reactants (CO and H2O), where
at low temperatures, the balance follows the opposite direction (CO2 and H2)
(OHayre et al., 2009). Therefore, at low temperatures, the reaction increases
its H2 yield (equilibrium), but on the other hand, at high temperatures the
reaction kinetics proceeds faster. To accomplish both tasks, the WGS process
may proceed in two (or more) stages, utilizing initially a high temperature
reactor, and afterwards another low temperature reactor, downstream of the
first. Typical low temperature shift (LTS) reactors and high temperature shift
(HTS) reactors operate at around 200-300C and 375-450C, respectively
(Kolb, 2008). Because of the temperature difference between the two shift
reactors, an intermediate cooling heat exchanger is required to maintain the
desired input temperature to the LTS reactor. The most common catalyst
material for LTS reactors is copper/zinc oxide (Kolb, 2008), while HTS
reactors incorporate iron oxide-based catalysts (Keiski et al., 1996; Kolb,
2008). A PBI-based PEMFC can tolerate up to 3% of CO-content, and
therefore a single stage WGS reactor is usually adequate for CO-cleaning
(Serenergy, 2011).
2.2.2 Fuel Cell Subsystem

The H2-rich reformate gas is fed to the anode side of the fuel cell stack, while
simultaneously compressed air is fed to the cathode side. The fuel cell reaction
produces DC electrical power and exhaust heat. The total system electrical
efficiency is lower than the fuel cell electrical efficiency, due to the parasitic
power required to operate pumps or compressors (OHayre et al., 2009).
Therefore a high efficiency can be maintained, only if the system can operate
near ambient pressure. The fuel cell efficiency increases with an increasing
voltage, but higher voltage values result into a decrease in power density.
Therefore in this case a larger fuel cell stack, with an associated higher
purchase cost, must be utilized for the same power output (Barbir, 2005).
Equivalently, part-load fuel cell efficiencies are higher than the
corresponding efficiencies at nominal load. It should be noted though, that a
high fuel cell efficiency does not necessary correspond to a high overall system
efficiency, because other system components may perform poorly under partload conditions. For example, the SMR reactor will be less efficient at partload due to the increase of heat losses, or balance-of-plant components may
require more power input (comparatively to power output) at these conditions.
40

The net electrical efficiency of the fuel cell subsystem R ,SUB is defined as the
ratio of the net electrical power of a fuel cell subsystem Pe ,SUB and the HHV of
H2 in the inlet gas (OHayre et al., 2009),
R,SUB

Pe,SUB
H

(5)

HHV,H 2

2.2.3 Thermal Management Subsystem

The thermal management subsystem recovers waste heat from the system for
both internal (system) use and also external use (OHayre et al., 2009), if
applicable. Therefore this subsystem must integrate the system streamlines in
such a way that will allow both smooth and efficient operation. This is
typically accomplished by the use of heat exchangers. The selection of a heat
exchanger network can be a very difficult process, because many important
parameters are involved. These parameters include the heat exchanger
purchase cost, the heat exchanger type, the exchanger heat transfer area, the
number of heat exchangers and also the heat exchanger integration with other
system components.
A fuel cell-based system can be very demanding in terms of thermal
integration, especially in the case of an integrated fuel processing subsystem
(Barbir, 2005). A series of heat exchangers will be required to either heat (e.g.
steam generation and fuel preheating) or cool (e.g. cooling of reformate fuel
between reactors and before the fuel cell anode inlet) the various system
streamlines. Typically heat can be recovered from the catalytic burner,
exothermic chemical reactor, reformate cooling processes and the fuel cell
stack exhaust. On the other hand, heat must be supplied to the steam generator,
fuel preheater, thermal storage tank, etc.
A very rigorous process for the configuration of heat exchangers is
required, in order for the system to accomplish an economically feasible
operational threshold, without the need for extensive external heating and/or
cooling. An optimum heat exchanger network can be designed with the aid of
advanced integration techniques, such as pinch analysis and process
integration, which allow the determination of an idealized configuration, based
on the system designers requirements. Therefore the objective function for
such a procedure may involve cost and other parameters, usually within a
constrained optimization regime (Kemp, 2007).

41

(This is an example of an empty page. A white rectangle is drawn on top of the page number.)

42

Description of Publications

In this chapter all published and submitted journal articles are introduced and
described in terms of hypothesis, methodology and results. In addition, a
description of the logical and scientific progression of the work is explained.

3.1 Paper I
In the first paper (I), the HT-PEMFC-based micro-CHP system is designed and
modeled in LabVIEWTM. A detailed literature review, including an
introduction to stationary fuel cell systems, micro-CHP and HT-PEMFC
technology, is outlined. Also, an introduction to the Danish micro-CHP project
is given, including state-of-the-art, project schedule and future objectives.
The system is designed and simulated in LabVIEWTM, with the intention of
providing easiness in user usage and control of the model, and to allow future
experimental testing capabilities with Data Acquisition hardware.
A representative averaged load profile of a single-family household in
Denmark is used to simulate the model within design and off-design
conditions. The system is divided into subVIs (subsystems) and connected and
controlled with the main VI. The model includes component models for the
fuel cell stack, SMR and WGS reactors, mixers, by-pass valves, heat
exchangers, combustor, steam generator and water pump. The model is then
validated and results are extracted to describe its characteristics for 25 to 100%
operational loads.

3.2 Paper II
In the second paper (II), the micro-CHP system is modeled in EES. The reason
is that modeling in EES is relatively easy and most importantly it includes
many built-in features, such as parametric tables, uncertainty analysis,

43

optimization methods, thermophysical property functions, etc. Therefore, this

topic is set as the basis for the remaining three research topics (III-V) of the
study plan.
A literature review on micro-CHP technology and other cogeneration
systems based on fuel cell technology, as found in the literature, is given. The
details on the system modeling are given in detail, including assumptions on
modeling simplifications. In this paper, a novel SMR reactor model is
integrated with the micro-CHP system. The compact SMR reactor (AhlstromSilversand & Odenbrand, 1999) is of the plate heat exchanger type and its
modeling details are explained. The other models are similar to the ones given
in paper I.
The system is then validated and a sensitivity analysis, using the built-in
EES uncertainty propagation tool, is applied to determine system performance
trends. A parametric study is then conducted using four decision variables,
found to be influential in the sensitivity analysis. These are the following: (a)
fuel cell operating temperature, (b) combustor output temperature, (c) anode
stoichiometric ratio, and (d) steam-to-carbon ratio. A detailed analysis of the
results of the parametric study is given, including efficiency variation against
different current density values. Also the variation in the chemical composition
of the reformate gas is illustrated and analyzed.

3.3 Paper III

In the third paper (III), the simulation model developed in II is used as a basis
for the study of conventional and combined improved strategies for the microCHP system. The paper includes a literature review on the different types of
micro-CHP systems available. These include PEMFC, SOFC, ICE (Internal
Combustion Engine), and SE-based systems. Their advantages and
disadvantages, with emphasis on their operational behavior and efficiency, are
given for each type. Further on, an introduction on the market penetration
projection for micro-CHP systems is analyzed. Finally, a literature review,
encompassing conventional and novel operational strategies, is explained in
detail.
The system modeling and layout is given. For the purposes of this study
topic (III), a thermal storage tank is modeled and coupled to the micro-CHP
system (Salcines, Estbanez, & Herrero, 2004), to provide greater operational
flexibility. Then, the three operational strategies used in this research study are
analyzed in detail. The three strategies are the following: heat-led, electricityled, and improved operation.

44

In the results and discussion section the averaged Danish load profile is
fulfilled (partly or fully) by use of the three operational strategies. Based on the
results from the two conventional operational strategies, the improved strategy
is formulated and applied to the simulation model. Emphasis is given on
constraining the system operation within high efficiency regimes, reduction of
frequent interaction with the grid, and avoiding the production of excess heat
that cannot be stored in the thermal storage tank. Also carbon dioxide
emissions are analyzed and compared with conventional systems. In addition,
heat losses from the thermal storage tank during different loads (and time
periods) are analyzed. Finally, an overall analysis of the three operational
strategies is presented and analyzed through comparisons.

3.4 Paper IV
After the development of the simulation model (paper II), and the investigation
of efficient operational strategies (paper III), the system is optimized using the
EES built-in GA optimization method (Godat & Marechal, 2003; Palazzi,
Autissier, Marechal, & Favrat, 2007; Weber, Marechal, Favrat, & Kraines,
2006). The purpose of this research study topic is the maximization of the net
electrical efficiency by variation of nine decision variables. In this system a
different subsystem simulation model for the SMR reactor (Georgopoulos,
2002) is modeled and coupled to the simulation model, since the previous
experimental model, used in study topics II-III does not allow simulation
flexibility (discretized model) and geometrical optimization. All other
components are modeled similarly as in II-III.
The paper includes the theoretical background of the optimization strategy,
including the modeling and optimizing assumptions used in the simulation
model, and the built-in EES min/max function. This function is based on
PIKAIA, which is a public domain, general purpose GA-based optimization
subroutine (Charbonneau, 2002). Internally, PIKAIA seeks to maximize a userdefined function f ( x) in a bounded n -dimensional space,

x x1 , x2 ,...xn , xk 0.0,1.0 k

(6)

Parameter values are restricted in the above range to allow greater flexibility
and adaptability across problem domains. Maximization is carried out on a
population made up of N p individuals, while the population size remains fixed
throughout the evolution. Instead of evolving the population until some
tolerance criterion is satisfied, the evolution is carried over a user-defined,
preset number of generations N g . Since breeding involves the production of
two offspring, the inner loop executes N p 2 times per generational iteration,
45

where N p is the population size. All parameter values defining the individual
members of the initial population are assigned a random number in the range
above, extracted from a uniform distribution of random deviates. This ensures
no initial bias is introduced by the initialization.
The values of the fixed parameters are given. Then the decision variables
are described, including their initial, minimum, maximum and the calculated
optimum values. The details on the optimization results including the number
of the needed generations and iterations to reach the optimum value are
described in detail. The behavioral pattern of the objective function and every
decision variable throughout the optimization procedure is illustrated and
analyzed in detail.

3.5 Paper V
The final study topic (paper V), includes a more advanced optimization
methodology, using process integration techniques. The purpose of this study
is to further increase the net electrical efficiency of the system using pinch
analysis (Kemp, 2007; Linnhoff & Flower, 1978a, 1978b; Linnhoff &
Hindmarch, 1983; Smith, 2005), and also redesign the HEN of the micro-CHP
system, using a MINLP problem formulation (Biegler et al., 1997; Grossmann,
2004; Ponce-Ortega, Jimenez-Gutierrez, & Grossmann, 2008; Viswanathan &
Grossmann, 1990; Yee & Grossmann, 1990). The objective function (PonceOrtega et al., 2008) is defined as the minimization of the total yearly cost,
which includes the cost of utilities and the fixed and variables costs of the
exchangers,

min

CCUqcu

iHPS

iHPS

iHPS

cu, j

jCPS

CHUqhu

jCPS

CF

jCPS kST

CF

jCPS

i , j i, j ,k

CF

iHPS

i , cu

zcui
(7)

zhu j
qi , j , k 1 hi , k 1 h j , k
C
i, j LMTD
kST
i, j ,k

where C is the area cost coefficient, CCU is the unit cost of cold utility, CHU
is the unit cost of hot utility, CF is the fixed charge for exchangers, CPS is the
{j|j} cold process stream, HPS is the {i|i} hot process stream, cu is the cold
utility, hu is the hot utility, h is the fouling heat transfer coefficient, qi , j , k is the
heat exchanged between hot process stream i and cold process stream j in stage

46

k, qcui is the heat exchanged between cold utility and hot stream i, qhu j is the
heat exchanged between hot utility and cold stream j, ST is the {k|k stage in the
superstructure, k=1, . . ., NOK}, zi,j,k is the set of binary variables for match (i,j)
in stage k, zcui is the set of binary variables for the match between the cold
utility and the hot stream i, zhuj is the set of binary variables for the match
between the hot utility and the cold stream j, is the exponent for area in the
cost equation and is a small number. Subscripts i and j denote the hot and
cold process streams, respectively, while k is the index for the stage (1,,
NOK) and temperature location (1,, NOK+1).
A general introduction on pinch analysis and process integration techniques
is given to explain the purpose of the study. Then, a literature review on fuel
cell-based systems using different methods of pinch analysis and process
integration (Autissier, Palazzi, Marechal, van Herle, & Favrat, 2007; Godat &
Marechal, 2003; Palazzi et al., 2007; Verda & Nicolin, 2010; Wallmark &
Alvfors, 2002; Weber et al., 2006) is given. Finally, the objectives of the
research study topic are given in steps. The simulation model is described in
detail. In general, the simulation model is the same as in IV, although there are
a few changes. The first modification is the removal of the condenser from the
system configuration. A condenser is not included, after personal
communication with the HT-PEMFC stack manufacturer (Serenergy) and the
Jlich Research Center (Korsgaard, 2011; Stolten, 2011)14. The second
modification is the inclusion of heat losses considerations for the fuel cell
stack, SMR reactor and combustor. This is done in order to make the system
simulation closer to an actual system, and also to investigate the extent of heat
loss rate from different system components.
The general pinch analysis and process integration methodology is
formulated and analyzed in detail, with all necessary steps followed in the
research study. The steps include definition of the cold and hot data set,
removal of all heat exchangers included in the initial configuration,
reevaluation of the HEN using process change techniques, and HEN
optimization to obtain a minimum total annual cost. The pinch analysis results
are first found for the initial stream data set. Based on these preliminary results
and by means of process change, the GA optimization method in EES is used
to optimize the system in terms of maximization of the net electrical efficiency.
The pinch analysis is then re-applied to the optimum stream data set, and used
in the MINLP optimization model in GAMS to minimize the total HEN annual
cost. Based on this result, a new HEN configuration is designed with all
14. Regarding the steam-content in the reformate fuel, the communication concluded that
steam could remain in the reformate fuel, without causing problems in the fuel cell stack
operation.
47

necessary data (stream flows, temperatures, number of heat exchangers, costs,

and heat loads).

3.6 Overall Evolution of the Configuration Topology

The configuration topology of the system has evolved differently in almost
every research topic for various reasons. Initially, in papers I-II, the system was
designed without a thermal storage tank (TST) (see Figure 8). Instead three
heat exchangers (IV, V, VI), were utilized to accomplish the heat exchange for
the thermal cogeneration of the system. In the third study topic (III), a TST
model replaced the three heat exchangers to reflect a more realistic design and
to offer operational flexibility in the effort of seeking an improved operational
strategy (see Figure 9). However, the use of a mixer, in a realistic system
configuration, can cause pressure drop and temperature compatibility problems
when the two flows are mixed (fuel cell stack exhaust and combustor flue gas
exhaust). This problem is solved in the fourth study topic (IV). In this case the
fuel cell exhaust flows into the combustor, while a heat exchanger (IV), after
the flue gas combustor exit, is used to increase the temperature of the fuel cell
stack exhaust (see Figure 10). A high incoming temperature achieves a higher
combustion efficiency (requires less extra fuel), and therefore the net electrical
efficiency of the system is increased. The final change in the system
configuration (paper V) was the removal of the condenser (see Figure 11), for
the reasons explained in the end of the previous subsection.
BypassValve
Mixer
PlateHeatExchanger
CH4

Methane
Reformate

Catalytic
Combustor
Hotfluegas
(in)

Preheated

Water

air

17

Air
Nodes 135

16

19

15

SteamGenerator

25

steam

H2O

CH4
steam
mixture

5
2

Desulfurizer

Naturalgas

I
1

CH4

WaterGas
Shift

8
II

28

10
III

13 14

Waterinjection

18

Preheater

20

Hotfluegas
(out)

24

4
Fuel

SteamMethane
Reformer

AirBlower

23
Reformatefuel
(H2Ocontentremoved)

Waterpump

29

26

Condenser/
Waterknockout

11
35

22
Exhaust

21
V

30
HotWater

Air(+H2 Ocontent)

34
IV
31

Closedcircuit
H2 Omixture
(Space Heating)

33
Exhaust

32
VI

exhaustmixture
(airwater,etc.)

AnodeCathode

27
HotWater

Depleted
fuel

12
Electricity
(togrid)

Figure 8 System configuration used in papers I-II.

48

BypassValve
Mixer
HeatExchanger
Methane
Reformate
Water
Air
Fluegas

Catalytic
Combustor

127 Nodes

17
Fluegas

13
16
AirBlower
SteamGenerator

14

25
6
CH4steam

5
Desulfurizer
2
Naturalgas

SteamMethane
Reformer

7
Reformatefuel

II

WaterGas
Shift

I 4

III

10

15

18
3

24
19

20

23
Condenser/
Waterknockout

Waterpump
11
27
21
Totalfluegas

26

Heatloss

Air

FuelCellStack
Anode

Exhaust

Cathode

12
Depleted fuel

Thermal
Storage
Tank

Heating
demand

Electricity
(togrid)

22
Exhaustfluegas

Figure 9 System configuration used in paper III.

BypassValve
Mixer
HeatExchanger
Methane
Reformate
Water
Air
Fluegas

13
Catalytic
Combustor

AirBlower

128 Nodes

14

17
Fluegas

16

SteamGenerator
15

25
19
6
CH4steam

5
Desulfurizer

SteamMethane
Reformer

I 4

II

18

2
Naturalgas

7
Reformatefuel

WaterGas
Shift

III

10

27

24
28

IV
26
Exhaust

23
Condenser/
Waterknockout

Waterpump
11
20

21
FuelCellStack
Anode

Heatloss

Air

Cathode

12
Depleted fuel

Thermal
Storage
Tank

Heating
demand

Electricity
(togrid)

22
Exhaustfluegas

Figure 10 System configuration used in paper IV.

49

BypassValve
Mixer
HeatExchanger
Methane
Reformate
Water
Air
Fluegas

13
Catalytic
Combustor

AirBlower

17
Fluegas

127 Nodes

14

16

SteamGenerator
15

25
19

2
1

6
CH4steam

5
Desulfurizer
3

I 4

SteamMethane
Reformer

7
Reformatefuel

18

II

WaterGas
Shift

III

27
10

24

Naturalgas

11

IV
26
Exhaust

23
Waterpump
21

20

Heatloss
FuelCellStack
Anode

Thermal
Storage
Tank

Air

Cathode

Heating
demand
12
Depleted fuel

Water
Tank

Electricity
(togrid)

22
Exhaustfluegas
WaterSeparator

Figure 11 System configuration used in paper V.

50

Summary of Principal Results and

Discussion

In this chapter a summary of the principal results obtained throughout the

evolution of the research project are given and analyzed. The analysis includes
the most significant trend lines obtained by the variation of independent and
dependent parameters, including decision and synthesis/design variables.
The LabVIEWTM system model (see paper I) is simulated at design and offdesign conditions. An averaged load profile from Danish consumption data
measurements, including the electricity and heating demand is used for
comparison with the simulated systems corresponding heat and power output.
These, rather rough, preliminary results were used in the subsequent
publications (namely papers II-III), to perform more specific calculations and
obtain more accurate results. Although the simulated model was not very
realistic and accurate at this stage, the obtained efficiencies verified the great
potential of the system. The maximum net electrical efficiency was calculated
at 45.4% while the maximum total efficiency was 95.2%.
The same system configuration, including a novel plate heat exchanger
SMR reactor, is then simulated in EES with more accurate and realistic
component/subsystem models (see paper II). A sensitivity analysis and
parametric study is performed to provide input for system optimization in the
subsequent study topics (IV-V). Four decision variables (steam-to-carbon ratio,
anode stoichiometric ratio, operating fuel cell temperature, and combustor
output temperature) are varied to determine whether they can cause significant
performance behavioral patterns in regards to efficiency. By observation, the
optimum value of the fuel cell operating temperature is 180 , since the fuel
cell stack operates more efficiently at elevated temperatures (see Figure 12).
On the other hand it should be observed that up to 160 the efficiency
increases rapidly, but beyond that value efficiency increases only slightly.
Therefore, operation beyond 160 is not justified, considering the faster

51

degradation of PBI-membranes at elevated temperatures (Bchi et al., 2009). A

similar pattern is observed with the combustion output temperature. On the
other hand, the other two parametric variables, steam-to-carbon ratio and anode
stoichiometric ratio have a more linear behavioral respond. Finally, the study
revealed the need to optimize geometric parameters, such as the WGS reactor
length.

Figure 12 Decision variables vs. combined cogeneration system efficiency.

To formulate an improved operational strategy, in terms of efficiency, the

effect of the application of conventional strategies (i.e., heat-led and electricityled) should be initially analyzed (see paper III). Then, based on the obtained
results from these strategies, their shortcomings and other disadvantages can be
monitored and eliminated, in the degree possible, by formulating an improved
strategy. The system is constrained within high net electrical efficiency
regimes, while the system must be shut-downed during periods of very low
demands to avoid both low efficiency regimes, but also the production of
excess heat that cannot be stored in the thermal storage tank and must be
exhausted (and thereby lost). The influence of operational strategies on the
proportion of electricity and heating demand met by the micro-CHP system is
illustrated in Figure 13. From an observation, it is obvious that the
electricity/heat demand is almost coincidental with the electricity/heat
production during mid-season (nodes 9-14, 22-25). During this period, only a
52

small amount of auxiliary heat is needed, while the import/export of electricity

to the grid is also kept at a minimum rate. On the other hand, during the winter
months most of the heat demand has to be fulfilled by external means.

Figure 13 Annual variation of electricity (production and demand) and heat (production, demand
and auxiliary).

The performance characteristics of the three operational strategies are

summarized in Table 4. Heat-led operation requires almost twice the fuel input,
when compared to the improved strategy, which produces an increased amount
of electricity overall. Nevertheless, electricity still has to be imported at periods
of low heat demand, such as the summer season. The electricity-led operation
requires significantly less fuel than the heat-led operation, but a high amount of
heat dumping is required. The merits of the improved operation are obvious in
all categories, although a significant amount of imported electricity is required
due to the system shutdown in the summer period and the constraining of the
system at a maximum load of 1 kWe. The most important parameter of the
study was the improvement of the average system efficiency, which is 85.9%
for the improved operational strategy, while the respective total system
efficiencies for the electricity-led and heat-led ones are 71.2 and 74.8%. It is
important to observe that the thermal efficiency (and subsequently the total
system efficiency) for the electricity-led operational is lowered significantly

53

due to the need for heat dumping. Further on, efficient operation has a positive
effect on CO2 emissions, which have been lowered to 2609 kgCO2, as
compared to 2931 and 4653 kgCO2 for the electricity- and heat-led operational
strategies, respectively.
Table 4 Performance of the three considered operational strategies: Overall comparison
Variable

Description [unit]

Electricityled

Heat-led

Improved

ELHV ,in

Total fuel (CH4) input [kWh]

15630

26475

13805

Eel , prod

Total annual electricity production [kWh]

4984

7164

4533

Eel ,imp

Total annual electricity import from the grid 0

[kWh]

753

1127

Eel,exp

Total annual electricity export to the grid

[kWh]

2933

676

Eheat , prod

Total annual heat production [kWh]

8136

12639

7334

Eheat ,aux

Total heat provided by external means

[kWh]

6502

5305

Eheat ,dump

Total annual amount of heat dumping [kWh] 1999

sys
el ,net

Average total system efficiency (%)

71.2

74.8

85.9

Average net electrical system efficiency (%) 31.9

27.1

32.8

th

Average thermal system efficiency (%)

39.3

47.7

53.1

eCO2

Total system CO2 emissions [kgCO2]

2931

4653

2609

By means of a GA-based optimization strategy, the model is improved in terms

of net electrical efficiency (see paper IV). The nine decision variables used in
the optimization process, shown in Table 5, were chosen with an initial value
as typically found in the literature for the kind of system under study. Their
allowable range of variation is chosen on the basis of component/process
operational and/or structural limitations. Therefore these constraints are based
on knowledge of the solution space and on observation of the behavioral
pattern of the optimization algorithm. The final optimum value of the objective
function has modified the initial values of the nine decision variables. The final
value of the steam-to-carbon ratio differs greatly when compared to the initial
one (4.00 vs. 2.91). This deviation suggests that a high amount of steam results
in significant system losses, because of the lack of heat recovery in the
condenser. Another significant change was in the value of the combustor flue
gas temperature, which was reduced from 1173 to 1095 K, while the SMR
reformate inlet temperature was increased from 400 to 530 K. This variation
results from the higher temperature in the SMR reactor inlet, which converts
hydrogen more effectively in a restricted reactor area.

54

The objective function was improved greatly, 40.9% compared to the initial
32.6%, suggesting that there was indeed a large space of improvement, with
respect to the selected objective function. The maximization evolution of the
Table 5 Initial, minimum, maximum and optimum values of the decision variables used in the
optimization procedure.
Variable

Description (unit)

Initial

Min

Max

Optimum

SC

Steam-to-carbon ratio

4.000

2.500

4.000

2.908

Hydrogen stoichiometry

1.500

1.500

1.550

1.509

Number of tubes in the SMR

275

275

300

297

SMR reactor length (m)

1.800

1.800

2.000

1.994

WGS reactor length (m)

0.350

0.350

0.500

0.465

Combustor output temperature (K)

1173

1050

1174

1095

SMR reformate inlet temperature (K)

400

399

530

530

WGS reformate inlet temperature (K)

499

470

500

474

Fuel preheater flue gas inlet temperature (K) 670

550

670

620

SMR
tubes
SMR
reactor
WGS
reactor
Comb
fg ,out
SMR
ref ,in
WGS
ref ,in
FP
fg ,in

objective function, in terms of generations, is shown in Figure 14. The

improvement accelerates quickly in the beginning of the optimization process,
while after the 50th generation the increase is slower. After the 280th generation
the maximization is marginal and almost constant towards the end. This flat
behavior indicates that the overall iterative optimization scheme has practically
converged.

Figure 14 Evolution of the objective function throughout the optimization procedure.

55

A more advanced optimization methodology, based on pinch analysis and

process integration techniques, is then formulated and applied to the microCHP simulation model (see paper V). The new methodology combines two
optimization strategies: (a) GA (see paper IV) and (b) MINLP. By means of a
systematic methodology involving selection of the stream data set, pinch
analysis and process change, the system is optimized in regards to net electrical
efficiency maximization. Initially, basic pinch analysis techniques are applied
to obtain an initial HEN configuration that encompasses the need for MER
(minimum energy requirement, or equivalently, maximum energy recovery).
An analysis of the CCs (composite curves) and the GCC (grand composite
curve) allows the study of process modifications that will further improve the
integrated system operation.

Figure 15 Hot and cold composite curves extracted from the initial process stream data set.

The next step is the application of the simultaneous optimization algorithm,

which is applied in an effort to reach an optimum HEN configuration, based on
total cost considerations. The CCs (see Figure 15) describe the total heating
and cooling demands of the system as a function of temperature intervals. The
overlap between the two curves represents the maximum amount of heat
recovery possible within the process. An observation in the upper right corner
indicates no external heating is required, but on the other hand, a significant
amount of cooling is required, as observed in the lower left part of the figure.
Since this analysis resulted in a threshold problem with a single pinch, no

56

redesigning of the original HEN configuration is required if only MER is

needed. On the other hand, process change techniques can be applied below the
pinch to: (a) decrease the total hot stream load and (b) increase the total cold
stream load. One of the streams allowing modification is the SMR reactor inlet
temperature. On the contrary, the WGS reactor, combustor and fuel cell stack
inlet/outlet temperature streams cannot be modified, since they are either fixed
or dependent. The GA simulation includes four independent design parameters,
shown in Table 6, with their range of variation.
Table 6 Design parameters with their range of variation used in the process change optimization.
Variable

Description (unit)

T6

Initial

Min

Max

Optimum

Inlet temperature of the methane599

steam mixture to the SMR reactor (K)

400

600

587

T11

Inlet temperature of the flue gas to

the methane preheater (K)

500

700

597

m 17

SMR reactor input flue gas mass flow 6.82E-3

rate (kg/s)

6.50E-3

6.87E-3

6.55E-3

SC

Steam-to-carbon ratio (-)

2.50

4.00

2.83

626

3.19

Pinch analysis is then repeated for the new stream data set obtained by the
optimization strategy. The procedure accomplished minimization of the cold
utility requirement from 1.78 to 1.54 kW. In a threshold problem it is necessary
to distinguish its characteristics. The system performance is then tested at partload operation to examine the response of the model at off-design conditions15.
The variation of efficiency at different net electrical power outputs is shown in
Figure 16. As the load decreases, the net electrical efficiency increases linearly,
while the thermal efficiency and total efficiency decrease because of heat
losses.
Since component16 temperatures and areas remain constant at all loads, heat
losses have an increasing effect at lower operational loads (see Figure 17).
Therefore, system operation at low, and especially critical, loads may not be
favored at periods of high demand on cogeneration heat. The greater loss is due
to the SMR reactor, which although it is heavily insulated, it contributes 50%
of the losses. This is due to the high temperature flows (reformate and flue gas)
occurring in the SMR reactor. Finally, the MINLP model is applied using
15. In a realistic system, part-load conditions may not allow smooth operation without the use
of BOP components, such as by-pass valves between heat exchangers.
16. Heat losses are considered for three components: (a) Fuel cell stack, (b) Combustor and (c)
SMR reactor. Heat losses for the WGS reactor are neglected, since the process is slightly
exothermic. Also, heat losses for heat exchanger units are not considered, because the HEN
configuration is not known a priori.
57

GAMS. The compact nature of the micro-CHP system does not favor the use
of stream splitting, and therefore this is restricted in the code. In addition to the
potentially more complex operation, stream splitting would require an
additional investment cost for a control valve.

Figure 16 Variation of efficiency at different operating loads.

Figure 17 System power output and component heat losses.

58

The grid diagram representation, shown in Figure 18, shows the heat transfer
operation of the optimum HEN. Hot streams are at the top running from left to
right, while cold streams are located at the bottom running from right to left.
Heat exchange matches are represented by vertical lines joining two arrows on
the two streams being matched.

Figure 18 Configuration after the application of the simultaneous HEN optimization.

Figure 19 Configuration of the proposed micro-CHP system after the application of the
simultaneous HEN optimization.

59

The coupling of the optimized HEN configuration with the micro-CHP system
is shown in Figure 19. The new configuration is significantly different than the
original one, due to the cost reduction requirements imposed in the objective
function. Therefore, the optimum configuration encompasses the need for heat
exchangers with a cheaper structure, such as a minimum heat transfer area. The
heat exchangers must allow a more cost efficient heat transfer distribution
along the streams, although the total number of heat exchangers remains the
same.
Overall, the efficiency was improved from an initial value of 31.9% to an
optimum value of 35.2% (at full-load). The highest net electrical efficiency
was 43.6%, while the highest total system efficiency was 91.1%. The MINLP
optimization strategy, managed to reach a minimized HEN total cost of
$8,147/year. The new HEN configuration is significantly different than the
initial one, indicating the level of improvement made after the application of
the optimization strategy. Finally, it should be noted that a more complete
analysis should include part-load results for the new HEN configuration. This
can be accomplished by modification of the original micro-CHP system
simulation model, to verify that HEN operability is indeed feasible.

60

Conclusions and Future Work

This thesis work has defined and completed a series of tasks to address the
main research question associated with the modeling, design and optimization
of a HT-PEMFC-based micro-CHP system. Components and subsystems are
modeled from first principles or modified previously developed models are
used to build up the total simulation system model. The obtained simulation
system models (see papers I-II) are validated successful using a comparable
model from the literature (Korsgaard et al., 2008).

5.1 Conclusions
A posteriori, it can be concluded that the structure of the five study topics
could have been completed in a slightly different order. After topic #2, topic #3
could have been left as the final topic of the research project. This is because
the system is optimized during topics #4-5, suggesting that better results could
have been obtained if an optimum system configuration was used to reach an
improved operational strategy. Further on, the first study topic proved to be
very time consuming and not so influential in the overall progression of the
research project. Therefore, if the project was repeated, this part would have
been omitted. If the two modeling methods, LabVIEWTM and EES-based
simulations, are compared, it can be concluded that EES modeling is easier and
possesses more capabilities, as illustrated throughout this research work.
Moreover, although LabVIEWTM simulations can provide a high degree of
easiness to the end user in terms of calculating manipulation, it is difficult to
model and modify highly complicated models, due to the graphical nature of
the program. Also LabVIEWTM has a limited number of parameters that can be
transferred from a subVI to the main VI (Virtual Instrument). Therefore,
LabVIEWTM modeling is more appropriate for single component dynamic
systems.

61

To provide a specific answer to the main research question, the following

points are used to summarize the benefits of the proposed system:
High efficiencies. The simulated system provides the ability of
reaching high net electrical system efficiencies 35 to 44% (see papers
IV-V). Also high overall efficiencies are obtained (up to 91% when
component heat losses are included). As compared to non-fuel cellbased micro-CHP systems, namely ICE and SE types, the proposed
system is not a heat engine, and thereby non-Carnot-limited.
Low CO2 emissions. This is due to the high efficiency obtainable by
the proposed system which reduces the fuel consumption
requirement, while the presence of a catalytic burner ensures no
traces of unburned fuel are left in the atmosphere.
Simplified fuel processor. The PBI fuel cell technology allows the
intake of lower quality hydrogen-rich reformate fuel. This fact
suggests a simpler fuel processing subsystem can be coupled, with
only a SMR and a single-stage temperature WGS reactor. This is
because the PBI-based fuel cell stack allows carbon monoxide
contents up to 5% (Serenergy, 2011).
Operational modularity. The proposed system can operate efficiently
within a wide range of operation.
Nafion-based systems can exhibit slightly higher fuel cell efficiencies.
Nevertheless, very high quality reformate gas is required, which makes them
less favorable when pure hydrogen is unavailable (as in the application under
study). This in turn increases the capital cost for building these systems.
Finally, heat engine-based systems are in general more mature technologies
and possess a lower capital cost. These are their main advantages, as compared
to the proposed system. Nevertheless, their heat-engine nature, with low
operational performances makes them unattractive for future development.
Finally, the proposed micro-CHP is capable of achieving high efficiencies,
which are comparable with the most efficient centralized systems (CCGT).
Nevertheless, further development is needed, in terms of system reliability and
realistic design. The most important barrier for the adoption of the system
under study is the high lifecycle cost. It could be argued that a solution with a
lower investment cost, than the proposed system, would be a purely network
grid/electric heat pump combination. But, in this scenario, the electrical load
demand would increase rapidly, since all the heating and electrical load
demands would be satisfied with electricity. Therefore, the increased need for
electricity would result in congestion of the distribution networks during peak
times. The possibility for network congestion will increase in the future, since
the increasing production of electricity by renewable energy sources, e.g. wind

62

power, is largely unpredictable and uncontrollable. In other words, there is a

mismatch between generation and demand. The proposed system can diminish,
or at least lower, the possibility of the aforementioned unpleasant situation,
since the greatest portion of the required heating and electricity demands are
produced and consumed onsite.

5.2 Future Work

All research study topics initially planned were completed successfully, but a
number of topics can be investigated further in future studies. The most
important topics are summarized below:
A greater input on the reliability of both the fuel cell stack and the
fuel processing subsystem is still needed. Great care should be given
in relation to operation of the system under harsh conditions for long
periods. These may include high amounts of carbon monoxide in the
reformate fuel, different steam-to-carbon ratios and operational
temperatures.
The development of a dynamic simulation model, based on
experimental data, for the micro-CHP system could provide input on
the effect of load variation, including system start-up and shut down,
on its performance.
The effect of a possible cooling load demand could be investigated
with the development of a heat pump subsystem, and subsequent
coupling with the micro-CHP system (Georgopoulos, 2002; Gunes &
Ellis, 2003).
Thermoeconomic optimization and exergy analysis could provide
insights into the overall synthesis, design, and operational problem,
since it accounts both for the quantity and quality of all energy
conversions present in a process. Its objective is to analyze the
judicious expenditure of exergy to reduce not just fuel costs, but also
total costs (Rancruel, 2005).

63

References
Ahlstrom-Silversand, A. F., & Odenbrand, C. U. I. (1999). Modelling catalytic
combustion of carbon monoxide and hydrocarbons over catalytically
active wire meshes. Chemical Engineering Journal, 73(3), 205-216.
doi:10.1016/S1385-8947(99)00029-7
Al-Sulaiman, F. A., Hamdullahpur, F., & Dincer, I. (2011). Trigeneration: A
comprehensive review based on prime movers. International Journal of
Energy Research, (March 2010), 233-258. doi:10.1002/er
Ang, S. M. C., Fraga, E. S., Brandon, N. P., Samsatli, N. J., & Brett, D. J. L.
(2011). Fuel cell systems optimisation Methods and strategies.
International Journal of Hydrogen Energy, 36(22), 14678-14703.
Elsevier Ltd. doi:10.1016/j.ijhydene.2011.08.053
Arsalis, A. (2007). Thermoeconomic modeling and parametric study of hybrid
SOFCgas turbinesteam turbine power plants ranging from 1.5 to 10
MWe. M.S. thesis, advisor: M.R. von Spakovsky, Virginia Polytechnic
Institute & State University, Blacksburg, Virginia.
Arsalis, A. (2008). Thermoeconomic modeling and parametric study of hybrid
SOFCgas turbinesteam turbine power plants ranging from 1.5 to 10
MWe. Journal of Power Sources, 181(2), 313-326.
doi:10.1016/j.jpowsour.2007.11.104
Autissier, N., Palazzi, F., Marechal, F., van Herle, J., & Favrat, D. (2007).
Thermo-Economic Optimization of a Solid Oxide Fuel Cell, Gas Turbine
Hybrid System. Journal of Fuel Cell Science and Technology, 4(2), 123.
doi:10.1115/1.2714564
BAXI. (2011a). Baxi SenerTec Dachs - Internal Combustion Engine. Retrieved
November 11, 2011, a from http://www.baxi.co.uk/products/DACHS.htm
65

BAXI. (2011b). Baxi Innotech. Retrieved October 21, 2011, b from

http://www.baxi-innotech.de
Barbir, F. (2005). PEM Fuel Cells: Theory and Practice. Elsevier.
Barclay, F. J. (2006). Fuel Cells, Engines and Hydrogen: An Exergy Approach
(1st ed.). Chichester, UK: Wiley.
Beckers, G. J. (2006). Development of a DCHP unit based on a Free Piston
Stirling Engine. Retrieved November 20, 2011, from
http://www.dgs.de/uploads/media/06_Ger_Beckers_ENATEC.pdf
Biegler, L. T., Grossmann, I. E., & Westerberg, A. W. (1997). Systematic
Methods of Chemical Process Design (1st ed.). Prentice Hall.
Brown, M. (2011). Cogeneration growth in Europe - Smaller systems have
better prospects. Cogeneration & On-Site Power Production, (JulyAugust), 13-17.
Bchi, F. N., Inaba, M., & Schmidt, T. J. (2009). Polymer Electrolyte Fuel Cell
Durability (pp. 81-82; 199-221; 447-463). New York, NY, USA:
Springer. doi:10.1007/978-0-387-85536-3
Calise, F., Dentice dAccadia, M., Palombo, A., & Vanoli, L. (2006).
Simulation and exergy analysis of a hybrid Solid Oxide Fuel Cell
(SOFC)Gas Turbine System. Energy, 31(15), 3278-3299.
doi:10.1016/j.energy.2006.03.006
Calise, F., Dentice dAccadia, M., Vanoli, L., & von Spakovsky, M. R. (2006).
Single-level optimization of a hybrid SOFCGT power plant. Journal of
Power Sources, 159(2), 1169-1185. doi:10.1016/j.jpowsour.2005.11.108
Callux. (2009). Callux, Praxistest Brennstoffzelle frs Eigenheim. Retrieved
February 2, 2011, from http://www.callux.net
Ceramic Fuel Cells Ltd. (2010). Fuel Cell Module for Highly Efficient
Electricity. Retrieved November 20, 2011, from
http://www.cfcl.com.au/Assets/Files/Gennex_Brochure_(ENG)_April_20
10.pdf
Charbonneau, P. (2002). An Introduction to Genetic Algorithms for Numerical
Optimization. Boulder, Colorado: High Altitude Observatory, National

66

Center for Atmospheric Research. Retrieved from

http://www.hao.ucar.edu/modeling/pikaia/pikaia.php
Chicco, G., & Mancarella, P. (2009). Distributed multi-generation: A
comprehensive view. Renewable and Sustainable Energy Reviews, 13(3),
535-551. doi:10.1016/j.rser.2007.11.014
Chong, E. K. P., & Zak, S. H. (2001). An Introduction to Optimization (2nd
ed.). New York, NY, USA: Wiley.
ClearEdge Power. (2011). ClearEdge5. Retrieved November 20, 2011, from
http://clearedgepower.com/sites/default/files/Commercial DS 05.26.11
v2.pdf
DONG Energy. (2011). Gas-fired power plants. Retrieved November 24, 2011,
from http://www.dongenergy.com
Danish Micro-CHP. (2009). Danish Micro Combined Heat and Power.
Retrieved August 3, 2011, from http://www.dk-mchp.eu/
De Paepe, M., & Mertens, D. (2007). Combined heat and power in a liberalised
energy market. Energy Conversion and Management, 48(9), 2542-2555.
doi:10.1016/j.enconman.2007.03.019
Delta Energy & Environment Ltd. (2010). Micro-CHP Markets Status and
Outlook. In D. Morgado (Ed.), Danish micro-CHP Workshop.
Snderborg, Denmark.
Dentice dAccadia, M., Sasso, M., Sibilio, S., & Vanoli, L. (2003). Microcombined heat and power in residential and light commercial
applications. Applied Thermal Engineering, 23(10), 1247-1259.
doi:10.1016/S1359-4311(03)00030-9
Dijkstra, S. (2010). Micro-CHP: Edging towards the mass market.
Cogeneration and On-Site Power Production, (August 2009), 1-4.
Energetix Group. (2010). Energetix Genlec Ltd unveils Kingston micro-CHP
boiler. Retrieved November 21, 2011, from
http://online.hemscottir.com/servlet/HsPublic?context=ir.access&ir_optio
n=RNS_NEWS&publicationGroup=RNS&item=374407331581355&ir_
client_id=5778

67

Energy Efficiency News. (2011). Vaillant and Honda launch micro-CHP

systems for homes. Retrieved November 21, 2011, from
http://www.energyefficiencynews.com/co-generation/i/3789/
Ferguson, A. (2004). Fuel cell modelling for building cogeneration
applications. Journal of Power Sources, 137, 30-42.
doi:10.1016/j.jpowsour.2004.05.021
Georgopoulos, N. G. (2002). Application of a Decomposition Strategy to the
Optimal Synthesis/Design and Operation of a Fuel Cell Based Total
Energy System. M.S. thesis, advisor: M.R. von Spakovsky, Virginia
Polytechnic Institute & State University, Blacksburg, Virginia.
Godat, J., & Marechal, F. (2003). Optimization of a fuel cell system using
process integration techniques. Journal of Power Sources, 118(1-2), 411423. doi:10.1016/S0378-7753(03)00107-1
Grossmann, I. E. (2004). Part II. Future perspective on optimization.
Computers & Chemical Engineering, 28, 1193-1218.
doi:10.1016/j.compchemeng.2003.11.006
Gummert, G. (2008). Baxi mCHP solutions: Heating and Electricity today and
tomorrow. Gas Industry mCHP Workshop. Paris, France.
Gunes, M. B., & Ellis, M. W. (2003). Evaluation of Energy, Environmental,
and Economic Characteristics of Fuel Cell Combined Heat and Power
Systems for Residential Applications. Journal of Energy Resources
Technology, 125(3), 208. doi:10.1115/1.1595112
Hawkes, A. D., & Leach, M. A. (2005a). Impacts of temporal precision in
optimisation modelling of micro-Combined Heat and Power. Energy,
30(10), 1759-1779. doi:10.1016/j.energy.2004.11.012
Hawkes, A. D., & Leach, M. A. (2005b). Solid oxide fuel cell systems for
residential micro-combined heat and power in the UK: Key economic
drivers. Journal of Power Sources, 149, 72-83.
doi:10.1016/j.jpowsour.2005.01.008
Hawkes, A. D., & Leach, M. A. (2007). Cost-effective operating strategy for
residential micro-combined heat and power. Energy, 32(5), 711-723.
doi:10.1016/j.energy.2006.06.001

68

Hawkes, A. D., & Leach, M. A. (2008). The capacity credit of micro-combined

heat and power. Energy Policy, 36(4), 1457-1469.
doi:10.1016/j.enpol.2007.12.022
Hawkes, A. D., & Leach, M. A. (2009). Modelling high level system design
and unit commitment for a microgrid. Applied Energy, 86(7-8), 12531265. Elsevier Ltd. doi:10.1016/j.apenergy.2008.09.006
Hawkes, A. D., Aguiar, P., Croxford, B., Leach, M. A., Adjiman, C., &
Brandon, N. P. (2007). Solid oxide fuel cell micro combined heat and
power system operating strategy: Options for provision of residential
space and water heating. Journal of Power Sources, 164(1), 260-271.
doi:10.1016/j.jpowsour.2006.10.083
Hawkes, A. D., Aguiar, P., Hernandez-Aramburo, C., Leach, M. A., Brandon,
N. P., Green, T., & Adjiman, C. (2006). Techno-economic modelling of a
solid oxide fuel cell stack for micro combined heat and power. Journal of
Power Sources, 156(2), 321-333. doi:10.1016/j.jpowsour.2005.05.076
Hubert, C., Achard, P., & Metkemeijer, R. (2006). Study of a small heat and
power PEM fuel cell system generator. Journal of Power Sources,
156(1), 64-70. doi:10.1016/j.jpowsour.2005.08.022
Inui, Y., Yanagisawa, S., & Ishida, T. (2003). Proposal of high performance
SOFC combined power generation system with carbon dioxide recovery.
Energy Conversion and Management, 44, 597-609.
Jahn, H., & Schroer, W. (2005). Dynamic simulation model of a steam
reformer for a residential fuel cell power plant. Journal of Power
Sources, 150, 101-109. doi:10.1016/j.jpowsour.2005.02.012
Jannelli, E., Minutillo, M., & Galloni, E. (2007). Performance of a Polymer
Electrolyte Membrane Fuel Cell System Fueled With Hydrogen
Generated by a Fuel Processor. Journal of Fuel Cell Science and
Technology, 4(4), 435. doi:10.1115/1.2756568
Kehlhofer, R., Bachmann, R., Nielsen, H., & Warner, J. (1999). Combined
Cycle Gas & Steam Turbine Power Plants (2nd ed.). Tulsa, Oklahoma:
PennWell.
Keiski, R. L., Salmi, T., Niemisto, P., Ainassaari, J., & Pohjola, V. J. (1996).
Stationary and transient kinetics of the high temperature water-gas shift
reaction. Applied Catalysis A: General, 137(2), 349-370.
doi:10.1016/0926-860X(95)00315-0
69

Kemp, I. C. (2007). Pinch Analysis and Process Integration: A User Guide on

Process Integration for the Efficient Use of Energy (2nd ed.).
Butterworth-Heinemann.
Kim, K., von Spakovsky, M. R., Wang, M., & Nelson, D. J. (2010). A hybrid
multi-level optimization approach for the dynamic synthesis/design and
operation/control under uncertainty of a fuel cell system. Energy.
Elsevier Ltd. doi:10.1016/j.energy.2010.08.024
Kolb, G. (2008). Fuel Processing for Fuel Cells (1st ed.). Weinheim,
Germany: Wiley-VCH.
Korsgaard, A. R. (2011). Personal communication concerning water-content in
the fuel cell stack anode inlet. Aalborg, Denmark.
Korsgaard, A. R., Nielsen, M. P., & Kr, S. K. (2008). Part one: A novel
model of HTPEM-based micro-combined heat and power fuel cell
system. International Journal of Hydrogen Energy, 33(7), 1921-1931.
doi:10.1016/j.ijhydene.2008.01.008
Korsgaard, A. R., Nielsen, M. P., Bang, M., & Kr, S. K. (2006). Modeling of
CO Influence in PBI Electrolyte PEM Fuel Cells. FUELCELL2006.
Irvine, CA.
Korsgaard, A. R., Refshauge, R., Nielsen, M. P., Bang, M., & Kr, S. K.
(2006). Experimental characterization and modeling of commercial
polybenzimidazole-based MEA performance. Journal of Power Sources,
162(1), 239-245. doi:10.1016/j.jpowsour.2006.06.099
Larminie, J., & Dicks, A. L. (2003). Fuel Cell Systems Explained (2nd ed.).
Chichester, UK: Wiley.
Linnhoff, B., & Flower, J. R. (1978a). Synthesis of heat exchanger networks: I.
Systematic generation of energy optimal networks. AIChE Journal,
24(4), 633-642. doi:10.1002/aic.690240411
Linnhoff, B., & Flower, J. R. (1978b). Synthesis of heat exchanger networks:
II. Evolutionary Generation of Networks with Various Criteria of
Optimality. AIChE Journal, 24(4), 642-654.
Linnhoff, B., & Hindmarch, E. (1983). The pinch design method for heat
exchanger networks. Chemical Engineering Science, 38(5), 745-763.

70

Lund, P. (2008). Analysis of advanced energy chain typologies. International

Journal of Energy Research, (May 2007), 144-153. doi:10.1002/er
Mench, M. M. (2008). Fuel Cell Engines (1st ed.). Hoboken, New Jersey:
Wiley.
OHayre, R., Colella, W., Cha, S.-W., & Prinz, F. B. (2009). Fuel Cell
Fundamentals (2nd ed.). Hoboken, New Jersey: Wiley.
Palazzi, F., Autissier, N., Marechal, F., & Favrat, D. (2007). A methodology
for thermo-economic modeling and optimization of solid oxide fuel cell
systems. Applied Thermal Engineering, 27, 2703-2712.
doi:10.1016/j.applthermaleng.2007.06.007
Pehnt, M., Cames, M., Fischer, C., Praetorius, B., Schneider, L., Schumacher,
K., & Voss, J.-P. (2006). Micro Cogeneration: Towards Decentralized
Energy Systems. Cogeneration. Springer.
Ponce-Ortega, J., Jimenez-Gutierrez, A., & Grossmann, I. E. (2008). Optimal
synthesis of heat exchanger networks involving isothermal process
streams. Computers & Chemical Engineering, 32(8), 1918-1942.
doi:10.1016/j.compchemeng.2007.10.007
Praetorius, B., Bauknecht, D., Cames, M., Fischer, C., Pehnt, M., Schumacher,
K., & Voss, J.-P. (2009). Innovation for Sustainable Electricity Systems.
Change (1st ed.). Heidelberg, Germany: Physica-Verlag.
Rancruel, D. F. (2005). Dynamic Synthesis/Design and Operation/Control
Optimization Approach applied to a Solid Oxide Fuel Cell based
Auxiliary Power Unit under Transient Conditions. PhD Dissertation,
advisor: M.R. von Spakovsky, Virginia Polytechnic Institute & State
University, Blacksburg, Virginia.
Raven, R., & Verbong, G. (2007). Multi-Regime Interactions in the Dutch
Energy Sector: The Case of Combined Heat and Power Technologies in
the Netherlands 1970-2000. Technology Analysis & Strategic
Management, 19(4), 491-507. doi:10.1080/09537320701403441
Ravindran, A., Ragsdell, K. M., & Reklaitis, G. V. (2006). Engineering
Optimization: Methods and Applications (2nd ed.). Hoboken, New
Jersey: Wiley.

71

Ropenus, S., Schrder, S. T., Costa, A., & Ob, E. (2010). Support Schemes
and Ownership Structures: The Policy Context for Fuel Cell Based
Micro-Combined Heat and Power. Ris, Denmark. Retrieved November
25, 2011, from
http://www.bigscience.dk/sitecore/content/Risoe_dk/Home/Knowledge_b
ase/publications/Reports/ris-r-1730.aspx?sc_lang=en
Salcines, D. S., Estbanez, C. R., & Herrero, V. C. (2004). Simulation of a
solar domestic water heating system with different collector efficiencies
and different volumen storage tanks. International Conference on
Renewable Energies and Power Quality. Barcelona.
Serenergy. (2011). serenergy.dk. Retrieved February 2, 2011, from
http://www.serenergy.dk/
Sieniutycz, S., & Jezowski, J. (2009). Energy Optimization in Process Systems
(1st ed.). Amsterdam: Elsevier.
Slowe, J. (2010). Can small make it big? The progress of micro-CHP in world
mass markets. Retrieved November 20, 2011, from
http://www.cospp.com/articles/print/volume-7/issue-4/features/cansmall-make-it-big-the-progress-of-micro-chp-in-world-massmarkets.html
Smith, R. (2005). Chemical Process Design and Integration. Chichester, UK:
Wiley.
Stolten, D. (2010). Hydrogen and Fuel Cells: Fundamentals, Technologies and
Applications (1st ed.). Weinheim, Germany: Wiley-VCH.
Stolten, D. (2011). Personal communication concerning water-content in the
fuel cell stack anode inlet. Aalborg, Denmark.
Thomas, B. (2008). Benchmark testing of Micro-CHP units. Applied Thermal
Engineering, 28(16), 2049-2054.
doi:10.1016/j.applthermaleng.2008.03.010
Tokyo Gas Ltd. (2011). Tokyo Gas and Panasonic to Launch New Improved
Ene-Farm Home Fuel Cell with World-Highest Generation Efficiency
at More Affordable Price. Retrieved November 20, 2011, from
http://www.tokyo-gas.co.jp/Press_e/20110209-01e.pdf

72

Vaillant. (2011). Vaillant Ecopower. Retrieved November 20, 2011, from

http://www.vaillant.de/Presse/Press-Releases/article/Vaillant_first_fullrange_supplier_for_combined_heat_and_power_in_the_small_output_ra
nge.html
Verda, V., & Nicolin, F. (2010). Thermodynamic and economic optimization
of a MCFC-based hybrid system for the combined production of
electricity and hydrogen. International Journal of Hydrogen Energy,
35(2), 794-806. Elsevier Ltd. doi:10.1016/j.ijhydene.2009.10.104
Viswanathan, J., & Grossmann, I. E. (1990). A combined penalty function and
outer approximation method for MINLP optimization. Computers &
Chemical Engineering, 14(7), 769-782.
Vogel, J., & Tsou, Y.-M. (2007). Plug Power: International Stationary Fuel
Cell Demonstration.
Wallmark, C., & Alvfors, P. (2002). Design of stationary PEFC system
configurations to meet heat and power demands. Journal of Power
Sources, 106, 83-92. doi:10.1016/S0140-6701(03)90808-3
Weber, C., Marechal, F., Favrat, D., & Kraines, S. (2006). Optimization of an
SOFC-based decentralized polygeneration system for providing energy
services in an office-building in Tky. Applied Thermal Engineering,
26(13), 1409-1419. doi:10.1016/j.applthermaleng.2005.05.031
Westner, G., & Madlener, R. (2011). Development of cogeneration in
Germany: A mean-variance portfolio analysis of individual technologys
prospects in view of the new regulatory framework. Energy, 36(8), 53015313. Elsevier Ltd. doi:10.1016/j.energy.2011.06.038
Woudstra, N., van der Stelt, T. P., & Hemmes, K. (2006). The Thermodynamic
Evaluation and Optimization of Fuel Cell Systems. Journal of Fuel Cell
Science and Technology, 3(2), 155. doi:10.1115/1.2174064
Xu, J., & Froment, G. F. (1989a). Methane Steam Reforming, Methanation and
Water-Gas Shift: I. Intrinsic Kinetics. AIChE Journal, 35(1), 88-96.
Xu, J., & Froment, G. F. (1989b). Methane Steam Reforming: II. Diffusional
Limitations and Reactor Simulation. AIChE Journal, 35(1), 97-103.

73

Yee, T. F., & Grossmann, I. E. (1990). Simultaneous optimization models for

heat integration-II. Heat exchanger netwrok synthesis. Computers &
Chemical Engineering, 14(10), 1165-1184.
Zhang, Jianlu, Xie, Z., Zhang, J., Tang, Y., Song, C., Navessin, T., Shi, Z., et
al. (2006). High temperature PEM fuel cells. Journal of Power Sources,
160, 872-891. doi:10.1016/j.jpowsour.2006.05.034
Zink, F., Lu, Y., & Schaefer, L. (2007). A solid oxide fuel cell system for
buildings. Energy Conversion and Management, 48(3), 809-818.
doi:10.1016/j.enconman.2006.09.010

74

Appendix
The appendix includes all journal articles mentioned in this thesis work (I-V).
Contents
Journal
Journal
Journal
Journal
Journal

Article
Article
Article
Article
Article

I
3
II 15
III 29
IV 47
V 61

APPENDIX

APPENDIX

Journal Article I

APPENDIX

APPENDIX

Energy 36 (2011) 993e1002

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Modeling and off-design performance of a 1 kWe HT-PEMFC

(high temperature-proton exchange membrane fuel cell)-based
residential micro-CHP (combined-heat-and-power) system for
Danish single-family households
Alexandros Arsalis*, Mads P. Nielsen, Sren K. Kr
Department of Energy Technology, Aalborg University, Pontoppidanstrde 101, 9220 Aalborg , Denmark

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 10 May 2010
Received in revised form
25 November 2010
Accepted 7 December 2010
Available online 8 January 2011

A novel proposal for the modeling and operation of a micro-CHP (combined-heat-and-power) residential
system based on HT-PEMFC (High Temperature-Proton Exchange Membrane Fuel Cell) technology is
described and analyzed to investigate its commercialization prospects. An HT-PEMFC operates at
elevated temperatures, as compared to Naon-based PEMFCs and therefore can be a signicant candidate for cogeneration residential systems. The proposed system can provide electric power, hot water,
and space heating for a typical Danish single-family household. A complete fuel processing subsystem,
with all necessary BOP (balance-of-plant) components, is modeled and coupled to the fuel cell stack
subsystem. The micro-CHP system is simulated in LabVIEW environment to provide the ability of Data
Acquisition of actual components and thereby more realistic design in the future. A part-load study has
been conducted to indicate performance characteristics at off-design conditions. The system is sized to
provide realistic dimensioning of the actual system.
2010 Elsevier Ltd. All rights reserved.

Keywords:
PBI membrane
HT-PEMFC systems
Micro-CHP
Residential systems
Fuel processing

1. Introduction
Fuel cell-based stationary power generation offers a great
market opportunity, because the fuel cell technology is capable of
achieving higher efciencies, with lower emissions as compared to
conventional power systems [1e3]. Residential fuel cell systems
can be grid-interconnected to allow power ow from/to the grid as
needed. This design offers greater exibility than a stand-alone
system, and is very attractive if the incoming power is produced by
renewable energy sources, such as wind power [3,4]. This means
that when cheap wind power can be produced, the fuel cell system
can operate at a minimum load and therefore reduce the fuel
consumption. A fuel cell-based micro-CHP (Combined-Heat-andPower) system converts on-site the chemical energy in a fuel, e.g.
natural gas, into electrical power and heat, as required by the
household demand. The range of energy demand for a Danish
household in terms of electricity and heat is 0.5e5.0 kWe, and
2e10 kWth, respectively. Therefore, for this system conguration,
a fuel-processing unit must be coupled with the fuel cell stack, to

* Corresponding author. Tel.: 45 9940 9240; fax: 45 9815 1411.

E-mail address: aar@et.aau.dk (A. Arsalis).

allow conversion of natural gas to hydrogen. Several BOP (BalanceOf-Plant) components are needed for the controlling and smooth
operation of the system, while heat exchangers are necessary for
the thermal management of the system. The thermal management
of the system includes heating/cooling of components (e.g. steam
reforming), and also heat recovery to satisfy the residential load
prole (e.g. space heating).
The operating temperature in a fuel cell stack is considered an
important factor to the efciency and the degradation of the
membrane. High operating temperatures reduce the cooling
requirements, simplify water management and lessen contamination problems. An HT-PEMFC (high temperature-proton exchange
membrane fuel cell) utilizes a PBI (Polybenzimidazole) membrane,
operating at temperatures between 160
C and 200
C. An HTPEMFC is therefore an ideal match for a micro-CHP system, because
not only the rates of electrochemical kinetics are enhanced and
water management and cooling is simplied, but also useful waste
heat can be recovered, and lower quality reformed hydrogen may
be used as fuel [5]. However, the frequent changes of demand, in
terms of electrical power and/or heat, require operation in varying
partial-load conditions. A simple and compact design, with efcient
adaptability to load changes must be accomplished, which is vital
for this type of application.

0360-5442/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2010.12.009

APPENDIX

994

A. Arsalis et al. / Energy 36 (2011) 993e1002

The Danish micro-CHP project is managed by a national consortium consisting of nine Danish energy companies. The consortium combines all the competencies necessary to develop, test, and
demonstrate micro-CHP systems. The project cooperates with
a wide range of political and specialist parties, such as the Danish
Energy Agency. It is also nancially assisted by the Danish Ministry
of Climate and Energy by 40% of the total cost of the project. It is
scheduled to run from 2006 to 2012, and includes fuel cell systems
based on three different technologies: LT-PEMFC (low temperatureproton exchange membrane fuel cell), HT-PEMFC and SOFC (Solid
Oxide Fuel Cell). It is divided into three phases, and two fuel types
are used: hydrogen and natural gas. The choice of fuel depends on
the fuel cell technology and its availability at the installation sites.
A preliminary assumption concerning the operational control of the
system, indicated that the electrical load must be fullled, but also
the design should secure the right amount of heat will be available
on demand for the space heating and hot water loads. Further on,
an auxiliary burner for peak production of thermal energy is under
consideration to be integrated to the end-user system [3,6].
The HT-PEMFC-based micro-CHP units are developed by Dantherm
Power and operate on natural gas. Experimental tests, with pure
hydrogen fuel, showed start-up times of 30e60 min, with an electrical
efciency of 40e58% (LHV (Lower Heating Value)-based). From the
operating point of the nalized design, an efciency of 50% is expected.
Further experimental tests and calculations showed a potential system
efciency of 85e90% (LHV-based) [6]. A fast adaption to load variations
was also observed. In the rst phase, Danish Micro Combined-Heatand-Power developed micro-CHP unit prototypes with PEMFC and
SOFC fuel cells. In the second phase, ten micro-CHP units are scheduled

to be installed and tested at selected consumer households in the

municipalities of Lolland and Snderborg, Denmark. In the third and
last phase, micro-CHP systems will be installed and demonstrated at
around 100 households in the two aforementioned municipalities.
This will allow the Danish micro-CHP system project to gather and
analyze realistic experiences related to installation, operational
procedures, maintenance needs and also consumer satisfaction. These
will be used for further system improvement [6].
2. System layout
The proposed micro-CHP system, shown in Fig. 1, is modeled in
LabVIEW to provide easiness in user usage of the program, and to
allow future experimental testing capabilities with Data Acquisition
hardware. Fig. 2, shows an indicative front panel of the fuel cell stack
subsystem. In order to synthesize and design the proposed microCHP system, the energy requirements for a representative residential building must be established. For the current research study, the
representative residential building is a typical Danish single-family
household (130 m2 house with four persons: two adults and two
children). A micro-CHP unit of approximately 1 kWe is considered.
The types of residential loads considered are the following:
(a) Electrical load;
(b) Space heating load;
(c) Hot water load.
Averaged daily power load curves for all three loads were
measured for 25 houses at a 15 min average time segment [4]. For

Fig. 1. Schematic of the proposed micro-CHP system.

APPENDIX

A. Arsalis et al. / Energy 36 (2011) 993e1002

995

Fig. 2. LabVIEW front panel of the Fuel Cell Stack subVI.

the purposes of this research study, the entire load prole, shown in
Table 1, has been simplied into three time segments, i.e. winter,
summer, and spring (autumn).
By observation, it is clear that the winter segment appears to be
the one with the most stringent for electrical and space heating
loads, while hot water load is most stringent during spring season.
Therefore the system must be designed to fulll the maximum
winter electrical load 0.95 kWe, since the proposed micro-CHP
system assumes an electricity-led operation. It should be noted that
at short time segments the electrical power demand will exceed
1 kWe. In this case, electrical power will be imported from the grid.
On the other hand, the remaining heat load demand can be satised
with external means, e.g. auxiliary burner. It should be noted that
these are beyond the scope of this study, and only the percentage of
heating load fullled from the proposed system is analyzed.
The various efciencies should be dened, before the modeling
details are given in the next section. The cogeneration thermal efciency is dened as the ratio of the actual waste heat recovered by heat
exchangers IV, V, and VI and the chemical energy input to the system,

hthermal cogen:

Q_ HEx

cogen:

(1)

_ CH4 ;in LHVCH4

m

The corresponding electrical efciency is dened as the ratio of

the net electrical power output of the system divided by the
chemical energy input to the system,
Table 1
Mean residential load requirements for a typical Danish household.
Residential load type

Time segment
Winter

Electrical load (We)

Space heating load (Wth)
Hot water load (Wth)

Summer

Spring

Mean

Max

Mean

Max

Mean

Max

540
1450
330

950
1930
1080

380
70
230

650
140
1420

460
600
310

920
1070
1620

helectrical net

_
W
electrical net
_ CH4 ;in LHVCH4
m

(2)

The combined cogeneration system efciency is dened as the

sum of the net electrical power output of the system and the actual
waste heat recovered divided by the chemical energy input to the
system,

hsystem


cogen:

_
W
electrical

net

Q_ HEx

cogen:

(3)

_ CH4 ;in LHVCH4

m

3. Modeling of the micro-CHP system

The system is designed initially at nominal load to investigate
operating behavior and geometric analysis of the most important
system components. All necessary components/subsystems, with
emphasis on the fuel processing subsystem, are modeled in detail
based on semi-empirical assumptions from the literature to allow
a realistic analysis of the system parameters. Then the system is
simulated at part-load conditions to investigate its behavior at offdesign and the effects of an electricity-led operation on the total
heat load fulllment. All the individual models are described in this
section in terms of input and output parameters and specic
assumptions for every particular model under study are also given.
The operational principle of the proposed system, shown in
Fig. 1, is the following: Natural gas is desulfurized at an acceptable
level; hydrogen sulde content must be removed to values below
1 ppm [7]. Water is pumped from the line into the steam generator,
where it is superheated with ue gas exhausted by the shell-side of
the SMR (steam methane reformer) reactor. Superheated steam and
preheated natural gas are mixed in the mixer, and then the steamfuel mixture is fed to the tube-side of the SMR reactor. The inputs of
the catalytic combustor are air, methane (necessary at start-up),
and reformate fuel (depleted fuel from the anode of the fuel cell

APPENDIX

996

A. Arsalis et al. / Energy 36 (2011) 993e1002

stack). Its main purpose is to provide heat for the endothermic

steam methane reforming reaction.
In the SMR reactor, the reforming reaction takes place. It should
be noted that the SMR process is based on reaction kinetics, and is
modeled as a PFR (plug ow reactor) with the presence of a catalyst,
with experimental data taken from [8]. In terms of heat exchanging,
the SMR is modeled as a shell-and-tube heat exchanger, where the
shell part carries the hot ue gas, and the tube part carries the
vaporized methane-water mixture. The produced hydrogen-rich
synthesis gas is mixed with water and enters the WGS (water gas
shift) reactor, which is modeled similarly to the SMR. It serves as
a reduction of the carbon monoxide content, which is converted to
carbon dioxide, into a tolerable level for the fuel cell stack. It also
contributes in further hydrogen production.
The reformate fuel enters the anode side of the fuel cell stack,
where a part of it is depleted and used in the combustor for the
SMR reaction described above. In the cathode of the fuel cell stack,
air enters and used for the fuel cell reaction converting the
chemical energy of reformate and air into electricity. A by-product
of this reaction is the hot mixture, which exits from the fuel cell
stack. This mixture is used in a plate heat exchanger to satisfy
a part of the residential heat load. A part of the residential heat
load is also satised by the exhausted ue gas at the natural
gas pre-heater. Flue-gas is then passed through open-loop and
closed-loop heat exchangers for water heating and space heating,
respectively.
3.1. Fuel cell stack
The fuel cell stack model is based on [3,9]. Initially, the model
calculates the required hydrogen feed based on the number of cells
specied in the model. At part-load the model is modied to have
the feed reformate fuel (hydrogen-rich gas) as an input to the
model, in order to allow the calculation of power output at offdesign conditions. Similarly, the current density is constant at fullload (0.2 cm2/A) and a calculated variable at part-load conditions.
The following assumptions were made:
 The model considers only the reaction of hydrogen with
oxygen (air). All other substances are considered inactive.
 The reformate fuel after the WGS Reactor/Plate Heat Exchanger
stage passes through a condenser/knock-out stage, where all
the water content in the fuel is removed. Nevertheless, some
water-content is considered because the moisture content in
the incoming air is included in the calculations.
 The temperature of the incoming reformate fuel is 160
C,
which is the fuel cell stack operating temperature.
 The cell active area is 49 cm2.
 Heat losses are neglected in the calculations.
The ohmic and diffusion resistances are given by,

Rohmic 0:0001667Tcell 0:2289

(4)

Rdiff 0:4306  0:0008203Tcell

(5)

hohmic iRohmic
The total cell voltage is given by,

Vcell V0  ha  hc  hohmic




i i
i
hc
Rdiff
ln 0
lair  1
4acathode F
i0
RTcell

Wcell Vcell i

(10)

The stack voltage is given by,

Vstack Vcell ncells

(11)

The fuel cell stack electrical power output is given by,

Pstack Acell ncells Wcell

(12)

Finally the current is dened as,

Istack

Pstack
Vstack

(13)

3.2. SMR reactor

The kinetic model of the SMR reactor is described analytically in
[9]. The assumed reactor material and its properties are described
in detail in [8]. The following assumptions were made for the SMR
reactor:
 The reaction kinetics is based on a generalized LangmuireHinshelwood kinetic model [8].
 The model considers a steady-state non-isothermal PFR, with
the presence of catalytic material in a PBR (packed-bed reactor)
arrangement.
 The SMR reactor is heat integrated, as shown in Fig. 3, with
a shell-and-tube heat exchanger design, where the external
shell-side wall is considered adiabatic.
 The reactor material is assumed inactive during the catalytic
reaction.
For the current research study only the following reactions are
considered:

DH298 206:2kJ=mol

CH4 H2 O#CO 3H2

CO H2 O#CO2 H2

DH298 41:1kJ=mol

CH4 2H2 O#CO2 4H2

DH298 164:9kJ=mol

(14)
(15)
(16)

The heat transfer model is based on [10], where the shell-andtube heat integration model is based on a transient model. For the
current research study, the model was simplied in a steady-state
one, where the partial differential equations are reduced into
ordinary differential equations. The convective heat transfer coefcient of the inside tube wall is based on a semi-empirical relation
for spherical packing [11].
The heat transfer coefcient for a packed-bed tubular reactor is
obtained from the following empirical correlation,

hw 2:03Re0:8
p



kref
6dp
exp 
Di
Di

(17)

while the shell-side heat transfer coefcient is dened as,

The Ohmic losses are given by,

(6)

(9)

The power density is given by,

The anode and cathode overpotentials are given by,



RTcell
i
ha
arcsinh
aanode F
2keh qH2

(8)

(7)




 


Deq Gs 0:55 Cp;s ms 1=3 ms 0:14
ks
hs 0:36
ms
mwall
Deq
ks

APPENDIX

(18)

A. Arsalis et al. / Energy 36 (2011) 993e1002

997

Fig. 3. Schematic of the SMR reactor, modeled as a shell-and-tube heat exchanger.

The ue gas temperature gradient is obtained by the energy

balance in the shell-side,

phs Do Tw  Ts
dTs

_ s Cp;s
dz
m

a plate heat exchanger is its compactness. These heat exchangers

fulll the following system needs:
 Fuel pre-heater: Flue gas, exhausted from the steam generators ue gas side, preheats the natural gas, before being mixed
with the superheated steam in the mixer.
 Water heater: The exhaust water mixture produced from the
fuel cell stack is used to heat tap water at a temperature high
enough for domestic use.
 Space heating heater: The exhaust water mixture produced
from the fuel cell stack is used to heat the uid (water mixture)
in a closed circuit for space heating.
 High temperature reformate fuel cooler: it is located between
the SMR and the WGS reactors to cool the reformate gas exiting
the SMR reactor. The reformate gas must reach a lower
temperature before entering the WGS reactor. On the cold-side
of the heat exchanger, the air absorbs the heat before entering
the combustor.
 Low temperature reformate fuel cooler: It is located after the
WGS reactor and cools the reformate gas with ambient air. This
is necessary to ensure the reformate fuel is at a low temperature before entering the condenser, and subsequently the fuel
cell stack. On the cold-side of the heat exchanger the ambient
air is heated, and then send to the high temperature reformate
fuel cooler for further heating, as explained above.

(19)

while the inside tube wall temperature is dened as,

Tw

Do hs Ts Di hw Tref
Do hs Di hw

(20)

3.3. WGS Reactor

The inlet composition for the WGS reactor is the sum of the exit
composition of the SMR reactor and the injected water. The heat
exchanger placed prior to the WGS reactor inlet, cools the WGS
inlet temperature to an appropriate level, as needed by the WGS
reactor. The CO (carbon monoxide) content, in terms of molar
composition, should be reduced to an acceptable level of 0.1%e0.2%.
The kinetic constant, based on a power law relationship, is at steady
state. The type of the converting catalyst and its deactivation are
described in detail in [12]. The kinetic model of the WGS reactor is
described in detail in [9,12]. The WGS reaction is,

CO H2 O#CO2 H2

(21)

The equilibrium constant given in [13],

KT exp



4400
 4:063
T

(22)

The kinetic power law t parameters are obtained from [12],



Ea
RT

KWGS k0 exp

The modeling of the heat exchangers is based on simple heat and

energy balances, where the heat transfer rate and the (overall heat
transfer coefcient-area)-value needed to fulll the heat exchange
for the given mass ow rates and inlet temperatures are calculated.

(23)
3.5. Mixers & by-pass valves

The CO extent of reaction is given by,

dxCO
3600rCO
urwiremesh
n_ CO;i
dz

(24)

The mixers and by-pass valves used in the micro-CHP system are
necessary for the operation and the regulation of the system. The
modeling of each mixer and by-pass valve is done with simple mass
and energy balances, assuming no pressure losses.

(25)

3.6. Steam generator

while the temperature gradient is dened as,

rwiremesh
dT

dz
rg cp;mass us  rCO dHr;1
3.4. Plate heat exchangers

In the current conguration several heat exchangers of the plate

type are used, where all ows are unmixed. The main advantage of

The steam generator model is divided into three sections:

economizer, evaporator, and superheater (see Fig. 4). In the water/
steam side of the SG (steam generator), water is pumped in the
economizer section by the water pump, and undergoes phase
changes until it becomes superheated steam. The steam is then

APPENDIX

998

A. Arsalis et al. / Energy 36 (2011) 993e1002

Fig. 4. Schematic of the Steam Generator.

mixed with preheated fuel in the mixer, at the same temperature. In

the ue gas side, ue gas exhausted from the SMR reactor is used to
heat the water following the reverse path, as compared to the water
side. The exhausted ue gas is then used to preheat the methane in
the fuel pre-heater.
3.7. Combustor
Although the primary purpose of the combustor is to produce heat
for the steam reforming reaction in the SMR reactor, it can be used to
produce additional heat to supplement the heat production from the
fuel cell stack. In this manner, the residential heat demand can be
fullled. Depleted fuel (hydrogen-rich gas) from the fuel cell is combusted with air in the catalytic combustor and the following assumptions concerning the operation of the combustor have been made:
 The ue gas temperature is regulated to remain constant at
1100 K. Therefore depending on the amount of the fuel, the
amount of air is regulated accordingly. The effect of NOxcontent has not been investigated in the current model.
 The ue gas ow rate is tested whether it is adequate to fulll
the SMR reaction, otherwise more fuel is needed to be combusted. In the latter case, natural gas from the line can be used
through the relevant by-pass valve.
 The air used in the combustor is twice preheated by two plate
heat exchangers, before and after the WGS reactor.
The relevant stoichiometric combustion reactions are given
below:

H2 0:5O2 3:76N2 /H2 O 0:53:76N2

(26)

CH4 2O2 3:76N2 /CO2 2H2 O 7:52N2

(27)

_ pump
_ pump
m
m
out
in

(28)



_ pump m
_ in hout  hin pump
W

(29)

_ in is the non-pressurized mass ow rate entering the

m
_ out the pressurized mass ow rate exiting the pump,
m
the pump work rate consumption, and hpump
; and hpump
out
in
are the corresponding enthalpies for the mass ow. The pressure
boost is around 0.2 bar, assuming that every component has
a pressure loss of 0.015 bar. This assumption is done in order to
have a fuel cell stack pressure (reformate fuel in the anode) slightly
above atmospheric, close to 1.05 bar.
where
pump,
_ pump
W

4. System validation
The validation of the SMR reactor, the WGS reactor, the HTPEMFC stack, and the overall system were compared with reference
models from the literature [3]. Before validation is explained, it
should be noted that mass balances for every component input/
output were performed; the mass balances veried the validity of
the results. All three individual subsystems showed only minor
discrepancies between the reference models and the models under
study.
Although an analytical comparison between the research under
study and a reference system is not possible, due to signicant
differences between the current system and the systems in the
literature, some values can be compared to investigate validity, and
also the sources of discrepancy. The chosen reference system is the
HT-PEMFC-based micro-CHP system developed by [3]. The results
are given in Table 2. The most signicant differences between the
two systems are the following:

3.8. Water pump

The water pump is used to pump water to the steam generator,
the WGS reactor (water injection), and to the water heater (plate
heat exchanger after the fuel cell stack exit). Since the thermodynamic states in the inlet are known and the outlet thermodynamic
states can be xed as desired, what is left is a calculation of the
pump power consumed. The corresponding mass and energy
balances are given by:

 Different heat integration techniques in the SMR reactor. The

reference system considers a simple heat integration model
without any consideration of the geometric inuence of a realistic
system. The current research study considers a shell-and-tube
heat integration model as explained in the preceding sections.
 Different chemical kinetics modeling in the SMR reactor. The
reference system considers a continuous stirred-tank reactor
(CSTR), based on chemical equilibrium, where the research
project under study considers a PFR, based on reaction kinetics,
as explained in the preceding section.

Table 2
Overall system validation.
Input values
Fuel cell active area, Acell
Results
_ net
Electrical power output, W
Korsgaard, 2008
Proposed system

10

135 cm2

Number of cells

65

Methane input, LHVCH4

1500 W

666 W
669 W

Net electrical efciency

Korsgaard, 2008
Proposed system

0.45
0.4455

Total system efciency

Korsgaard, 2008
Proposed system

0.88
0.95

APPENDIX

A. Arsalis et al. / Energy 36 (2011) 993e1002

999

Table 3
Simulation results for full-load and part-load operation.
Node

T(
C)

Full-load operation
1
10.0
7
602.5
8
250.0
9
296.6
10
160.0
11
160.0
12
160.0
16
85.3
17
820.0
18
698.3
19
597.2
20
592.5
21
469.4
22
313.6
23
10.0
25
200.0
27
60.0
30
60.0
33
90.8
75%-load operation
1
10.0
7
677.2
8
250.0
9
323.9
10
160.0
11
160.0
12
160.0
16
88.3
17
718.0
18
678.2
19
593.0
20
589.0
21
422.6
22
316.2
23
10.0
25
200.0
27
60.0
30
60.0
32
160.0
33
64.0
50%-load operation
1
10.0
7
588.3
8
250.0
9
290.2
10
160.0
11
160.0
12
160.0
16
60.6
17
610.0
18
588.6
19
514.5
20
511.0
21
328.9
22
95.3
23
10.0
25
200.0
27
60.0
30
60.0
32
160.0
33
94.6
25%-load operation
1
10.0
7
577.6
8
250.0
9
285.5
10
160.0
11
160.0
12
160.0
16
56.7
17
585.0
18
577.6

_
mkg=s

xCH4

xCO

xCO2

xO2

xN2

xair

xH2 O

xH2

5.530E-05
2.587E-04
2.587E-04
3.259E-04
3.259E-04
1.523E-04
4.569E-05
4.768E-03
4.841E-03
4.841E-03
4.841E-03
4.841E-03
4.841E-03
4.841E-03
4.871E-03
2.034E-04
6.000E-04
4.000E-03
1.695E-03

1.000
0.031
0.031
0.026
0.026
0.043
0.043
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.007

0.000
0.048
0.048
0.003
0.003
0.005
0.005
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.001

0.000
0.088
0.088
0.112
0.112
0.187
0.187
0.000
0.006
0.006
0.006
0.006
0.006
0.006
0.000
0.000
0.000
0.000
0.031

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.195
0.195
0.195
0.195
0.195
0.195
0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.777
0.777
0.777
0.777
0.777
0.777
0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.809

0.000
0.336
0.336
0.402
0.402
0.000
0.000
0.000
0.022
0.022
0.022
0.022
0.022
0.022
1.000
1.000
1.000
1.000
0.152

0.000
0.498
0.498
0.457
0.457
0.764
0.764
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

3.440E-05
1.609E-04
1.609E-04
2.047E-04
2.047E-04
1.044E-04
3.132E-05
3.538E-03
3.589E-03
3.589E-03
3.589E-03
3.589E-03
3.589E-03
3.589E-03
1.170E-03
1.265E-04
4.000E-04
2.000E-03
1.224E-03
1.224E-03

1.000
0.007
0.007
0.006
0.006
0.009
0.009
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.001
0.001

0.000
0.076
0.076
0.006
0.006
0.010
0.010
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.002
0.002

0.000
0.077
0.077
0.122
0.122
0.190
0.190
0.000
0.005
0.005
0.005
0.005
0.005
0.005
0.000
0.000
0.000
0.000
0.030
0.030

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.197
0.197
0.197
0.197
0.197
0.197
0.000
0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.778
0.778
0.778
0.778
0.778
0.778
0.000
0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.813
0.813

0.000
0.306
0.306
0.356
0.356
0.000
0.000
0.000
0.020
0.020
0.020
0.020
0.020
0.020
1.000
1.000
1.000
1.000
0.153
0.153

0.000
0.534
0.534
0.509
0.509
0.791
0.791
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

2.730E-05
1.277E-04
1.277E-04
1.602E-04
1.602E-04
7.176E-05
2.153E-05
3.314E-03
3.345E-03
3.345E-03
3.345E-03
3.345E-03
3.345E-03
3.345E-03
4.733E-03
1.004E-04
2.600E-04
1.800E-03
7.782E-04
7.782E-04

1.000
0.041
0.041
0.035
0.035
0.060
0.060
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.010
0.010

0.000
0.041
0.041
0.003
0.003
0.004
0.004
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.001
0.001

0.000
0.088
0.088
0.106
0.106
0.184
0.184
0.000
0.004
0.004
0.004
0.004
0.004
0.004
0.000
0.000
0.000
0.000
0.031
0.031

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.199
0.199
0.199
0.199
0.199
0.199
0.000
0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.781
0.781
0.781
0.781
0.781
0.781
0.000
0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.807
0.807

0.000
0.355
0.355
0.423
0.423
0.000
0.000
0.000
0.015
0.015
0.015
0.015
0.015
0.015
1.000
1.000
1.000
1.000
0.152
0.152

0.000
0.475
0.475
0.433
0.433
0.751
0.751
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

1.350E-05
6.314E-05
6.314E-05
7.904E-05
7.904E-05
3.447E-05
1.034E-05
1.708E-03
1.722E-03
1.722E-03

1.000
0.048
0.048
0.041
0.041
0.072
0.072
0.000
0.000
0.000

0.000
0.036
0.036
0.002
0.002
0.004
0.004
0.000
0.000
0.000

0.000
0.088
0.088
0.103
0.103
0.183
0.183
0.000
0.004
0.004

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.200
0.200

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.782
0.782

0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000

0.000
0.367
0.367
0.436
0.436
0.000
0.000
0.000
0.014
0.014

0.000
0.461
0.461
0.418
0.418
0.742
0.742
0.000
0.000
0.000

(continued on next page)

APPENDIX

11

1000

A. Arsalis et al. / Energy 36 (2011) 993e1002

Table 3 (continued )
Node

T(
C)

_
mkg=s

xCH4

xCO

xCO2

xO2

xN2

xair

xH2 O

xH2

19
20
21
22
23
25
27
30
32
33

506.2
502.9
148.3
93.3
10.0
200.0
60.0
60.0
160.0
90.9

1.722E-03
1.722E-03
1.722E-03
1.722E-03
4.666E-03
4.964E-05
1.300E-04
9.000E-04
3.673E-04
3.673E-04

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.012
0.012

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.001
0.001

0.004
0.004
0.004
0.004
0.000
0.000
0.000
0.000
0.031
0.031

0.200
0.200
0.200
0.200
0.000
0.000
0.000
0.000
0.000
0.000

0.782
0.782
0.782
0.782
0.000
0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.805
0.805

0.014
0.014
0.014
0.014
1.000
1.000
1.000
1.000
0.152
0.152

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

Table 4
Overall power-to-heat performance for the adopted load prole.

Fig. 5. Current density and HT-PEMFC efciency vs. Load.

 Some other differences in the layout of the system components

can be noted as well; a steam generator and fuel pre-heater are
modeled in the current system under study, where a mixture
vaporizer was modeled in the reference model. Also different
heating/cooling between streams is used (e.g. air coolers, etc.).
Nevertheless of the differences between the two systems, the
validation has shown a good agreement of the two models, verifying the validity of the research project in study.
5. Results and discussion: On-design and off-design system
performance
Off-design operational conditions and simulation at these
conditions provides valuable information on the operation of the
component/system, particularly on its range of applicability.
Therefore, it is necessary to analyze the amount of electricity and
heat produced by the micro-CHP system, in terms of size, part-load

Fig. 6. Electrical Load vs. Total Cogeneration Heat Load.

12

Load

_
W
electrical net

Q_ HEx cogen:

Total cogeneration
Heat load fullled (%)

Winter, mean
Winter, max
Summer, mean
Summer, max
Spring, mean
Spring, max

540
950
380
650
460
920

840
1560
580
1000
700
1470

47.2
51.8
193.0
64.1
76.9
54.6

characteristics, etc. The results for all the nodes, shown in Fig. 1, are
given with the following parameters: temperature, mass ow rate,
and mole fractions. In Table 3, the results for full-load, 75%, 50% and
25%-load operation are given. The fuel cell stack operates at
a constant temperature of 160
C to avoid fast degradation of the
membrane, if operated at higher temperatures [14,15]. Also the
temperature is kept constant, because rapid and frequent variations
of the operating temperature would contribute to faster deterioration of the membrane [5,14]. The corresponding fuel cell operating pressure is kept at a value slightly above atmospheric, since an
increase of pressure would not contribute signicantly in the
system efciency, and also the degradation of the membrane would
be faster [14,15].
The current density variation for the given loads and the HTPEMFC stack efciency (LHV) variation for the given loads are given
in Fig. 5. The current density is varied from 0.19 A/cm2 at full-load
linearly up to 0.04 A/cm2 at 25% load. The efciency follows an
analogous path, increasing at part-load to 45.5%.
The net electrical power output at full-load and part-loads vs.
the corresponding cogeneration heat loads is shown in Fig. 6. The
systems useful heat capacity was analyzed in Section 2. For this
load prole the net electrical power output, along with the corresponding total cogeneration heat output are given in Table 4. In the

Fig. 7. Mass ow variation of hydrogen, methane and carbon monoxide at full-load

operation.

APPENDIX

A. Arsalis et al. / Energy 36 (2011) 993e1002

6. Conclusions

Table 5
Component geometries for the proposed micro-CHP system.
Geometric variable
description
ncells
DSMR
i
DSMR
o

Number of cells
SMR inlet
tube diameter
SMR outlet
tube diameter

Value

Geometric variable
description

Value

176
0.1016 m

LSMR
DWGS
i

0.8 m
0.13 m

0.1322 m

LWGS

SMR tube length

WGS tube
diameter
WGS tube length

0.45 m

same table, the percentage of heat output fullling the aforementioned load prole at every time segment is also given. As expected,
the combined heat load is satised better in the summer segment.
It should be noted that there would still be a decit of heat at some
instances in the summer if no heat storage was provided. In the case
of the other two time segments, winter and spring, an auxiliary
burner will be required at some instances, even if a heat storage
system is used. More specically for the winter segment, the heat
load can be fullled at around 50% in most instances. The system
should avoid operation at part-loads below 20% because then
higher selected nominal cell voltage does not necessarily mean
higher operating efciency; at very low current densities where
parasitic losses (including gas permeation through the polymer
membrane) may not be negligible [16].
The mass ow variation of hydrogen, methane, and carbon
monoxide throughout the proposed micro-CHP system is shown in
Fig. 7. Mole fractions have not been used (although it could be done
with mole fractions at dry basis), because it would appear as if the H2
content has decreased at the exit of the WGS reactor. This is because
of the water addition at the reactors inlet, which alters the total
mass ow rate (and therefore the mole fractions) exiting the WGS
reactor. By observation the SMR reactor is very efcient, converting
hydrogen to 2.01E-5 kg/s for a methane input of 5.52E-5 kg/s. This
suggests that the current conguration is very efcient and appropriate for the application under study, provided this can be proved
with experimental calculations in the future. The CO content is
reduced from 2.71E-5 kg/s, at the SMR outlet, to 0.21E-5 kg/s. The
HT-PEMFC stack is assumed to withstand CO-contents in the inlet
reforming fuel of up to 10,000 ppm. The CO-content was maintained
within this limit at full- and part-load. At full-load it was calculated
to be 5480 ppm (see Table 3). This result veries that the current
design of the fuel processing system, although simple, it can be
adequate for the needs of the fuel cell stack, in terms of CO-content
removal.
The HT-PEMFC stack, SMR reactor and the WGS reactor geometries for the proposed micro-CHP system are given in Table 5. The
UA-values for the heat exchangers used in the proposed micro-CHP
system are given in Table 6. These component values satisfy
a compact, yet efcient micro-CHP system, as required for this
application [4].
Table 6
UA-values [W/(m2-
C)] for the heat exchangers used in the proposed micro-CHP
system.
Geometric variable
description
UAI
UAII
UAIII
UASG,sup
UASG,ev

Natural gas
pre-heater
SMR/WGS cooler

Value

Geometric variable
description

0.0538784 UAIV
0.720725

UAV

WGS/HT-PEMFC
0.548661
cooler
Superheater
0.107073
(steam generator)
Evaporator
0.843147
(steam generator)

UAVI
UASG,ec

1st Cogeneration
heat exchanger
2nd Cogeneration
heat exchanger
3rd Cogeneration
heat exchanger
Economizer
(steam generator)

1001

Value
1.41641
2.36245
1.39233
0.0896212

A proposed micro-CHP system, with an applied load prole for

a Danish household, was presented to indicate the high potential of
HT-PEMFC technology as a candidate system in decentralized
systems. It can provide a basis for further experimental and data
acquisition testing in the developing process of the end-user
product. An electrical efciency varying from 45.4% (25%-load) to
38.8% (full-load) was calculated. The corresponding total efciency
was around 95.2%.
One of the most important advantages of the proposed system is
the simplicity of the fuel processing system. An equivalent LTPEMFC or SOFC fuel processing subsystem is more complex because
it requires extra components, such as dual temperature WGS
reactors, or PrOx devices in order to be suitable for the relevant fuel
cell stack [2]. Simplicity is necessary, because the micro-CHP
system must maintain a high level of compactness. The fuel processing system provides a high rate of methane conversion and
acceptable CO-removal, making it appropriate for integration with
an HT-PEMFC stack. Inaccuracies were inevitable, especially in the
fuel processing system, due to the lack of matching semi-empirical
functions, which are due in the future, for small scale fuel processors for fuel cell systems. The modeling used in the SMR and WGS
reactors is based on reaction kinetics, which provide more realistic
results and also it can be easy to adjust the modeling when new
experimental kinetic data become available.
The obtained efciencies for the proposed system, veried the
great potential of the HT-PEMFC technology in micro-CHP systems.
A lot of ground for improvement is necessary before the end-user
system is available; the modeling results have to be veried by
experimental testing to investigate sources of discrepancy and
other hurdles, such as incompatibility of the subsystems. As a novel
technology, HT-PEMFC-based micro-CHP systems have the disadvantage of lacking in operational data. This means that it can enter
the CHP market in a gradual process (as explained in the introduction), with extended testing of the units in real conditions to
investigate sources of malfunction and possible operational failures. Therefore, although the proposed technology is highly
promising, in terms of efciency, maintenance, and performance in
general, it has yet to be proved that the actual efciencies, operating and maintenance costs will be within the limits set by projected ones to maintain competitiveness with the other micro-CHP
candidates. Once more data are available from the demonstration
systems of the Danish micro-CHP project, more realistic conclusions will be drawn on the full potential of the proposed system.
Moreover, one of the main barriers of the proposed technology,
that should be eliminated, is the currently high installation cost of
the units. The installation cost of micro-CHP units can have a strong
impact on the commercialization prospects of the technology, since
new units will be installed on existing or new households. In the
latter case, the replacement of gas-red furnaces should be justied
with signicantly higher efciencies, which in turn correspond to
reduced operating costs. The competitiveness of the generated
electricity cost, as compared with electricity grid market, can be
kept high with an efcient operational strategy. In this way excess
electricity will almost never have to be produced (and sold to the
grid). On the other hand, electricity can be purchased from the grid
to supplement electricity shortages from the micro-CHP system
(e.g. during the summer season). This will eliminate the problem of
producing excess heat, which cannot be stored.
It was shown that the system efciency increased at part-load
conditions. The effect of BOP components at part-load is not so
signicant and therefore, even though they operate at lower efciencies as the systems load is reduced, the combination of
increased fuel cell efciency and lower supporting component

APPENDIX

13

1002

A. Arsalis et al. / Energy 36 (2011) 993e1002

efciencies can maintain high efciencies as the load is reduced.

This gives a signicant leverage to the proposed micro-CHP system
as compared to conventional micro-CHP systems, such as Stirling
engines, micro-turbines, etc. The latter will typically experience
a signicant drop-off in efciency at part-load [17]. This gives the
proposed system a fuel cost advantage, since this application
requires a signicant amount of part-load operation.
The current results do not consider an exergy analysis which can
be used to provide a more complete picture in the overall design and
synthesis of the system, since it accounts both for the quantity and
quality of all energy conversions present in a process. In that manner
a more cost-effective and better performing system could be achieved, to reduce not just operational (fuel) costs, but also total (incl.
capital) costs. Also an optimization process needs to be applied to
verify the validity of the results. Specically, the high thermal efciency achieved by the high recovery of exhaust gases might be
unrealistic, if components of reasonable size and efciency (e.g. heat
exchangers) cannot be coupled in the micro-CHP system. When
there is a very high temperature difference between the heat
exchange streams, these have to be matched in order for the nal
temperature of one to be close to the initial temperature of the other.
Also the heat exchange is more efcient if the ow heat capacities of
the streams are similar. Otherwise streams with high heat capacities
might need to be split in several stream ows. Therefore alternative
designs of system/component congurations should be used with
the aid of heat integration methods, such as pinch analysis.
Finally, LabVIEW was chosen as the modeling tool for this
research study for the reasons explained in the introduction.
Although the end result can provide a high degree of easiness to the
program user in terms of calculations, the modeling procedure
proved to be very time-consuming. There is a great difculty in
adjusting and modifying highly complicated models, due to the
graphical modeling nature used in LabVIEW. Also LabVIEW has
a limited number of parameters that can be transferred from
a subVI to the main VI (Virtual Instrument). Therefore, LabVIEW
modeling is more appropriate for single component dynamic
systems. Nonetheless, the current system can provide a good tool
for experimental testing in the future.
Acknowledgments
The authors would like to acknowledge the support of Danfoss and
Dantherm Power throughout the realization of this research study.
References
[1] Arsalis A, von Spakovsky MR, Calise F. Thermoeconomic modeling and parametric study of hybrid solid oxide fuel cell-gas turbine-steam turbine power
plants ranging from 1.5 MWe to 10 MWe. Journal of Fuel Cell Science and
Technology 2009;6(1):011015.
[2] OHayre R, Colella W, Cha S, Prinz FB. Fuel cell fundamentals. Wiley; 2009.
[3] Korsgaard AR, Nielsen MP, Kr SK. Part one: a novel model of HTPEM-based
micro-combined heat and power fuel cell system. International Journal of
Hydrogen Energy 2008;33(7):1909e20.
[4] Korsgaard AR, Nielsen MP, Kr SK. Part two: control of a novel HTPEM-based
micro combined heat and power fuel cell system. International Journal of
Hydrogen Energy 2008;33(7):1921e31.
[5] Zhang J, Xie Z, Tang Y, Song C, Navessin T, Shi Z, et al. High temperature PEM
fuel cells. Journal of Power Sources 2006;160(2):872e91.
[6] Pedersen AH, Balslev P. Demonstration of mCHP based on Danish fuel cells.
Lucerne, Switzerland: European Fuel Cell Forum 2009; 2009.
[7] Kolb G. Fuel processing for fuel cells. Weinheim, Germany: Wiley-VCH; 2008.
[8] Xu J, Froment GF. Methane steam reforming, methanation and water-gas
shift: 1. intrinsic kinetics. AIChE Journal 1989;35(1):88e96.
[9] Nielsen MP. Modeling of proton exchange membrane fuel cell systems. Aalborg, Denmark: PhD Dissertation, Aalborg University; 2005.

14

[10] Kim K. Dynamic proton exchange membrane fuel cell system synthesis/design
and operation/control optimization under uncertainty. Blacksburg, VA: PhD
Dissertation, Virginia Tech; 2008.
[11] Li C, Finlayson BA. Heat transfer in packed bedsea reevaluation. Chemical
Engineering Science 1977;32:1055e66.
[12] Keiski RL, Salmi T, Niemisto P, Ainassaari J, Pohjola VJ. Stationary and transient
kinetics of the high temperature water-gas shift reaction. Applied Catalysis A:
General 1996;137(2):349e70.
[13] Davies J, Lihou D. Optimal design of methane steam reformer. Chemical and
Process Engineering 1971;52:71e80.
[14] Korsgaard AR, Refshauge R, Nielsen MP, Bang M, Kr SK. Experimental
characterization and modeling of commercial polybenzimidazole-based MEA
performance. Journal of Power Sources 2006;162(1):239e45.
[15] Bchi FN, Inaba M, Schmidt TJ. Polymer electrolyte fuel cell durability. New
York, NY, USA: Springer; 2009.
[16] Barbir F. PEM fuel cells: theory and practice. Elsevier; 2005.
[17] Peacock A, Newborough M. Impact of micro-CHP systems on domestic
sector CO emissions. Applied Thermal Engineering 2005;25(17e18):
2653e76.

Nomenclature
Acell: Fuel cell active area (cm2)
Cp,s: Specic heat of shell-side gas in SMR (kJ/(kmol.K))
dp: Catalyst diameter, SMR reactor (m)
dHr,1: Enthalpy of reaction (kJ/mol)
Di: Inlet diameter (m)
Do: Outlet diameter (m)
Deq: Equivalent diameter (m)
F: Faraday constant (C/mol)
Gs: Mass ux (kg/(m2.s))
hs: Shell-side heat transfer coefcient (W/m2.K)
hw: Tube-side heat transfer coefcient (W/m2.K)
i: Current density (A/cm2)
Istack: Fuel cell stack current (A)
i0: Exchange current density (A/cm2)
keh: H2 electro-oxidation rate constant (A/cm2)
ki: Thermal conductivity of species i (W/K.m)
KT: Equilibrium constant
KWGS: Kinetic power law t parameter (m^1.92/kmol^-0.36.kg.s)
LHVi: Lower heating value of species i (J/kg)
_ i : Mass ow rate of species i (kg/s)
m
ncells: Number of cells in the fuel cell stack
n_ CO : CO molar feed rate (kmol/h)
Pstack: Fuel cell stack electrical power output (W)
Q_ i : Heat transfer rate of component i (W)
rCO: CO reaction rate (kmol/kg.s)
R: Ideal gas constant (J/K.mol)
Rdiff: Diffusion resistance (Ohm.cm2)
Rohmic: Ohmic resistance (Ohm.cm2)
Rep: Reynolds number
Tcell: Fuel cell operating temperature (oC)
Ti: Temperature of species i (oC)
Vcell: Total cell voltage (V)
Vstack: Fuel cell stack voltage (V)
V0: Open circuit voltage (V)
us: Supercial velocity (m/s)
Wcell: Power density (W/m2)
_
W
electrical : Electrical power output (W)
z: Distance along the reactor length (m)
ai: Charge transfer coefcient
ha: Anode overpotential (V)
hc: Cathode overpotential (V)
hohmic: Ohmic losses (V)
hi: Efciency of component/system i
qH2 : H2 adsorption/desorption constant
lair: Air stoichiometry, cathode side
mi: Viscosity of species i (kg/s.m)
rg: Density of the reformate gas (kg/m3)
rwiremesh: Density of the wiremesh catalytic material (kg/m3)
u: WGS reactor cross-sectional area (m)
Subscripts/superscripts
cogen.: Cogeneration
HEx: Heat exchanger
in: Input stream
out: Output stream
ref: Reformate gas
s: Flue gas
thermal: Thermal power

APPENDIX

Journal Article II

APPENDIX

15

16

APPENDIX

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Modeling and parametric study of a 1 kWe HT-PEMFC-based

residential micro-CHP system
A. Arsalis*, Mads P. Nielsen, Sren K. Kr
Department of Energy Technology, Aalborg University, Pontoppidanstrde 101, 9220 Aalborg , Denmark

article info

abstract

Article history:

A detailed thermodynamic, kinetic and geometric model of a micro-CHP (Combined-Heat-

Received 27 September 2010

and-Power) residential system based on High Temperature-Proton Exchange Membrane

Received in revised form

Fuel Cell (HT-PEMFC) technology is developed, implemented and validated. HT-PEMFC

17 January 2011

technology is investigated as a possible candidate for fuel cell-based residential micro-CHP

Accepted 22 January 2011

systems, since it can operate at higher temperature than Nafion-based fuel cells, and

Available online 2 March 2011

therefore can reach higher cogeneration efficiencies. The proposed system can provide
electric power, hot water, and space heating for a typical Danish single-family household.

Keywords:

A complete fuel processing subsystem, with all necessary balance-of-plant components, is

PBI

modeled and coupled to the fuel cell stack subsystem. The micro-CHP systems synthesis/

HT-PEMFC system

design and operational pattern is analyzed by means of a parametric study. The parametric

Micro-CHP

study is conducted to determine the most viable system/component design based on

Residential system

maximizing total system efficiency, without violating the requirements of the system. Four

Parametric study

decision variables (steam-to-carbon ratio, fuel cell operating temperature, combustor

Fuel cell system

temperature and hydrogen stoichiometry) were parameterized within feasible limits to

provide insight on their effect on the overall performance of the proposed system under
study and also to provide input on more efficient design in the future. The system is
designed to provide maximum loads of 1 kWe and 2 kWth. A sensitivity analysis is applied
to investigate the influence of the most important parameters on the simulated performance of the system.
Copyright 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.

1.

Introduction

Stationary fuel-cell-based systems are designed to convert the

chemical energy in a fuel into both electrical power and useful
heat. In a residential application of the combined-heat-andpower technology, a residential micro-CHP system provides
electricity and heat (space heating and hot water) for a building
[1]. Fuel cell-based stationary power generation technology
is a very promising one because it is capable of achieving
higher efficiencies, with lower emissions as compared to

conventional power systems. Further on, fuel cell residential

systems have simple routine maintenance requirements, quiet
operation, and low emissions as compared to conventional
systems [2]. Combustion-based systems, such as the internal
combustion engine technology, are not suitable for micro-CHP
applications mainly due to their high thermal-to-energy ratio
(TER). Recent improvements and developments in the fuel cell
technology research area have reduced the cost of the fuel cell
technology making it the primary candidate for residential
cogeneration applications.

* Corresponding author.
E-mail address: aar@et.aau.dk (A. Arsalis).
0360-3199/$ e see front matter Copyright 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2011.01.121

APPENDIX

17

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

Since hydrogen is not readily available, a fossil fuel is

generally used in the operation of the fuel cell system. In the
Danish micro-CHP infrastructure gas-fired furnaces can be
replaced with fuel cell-based micro-CHP systems [3]. In this
case, a fuel-processing unit must be coupled with the fuel cell
stack, to allow conversion of natural gas to hydrogen. Residential fuel cell systems can be grid-interconnected to allow power
flow from/to the grid as needed. This design offers greater
flexibility than a stand-alone system, and is very attractive if the
incoming power is produced by renewable energy sources, such
as wind power [4,5]. This means that when cheap wind power
can be produced, the fuel cell system can operate at a minimum
load and therefore reduce the fuel consumption.
During fuel cell development, high operating temperature is always preferred. High operating temperatures
reduce the cooling requirements, cogeneration of heat and
electricity for stationary applications becomes more efficient, the contamination problem is lessened and water
management is easier. Further on, CO tolerances up to 3%
can be achieved [6] An HT-PEMFC utilizes a PBI (polybenzimidazole) membrane, operating at temperatures
between 160
Ce200
C. These membrane exchange areas
are therefore especially suitable for reformed-hydrogenbased polymer electrolyte fuel cells. No external humidification of the gases is necessary, which leads to a significant
reduction of complexity and cost. [6].

1.1.

Background and research motivation

Several research groups investigated the modeling and operation of fuel cell-based micro-CHP systems throughout the
recent years. Georgopoulos [7] studied the design, synthesis
and operational optimization of a PEMFC-based micro-CHP
system using a decomposition methodology, while considering different system configurations. Ferguson and Ugursal
[8] studied operating strategies for different PEMFC sizes and
their effect on the performance of a cogeneration system,
while demonstrating that these are critical factors affecting
the performance of these systems. Obara [9] studied the
heating network of residential fuel cell-based micro-CHP
systems showing how the hot-water piping network can be
optimized. Godat and Marechal [10] applied process integration techniques on a PEMFC-based micro-CHP to identify its
optimal operating conditions along with the fuel cells optimal
process structure. Braun [11] studied a SOFC-based residential
system in terms of fuel supply and concluded that the efficiency performance advantages of methane-fueled SOFC
systems compared to hydrogen-based SOFC systems can be as
high as 6%.
Finally, Korsgaard et al. [4,5] investigated the modeling
prospects of a higher temperature PEMFC-based micro-CHP
system. The system was modeled in MATLAB Simulink to
enable a dynamic modeling of the system with the application
of different control strategies. Based on the promising results
extracted from this latter research study, the current research
work investigates the modeling of such a high temperature
PEMFC-based micro-CHP system to investigate in greater
detail and precision the design/synthesis of key system
components, such as the reactors, heat exchangers and
combustor. Further on, the modeling of a novel plate heat

18

5011

exchanger SMR reactor and its coupling with the HT-PEMFC

subsystem provides a novel proposal in the field of fuel cellbased micro-CHP systems. Finally, the modeling in Engineering Equation Solver (EES) allows the parametric variation
and sensitivity analysis of key variables to investigate significant optimization trends.

2.

System layout

The proposed micro-CHP system, shown in Fig. 1, is modeled

in EES to provide easiness in the parametric analysis of
selected decision variables. Fig. 1 shows the temperatures and
mass flow rates at nominal condition. The model contains 35
state points (nodes). In order to synthesize and design the
proposed micro-CHP system, the energy requirements for
a representative residential building must be established. For
the current research study, the representative residential
building is a typical Danish single-family household (130 m2
house with four persons) [3]. A micro-CHP unit of approximately 1 kWe is considered. The types of residential loads
considered are the following:
(a) Electrical load;
(b) Space heating load;
(c) Hot water load.
The system is designed to fulfill a maximum electrical load
of 1 kWe, while the maximum thermal load is set at 2 kWe. It
should be noted that at short time segments the load profile
demands exceed these values. In this case, extra electrical
power can be imported from the grid, while the thermal load
can be satisfied by external means, e.g. auxiliary burner.
The systems performance is analyzed in terms of efficiencies defined here; the cogeneration thermal efficiency is
defined as the ratio of the recovered waste heat (used for the
thermal load profiles) and the chemical energy input to the
system,
hther cogen

Q_ HEx cogen
_ CH4 ;in LHVCH4
m

(1)

The net system electrical efficiency is defined as the ratio of

the net electrical power output of the system divided by the
chemical energy input to the system,
hel net

_ el
W
net
_
mCH4 ;in LHVCH4

(2)

Finally, the combined cogeneration system efficiency is

defined as the sum of the two efficiencies above,



hsys

cogen

3.

_ el Q_ HEx
W
net
cogen
_ CH4 ;in LHVCH4
m

(3)

System modeling

EES was selected as the modeling tool for this research study
because it includes many built-in mathematical and thermophysical property functions, while selected input variables

APPENDIX

5012

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

Fig. 1 e Schematic of the proposed micro-CHP system.

can be varied in a parametric table included in EES. The basic

function provided by EES is the numerical solution of a set of
algebraic equations. Also the built-in uncertainty analysis is
used for the sensitivity analysis of the model. System simulation solutions were obtained by solving a total of 1079
equations in EES.
All necessary components/subsystems, with emphasis on
the fuel processing subsystem, are modeled in detail based on
semi-empirical assumptions from the literature to allow
a realistic analysis of the system parameters. All the individual models are described in this section in terms of input
and output parameters, and specific assumptions for every
particular model under study are also outlined. The overall
system assumptions including modeling simplifications are
the following:
 For the purpose of calculation simplicity, natural gas is
assumed to behave as methane.
 Reformate gas is provided to the fuel cell stack at the
required operating fuel cell temperature.
 Heat losses in the system are neglected.
 Complete fuel oxidation is assumed in the combustor.
 The system and component performance are calculated
only for steady-state conditions.
 Pressure drops in heat exchangers are considered to be
0.01 bar for every heat exchanger (hot side/cool side).
 Inverter losses are not accounted in the calculations.
The operational principle of the proposed system, shown
in Fig. 1, is the following: Natural gas is desulfurized at an

acceptable level. Water is pumped from the line into the

steam generator, where it is superheated with flue gas
exhausted by the SMR reactor. Superheated steam and preheated natural gas are mixed in the mixer, and then the
steam-fuel mixture is fed to the SMR reactor. The inputs of the
catalytic combustor are air, methane, and reformate fuel
(depleted fuel from the anode-side of the fuel cell stack). Its
main purpose is to provide heat for the endothermic steam
methane reforming reaction.
The reforming reaction in the SMR reactor is calculated on
the basis of chemical equilibrium with the presence of
a catalytic material. In terms of heat exchanging, the SMR is
modeled as a series of plate heat exchangers, with two inputs/
outputs for the reformate gas and the flue gas. The produced
hydrogen-rich synthesis gas enters the WGS reactor where it
is mixed with water. It serves as a reduction of the carbon
monoxide content into a tolerable level for the PBI fuel cell
material. It also contributes in further hydrogen production.
Prior to entering the fuel cell stack, the water content of the
reforming gas is removed by means of a condenser/knock-out
stage. In the fuel cell stack a part of the fuel is depleted and
used in the combustor, while in the cathode side air enters
and used for the fuel cell reaction converting the chemical
energy of reformate and air into electricity. The hot mixture
by-product of the fuel cell reaction exiting the stack is used in
a plate heat exchanger to satisfy a part of the residential heat
load. The additional heat demand of the residential load is
fulfilled by the exhausted flue gas (natural gas preheater), by
means of open-loop and closed-loop heat exchangers for
water heating and space heating, respectively.

APPENDIX

19

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

3.1.

Fuel cell stack

The fuel cell stack model is based on [4,12,13]. The fuel cell stack
is sized in a design point that will fulfill the required electrical
power output of the micro-CHP system by adjusting the number
of fuel cells in the stack and the current density at the nominal
load. The model considers only the reaction of hydrogen with
oxygen, while all other species are considered inactive.
The ohmic and diffusion resistances are based on the
following linear regressions with experimental values found
in [12],
Rohmic 0:0001667Tcell 0:2289

(4)

Rdiff 0:4306  0:0008203Tcell

(5)

where Tcell is the fuel cell operating temperature.

The anode and cathode overpotentials are given from the
following expressions, respectively,
ha

hc



RTcell
i
1
sinh
aanode F
2keh qH2

(6)





RTcell
i0 i
i
Rdiff
ln
4acathode F
i0
lair  1

(7)

5013

a compact design and therefore better adaptability in the

micro-CHP system. A schematic of the reactor is shown in
Fig. 2. Significantly less catalyst volume is generally required
because of the highly improved utilization of the catalyst (as
compared to a tubular fixed-bed reactor), which originated
from the elimination of mass transfer limitations [14].
The SMR reactor model assumes a one-dimensional
steady-state model, where the governing equations include
energy balances, momentum balances and continuity equations. The model assumes two unmixed working mediums,
the reformate gas and the flue gas, with an inlet and outlet
flow for each. The model assumes ideal gas behavior with the
presence of a wired-mesh catalyst. The wire mesh catalyst
geometry for the modified Ergun equation incorporates the
sphericity of the catalyst used to estimate the reformate side
pressure losses [15]. Enthalpies of reaction are calculated on
a chemical equilibrium based on the law of mass action. The
chemical equilibrium composition calculations are based on:

nie nio

2
X

ni;j xj

i1

where ai is the charge transfer coefficient, lair is the cathode

air stoichiometric ratio, keh is the electrooxidation rate of
hydrogen, qH2 is the surface coverage of hydrogen, i is the
current density, and i0 is the exchange current density.
The ohmic losses are given by,
hohmic iRohmic

(8)

Finally, the total cell voltage is given by,

Vcell V0  ha  hc  hohmic

(9)

where V0 is the open circuit voltage.

The fuel cell mass balance requires that the sum of all mass
inputs must be equal to the sum of all mass outputs. The
inputs are the flows of reformate fuel and air (including the
water vapor present in the air input). For calculation
simplicity, all the water content from the reformate fuel is
removed in the condenser/water knockout stage. The outputs
are the flows of depleted fuel and the fuel cell reaction water/
air mixture exhaust,
X

_i
m


in

X

_i
m

(10)

out

where i are the incoming/outgoing species.

The fuel cell energy balance requires that the sum of all
energy inputs must be equal to the sum of all energy outputs,
X

Hi in WFC;el:

Hi out

(11)

The inputs are the enthalpies of all the flows into the fuel
cell stack, while the outputs are the electric power produced
and enthalpies of all the flows out of the fuel cell.

3.2.

Steam methane reformer (SMR) reactor

An SMR reactor of the plate heat exchanger type [14] filled

with catalyst particles was selected because it allows

20

APPENDIX

Fig. 2 e Schematic description of the SMR Reactor.

(12)

5014

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

where nie is the equilibrium moles of species Sj , nio is the initial

mole number of species Sj , ni;j is the stoichiometric coefficient,
and xj is the extent of reaction.
The model considers only the following reactions:
CH4 H2 O#CO 3H2
CO H2 O#CO2 H2

DH298 206:2kJ=mol
DH298 41:1kJ=mol

(13)
(14)

The heat transfer was calculated as a single total heat

transmission coefficient in each element. The heat conduction of the thin plate material and the plane heat transfer
(axially) in the plates are assumed to be of minor importance
and therefore are neglected in the calculations. It is assumed
the reformate reactants consist of pure methane and water
vapor. The heat exchanger reactor was discretized into 14
individual cells on either side of the heat exchanger. The
number of cells was increased until a sufficient grid consistency was obtained.

3.3.

Water gas shift (WGS) reactor

The inlet composition for the WGS reactor is the sum of the exit
composition of the SMR reactor and the injected water. It has
been proved experimentally [14] that an addition of water in the
WGS reactor can increase the carbon monoxide conversion as
a result of the higher S/C, despite the higher space velocity.
Therefore this method will allow a smaller reactor area, which
is favored for a micro-CHP system design. Thus, when the WGS
reactor is operated at an elevated temperature, will allow size
reduction. Moreover, lesser time and energy demand for startup are required [14]. The CO content should be reduced to an
acceptable level (for the fuel cell stack) of 0.1%e0.2%. The
kinetic constant is based on a power law relationship and
calculated at steady state. The kinetic model of the WGS reactor
is described in detail in [16]. The water gas shift reaction (Eq.
(12)) is moderately exothermic and its thermodynamic equilibrium is calculated according to the following expression,


4400
 4:063
KT exp
T

(15)

divided into three sections: economizer, evaporator and

superheater. In the water/steam side of the SG, water is pumped
in the economizer section by the water pump, and undergoes
phase changes until it becomes superheated steam. The steam
is then mixed with preheated fuel in the mixer, at the same
temperature. In the flue gas side, flue gas exhausted from the
SMR reactor is used to heat the water following the reverse path,
as compared to the water side. The exhausted flue gas is then
used to preheat the methane in the fuel preheater. The
modeling of the heat exchangers is based on the e-NTU method,
with simple heat and energy balances, where the heat transfer
rate and the UA-value needed to fulfill the heat exchange for the
given mass flow rates and inlet temperatures are calculated.
The UA-values for the proposed system are given in Table 1.

3.5.

Combustor

Depleted fuel (hydrogen-rich gas) from the fuel cell and additional natural gas from the line are combusted with air in the
catalytic combustor to fulfill the SMR reaction and the thermal
load residential demand. The adiabatic flame temperature of
the combustor unit is limited by the combustor material. The
amount of fuel input to the system is a model input, while the
air is regulated according to the fuel input with a fixed air
stoichiometry. The effect of NOx-content has not been investigated in the current model.
The mass flow rate of the flue gas must be high enough to
fulfill the required thermal load demand. Since the depleted
fuel from the fuel cell stack anode is not sufficient, more fuel is
added to the combustor from the fuel line throughout a bypass valve. The air used in the combustor is twice preheated
by the two plate heat exchangers situated before and after the
WGS reactor. The relevant stoichiometric combustion reactions are given below:
H2 0:5O2 3:76N2 /H2 O 0:53:76N2

(19)

CH4 2O2 3:76N2 /CO2 2H2 O 7:52N2

(20)

CO 0:5O2 3:76N2 /CO2 0:53:76N2

(21)

where KT is the equilibrium constant given in [17]

The kinetic power law fit parameters are obtained
from [16],

4.



Ea
KWGS k0 exp
RT

(16)

The proposed micro-CHP system must be validated to investigate possible modeling errors and other sources of

(17)

Table 1 e UA-values [W/(m2-oC)] for the heat exchangers

used in the proposed micro-CHP system.

System validation

The CO extent of reaction is given by,

dxCO
3600rCO
urwiremesh
n_ CO;i
dz

Geometric variable description

while the temperature gradient is defined as,

dT
rwiremesh

rg cp;mass us  rCO dHr;1

dz

3.4.

(18)

Plate heat exchangers

In all plate heat exchangers in the proposed micro-CHP system

all flows are unmixed. The main advantage of a plate heat
exchanger is its compactness. The steam generator (SG) is

UAI
UAII
UAIII
UASG;sup
UASG;ev
UAIV
UAV
UAVI
UASG;ec

APPENDIX

Natural gas preheater

SMR/WGS Cooler
WGS/HT-PEMFC Cooler
Superheater (steam generator)
Evaporator (Steam generator)
1st Cogeneration heat exchanger
2nd Cogeneration heat exchanger
3rd Cogeneration heat exchanger
Economizer (steam generator)

Value
0.0538784
0.720725
0.548661
0.107073
0.843147
1.41641
2.36245
1.39233
0.0896212

21

5015

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

Table 2 e Overall system validation.

Input values
135 cm2
175
C
65
1500 W

Fuel cell active area, Acell

Fuel cell operating temperature, Tcell
Number of cells
Methane Input,LHVCH4

Results
_ net
Electrical power output, W
Korsgaard, 2008
Proposed system

Net electrical efficiency

666 W
616 W

Korsgaard, 2008
Proposed system

discrepancy. The validation of the proposed system with

a reference model from the literature cannot be exact, because
of the significant differences in the configuration of the two
systems. The most relevant system from the literature was

Table 3 e Sensitivity analysis of the proposed system

using uncertainty propagation.
Input parameter description
l16

Stoichiometry of air fed

to the combustor
Pressure in the combustor bar
pcomb
Fuel cell active area cm2
Acell
Charge transfer coefficient
aanode
in the anode
Stoichiometry of air fed to the fuel
l35
cell stack
lH 2
Stoichiometry of hydrogen fed
to the fuel cell stack
Number of fuel cells
ncell
Pressure in the fuel cell stack bar
pcell
_ 17
m
Mass flow rate of flue gas fed
to the SMR reactor kg=s
_ 26
m
Mass flow rate of water fed to heat
exchanger VI kg=s
_ 29
m
Mass flow rate of water fed to heat
exchanger V kg=s
_3
m
Mass flow rate of methane fed to fuel
preheater kg=s
_ 31
m
Mass flow rate of water fed to heat
exchanger IV kg=s
T1
Methane input temperature C
Methane-steam mixture SMR input
T4
temperature C
Reformate gas WGS input
T8
temperature C
T10
Fuel cell operating temperature C
Combustor air input temperature C
T13
Combustor air preheated
T14
temperature C
T17
Flue gas SMR input temperature C
Flue gas heat exchanger IV input
T34
temperature C
T35
Air fed to fuel cell stack input
temperature C
Selected output parameter description
hsys cogen

22

Combined cogeneration
system efficiency

Uncertainty
(%)
5.42
0.00
0.23
0.02
0.00

Total system efficiency

0.45
0.41

Korsgaard, 2008
Proposed system

0.88
0.75

the one developed by [4]. The most significant differences, in

terms of system configuration, between the two systems are
the following:
 The modeling assumptions regarding the chemical kinetics
follow a different approach; the reference system considers
a continuous stirred-tank reactor, where the proposed
system considers a plate heat exchanger reactor, as explained
in the preceding section.
 There are some significant differences in the thermal
management of the proposed system, where a steam
generator and a fuel preheater are modeled instead of
a mixture vaporizer arrangement in the reference system. In
addition, different heating/cooling between streams is used
(e.g. air coolers, etc.).
The validation results are shown in Table 2. By observation,
the validation has shown a good agreement of the two models,
verifying the validity of the research project under study,
nevertheless of the differences between the two systems.

14.92
0.23
0.00
4.67

5.

Sensitivity analysis

0.63

A multiple parameter sensitivity analysis using the built-in

EES uncertainty propagation tool was applied to determine
the system performance for the selected parameter range [18].
The results show how the outputs of the model vary with
variations in the input values. When selected system variablesXN are varied, their effect on the combined cogeneration
system efficiency is calculated, using the law of uncertainty
propagation. The uncertainty of a calculated variable is

3.10
1.63
0.11
0.00
0.00
0.39
25.27
7.32
7.58

Table 4 e Description and values of the parameters held

fixed during the parametric study.

1.51
26.98
0.00
Variable 
Uncertainty
75:32  6:74

Fixed parameter description

Value

ncell
Acell
T6
T8
LSMR
wSMR
LWGS
dWGS

185
45.16
126
250
0.49
0.09
0.45
0.13

APPENDIX

Number of fuel cells

Fuel cell active area cm2
SMR inlet reformate mixture temperature C
WGS inlet reformate mixture temperature C
SMR reactor plate length m
SMR reactor plate width m
WGS reactor tube length m
WGS reactor tube diameter m

5016

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

Table 5 e Description, initial values and range of the

decision variables used in the parametric study.
Decision variable description
Tcell
Tcomb
lH2
S=C

Initial Value

Fuel cell operating

temperature C
Combustor output
temperature C
Anode stoichiometric ratio
Steam-to-carbon ratio

Range

160

150e180

900

700e1000

1.3
4.0

1.3e1.6
2.5e4.5

determined as a function of the uncertainties of each variableUX . By assuming the individual calculations are uncorrelated and random (neglecting covariance), the uncertainty
of the efficiency can be expressed as,
v
u  2
uX vhi
U2X
Uhi t
vXi
i

where hi f X1 ; X2 ; .; XN

(22)

The results of the uncertainty propagation are shown in

Table 3. Every variable is set with a fixed standard uncertainty
of 5%, since higher values caused simulation inconsistencies
and failure to reach a solution convergence point. The result
shows that the combined cogeneration system efficiency is
affected by an uncertainty value of 6.74%, for an output
value of 75.32% efficiency. This value is moderately high,
which means that some of the input parameters can cause

a highly considerable effect on the model performance when

varied, while others have little or no effect.
By observation it can be concluded that the larger sources
of uncertainty are three parameters, namely the fuel cell stack
operating temperature, the flue gas heat exchanger IV input
temperature and the stoichiometry of hydrogen fed to the fuel
cell stack. The flue gas heat exchanger IV input temperature is
the temperature that regulates the thermal power output of
the space heating. If the corresponding output temperature is
kept constant, a high input temperature will result in a higher
thermal power output. This will increase the overall cogeneration thermal efficiency of the system, which in turn will
increase the overall system efficiency. As expected the
increase in fuel cell temperature will increase the fuel cell
electrical efficiency, which in turn will increase the system
total efficiency. Similarly, for the hydrogen stoichiometry, if
more less fuel is depleted to the combustor and more electrical power output is produced in the fuel cell stack will result
in an increase in the system total efficiency. These latter two
parameters are investigated further in the next section.
On the other hand some input system parameters have
insignificant effect on the system performance; these include
the combustor pressure, the fuel cell stack pressure, the fuel
cell air stoichiometry, and the temperature of the methanesteam mixture SMR input. As verified by the literature [6], the
fuel cell pressure for a PBI-membrane has a minor effect on
the fuel cell performance. This can have a positive effect on
the electrical efficiency because little pressure boost is

Table 6 e Parametric results for the proposed micro-CHP system.

Decision variables

Objective functions

Power output (kW)

S=C

lH2

Tcell

Tcomb

hsys cogen

hel FC

hel net

hther cogen

_ el
W
FC

2.5
3
3.5
3.75
4
4.25
4.5
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4

1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.35
1.4
1.45
1.5
1.55
1.6
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3

433
433
433
433
433
433
433
433
433
433
433
433
433
433
423
428
433
438
443
448
453
433
433
433
433
433
433
433

1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
1173
973
1023
1073
1123
1173
1223
1273

0.7999
0.8166
0.8256
0.8285
0.8308
0.8327
0.8522
0.8308
0.8497
0.8704
0.8932
0.9182
0.9455
0.9755
0.7163
0.8206
0.8308
0.8366
0.8406
0.8436
0.8459
0.7663
0.7974
0.8208
0.8301
0.8308
0.8299
0.829

0.3444
0.3628
0.372
0.375
0.3774
0.3793
0.3765
0.3774
0.366
0.3552
0.345
0.3354
0.3264
0.3178
0.2263
0.364
0.3774
0.3851
0.3904
0.3944
0.3974
0.4048
0.3948
0.3854
0.3798
0.3774
0.3761
0.375

0.2455
0.2621
0.271
0.274
0.2762
0.2781
0.2923
0.2762
0.2765
0.2774
0.2788
0.2809
0.2836
0.2869
0.1618
0.266
0.2762
0.282
0.2861
0.289
0.2913
0.2123
0.2431
0.2663
0.2755
0.2762
0.2754
0.2744

0.5545
0.5545
0.5546
0.5546
0.5546
0.5546
0.5599
0.5546
0.5732
0.5931
0.6144
0.6373
0.662
0.6886
0.5546
0.5546
0.5546
0.5546
0.5546
0.5546
0.5546
0.5541
0.5543
0.5545
0.5546
0.5546
0.5546
0.5546

933.1
994
1026
1037
1046
1052
1094
1046
1014
983.9
955.8
929.2
904.1
880.3
626.9
1008
1046
1067
1082
1092
1101
812.5
924.8
1009
1043
1046
1042
1039

APPENDIX

_ el
W
net
897.8
958.7
991.1
1002
1010
1017
1059
1010
978.4
948.6
920.5
893.9
868.8
845
591.6
972.8
1010
1031
1046
1057
1065
777.1
889.4
974
1007
1010
1007
1004

23

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

required for the fuel cell stack (similarly for the combustor).
The most interesting parameter in this case with little effect
on system performance is the temperature of the methanesteam mixture SMR input. Again a positive effect is observed,
this time on the overall system thermal efficiency; since a low
temperature for the methane-steam mixture is sufficient,
meaning a minimum amount of heat from the flue gas will be
extracted for heating it prior to entrance to the SMR reactor. In
turn the flue gas heat will be utilized in the cogeneration heat
exchanger IV and V, producing a higher thermal cogeneration
output. This will increase the thermal power cogeneration
output of the system, which in turn will increase the overall
system efficiency.

6.

Parametric study: results and discussion

The purpose of the parametric study is to investigate,

throughout variation of influential decision variables, the
maximization of the hsys cogen . The effect of four decision
variables, namely, steam-to-carbon ratioS=C, anode stoichiometric ratiolH2 , operating fuel cell temperatureTcell
and combustor output temperatureTcomb is studied, within
the required constraints. In all cases considered, the fuel cell
active area, and the geometries of the heat exchanger and
reactors were held constant, to remain within realistic
dimensions for a micro-CHP system application. In addition,
the number of fuel cells is held constant to provide a design

basis for the system. Finally, the SMR and WGS reactors inlet
reformate mixture temperatures were held constant to
preserve the validity of the experimental results. A description
of the fixed parameters along with their values is given in
Table 4.
The S=C is varied from 2.5 to 4.5 in 0.5 increments. Values
below 2.5 are not included in order to avoid problems of
carbon deposition and coke formation on the anode of the fuel
cell stack as reported in [14,19]. On the other hand, values
above 4.5 are not included, since a high S=C favors hydrogen
production but overall process economics favor a low steamto-carbon ratio, because exit gases include steam, which is not
utilized (condensed in the water knockout stage) [14]. Fuel cell
stacks are not usually operated with a lH2 value below 1.25,
when hydrogen-rich reformate gas is used as fuel. In addition,
since a reasonable amount of fuel is required for the
combustor, then extremely low values are not appropriate
[20]. As an upper limit, the lH2 must remain lower than 1.7 to
avoid the risk of fuel depletion in the fuel cell stack [4,11]. The
Tcell is varied from 150
C to 180
C in 5
C increments. A value
beyond 180
C is not used because it would exceed the operating limit of the fuel cell stack. The Tcell is dictated by the
materials set and cell design, which sets the conduction
properties and durability of the fuel cell tri-layer [6]. The Tcomb
is limited to a maximum value of 1273 K for two reasons: (1) to
minimize heat losses in the combustor, (2) a higher temperature could cause severe damages to the SMR reactor, since it
has not been tested experimentally for higher temperatures.

Fig. 3 e Decision variables vs. combined cogeneration system efficiency.

24

5017

APPENDIX

5018

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

Fig. 6 e Chemical composition of the reformate gas at

different stages of the fuel processing subsystem.

Fig. 4 e Efficiency vs. current density for various anode

stoichiometric ratios.

The initial (setting) values are chosen on the basis of

a typical micro-CHP system of the kind found in [4,5]. A
description of the key decision variables is given in Table 5,
along with their range of variation and corresponding initial
values.
The results for the parametric study are shown in Table 6,
which includes the objective functions along with the electrical power outputs. It should be noted that the hther cogen is
set to remain constant at 2 kWth. The trend lines of the four
decision variables vs. hsys cogen are shown in Fig. 3.
From an observation of the trend lines shown in Fig. 3, the
following assumptions can be made. The S=C reaches an
optimum efficiency at 4.5, which is the highest tested value.
This value is not realistic since values above 4 violate the
minimum allowable system exhaust temperature (T22) and
therefore they must be rejected. Also, the HT-PEMFC stack is
assumed to withstand CO -contents in the H2 -rich gas of up to
10,000 ppm. This value is exceeded when the S=C is reduced
below 3.75 due to the higher carbon content.
The optimum value for the lH2 is 1.3 and not above, because
at higher values although the hsys cogen is higher, the net

Fig. 5 e Variation of efficiency and mole fractions for

different values of steam-to-carbon ratio.

electrical power decreases (along with current density) below

the minimum electrical load requirements of the system.
Although a part-load study is not part of this study, it can be
concluded that the lH2 value can be used to regulate the load
according to a residential load profile.
The optimum value for the Tcell appears to be 180
C,
because the fuel cell stack operates more efficiently at
increased temperatures as verified by the literature [6]. On the
other hand, it is clearly observed that the benefit in terms of
efficiency at temperatures beyond 160
C is rather insignificant, considering the fact that a PBI-membrane will degrade
much faster at elevated temperatures, as proved experimentally by [6,21].
Finally, the optimum value for the Tcomb is observed at
around 827
C, but since values below 850
C violate the
minimum allowable system exhaust temperature (T22) this
value is not realistic. Therefore the optimum value is 900
C.
Fig. 4 presents the variation of efficiency vs. current density
for various lH2 at constant Tcell . A decrease in the amount of lH2
decreases the average concentration of H2 , thereby reducing
the theoretical voltage. The net electrical efficiency decreases
monotonically with increasing current density, which is
consistent to the fuel cell polarization curve [12,13].
Fig. 5 presents the variation in efficiency and the mole
fractions of different species for various S/C values. Although
it appears the H2 generation increases as S/C (SMR reactor
outlet) decreases, this not valid because of the water content.
This is clearly observed when the water passes through the
water knockout stage, prior to fuel cell anode inlet.
Fig. 6 shows the chemical composition of the reacting
species throughout different stages of the fuel processing
subsystem. Almost all CH4 entering the SMR reactor is converted into H2 -rich reformate gas. The CO mole fraction
appears slightly increased, which is justified similarly as in
Fig. 5.
Fig. 7 presents the mole fraction variation forCO, CH4 , CO2,
H2O and H2 . CO is generated throughout the steam methane
reforming process. The reformate gas at the SMR reactor
outlet has a mole fraction value of 0.07621, while at the WGS
reactor outlet it is further reduced at 0.008574.

APPENDIX

25

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

5019

Fig. 7 e Variation of H2, CH4, CO2, H2O and CO throughout the SMR and WGS reactors.

7.

Conclusions

A proposed micro-CHP system based on PBI-membrane fuel

cell technology is modeled and analyzed, by means of
a sensitivity analysis and a parametric study, to investigate
the potential of this technology as a candidate application for
residential micro-CHP systems. The highest combined
cogeneration system efficiency is calculated with a value of
83.08%. The corresponding net system electrical and cogeneration thermal efficiencies are 27.62% and 55.46%, respectively, while the fuel cell stack efficiency is 37.74%.
One of the most important advantages of the proposed
system is the simplicity of the fuel processing system. An
equivalent Nafion-based-PEMFC or SOFC-based fuel processing subsystem is more complex (and therefore more costly)
because it requires additional components, such as high- and
low-temperature WGS reactors, or PrOx devices to be coupled
with the fuel cell stack [4]. The fuel processing subsystem
provides an adequate rate of CH4 conversion and acceptable
CO-removal, making it appropriate for integration with an HTPEMFC stack. In addition, the obtained efficiencies for the
proposed system, verified the promising potential of the HTPEMFC technology for micro-CHP system applications. A lot of
ground for improvement is necessary before the end-user
system is available; the modeling results have to be verified by
experimental testing to investigate sources of discrepancy
and other hurdles, such as incompatibility of the subsystems.
The modeled system is very complex, with a high amount
of decision and other variables, which make the parametric
study very limited, because of the constraints and limits set by
the system design. However, a clear indication of the trend
lines shown by the parametric study can give feedback for
further investigation using a more advanced optimization
methodology with multivariable optimization, which should
also include part-load operation. A finalized optimum system
will not be the system that operates optimally at full load, but

26

the one operating at an average load (as required by the residential load demand). A very significant, and still unexplored,
area of optimization is the geometry of the system. The
parametric study suggested the SMR and WGS reactors must
be sized and simulated in greater detail, in order to achieve the
geometry which will provide the highest amount of H2
conversion. This will lead to an even higher combined
cogeneration system efficiency and therefore lower the fuel
consumption.
Finally, EES was chosen as the modeling tool for this
research study for the reasons explained in the introduction.
The modeling was completed within a small amount of time,
taking advantage of the nature and the buildin capabilities of
the software (such as parametric tables, graphs, NewtoneRaphson method of solving algebraic equations, and
thermophysical properties).

Acknowledgments
The authors would like to acknowledge the support of Danfoss
and Dantherm Power throughout the realization of this
research study.

references

[1] OHayre R, Colella W, Cha S, Prinz FB. Fuel cell fundamentals.

Wiley; 2009.
[2] Gunes MB, Ellis MW. Evaluation of energy, environmental,
and economic characteristics of fuel cell combined heat and
power systems for residential applications. Journal of Energy
Resources Technology 2003;125(3):208.
[3] Pedersen AH, Balslev P. Demonstration of mCHP based on
Danish fuel cells. In: Proceedings of the European fuel cell
Forum; 2009 [Lucerne, Switzerland].

APPENDIX

5020

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 5 0 1 0 e5 0 2 0

[4] Korsgaard AR, Nielsen MP, Kr SK. Part one: a novel model of
HTPEM-based micro-combined heat and power fuel cell
system. International Journal of Hydrogen Energy 2008;33(7):
1909e20.
[5] Korsgaard AR, Nielsen MP, Kr SK. Part two: control of
a novel HTPEM-based micro combined heat and power fuel
cell system. International Journal of Hydrogen Energy 2008;
33(7):1921e31.
[6] Buchi FN, Inaba M, Schmidt TJ. Polymer electrolyte fuel cell
durability. New York, NY, USA: Springer; 2009.
[7] Georgopoulos N. Application of a Decomposition Strategy to
the Optimal Synthesis/Design and Operation of a Fuel Cell
Based Total Energy System, M.S. Thesis, 2002, Virginia Tech,
Blacksburg, VA, USA.
[8] Ferguson A, Ugursal VI. Fuel cell modelling for building
cogeneration applications. Journal of Power Sources 2004;
137:30e42.
[9] Obara S. Dynamic characteristics of a PEM fuel cell system
for individual houses. International Journal of Energy
Research 2006;30(15):1278e94.
[10] Godat J, Marechal F. Optimization of a fuel cell system using
process integration techniques. Journal of Power Sources
2003;118:411e23.
[11] Braun R. Optimal Design and Operation of Solid Oxide Fuel
Cell Systems for Small-scale Stationary Applications,
Doctoral Dissertation, 2002, University of
WisconsineMadison, USA.
[12] Korsgaard AR, Nielsen MP, Bang M, Kr SK. Modeling of CO
influence in PBI electrolyte PEM fuel cells. In: Proceedings of
the FUELCELL; 2006 [Irvine, CA, USA].
[13] Korsgaard AR, Refshauge R, Nielsen MP, Bang M, Kr SK.
Experimental characterization and modeling of commercial
polybenzimidazole-based MEA performance. Journal of
Power Sources 2006;162(1):239e45.
[14] Kolb G. Fuel processing for fuel cells. Weinheim, Germany:
Wiley-VCH; 2008.
[15] Ahlstrom-Silversand AF, Odenbrand CUI. Thermally
sprayed wire-mesh catalysts for the purification of flue
gases from small-scale combustion of bio-fuel catalyst
preparation and activity studies. Applied Catalysis A 1997;
153:177.
[16] Keiski RL, Salmi T, Niemisto P, Ainassaari J, Pohjola VJ.
Stationary and transient kinetics of the high temperature
water-gas shift reaction. Applied Catalysis A: General 1996;
137(2):349e70.
[17] Davies J, Lihou D. Optimal design of methane steam
reformer. Chemical and Process Engineering 1971;52:
71e80.
[18] F-Chart Software, 2010, EES Manual, www.fchart.com, v8.
590 edition.
[19] Larminie J, Dicks A. Fuel cell systems explained. Chichester,
UK: John Wiley & Sons; 2003.
[20] Barbir F. PEM fuel cells: theory and practice. Elsevier; 2005.
[21] Rasmussen PL, Nielsen MP, Kr SK, Andreasen SJ.
Experimental study and modeling of degradation

phenomena in HTPEM fuel cell stacks for use in CHP

systems. In: Proceedings of the hydrogen and fuel cells
conference; 2009. Vancouver, Canada.

Nomenclature
dHr;1 : Enthalpy of reaction (kJ/mol)
Ea : Activation energy (kJ/mol)
F: Faraday constant (C/mol)
i: Current density (A/cm2)
i0 : Exchange current density (A/cm2)
keh : Electrooxidation rate of hydrogen (A/cm2)
k0 : Pre-exponential factor
KT : Equilibrium constant
KWGS : Kinetic power law fit parameter (m^1.92/kmol^-0.36.kg.s)
LHVi : Lower heating value of species i (J/kg)
_ i : Mass flow rate of species i (kg/s)
m
n_ CO : CO molar feed rate (kmol/h)
Q_ i : Heat transfer rate of component i (W)
rCO : CO reaction rate (kmol/kg.s)
R: Ideal gas constant (J/K.mol)
Rdiff : Diffusion resistance (Ohm.cm2)
Rohmic : Ohmic resistance (Ohm.cm2)
S=C: Steam-to-carbon ratio
Tcell : Fuel cell operating temperature (
C)
Tcomb : Combustor output temperature (
C)
Ti : Temperature of species or state point i (
C)
Vcell : Total cell voltage (V)
V0 : Open circuit voltage (V)
us : Superficial velocity (m/s)
_ i : Work rate of component/subsystem i (W)
W
ai : Charge transfer coefficient
ha : Anode overpotential (V)
hc : Cathode overpotential (V)
hohmic : Ohmic losses (V)
hi : Efficiency of component/system i
qH2 : Surface coverage of hydrogen
lair : Cathode air stoichiometric ratio
lH2 : Anode stoichiometric ratio
rg : Density of the reformate gas (kg/m3)
rwiremesh : Density of the wire mesh catalytic material (kg/m3)
u: Water gas shift reactor cross-sectional area (m)
Subscripts/superscripts
cogen: Cogeneration
el:
Electrical power
FC:
Fuel cell subsystem
HEx: Heat exchanger
in:
Input stream
net:
Net amount
out:
Output stream
sys:
System
ther: Thermal power

APPENDIX

27

28

APPENDIX

Journal Article III

APPENDIX

29

30

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

ON A PBI FUEL CELL-BASED RESIDENTIAL SYSTEM FOR
DANISH SINGLE-FAMILY HOUSEHOLDS
ALEXANDROS ARSALIS, MADS P. NIELSEN, SREN K. KR
Abstract. A proposed residential energy system based on the PBI (Polybenzimidazole) fuel cell technology is analyzed in terms of operational performance.
Conventional operational strategies, such as heat-led and electricity-led, are
applied to the simulated system to investigate their performance characteristics. Based on these findings, an improved operational strategy is formulated and applied in an attempt to minimize the shortcomings of conventional
strategies. System parameters, such as electrical and thermal efficiencies, heat
dumping, and import/export of electricity, are analyzed. The applied load
profile is based on average data for a single-family household in Denmark
and includes consumption data for electricity and heat demands. The study
analyzes the potential of the proposed system on market penetration in the
area of residential heat-and-power generation and whether this deployment
can be justified as compared to other micro-CHP system technologies. The
most important findings of this research study indicate that in comparison
to non-fuel cell-based micro-CHP systems, such as Stirling Engine-based systems, the proposed system has significantly higher efficiencies. Moreover, the
lower heat-to-power ratios allow the system to avoid high thermal surpluses
throughout the whole annual operational profile.

1. Introduction
The application of fuel cell technology in residential micro-CHP (Combined Heat
and Power) systems has been gaining an increased amount of interest during the
recent years, which is mainly due to their promising efficiency performance. Many
advantages can be listed for these systems, as well for the fuel cell itself and the
system in general. The main advantages of fuel cells include, among others, high
electrical efficiencies and low greenhouse gas emissions [1-4]. In addition to these advantages, the micro-CHP technology can provide increased cogeneration efficiencies,
due to the thermal efficiency utilization. Also on-site power-and-heat production
eliminates transmission and distribution losses, which are inevitable for centralized
systems [5-7]. Another important aspect of fuel cell-based micro-CHP technology is
the fact that fuel cell systems can operate efficiently at part-load operation, which
provides greater flexibility in adopting an operational strategy. Further on, fuel cells
possess the capability of responding rapidly to load changes. As a consequence of
high efficiency, fuel cell systems can offer the potential of reduced operational costs,
which has been becoming extremely influential in the recent years due to the rapidly
increasing cost of fossil fuels.
In the current research study, different operational strategies are examined to
investigate the response of the system to changing loads, primarily with respect to
efficiency. The overall goal of this research work is to investigate the application of
conventional operational strategies such as heat-led and electricity-led operation [6],
Received by the editors 2011.
Key words and phrases. PBI; HT-PEMFC system; micro-CHP; residential system; operational
strategy; fuel cell system.
1

APPENDIX

31

A. ARSALIS ET AL.

and then formulate an improved operational strategy, which provides a higher average net electrical (and total) efficiency for the system. In this manner a comparison
can be made with the conventional operational strategies, drawing assumptions for
discussion and analysis. Also a thermal storage tank is modeled and coupled to the
system to increase its operational flexibility. Finally, an investigation of the heat
losses exhibited by the thermal storage tank is carried out throughout the whole
annual operational load profile. This is done to monitor the periods exhibiting the
highest losses. The applied load profile is based on averaged consumption data for
a typical single-family household in Denmark [5,6].
2. Literature Review
Residential micro-CHP systems can be categorized in four main types, based
on their respective prime mover technology: Stirling engines (SE), internal combustion engines (ICE), proton exchange membrane fuel cells (PEMFC), and solid
oxide fuel cells (SOFC). SE- and ICE-based micro-CHP systems are considered
to be the most mature technologies, since they are commercially available from a
number of manufacturers [8-11]. The operating principle of the SE is based on an
external combustion engine with an internal piston being driven via a temperature
difference between the ends of a cylinder. SE residential systems typically have
low electrical efficiencies ranging from 0.07 to 0.15 (based on LHV) [9,12], while
their respective overall efficiencies are comparable to that of a condensing boiler. In
terms of operational behavior, they have rapid startup times and quick load changing responses [12]. The operating principle of an ICE micro-CHP system is based
on combustion occurring inside a cylinder driving a piston to create mechanical
work, which is converted to electrical energy in a generator. Electrical efficiencies
are typically at around 0.25 (based on LHV), with similar operational behavior as
SE-based systems [12,13].
Fuel cell-based micro-CHP systems is the least mature technology, since a number of technical and economic issues must be resolved before its potential can be
realized. Nevertheless, it is the most promising technology because it has the potential of achieving high electrical and overall efficiencies [2,5,12,14]. Startup times
are rather slow for the SOFC type [15,16], while for PEMFCs are significantly
faster. On the other hand, low-temperature PEMFCs (Nafion) have a very low
heat-to-power ratio. They are also very sensitive when operated with reformate
fuel, requiring extensive carbon monoxide reduction. A relatively new technology
is HT-PEMFCs, which operate at temperatures at around 160o C and they are relatively tolerant to carbon monoxide poisoning [5,17,18]. Fuel cell systems are able
to respond to load changes rapidly, although a regular load changing operational
pattern will deteriorate the fuel cell membrane, and therefore reduce its lifetime.
Electrical efficiencies can reach 0.40 (based on LHV), while the overall efficiency is
as high as the ICE and SE types at nominal load. In addition, the fuel cell electrical
efficiency is significantly higher at part-load, because of the higher cell voltage at
lower loads [17].
2.1. Market penetration projections. Micro-CHP technology has the potential to reduce transmission and distribution losses that typically occur when a large
remotely located power station transmits electricity to a household [5]. This suggests that onsite produced and consumed electricity will have negligible losses,
while imported/exported electricity will still have the usual losses. Therefore an
effort should be made to choose an operational pattern which will minimize grid
interaction. According to the Danish Energy Association [19], the fleet of central
power plants in Denmark is relatively old, and the newest power plant was built in
2002. This fact in connection to the new initiative to promote the fuel cell-based

32

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

micro-CHP technology in Denmark, where gas-fired boilers will be replaced with

fuel cell-based micro-CHP systems, may change the residential heat-and-power consumption in the near future. The Danish micro-CHP project is explained in detail
in [14].
One significant uncertainty in fuel cell-based micro-CHP technology is the type
of system configuration and operating modes of micro-CHP systems, which must be
efficient and also practical [20]. Although the capital cost of a micro-CHP system
is greater than the one for a conventional domestic boiler, a micro-CHP system
provides the opportunity to recover expenditure. This is because the import of
electricity from the network grid is minimized, while surplus electricity can be
exported [6]. In addition, residential micro-CHP systems can satisfy the heatand-power demand more efficiently than conventional configurations and therefore
decrease fuel consumption.
2.2. Operational strategies in the literature. A common practice is to operate
the micro-CHP system during times of increased load demand, while the system is
switched off during low demand. This practice is more common for heat engines
(i.e. ICE, SE), because part-load efficiencies are very low, as explained above. On
the other hand, daily on/off operation results in slower responses from the system
(with associated losses), and also decreases the lifetime of the system [10,21]. Since
a low heat-to-power ratio is available for fuel cell-based systems, the system can
be operated continuously without any efficiency reduction, or the need to dump
heat, provided a thermal energy system is coupled to the system. However, it
should be noted that if the stack is continuously operated at a part-load below
25%, the higher selected nominal cell voltage does not necessarily mean higher
operating efficiency. This is due to operation at very low current densities, where
parasitic losses, including gas permeation through the polymer membrane, may not
be negligible [22].
A number of operational strategies applied to micro-CHP systems are available
in the literature. These are primarily simple strategies such as heat-led operation,
where the system operates in accordance to the heating load demand. Similarly, an
electricity-led operation operates in accordance to the electrical load demand. In
1991, an extensive investigation of energy consumption in single-family households
in Denmark was performed. In this study the consumption data were measured as
15 minute averages. The study included 25 different households. The consumption
data included electricity, space heating and hot water consumptions for a whole
year, yielding a total dataset of 335,000 points [6]. According to the Danish
Energy Association [19], the average electricity consumption for a single-family
household is 3,960 kWh, while the consumption in 1991 was 4,849 kWh. Although
a safe and accurate comparison (based on the circumstances) cannot be made based
on those two values, a decrease in electricity consumption is explained by the use of
more efficient equipment and changes in household structures during recent years.
Korsgaard et al. [6] applied the aforementioned load profile in a dynamic HTPEMFC-based micro-CHP system using the average consumption of the 25 houses.
The investigated operational scenarios were heat-led operation and combined heatand-power operation with different thermal storage tank sizes.
The heat and power demand can be significantly different throughout the fluctuations of the consumption data. This is due to a number of reasons, such as the
weather season (e.g., summer vs. winter vs. mid-season) or the time period (e.g.,
day vs. night, or weekday vs. weekend day). An averaged representation of the
consumption data for an averaged single-family household in Denmark, including
the electricity load and the total heating load, is shown in Fig. 1. By observation,
the electrical demand is almost constant throughout the year, with a slightly lower

APPENDIX

33

A. ARSALIS ET AL.

Figure 1. Averaged consumption data for an averaged singlefamily household in Denmark.

demand during the summer season. On the other hand, the heating load demand is
varied significantly, but still almost linearly, with a high demand during the winter
months and a very low demand during the summer months. The total electrical
demand throughout the year is 4,975 kWh, where the total heating demand is
12,613 kWh.
For a more thorough investigation of the load profile, three representative days
are selected from the annual consumption dataset, for winter, mid-season and summer. These are shown in Fig. 2, where the first day is the 25th of January (winter),
the second day is the 29th of April (mid-season), and the third day is the 21st of
August (summer). It is interesting to observe the difference in the variation of the
heat-to-power ratio: Spring and summer days suggest that a heat-led operation is
needed to avoid the need for heat dumping, where the winter day suggests that
the heat provided by the system needs to be supplemented by the auxiliary heater,
if grid interaction is desired to be kept at a minimum rate. In general, across a
year there is less seasonal variation in the electricity requirement than in the heat
requirement, as suggested in the literature [6]. The above observations suggest
that conventional strategies, such as heat-led and electricity-led, will be insufficient
if the system is desired to deliver efficiently. Therefore some form of intelligent
operational control is critical to the success of the proposed micro-CHP system.
3. Methodology analysis: operational strategies
The operational strategies considered in this research work initially consist of
two simple conventional operational methods: Heat-led and electricity-led. Based
on the results extracted from those two methods, a third improved method can be
formulated, which is a combination of the two conventional methods. The success
of the improved operation will be analyzed with respect to important operational
parameters such as electrical and overall efficiencies, carbon dioxide emissions, fuel
(methane) consumption and heat dumping. The number of on/off operations is also
investigated to illustrate the behavior of the system for every adopted strategy. The

34

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

Figure 2. Consumption data of three representative days in the

winter, mid-season and summer.

APPENDIX

35

A. ARSALIS ET AL.

extensive modulation range capability of a fuel cell-based system allows variation

of the load according to the needs of the adopted operational strategy.
The operational modes can have significant effects on the energy and environmental system performance. A heat-led operational strategy promises high cogeneration efficiencies. However, this cogeneration efficiency is due to the increased
thermal efficiency, and not due to a high electrical efficiency. A low electrical efficiency will decrease the competitiveness of the proposed micro-CHP system and
therefore not justify its selection over a gas-fired boiler or other micro-CHP system
types. On the other hand, electricity-led operational strategies can promise minimum grid interaction and therefore avoid transmission losses, but heat dumping
will be needed in many instances. Therefore a combined strategy must be formulated. Finally, interaction with the network grid offers the ability to meet marginal
electricity demand and reduction of peak loads.
3.1. Heat-led operation. In this research work, the heat-led operation is defined
as an operational strategy attempting to force the micro-CHP system to operate
in a manner that will fully satisfy the heating load demand, without exceeding
it. It should be noted that the heating load demand is a simplified approach to
combine the space heating demand and the hot water supply, as suggested by the
averaged consumption data of the Danish load profile for a single-household. The
exact configuration of the space heating and hot water components are beyond the
scope of this research work. This configuration assumes that the electrical load
demand is satisfied with imported electricity from the grid, while in operational
instances of electricity production beyond the demand of the household electricity
is exported to the grid. The advantage of this operational strategy is the avoidance
of utilizing a thermal storage tank, which results in capital cost savings in terms of
components (tank, piping, etc.) and insulation.
In an actual thermal energy storage arrangement, the heat inside the tank cannot
be assumed to have an average temperature. In other words, heat will be stratified,
with temperature decreasing from top to bottom. Therefore space heating return
must enter the tank from the bottom and exit from the top. On the other hand,
for tap water heating, an internal heat exchanger must be included inside the tank,
since hot water demand is required more instantly than space heating demand.
3.2. Electricity-led operation. Equivalently to the heat-led operational strategy, electricity-led operation attempts to fully satisfy the electrical load demand.
This configuration assumes that the heating load demand is satisfied with external
heating, as described in the previous section, when necessary, while the thermal
storage tank will limit, to the minimum possible extent, the need for heat dumping
in the atmosphere.
3.3. Improved operation. The improved operation is defined as a combination
of the heat-led and electricity-led operational strategies. The improved strategy
will therefore attempt to restrict a number of parameters that reduce the electrical
and cogeneration efficiencies of the system. These include avoidance of operation at
low efficiency regimes and therefore constraining the power output of the system at
a minimum and a maximum power output value. A thermal storage tank will also
assist in the increase of the operational flexibility of the system. Another considered
restriction in this strategy is the avoidance of operating the system in operational
points which will need heat dumping to the atmosphere.
A frequent design problem is that of maximizing daily savings on a summer
day, due to the need of the micro-CHP system to be shut down once the heat
demand has been met. Any heat generation in excess of this heat limit is surplus
[21]. Also special attention is given to the number of turn on/off for the system.

36

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

Finally grid interaction should be minimized not only because of technical reasons
(transmission losses, etc.), but also because of economic reasons (e.g., taxation
imposed for purchasing/selling electricity). Ideally, electricity should be imported
and exported according to a combination of fuel prices, electricity import/export
prices, and their interaction with the electrical/overall efficiency profiles [6]. Based
on the above, some general preliminary assumptions can be made on the nature of
this operational strategy: it is expected that the system will follow an electricity-led
operational strategy in the winter months, while the system will follow a heat-led
strategy in the summer months. It is also possible to shut down the system in some
instances during the summer months.
4. System layout and modeling
4.1. HT-PEMFC-based micro-CHP system design, layout and modeling.
The developed model is based on the PBI (Polybenzimidazole) HT-PEMFC technology and is described in detail in [23]. It is noted that the system should be
considered coupled with a grid network to import/export electricity as needed in
case a purely electricity-led operation is not followed. Equivalently, a thermal storage tank and an auxiliary burner are coupled to the system to satisfy the heating
load demand in case a purely heat-led strategy is not followed. The operating principle of the system, shown in Fig. 3, is the same as in [23], with the exception of
recovering the thermal energy of the flue gas utilizing a thermal storage tank, instead of heat exchangers. The fuel processing subsystem consists of a desulfurizer,
steam methane reforming (SMR) reactor, water gas shift (WGS) reactor, combustor and water-knock out device. Syngas is directed to the fuel cell stack anode. The
combustor outlet flue gas is used in the SMR reactor, the fuel preheater and the
steam generator. The fuel cell stack exhaust mixture is used to preheat the natural
gas and subsequently it is mixed with the combustor outlet flue gas for heat recovery
in the thermal storage tank. The system, modeled in the commercially available
software EES (Engineering Equation Solver), includes 27 state points. Figure 3
indicates the temperature and mass flow rate for every node at 1 kWe net power
output and current density 0.2 A/cm2 . The net electrical efficiency and the total
system efficiency at these conditions are 0.291 and 0.829, respectively. It should
be noted that the operating pressure of the system is assumed to be slightly above
atmospheric conditions.
4.2. HT-PEMFC. The HT-PEMFC model considers only the reaction of hydrogen with oxygen, while all other species are considered inactive. The ohmic and
diffusion resistances are based on linear regressions evaluated experimentally in [24],
(1)

Rohmic = 0.0001667Tcell + 0.2289

(2)

Rdif f = 0.4306 0.0008203Tcell

where Tcell is the fuel cell operating temperature (K). The anode and cathode
overpotentials are given by the following expressions, respectively,


i
RTcell
1
sinh
(3)
a =
anode F
2keh H2
(4)

c =

RTcell
ln
4cathode F

i0 + i
i0

+ Rdif f

i
air 1

where i is the charge transfer coefficient, air is the cathode air stoichiometric
ratio, keh is the electro-oxidation rate of hydrogen, H2 is the surface coverage of

APPENDIX

37

A. ARSALIS ET AL.

Figure 3. Configuration of the proposed micro-CHP system.

hydrogen, i is the current density, and i0 is the exchange current density. The
ohmic losses are given by,
(5)

ohmic = iRohmic

Therefore, the total cell voltage can be formulated as,

(6)

Vcell = V0 a c ohmic

where V0 is the open circuit voltage.

4.3. Fuel processing subsystem. The fuel processing subsystem consists of the
SMR reactor and a single-stage WGS reactor. The SMR reactor model is based
on an experimental configuration of the compact plate heat exchanger type, which
is filled with s wire mesh catalyst material. The governing equations, in the onedimensional steady-state model, consist of energy balances, momentum balances
and continuity equations. The model assumes two unmixed working medium, the
reformate gas and the flue gas, with an inlet and outlet for each. The model
considers only the following reactions:
(7)

CH4 +H2 O
CO + 3H2

(8)

CO+H2 O
CO2 +H2

0 = 206.2kJ/mol
h
rxn
0 = 41.1kJ/mol
h
rxn

Heat transfer rates are evaluated in each element as a single total heat transmission
coefficient. The reactor is discretized into 14 individual cells on either side of the
heat exchanger. The number of cells was increased until a sufficient grid consistency
was obtained. The WGS reactor receives the cooled reformate gas and its purpose
is to reduce the CO content to a fuel cell tolerable percentage, which is up to 3%
(per mole basis). The thermodynamic equilibrium constant of the WGS reaction

38

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

(Eq. (8)), is calculated as follows [25]:



4400
(9)
KT = exp
4.063
TW GS

where TW GS is the average operating temperature of the WGS reactor (K). The
carbon monoxide conversion is given by:
dxCO
rCO
(10)
=w
dz
n CO
where is the cross-sectional area of the WGS reactor (m2 ), w is the density of
the wire mesh catalytic material (kg/m3 ), rCO is the carbon monoxide reaction
rate (kmol/kg-s), and n CO is the molar flow rate of carbon monoxide (kmol/s).
4.4. Thermal storage tank. The model for the thermal storage tank assumes
simple heat transfer balances of the three working mediums (flue gas, tank water,
cogeneration water), while heat losses are accounted only to occur from the top
surface of the tank as in [5,26]. It is assumed that auxiliary heating is available
(although not modeled in detail), in case the heating load demand exceeds the
provided heat transfer from the flue gas. The auxiliary heating can be provided
by means of an auxiliary burner or an electric circuit inside the tank as in [26].
Therefore the tank temperature can be formulated as a function of the heat transfer
rates and the tank capacity:
Z 

1
Q f luegas + Q aux Q heatingdemand Q loss dt
(11)
Ttank =
mtank cp,tw

where Ttank is the tank temperature, mtank is the mass of the tank and cp,tw is the
average heat capacity of the water in the tank. Finally the auxiliary heat transfer
rate input is defined as,
(12)
Q aux = m
tank cp,tw (60 Ttank )

suggesting that the auxiliary heater will be utilized when the tank temperature
drops below 60o C. The set point temperature for the tank is 60o C, while the
temperatures for the incoming and outgoing residential water streams are 15 and
55o C, respectively.
4.5. General remarks. The proposed system is assumed to be able to fulfill any
heat- or electrical-load from low to peak values, based on the adopted Danish load
profile as explained above. Also the supply/demand matching performance of the
micro-CHP system will be affected by the transient characteristics of the micro-CHP
system (start/stop, operational strategy selection). The current system simulation
assumes a perfect startup/shutdown characteristic since a detailed pattern of the
start/stop behavior is not available. However it should be noted that a purely
electricity-led operational strategy suggests the system will never be shut down
since the electrical load demand is never zero. A shutdown/startup is needed in
the case of adopting a purely heat-led strategy, and also for a combined heat- and
electricity-led operational strategy. Finally, the analysis assumes a household with
an area of 130 m2 , inhabited by four persons. The thermal storage tank size is
assumed to have a volume of 250 L as defined in [6].
5. Results and Discussion
5.1. Application and analysis of conventional operational strategies. Before an improved strategy can be formulated, the two conventional operational
strategies are applied to investigate trends on the behavior of the system, which
can be used later on the improved strategy. Table 1 shows the calculated results
for the minimum, maximum, and average values for net electrical power output,

APPENDIX

39

10

A. ARSALIS ET AL.

Table 1. Calculated results of conventional operational strategies,

including efficiencies, net electrical power output and heat output.
Operational
strategy

min
Wel,net

Welavg
, net

max
Wel,net

Q hmin

Q havg

Q hmax

min
sys

avg
sys

max
sys

elmin

elavg

elmax

thmin

thavg

thmax

Electricity-led

162

567

2211

267

926

5056

0.536

0.850

0.891

0.163

0.326

0.345

0.373

0.524

0.546

Heat-led

816

2453

1439

6186

0.478

0.784

0.891

0.134

0.291

0.345

0.345

0.493

0.546

total cogeneration heat transfer rate, total efficiency, thermal efficiency, and net
electrical efficiencies. The net electrical efficiency (net,el ), the cogeneration thermal efficiency (uh ), and the total system efficiency (sys ) are used to monitor the
performance of the proposed micro-CHP system. The net electrical efficiency is
el ) divided by the chemical energy input to the
defined as the net power output (W
system (m
CH4,in LHVCH4 ). The net power output is the remaining power generated
by the fuel cell stack after parasitic losses from balance-of-plant components have
been subtracted. The cogeneration thermal efficiency is defined as the recovered
useful heat (Q uh ) divided by the chemical energy input to the system. The total
system efficiency is the sum of the net power output and the recovered useful heat
divided by the chemical energy input to the system.
el
W
m
CH4,in LHVCH4

(13)

net,el =

(14)

uh =

Q uh
m
CH4,in LHVCH4

(15)

sys =

el + Q uh
W
m
CH4,in LHVCH4

The power and heat outputs suggest the system will be required to operate over a
large range of operation. Although fuel cell systems are capable of achieving high
efficiencies over a large range of operation, the system has to be constrained to
avoid operation at low efficiency regimes. Low net electrical efficiencies are caused
by high current densities. For example, if the fuel cell stack follows a purely heat-led
operational strategy, and since heat and electricity production are interrelated, during time periods of high heat demand the high current density will cause a decrease
in fuel cell stack efficiency. The three aforementioned representative days are used
with the applied operational strategies and the results are shown in Fig. 4. From
an observation on these figures it can be concluded that the relation between electricity and heat production is quite linear. Also it can be concluded that adoption
of a heat-led strategy will require frequent utilization of the grid (import/export),
while a thermal storage tank/external burner will be needed when an electricity-led
operation is adopted.
5.2. Formulation and application of an improved strategy. The application
of the improved strategy is formulated on the basis of the following assumptions:
The annual load profile is split into 35 time periods, to locate the periods which require electricity-led operation, heat-led operation or system
shutdown.
The net electrical power output of the micro-CHP system is constrained
for a maximum value of 1 kWe. This is done to avoid operation at low
efficiency regimes.

40

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

11

Figure 4. Influence of conventional operational strategies (left:

electricity-led; right: heat-led) on system output vs. household
demand for three representative days.

An electricity-led operation is followed during most of the year, primarily

to reduce interaction with the grid. This strategy is followed provided the
average heat production does not exceed the average heat demand.
A heat-led operation is followed when the heat production would exceed the
heat demand, if the electricity-led operation was followed. In this manner
heat dumping is avoided.
The system is shut down when the average heat demand is lower than
500 W. The reason is again to avoid operation of the system in low efficiency regimes. Also if the fuel cell stack is continuously operated at
extremely low current densities, the lifetime of the fuel cell stack would be
significantly reduced. The reason is that low current densities correspond to
high voltages, which result in a decrease of power density. Therefore, parasitic losses, such as gas permeation through the polymer membrane, may
not be negligible. The current pattern of the improved strategy requires
only a single turn on/off throughout the year.
The influence of operational strategies on the proportion of electrical and heat demand met by the micro-CHP system is given in Fig. 5. From an observation, it can
be concluded that the electricity/heat demand is almost coincidental with the electricity/heat production during mid-season (nodes 9-14, 22-25). During this period
very little auxiliary heat is needed, while the import/export of electricity to the grid
is also kept at a minimum rate. On the other hand, during the winter months most
of the heat demand has to be fulfilled by external means. Also Fig. 5 indicates the
type of operation chosen for the respective averaged time period (e: Electricity-led
operation, h: Heat-led operation, s: System shutdown). These operational modes
are chosen on the basis of the assumptions given above.

APPENDIX

41

12

A. ARSALIS ET AL.

Figure 5. Annual variation of electricity (production and demand) and heat (production, demand and auxiliary).
The annual variation of imported and exported electricity for the improved operational strategy is shown in Fig. 6. A negative value for the import/export curve
indicates electricity is exported during that averaged time period, while a positive
value indicates import of electricity. The goal of keeping these values at a minimum
range is primarily fulfilled for the export part, where a significant amount has to
be imported during the shutdown of the system. It should be noted that small
amounts of electricity import, primarily during the winter season, are caused due
to the constraining of the maximum power output of the system during electricityled operation. If a purely electricity-led operation is followed during the summer
period, where the heating load demand is rather limited, then there will be a need
for heat dumping, since the TST will be unable to fully store the incoming heat.
If the heat demand vs. heat production is calculated for every averaged time period, it is observed that during the periods 10 to 25, there is an excess of heat
production as compared to heat demand. Therefore the total heat dumping value
for electricity-led operation is the sum of the differences between heat recovered
by the micro-CHP system and heat demanded by the household for the aforementioned periods. The rate of heat dumping for electricity-led operation is illustrated
in Fig. 7.
5.3. Thermal storage tank heat losses. The thermal storage tank can exhibit
variable heat losses depending on the operational profile of the system, the external
heater and the heat demand. During heat-led operation, where all the heat demand
is supplied by the micro-CHP system, heat losses increase rapidly. The reason is
due to the fact that the micro-CHP system has to supply all the required heat
demand, while the external heater is not utilized. In other words, the thermal

42

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

13

Figure 6. Annual variation of imported and exported electricity

for the improved strategy.

Figure 7. Averaged rate of heat dumping throughout the year

for the electricity-led operational strategy.
storage tank operates at critical conditions, which increase the heat loss rate. The
total heat losses for annual operation are 23.4 kWh.
5.4. Overall comparison of the three operational strategies. All the considered operational strategies are analyzed and their overall results are given in

APPENDIX

43

14

A. ARSALIS ET AL.

Table 2. Overall performance comparison of the three considered

operational strategies.
Parameter

Operational strategy

Variable

Description (unit)

Electricity-led

Heat-led

Improved

ELHV ,in

Total fuel (CH4) input (kWh)

15630

26475

13805

Eel , prod

Total annual electricity production (kWh)

4984

7164

4533

Eel ,imp

Total annual electricity import from the grid (kWh)

753

1127

Eel,exp

Total annual electricity export to the grid (kWh)

2933

676

Eheat , prod

Total annual heat production (kWh)

8136

12639

7334

Eheat , aux

Total heat provided by external means (kWh)

6502

5305

Eheat ,dump

Total annual amount of heat dumping (kWh)

1999

el ,net

Average net electrical system efficiency (%)

31.9

27.1

32.8

th

Average thermal system efficiency (%)

39.3

47.7

53.1

sys

Average total system efficiency (%)

71.2

74.8

85.9

Table 2, which summarizes the effect of the applied strategies on system performance. Utilizing the heat-led operation requires almost twice fuel input to the
micro-CHP system (compared to the improved strategy), which produces an increased amount of electricity overall. Nevertheless electricity has to be imported
at periods of low heat demand, such as the summer season. The electricity-led
operation requires significantly less fuel than the heat-led operation, but a high
amount of heat dumping is required. The merits of the improved operation are
observed in all categories, although a significant amount of imported electricity is
required due to the system shutdown in the summer period and the constraining
of the system at 1 kWe. The most important parameter of the study is the improvement of the average total system efficiency, which is 85.9% for the improved
operational strategy, while the respective total system efficiencies for electricity-led
and heat-led operations are 71.2 and 74.8%, respectively. It is also important to
observe that the thermal efficiency, and subsequently the total system efficiency, for
the electricity-led strategy is significantly low due to the need for heat dumping.
6. Conclusions
A proposed residential micro-CHP system, based on PBI fuel cell technology, is
analyzed in terms of operational performance. The applied load profile consists of
the electrical and heat demands for a single-family household in Denmark. The
results indicate that the proposed micro-CHP system has the potential to maintain high efficiencies throughout the annual load profile, if an improved strategy
is applied. The improved parameters include the net electrical efficiency, thermal
(cogeneration) efficiency, and in effect the total system efficiency. Also, heat dumping is avoided by use of a thermal storage tank. The study also indicates that the
system has to be coupled to the grid network and an external burner, to prohibit
system operation at low efficiency regimes, and within critical values of current density. The coupling of the system to the grid also allows the system to be shutdown

44

APPENDIX

APPLICATION OF AN IMPROVED OPERATIONAL STRATEGY

15

in periods of low demand, such as the summer period. This means that only a single
turn on/off is required, which in effect can assist in maintaining a high lifetime for
the fuel cell stack, and also minimize losses caused by frequent starts/stops.
The average net electrical efficiency of the system is 32.8%, but it is expected that
higher efficiencies will be achieved, when the experimental SMR reactor, utilized in
this study, has been optimized. The average total system efficiency is 85.9%, which
verifies the ability of such a system to achieve high total system efficiencies. Nevertheless, a market penetration breakthrough of the proposed system into the area
of residential prime movers will only be justified when it has met a threshold point,
in terms of total life cycle cost. In comparison to non-fuel cell-based micro-CHP
systems, such as SE- and ICE-based systems, the proposed system has significantly
higher net electrical efficiencies. In addition, the system allows lower heat-to-power
ratios, which aid the system to avoid high thermal surpluses throughout annual
operation. The improved operation also targets into minimizing grid interaction,
which can be justified by economic reasons, taxation of purchased/sold electricity,
and technical reasons such as distribution losses. Although the exported electricity
has been reduced to 676 kWh, the imported electricity remains significantly high
at 1,127 kWh. This is caused by the inevitable constraining of the net electrical
power output of the system and the system shutdown.
Acknowledgment
The authors would like to acknowledge the support of Danfoss and Dantherm
Power throughout the realization of this research study.
References
[1] R. OHayre, W. Colella, S.-W. Cha, F.B. Prinz, Fuel Cell Fundamentals, 2nd
ed., Wiley, Hoboken, New Jersey, 2009.
[2] L. Barelli, G. Bidini, F. Gallorini, A. Ottaviano, An energeticexergetic comparison between PEMFC and SOFCbased microCHP systems, International
Journal of Hydrogen Energy. 36 (2011) 32063214.
[3] L. Barelli, G. Bidini, F. Gallorini, A. Ottaviano, Dynamic analysis of PEMFCbased CHP systems for domestic application, Applied Energy. 91 (2012) 13-28.
[4] V. Dorer, A. Weber, Energy and CO2 emissions performance assessment of
residential micro-cogeneration systems with dynamic whole-building simulation
programs, Energy Conversion and Management. 50 (2009) 648-657.
[5] A.R. Korsgaard, M.P. Nielsen, S.K. Kr, Part one: A novel model of HTPEMbased micro-combined heat and power fuel cell system, International Journal of
Hydrogen Energy. 33 (2008) 1921-1931.
[6] A.R. Korsgaard, M.P. Nielsen, S.K. Kr, Part two: Control of a novel HTPEMbased micro combined heat and power fuel cell system, International Journal of
Hydrogen Energy. 33 (2008) 1921-1931.
[7] B.V. Mathiesen, H. Lund, Comparative analyses of seven technologies to facilitate the integration of fluctuating renewable energy sources, Engineering and
Technology. 3 (2009) 190- 204.
[8] M. De Paepe, P. DHerdt, D. Mertens, Micro-CHP systems for residential applications, Energy Conversion and Management. 47 (2006) 3435-3446.
[9] B. Thomas, Benchmark testing of Micro-CHP units, Applied Thermal Engineering. 28 (2008) 2049-2054.
[10] H. Onovwiona, V. Ismetugursal, A. Fung, Modeling of internal combustion engine based cogeneration systems for residential applications, Applied Thermal
Engineering. 27 (2007) 848-861.

APPENDIX

45

16

A. ARSALIS ET AL.

[11] K. Alanne, N. S
oderholm, K. Siren, I. Beausoleil-Morrison, Techno-economic
assessment and optimization of Stirling engine micro-cogeneration systems in
residential buildings, Energy Conversion and Management. 51 (2010) 26352646.
[12] M. Dentice dAccadia, M. Sasso, S. Sibilio, L. Vanoli, Micro-combined heat
and power in residential and light commercial applications, Applied Thermal
Engineering. 23 (2003) 1247-1259.
[13] F. Caresana, C. Brandoni, P. Feliciotti, C.M. Bartolini, Energy and economic
analysis of an ICE-based variable speed-operated micro-cogenerator, Applied
Energy. 88 (2011) 659-671.
[14] A. Arsalis, M.P. Nielsen, S.K. Kr, Modeling and off-design performance of
a 1 kWe HT-PEMFC (high temperature-proton exchange membrane fuel cell)based residential micro-CHP (combined-heat-and-power) system for Danish singlefamily households, Energy. 36 (2011) 5010-5020.
[15] R.J. Braun, Techno-economic optimal design of solid oxide fuel cell systems for
micro-combined heat and power applications in the U.S., Journal of Fuel Cell
Science and Technology. 7 (2010).
[16] T. Wakui, R. Yokoyama, K.-ichi Shimizu, Suitable operational strategy for
power interchange operation using multiple residential SOFC (solid oxide fuel
cell) cogeneration systems, Energy. 35 (2010) 740-750.
[17] J. Zhang, Z. Xie, J. Zhang, Y. Tang, C. Song, T. Navessin, et al., High temperature PEM fuel cells, Journal of Power Sources. 160 (2006) 872-891.
[18] S.J. Andreasen, S.K. Kr, Modelling and evaluation of heating strategies for
high temperature polymer electrolyte membrane fuel cell stacks, International
Journal of Hydrogen Energy. 33 (2008) 4655-4664.
[19] Danish Energy Association, Danish Electricity Supply08 - Statistical Survey,
2009.
[20] M. Houwing, A. Ajah, P. Heijnen, I. Bouwmans, P. Herder, Uncertainties in the
design and operation of distributed energy resources: The case of micro-CHP
systems, Energy. 33 (2008) 1518-1536.
[21] A. Ferguson, Fuel cell modelling for building cogeneration applications, Journal
of Power Sources. 137 (2004) 30-42.
[22] F. Barbir, PEM Fuel Cells: Theory and Practice, Elsevier, 2005.
[23] A. Arsalis, M.P. Nielsen, S.K. Kr, Modeling and parametric study of a 1 kWe
HT-PEMFC-based residential micro-CHP system, International Journal of Hydrogen Energy. 36 (2011) 5010-5020.
[24] A.R. Korsgaard, R. Refshauge, M.P. Nielsen, M. Bang, S.K. Kr, Experimental
characterization and modeling of commercial polybenzimidazole-based MEA
performance, Journal of Power Sources. 162 (2006) 239-245.
[25] R.L. Keiski, T. Salmi, P. Niemisto, J. Ainassaari, V.J. Pohjola, Stationary and
transient kinetics of the high temperature water-gas shift reaction, Applied
Catalysis A: General. 137 (1996) 349-370.
[26] D.S. Salcines, C.R. Estebanez, V.C. Herrero, Simulation of a solar domestic
water heating system with different collector efficiencies and different volumen
storage tanks, in: International Conference on Renewable Energies and Power
Quality, Barcelona, 2004.
Department of Energy Technology, Aalborg University, Pontoppidanstrde 101,
9220, Aalborg ., Denmark
E-mail address: aar@et.aau.dk

46

APPENDIX

Journal Article IV

APPENDIX

47

48

APPENDIX

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Modeling and optimization of a 1 kWe HT-PEMFC-based

micro-CHP residential system
Alexandros Arsalis*, Mads P. Nielsen, Sren K. Kr
Department of Energy Technology, Aalborg University, Pontoppidanstrde 101, 9220 Aalborg ., Denmark

article info

abstract

Article history:

A high temperature-proton exchange membrane (HT-PEMFC)-based micro-combined-

Received 11 August 2011

heat-and-power (CHP) residential system is designed and optimized, using a genetic

Received in revised form

algorithm (GA) optimization strategy. The proposed system consists of a fuel cell stack,

11 October 2011

steam methane reformer (SMR) reactor, water gas shift (WGS) reactor, heat exchangers,

Accepted 18 October 2011

and other balance-of-plant (BOP) components. The objective function of the single-

Available online 16 November 2011

objective optimization strategy is the net electrical efficiency of the micro-CHP system.
The implemented optimization procedure attempts to maximize the objective function by

Keywords:

variation of nine decision variables. The value of the objective function for the optimum

PBI

design configuration is significantly higher than the initial one, with a 20.7% increase.

HT-PEMFC

Copyright 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

Micro-CHP

reserved.

Residential system
Optimization
Fuel cell system

1.

Introduction

A residential application of combined-heat-and-power (CHP)

technology is a micro-CHP system, providing electricity and
heat (hot water and space heating) for a detached singlefamily household. Such a system is designed to convert the
chemical energy in a fuel into both electrical power and
useful heat [1e5]. Micro-CHP systems operating on natural
gas must be coupled with a fuel processing unit, to allow
conversion of natural gas to hydrogen. Balance-of-plant (BOP)
components are also needed to perform various necessary
tasks, such as air compressing or water pumping, while heat
exchangers are necessary for the thermal management of
the system. The thermal management of the system includes
heating/cooling of components (e.g. steam reforming), and
also heat recovery to satisfy the residential load profile (e.g.
space heating). A thermal storage tank is coupled with the

system to provide greater operational flexibility during

transient load demands.
Micro-CHP systems can be categorized into combustionand fuel cell-based. Although combustion-based technologies
are more mature and currently available in the market, fuel
cell-based systems are considered more promising for
a number of reasons. Combustion-based systems, such as the
internal combustion engine technology, are not suitable for
micro-CHP applications mainly due to their high thermal-toelectric ratio (TER) [1], and also due to their low efficiencies
at part-load operation. Fuel cell-based stationary power
generation technology is capable of achieving high efficiencies, with lower emissions as compared to combustion-based
systems. Also these systems have simple routine maintenance requirements and quiet operation [1,6,7].
The operating temperature in a fuel cell stack is an
important element for the efficiency and the degradation of

* Corresponding author.
E-mail address: aar@et.aau.dk (A. Arsalis).
0360-3199/$ e see front matter Copyright 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2011.10.081

APPENDIX

49

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

the membrane. Fuel cell technologies operating at high

temperatures can lessen the cooling requirements, simplify
water management and reduce contamination problems.
High temperature-proton exchange membrane fuel cells (HTPEMFC) utilize a Polybenzimidazole (PBI) membrane, which
operates at temperatures between 150 and 200
C. It is therefore an ideal match for a micro-CHP system, because not only
the rates of electrochemical kinetics are enhanced and water
management and cooling is simplified, but also useful waste
heat can be recovered, and lower quality reformed hydrogen
may be used as fuel [8e11].
A global optimization strategy is usually desirable for
multi-component systems, such as the micro-CHP system
under study, because the global maximum is not just the best
solution to the optimization problem, but also because local
maxima can severely confound the interpretation of the
results of studies investigating the effects of model parameters [12]. A stochastic method, such as genetic algorithms
(GA), can solve a problem with a systematic multi-start
approach with random sampling.

2.

System layout

The proposed system, shown in Fig. 1, is modeled in the

commercially available software EES (Engineering Equation
Solver). Natural gas, used as the system fuel input, is converted to a hydrogen-rich mixture, throughout a fuel processing series of steps. The fuel processing subsystem

includes a desulfurizer, steam methane reforming (SMR)

reactor, water gas shift (WGS) reactor and combustor. The fuel
cell stack receives the reformate fuel, without any water
content since this is removed by the condenser/water-knock
out stage. The fuel cell stack exhaust mixture is then fed to
the combustor, since it contains a large amount of air mixture,
which can be used in the combustion process. The exhaust
mixture also contains small traces of unreacted methane and
carbon monoxide, which are combusted in the combustor. If
additional air and/or natural gas are needed, they are provided
by the air blower and the fuel line, respectively. The
combustor output flue gas is used in the SMR reactor, the fuel
preheater, the steam generator, and finally in the thermal
storage tank for cogeneration purposes. The thermal
management of the system is provided by the use of four heat
exchangers.
The model includes 28 state points (nodes). In order to
synthesize and design the proposed system, the energy
requirements for a representative residential building must be
established. For the current research study, the representative
residential building is a typical Danish single-family household (130 m2 house with four persons). A micro-CHP unit of
approximately 1 kWe is considered. The type of residential
loads considered, include the electrical load, the space heating
load and the hot water load. The mean residential load
requirements are based on [4,13] and are shown in Table 1.
The current research study assumes the electrical load and
the combined heating load as dependent variables, calculated
with respect to the values of the independent parameters

Fig. 1 e Configuration of the proposed micro-CHP system.

50

2471

APPENDIX

2472

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

Table 1 e Mean residential load requirements for

a Danish single-family household.
Residential Load Type


RTcell
i
1
sinh
aanode F
2keh qH2

(3)

hc







RTcell
i0 i
i
Rdiff
ln
4acathode F
i0
lair  1

(4)

Time Segment
Winter

Summer

Spring

Mean Max Mean Max Mean Max

Electrical Load (We)
Space Heating Load (Wth)
Hot Water Load (Wth)

ha

540
1450
330

950
1930
1080

380
70
230

650
140
1420

460
600
310

920
1070
1620

where ai is the charge transfer coefficient, lair is the cathode

air stoichiometric ratio, keh is the electro-oxidation rate of
hydrogen, qH2 is the surface coverage of hydrogen, i is the
current density, and i0 is the exchange current density.
The ohmic losses are given by,
hohmic iRohmic

used in the optimization study. The performance of the

system is analyzed in terms of electrical efficiency, which is
the objective function of the optimization problem. EES was
selected as the modeling tool for this research study, because
it includes many built-in mathematical and thermo-physical
property functions, while an optimization capability in the
software, so-called min/max, can minimize or maximize
a single parameter, while varying up to twenty independent
parameters (decision variables).

3.

In this section, the modeling assumptions of the main

components (fuel cell stack, SMR reactor, WGS reactor) used in
the proposed micro-CHP system are described and analyzed.
As shown below, the modeling characteristics are based on
theoretical, empirical, and experimental assumptions, to
reflect a realistic system configuration. The current study does
not consider component heat losses, or a detailed operational
pattern of the thermal storage tank (assumed to be a heat
exchanger).

3.1.

Therefore, the total cell voltage can be formulated as,

Vcell V0  ha  hc  hohmic

Fuel cell stack

The fuel cell stack model is based on the modeling assumptions previously published by some of the authors in [4,10].
The model considers only the reaction of hydrogen with
oxygen, while all other species are considered inactive. The
ohmic and diffusion resistances are based on linear regressions evaluated experimentally as follows,
Rohmic 0:0001667Tcell 0:2289

(1)

Rdiff 0:4306  0:0008203Tcell

(2)

where Tcell is the fuel cell operating temperature.

For the research work under study, it is assumed optimization takes place only at nominal load (1 kWe). Therefore the
fuel cell is not expected to operate at very low current densities, where hydrogen crossover might have significant effects.
In other words, the steady-state nature of the simulation does
not consider start-up effects that may result in hydrogen
crossover [14].
The anode and cathode overpotentials are given from the
following expressions, respectively,

(6)

where V0 is the open circuit voltage.

The polarization curves of the fuel cell simulation model
are compared with experimental data, found in [10], for
different CO concentrations (1000, 10000, and 20000 ppm). By
observation of Fig. 2, it can be concluded that the simulation
model is in good agreement with the experimental data at all
CO concentrations.

3.2.

Modeling of the micro-CHP system

(5)

SMR reactor

The SMR reactor receives the methaneesteam mixture to be

converted by the steam methane reforming reaction to
a hydrogen-rich reformate gas. The SMR reactor is based on
the modeling assumptions found in [7]. The highly endothermic reaction requires external heating by means of a high
temperature flue gas, provided by the combustor. For
modeling simplification purposes, both gases are assumed to
behave ideally throughout the whole reactor length, assuming
a plug flow pattern. All tubes in the reactor behave independently and a uniform temperature exists throughout each
catalyst particle, which is the same as the gas temperature in
that section of the catalytic bed. Carbon deposition and bed
pressure drops are neglected. The chemical equilibrium
equation is given by means of a chemical equilibrium
constant of the overall reaction, which is defined as,


KSMR

1=2 
1=2 
7=2

2
yCO;o
yCO2 ;o
yH2 ;o
peq



3=2
p0
yCH4 ;o yH2 O;o

(7)

where yi is the chemical equilibrium mole fraction of species

i,p0 is the atmospheric pressure, and peq is the reformate gas
mixture equilibrium pressure.
The length of the reactor is calculated by means of a plug
flow design equation,

LSMR

n_ CH4 ;i
ntubes rB A

XCH4 ;eq

dXCH4
rCH4

(8)

where n_ CH4 ;i is the inlet molar flow rate of methane (kmol/s),

ntubes is the number of tubes, rB is the catalyst bulk density (kg/
m3), A is the cross-sectional area of a single tube (m2), XCH4 is
the kinetic methane conversion, and rCH4 is the demethanation reaction rate (kmol/kg-h).

APPENDIX

51

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

2473

Fig. 2 e Fuel cell model validation.

3.3.

WGS reactor

thermodynamic equilibrium of the slightly exothermic

wateregas shift reaction is calculated as follows [16],

The WGS reactor receives the hydrogen-rich reformate gas

from the SMR reactor and its main purpose is to reduce the
carbon monoxide-content, while simultaneously achieving
a slight increase in the hydrogen-content. Although experimental tests [15] indicate that an addition of water in the WGS
reactor can increase the carbon monoxide conversion, as
a result of the higher steam-to-carbon ratio, this is not followed in the current study. The reason is that steam has to be
removed from the reformate gas, prior to the fuel cell stack
anode inlet. Therefore, this would result in an increase of the
thermal energy loss in the water knockout stage. The carbon
monoxide-content should be reduced at an acceptable level
for the fuel cell stack, which is less than 2% (on a mole
fraction-basis). The kinetic constant is based on a power law
relationship
and
calculated
at
steady-state.
The

Table 2 e Values of the fixed parameters in the

optimization procedure.
Variable
Tcell
Acell
ncells
dWGS
i
rw
rB

52

Description (unit)

Value

Fuel cell operating temperature (
C)

Fuel cell active area (cm2)
Number of cells in the fuel cell stack
Inlet tube diameter of the WGS reactor (m)
Density of the wire mesh catalytic
material of the WGS reactor (kg/m3)
Catalyst bulk density of the
SMR reactor (kg/m3)

160
45.16
267
0.13
1385
1200

KT exp


4400
 4:063
T

(9)

where KT is the equilibrium constant given in [17], and T is the

reactor temperature.
The kinetic power law fit parameters are obtained from
[16],



Ea
KWGS k0 exp
RT

(10)

where k0 is the pre-exponential factor, Ea is the activation

energy (kJ/mol), and R is the ideal gas constant (J/K-mol).
The carbon monoxide conversion is given by,
dxCO
3600rCO
urw
n_ CO
dz

(11)

where u is the water gas shift reactor cross-sectional area (m),

rw is the density of the wire mesh catalytic material (kg/m3),
rCO is the carbon monoxide reaction rate (kmol/kg-s), and n_ CO
is the molar flow rate of carbon monoxide (kmol/s).
The temperature gradient is defined as,
dT
rw

dz rg cp;mass us  rCO dHr;1

(12)

while rg is the density of the reformate gas (kg/m3), us is the

superficial velocity (m/s), and dhr,1 is the enthalpy of reaction
(kJ/mol).

APPENDIX

2474

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

Table 3 e Initial, minimum, maximum and optimum values of the decision variables used in the optimization procedure.
Variable
S/C
lH2
nSMR
tubes
LSMR
reactor
WGS
Lreactor
TComb
fg;out
TSMR
ref;in
TWGS
ref;in
TFP
fg;in

4.

Description (unit)

Initial

Min

Max

Optimum
1st simulation

Optimum
2nd simulation

Steam-to-carbon ratio
Hydrogen stoichiometry
Number of tubes in the SMR
SMR reactor length (m)
WGS reactor length (m)
Combustor output temperature (K)
SMR reformate inlet temperature (K)
WGS reformate inlet temperature (K)
Fuel preheater flue gas inlet temperature (K)

4.000
1.500
275
1.800
0.350
1173
400
499
670

2.500
1.500
275
1.800
0.350
1050
399
470
550

4.000
1.550
300
2.000
0.500
1174
530
500
670

2.908
1.509
297
1.994
0.465
1095
530
474
620

3.002
1.502
298
1.996
0.441
1073
530
472
614

Methodology of the optimization strategy

hel;net

The mathematical model involves a number of state variables

describing the physical state of the system. The state variables
/
/
consist of the degrees of freedom, x and the unknowns, y .
The degrees of freedom are the decision, or independent,
variables determined by the user, while the unknowns are the
dependent variables computed by solving the model equations once the degrees of freedom are fixed [18]. The objective
function, f is optimized
subject to a set of equality and
/
/
inequality constraints, h and g , respectively [7]. Therefore
the general mathematical formulation of the optimization
problem can be expressed as follows:
/ / 
x ;y 0
max f x ; y subject to / / / 
g x ;y 0
/

(13)

The fitness is defined as the set of values of the objective

function as quantified by the value of the variable that is to be
minimized or maximized. The probability an individual, in
every given population, has to be selected for breeding in the
next generation is an increasing function of its fitness. For the
current study, the objective function is the net electrical efficiency, hel, net which is defined as the ratio of the net electrical
power output of the system divided by the chemical energy
input to the system,

_ el;net
W
_
mCH4 ;in LHVCH4

(14)

The simulation model is modeled in EES and is divided in

different sections (so-called modules) for every simulated
component. The EES min/max function, described in detail in
[19], computes the optimum objective function, based on the
input independent variables. The choice of decision variables
is based on their assumed influence on the system performance and their range of variation is based on the component/process limitations. The decision variables include
operational and geometric variables. An operational variable
may be a temperature, while a geometric variable may be the
length of a reactor. Therefore the optimum performance is
determined throughout variables affecting operational and
structural changes of the system.
The nature of this optimization problem represents
a non-linear programming problem. This complexity is due
to the high degree of coupling between the components (e.g.
the combustor with the fuel cell stack and the SMR reactor),
and also the high number of independent variables.
Therefore it cannot be solved with deterministic, gradient
search methods, but only with a stochastic method [20,21].
A typical stochastic method, such as a Genetic Algorithm
(GA) optimization strategy will be able to avoid local
optimum traps and reach the global optimum, because it

Fig. 3 e Evolution of the objective function throughout the optimization procedure.

APPENDIX

53

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

2475

Fig. 4 e Behavioral pattern of the steam-to-carbon ratio at the design point throughout the optimization procedure.

operates on a whole population of points (strings), rather

than single points [20e22]. The initial population of individuals in the GA is set at 64, while the number of generations is set at 512. The maximum mutation rate is set at
a rather high value (0.4375), to allow the algorithm to search
more aggressively for an optimum at locations distant from
the current optimum, since lower values focus the search
more around the current optimum [19]. The GA is stopped
when no further improvement of the objective function for
a number of generations is observed. In summary, the
methodology of the genetic algorithm consists of the
following steps:
1. Generate a random population and evaluate the fitness of
each member.
2. Define the termination conditions.
3. Select parents if the crossover condition is met, select
crossover parameters and apply crossover.

4. Select member if the mutation condition is met, select

mutation parameters and apply mutation.
5. Evaluate the fitness of the offspring and update the
population.
6. Repeat steps 2e6, until the termination conditions are
satisfied.
The optimization model consists of a number of fixed
parameters, shown in Table 2. Some of the geometric variables, such as the WGS reactor diameter, must be kept fixed to
avoid unrealistic results and therefore misleading assumptions. Some thermodynamic properties must be kept fixed as
well, because variation of these parameters would tend to
maximize or minimize. For example, the fuel cell operating
temperature is kept fixed as suggested in [10], because obviously the highest allowable value would yield toward the
maximization of the objective function. The active cell area
and the number of cells are fixed as suggested by [23]. The

Fig. 5 e Behavioral pattern of the hydrogen stoichiometry at the design point throughout the optimization procedure.

54

APPENDIX

2476

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

Fig. 6 e Behavioral pattern of the number of tubes in the SMR reactor at the design point throughout the optimization
procedure.

values for the SMR and WGS reactor parameters are selected
from [7] and [16], respectively.

5.

Results and discussion

In this section the optimum configuration of the proposed

micro-CHP system is presented, and then compared to the
results obtained prior to the application of the optimization
strategy. In addition, the optimization process is analyzed to
indicate its main characteristics.
The decision variables, shown in Table 3, were chosen with
an initial value as typically found in the literature for the kind
of system under study. Their allowable range of variation is
chosen on the basis of component/process operational and/or
structural limitations. Therefore these constraints are based
on knowledge of the solution space and on observation of the

behavioral pattern of the optimization algorithm. The

optimum values, shown in Table 3, are the ones used to
evaluate the global optimum objective function. The optimum
values are evaluated after two sequential simulation runs;
after the first simulation run is completed a second run
repeats the optimization, using the optimum values of the
first optimization as the new initial values. The purpose of this
procedure is to ensure that the best possible value of the
objective function is reached. In addition, the variation of the
nine independent variables is more stable in the second
simulation, as compared to the first simulation, while the
optimization evolves toward the final generations.
The genetic algorithm required 508 generations until the
optimum value for the objective function was reached, while
the simulation for the optimization strategy was performed in
33,346 iterations. By comparison of the optimum and initial
design configurations, it can be concluded that the initial

Fig. 7 e Behavioral pattern of the SMR reactor length at the design point throughout the optimization procedure.

APPENDIX

55

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

2477

Fig. 8 e Behavioral pattern of the WGS reactor length at the design point throughout the optimization procedure.

configuration is significantly different than the optimum one.

The objective function has been significantly improved from
0.3264 to 0.4117, as shown in Fig. 3, in other words a 20.7%
increase. The improvement is more noticeable in the beginning of the optimization process, since after a few generations
the objective function begins to increase rapidly, while after
the 50th generation the increase is slower. After the 280th
generation the maximization is marginal and almost constant
toward the end. This flat behavior indicates that the overall
iterative optimization scheme has practically converged.
The response of the nine decision variables (independent
parameters) used in the optimization procedure is illustrated
in Figs. 4e12. A general trend for all nine decision variables is
the aggressive behavior of the optimization procedure in the
beginning (0e50 generations), varying throughout the whole
allowable range, in an effort of the GA to search randomly as
explained in the previous section. In Fig. 4, the optimum value

of the steam-to-carbon ratio deviates significantly from the

initial guess, which suggests that a high amount of steam will
result in significant system losses because of the condenser,
as explained in section 3. In Fig. 5, the hydrogen stoichiometry
exhibits an almost flat behavior very early (after the 15th
generation), which suggests that the initial value was a good
guess. In Fig. 6, the number of tubes varies throughout the
whole procedure, suggesting the existence of many local
optimums, significantly distant from each other. If a cost
parameter was involved in the optimization strategy, the
selected optimum would probably move to a lower value
(270th generation). In Fig. 7, the value of the SMR reactor
length tends to reach the maximum allowable value, while
exhibiting a very linear behavior at an early stage of the
optimization (22nd generation). In Fig. 8, the behavior of the
value for the WGS reactor follows a similar pattern as Fig. 6,
since these two parameters are interrelated. Equivalently to

Fig. 9 e Behavioral pattern of the combustor output temperature at the design point throughout the optimization procedure.

56

APPENDIX

2478

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

Fig. 10 e Behavioral pattern of the SMR reformate inlet temperature at the design point throughout the optimization
procedure.

the latter, is the behavior of the combustor output temperature value (shown in Fig. 9), which is interdependent to the
SMR reactor length. Obviously when a larger reactor area is
available, more flue gas can be utilized by the endothermic
reaction in the SMR reactor. Another interrelated parameter is
the SMR reformate inlet temperature (shown in Fig. 10). Its
value tends to maximize, because a higher temperature in the
inlet of the SMR reactor will convert hydrogen more effectively in a restricted reactor area. The initial value was very
low, suggesting that this was another source of low efficiency
in the initial configuration. In Fig. 11, the optimum WGS
reformate inlet temperature value deviates significantly from
the initial one, tending to minimize, suggesting the WGS
reaction is more efficient at lower values. Finally, Fig. 12
illustrates the behavioral pattern of the fuel preheater flue
gas inlet temperature. Interestingly the optimum value

reaches an optimum very early (50th generation), which is

around the middle of the allowable range of variation.
Table 4 shows the values of the thermodynamic properties
prior and after the optimization for all state points. After an
observation to this table, a variety of interesting comments
can be made. Beginning from the temperature variation, the
value of the SMR reformate outlet temperature is significantly
higher than the one for the initial configuration, as a consequence of the associated higher value of the SMR reformate
inlet temperature, discussed above. The value of the fuel cell
flue gas temperature, after preheated twice by heat
exchangers II and III, is lower than the initial configuration
because of the reduction of steam-to-carbon ratio in the
optimum configuration. The latter affects the output
combustor temperature, and in effect the SMR reactor outlet
flue gas temperature.

Fig. 11 e Behavioral pattern of the WGS reformate inlet temperature at the design point throughout the optimization
procedure.

APPENDIX

57

Table 4 e Values of the thermodynamic properties for all state points for the initial and optimum configurations.
Ti (K)

Node

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

298
298
298
420
420
420
845
499
566
433
433
433
298
298
468
1014
1173
1016
666
298
487
298
298
298
420
432
564
670

298
298
298
530
530
530
942
472
561
433
433
433
298
298
458
890
1073
909
605
298
464
298
298
298
530
432
553
614

initial
2.833E-05
2.834E-06
2.550E-05
2.550E-05
2.550E-05
1.400E-04
1.400E-04
1.400E-04
1.400E-04
1.400E-04
6.698E-05
2.679E-05
2.928E-04
2.928E-04
1.182E-03
1.182E-03
1.500E-03
1.500E-03
1.500E-03
1.125E-03
1.500E-03
1.500E-03
1.145E-04
1.145E-04
1.145E-04
1.182E-03
1.182E-03
1.500E-03

yCH4

yCO

yCO2

yH2

yH2 O

yN2

yO2

optimum initial optimum initial optimum initial optimum initial optimum initial optimum initial optimum initial optimum
2.550E-05
9.501E-10
2.550E-05
2.550E-05
2.550E-05
1.114E-04
1.114E-04
1.114E-04
1.114E-04
1.114E-04
7.644E-05
3.840E-05
1.038E-04
1.038E-04
1.408E-03
1.408E-03
1.550E-03
1.550E-03
1.550E-03
1.349E-03
1.550E-03
1.550E-03
8.595E-05
8.595E-05
8.595E-05
1.408E-03
1.408E-03
1.550E-03

1.0000
1.0000
1.0000
1.0000
1.0000
0.2000
0.0411
0.0411
0.0411
0.0411
0.0678
0.0678
0.0000
0.0000
0.0060
0.0060
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0060
0.0060
0.0000

1.0000
1.0000
1.0000
1.0000
1.0000
0.2499
0.0140
0.0140
0.0140
0.0140
0.0177
0.0177
0.0000
0.0000
0.0013
0.0013
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0013
0.0013
0.0000

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0568
0.0568
0.0031
0.0031
0.0051
0.0051
0.0000
0.0000
0.0005
0.0005
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0005
0.0005
0.0000

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0787
0.0787
0.0099
0.0099
0.0126
0.0126
0.0000
0.0000
0.0009
0.0009
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0009
0.0009
0.0000

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0568
0.0568
0.1104
0.1104
0.1823
0.1823
0.0000
0.0000
0.0162
0.0162
0.0328
0.0328
0.0328
0.0000
0.0328
0.0328
0.0000
0.0000
0.0000
0.0162
0.0162
0.0328

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0787
0.0787
0.1474
0.1474
0.1864
0.1864
0.0000
0.0000
0.0136
0.0136
0.0284
0.0284
0.0284
0.0000
0.0284
0.0284
0.0000
0.0000
0.0000
0.0136
0.0136
0.0284

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.3973
0.3973
0.4510
0.4510
0.7447
0.7447
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.5506
0.5506
0.6193
0.6193
0.7834
0.7834
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000

0.0000
0.0000
0.0000
0.0000
0.0000
0.8000
0.4481
0.4481
0.3944
0.3944
0.0000
0.0000
0.0000
0.0000
0.0995
0.0995
0.1300
0.1300
0.1300
0.0000
0.1300
0.1300
1.0000
1.0000
1.0000
0.0995
0.0995
0.1300

0.0000
0.0000
0.0000
0.0000
0.0000
0.7501
0.2781
0.2781
0.2094
0.2094
0.0000
0.0000
0.0000
0.0000
0.0973
0.0973
0.1424
0.1424
0.1424
0.0000
0.1424
0.1424
1.0000
1.0000
1.0000
0.0973
0.0973
0.1424

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.7900
0.7900
0.6935
0.6935
0.6928
0.6928
0.6928
0.7900
0.6928
0.6928
0.0000
0.0000
0.0000
0.6935
0.6935
0.6928

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.7900
0.7900
0.7006
0.7006
0.6797
0.6797
0.6797
0.7900
0.6797
0.6797
0.0000
0.0000
0.0000
0.7006
0.7006
0.6797

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.2100
0.2100
0.1844
0.1844
0.1444
0.1444
0.1444
0.2100
0.1444
0.1444
0.0000
0.0000
0.0000
0.1844
0.1844
0.1444

0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.2100
0.2100
0.1862
0.1862
0.1495
0.1495
0.1495
0.2100
0.1495
0.1495
0.0000
0.0000
0.0000
0.1862
0.1862
0.1495

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

initial optimum

_ i kg=s
m

2479

58

APPENDIX

2480

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

Fig. 12 e Behavioral pattern of the fuel preheater flue gas inlet temperature at the design point throughout the optimization
procedure.

Further on, the mass flow rate variation indicates some

significant changes in the flow pattern of the system. A
parameter associated with the net electrical efficiency
(objective function), is the value of the methane mass flow
rate. Clearly this value decreases (state point 1), because of the
reduction in additional fuel supply from the line (state point
2). A reduction in state points 6e11, is justified by the
decreased steam-to-carbon ratio, as explained above. The
depleted fuel increases because of the increased hydrogen
stoichiometry in the optimum configuration. Also, the
increased amount of fuel cell stack exhaust (state point 16),
reduces the need for extra air supply (state points 13e14).
The most significant parameters from the mole fractions
are the carbon monoxide and hydrogen species. In the
optimum configuration, the carbon monoxide-content is
significantly higher in the reformate gas (state points 7e11),
due to the lowered steam-to-carbon ratio, but still within the
limitations of the PBI-based fuel cell technology. The most
significant state point, for comparison purposes, is the
hydrogen variation in state point 11, since it is at a dry basis.
The increase from 0.7447 to 0.7834, suggests a better

Fig. 13 e Net electrical power output, heat cogeneration

output and system losses for the optimum configuration.

utilization of the SMR reactor in the optimum configuration,

which is due to the improved combination of geometries
(number of tubes, reactor length) and inlet temperatures
(reformate and flue gas).
Fig. 13 illustrates the pattern of energy output/loss for the
proposed micro-CHP system. This figure suggests that the
highest system loss is due to the condenser (7.8%), while the
losses from the exothermic reaction in the WGS reactor and
the parasitic losses from the BOP-components (air blower and
water pump) are rather insignificant (1.3%). However, it should
be noted that a more realistic model, with heat loss considerations and other component coupling-related losses would
result in a further decrease of the total system efficiency.

6.

Conclusions

A GA-based optimization strategy was applied to a HT-PEMFCbased micro-CHP system. The value of the net electrical efficiency (objective function) obtained is 0.412, while the values
of the cogeneration thermal efficiency and total system efficiency are 0.497 and 0.91, respectively. The results of this
research study indicate that the proposed micro-CHP system
must focus on all system components, paying special attention at their geometric dimensions and operational variables.
The applied global optimization strategy successfully
managed to locate a system configuration with a significantly
higher net electrical efficiency than the one obtained initially.
The choice on the values of the independent variables is still
within the feasible limits of operating the system. The theoretical high tolerance of PBI technology on carbon monoxidecontent in the reformate fuel fed to the fuel stack anode,
allows the use of a low steam-to-carbon ratio as indicated by
the results. A more in-depth study for the water knockout/
condenser stage might be necessary in future studies to
minimize the losses occurred in that stage.
Finally, the applied global optimization strategy avoids the
typical trap of designing a fuel cell-based system with a high

APPENDIX

59

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 2 4 7 0 e2 4 8 1

fuel cell stack efficiency, but low net electrical efficiency. This
is accomplished by taking into account the geometric and
operational parameters of the fuel processing components,
such as the SMR and WGS reactors. A rigorous variation of
these parameters shifts the results toward the maximization
of the net electrical efficiency. Although this study has not
included cost-related parameters, such as the unit cost of fuel,
it should be noted that a high net electrical efficiency has a high
impact factor on the reduction of the system operational cost.

Acknowledgments
The authors would like to acknowledge the support of Danfoss
and Dantherm Power throughout the realization of this
research study.

references

[1] Hawkes AD, Leach MA. The capacity credit of microcombined heat and power. Energy Policy 2008;36(4):1457e69.
[2] Hawkes AD, Leach MA. Modelling high level system design
and unit commitment for a microgrid. Applied Energy 2009;
86(7e8):1253e65.
[3] Hawkes AD, Leach MA. Cost-effective operating strategy for
residential micro-combined heat and power. Energy 2007;
32(5):711e23.
[4] Korsgaard AR, Nielsen MP, Kr SK. Part one: a novel model of
HTPEM-based micro-combined heat and power fuel cell
system. International Journal of Hydrogen Energy 2008;33(7):
1921e31.
[5] Peacock A, Newborough M. Impact of micro-CHP systems on
domestic sector co emissions. Applied Thermal Engineering
2005;25(17e18):2653e76.
[6] Gunes MB, Ellis MW. Evaluation of energy, environmental,
and economic characteristics of fuel cell combined heat and
power systems for residential applications. Journal of Energy
Resources Technology 2003;125(3):208.
[7] Georgopoulos NG. Application of a decomposition strategy to
the optimal synthesis/design and operation of a fuel cell
based total energy system. Virginia Tech; 2002.

60

2481

[8] Bergmann A, Gerteisen D, Kurz T. Modelling of co poisoning

and its dynamics in HTPEM fuel cells. Fuel Cells 2010;10(2):
278e87.
[9] Zhai Y, Zhang H, Zhang Y, Xing D. A novel H3PO4/nafionePBI
composite membrane for enhanced durability of high
temperature PEM fuel cells. Journal of Power Sources 2007;
169(2):259e64.
[10] Korsgaard AR, Nielsen MP, Bang M, Kr SK. Modeling of co
influence in PBI electrolyte PEM fuel cells. In: Fuelcell 2006;
2006. Irvine, CA.
[11] Mocoteguy P, Ludwig B, Scholta J, Barrera R, Ginocchio S.
Long term testing in continuous mode of HT-PEMFC based
H3PO4/PBI celtec-p MEAS for m-CHP applications. Fuel Cells
2009;9(4):325e48.
[12] Ravindran A, Ragsdell KM, Reklaitis GV. Engineering
optimization: methods and applications. Hoboken, New
Jersey: Wiley; 2006.
[13] Korsgaard a, Nielsen M, Kar S. Part two: control of a novel
htpem-based micro combined heat and power fuel cell
system. International Journal of Hydrogen Energy 2008;33(7):
1921e31.
[14] Barbir F. Pem fuel cells: theory and practice. Elsevier; 2005.
[15] Kolb G. Fuel processing for fuel cells. Weinheim, Germany:
Wiley-VCH; 2008.
[16] Keiski RL, Salmi T, Niemisto P, Ainassaari J, Pohjola VJ.
Stationary and transient kinetics of the high temperature
water-gas shift reaction. Applied Catalysis A: General 1996;
137(2):349e70.
[17] Davies J, Lihou D. Optimal design of methane steam
reformer. Chem. Proc. Eng 1971;52:71e80.
[18] Marechal F, Palazzi F, Godat J, Favrat D. Thermo-economic
modelling and optimisation of fuel cell systems. Fuel Cells
2005;5(1):5e24.
[19] F-Chart Software. Ees manual. www.fchart.com, V8.590
Edition: 2010.
[20] Pelster S, Favrat D, von Spakovsky MR. The thermoeconomic
and environomic modeling and optimization of the
synthesis, design, and operation of combined cycles with
advanced options. Journal of Engineering for Gas Turbines
and Power 2001;123(4):717.
[21] Calise F, Dentice dAccadia M, Vanoli L, von Spakovsky MR.
Single-level optimization of a hybrid SOFCeGT power plant.
Journal of Power Sources 2006;159(2):1169e85.
[22] Sieniutycz S, Jezowski J. Energy optimization in process
systems. Amsterdam: Elsevier; 2009.
[23] Serenergy. Serenergy.dk 2011.

APPENDIX

Journal Article V

APPENDIX

61

62

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED

MICRO-CHP SYSTEM BY FORMULATION AND APPLICATION
OF A PROCESS INTEGRATION METHODOLOGY
ALEXANDROS ARSALIS, MADS P. NIELSEN, SREN K. KR
Abstract. A 1 kWe micro-CHP system based on high temperature PEMFC
technology is modeled and optimized by formulation and application of a process integration methodology. The system can provide heat and electricity
for a single-family household. It consists of a fuel cell stack, a fuel processing subsystem, heat exchangers, and balance-of-plant components. Since high
temperature PEMFC technology allows the use of synthesis gas with relatively
high carbon monoxide contents, as compared to Nafion-based fuel cell stacks,
the fuel processing subsystem is relatively simple, consisting only of steam
methane reforming and a water gas shift reactors. The optimization methodology involves system optimization attempting to maximize the net electrical
efficiency, and then by use of a MINLP problem formulation, the heat exchange
network annual cost is minimized. The results show the high potential of the
proposed model since high efficiencies are accomplished. The net electrical
efficiency and total system efficiency (based on LHV) vary from 35.2 to 43.6%,
and 65.2 to 91.1%, respectively. The minimized total annual cost of the HEN
is $8,147/year.

1. Introduction
Combined-heat-and-power (CHP) is a well-known technology for large-scale stationary plants. These systems are designed to convert the chemical energy in a fuel
into both electrical power and useful heat. During recent years, another application
of the CHP technology has evolved, which includes smaller scale applications such
as a residential micro-CHP system. This system can provide electricity and heat
(hot water and space heating) for a detached single-family household. Systems
operating on natural gas must be integrated with a fuel processing subsystem, to
allow conversion of natural gas to hydrogen. Heat exchangers are also used for the
thermal management of the system, while balance-of-plant components perform
various necessary tasks, such as air compressing or water pumping. The thermal
management of the system includes heating and cooling processes (e.g. preheating
of the steam-methane mixture, or cooling of the reformate mixture prior to the
water gas shift reaction), and also heat recovery to fulfill a residential load profile
(e.g. space heating). Finally, a thermal storage tank can be coupled to the system
to provide greater operational flexibility during transient heating load demands.
Pinch analysis and process integration is a method used to analyze the heating
and cooling processes of a system and then assist in identifying energy saving opportunities, through a more efficient heat exchange network (HEN) design [1-10].
An optimum HEN design will allow cost reductions, both capital and operating
ones [1,11,12]. The objective is to assist process engineers to identify and model
HEN configurations which are initially considered unknown. The technique can
be summarized in two main steps: the targeting phase, aiming at identifying the
Received by the editors 2011.
Key words and phrases. PBI; heat exchange network; micro-CHP; optimization; process integration; fuel cell system.
1

APPENDIX

63

A. ARSALIS ET AL.

minimum energy requirement (or equivalently maximum energy recovery) (MER)

of the system by calculating the possible energy recovery from the hot streams to
heat up the cold streams of the system, and the HEN synthesis that will implement
the targeted energy recovery.
Pinch analysis and process integration for fuel cell-based systems has been applied in several research studies found in the literature. Wallmark [11] studied the
process integration of a low temperature PEMFC (proton exchange membrane fuel
cell) system for Swedish residential applications, while Verda [13] performed simultaneous thermo-economic optimization and process integration for a biogas-fuelled
MCFC (molten carbonate fuel cell)-based hybrid system for the combined production of electricity and hydrogen. Godat [3] investigated the optimization of a low
temperature PEMFC system using process integration techniques. The improvement in the heat integration of the system has resulted in an electrical efficiency
increase of 14%. Marechal [6] studied a general methodology for the optimization
of fuel cell-based systems, including process integration. This study combined the
use of process simulation and process integration techniques to calculate thermoeconomic performances used in a multi-objective optimization framework. Palazzi
[14] investigated a methodology applied to planar SOFC (solid oxide fuel cell)-based
systems, where the system is integrated using the pinch-based methods that rely on
optimization techniques. This defines the minimum energy required and sets the
basis for an ideal HEN design. For the optimized design electrical efficiencies up
to 44% have been reported. Finally, Autissier [15] investigated larger scale SOFCmicro-GT (gas turbine) systems combining pinch analysis and process integration
techniques, with multi-objective thermo-economic optimization, resulting in total
efficiencies up to 70%.
The objective of the current research study is to examine the possibility of designing an optimum HEN for a HT (high temperature)-PEMFC-based micro-CHP
system. The research work analyzed in the following sections includes: (1) description of the modeling assumptions for the initial configuration of the micro-CHP
system, (2) Identification of the hot and cold streams and application of pinch
analysis and process integration techniques, (3) HEN design for MER, (4) HEN optimization by application of a Mixed-Integer Non-Linear Programming (MINLP)
approach based on [16]. The latter defines a super-configuration of the HEN allowing simultaneous cost and external heating/cooling utility minimizations.
2. System description
The initial configuration of the proposed micro-CHP system is shown in Fig. 1.
The system is fuelled with natural gas (assumed to behave as methane for calculation simplicity), which is converted to a hydrogen-rich reformate mixture. This
is accomplished by use of a fuel processing subsystem, which includes a desulfurizer, steam methane reforming (SMR) reactor, water gas shift (WSG) reactor and
catalytic combustor. The reformate fuel is fed to the anode of the fuel cell stack
for electricity production. The fuel cell stack exhaust mixture is then fed to the
combustor, since it contains a large amount of air mixture, which can be used in
the combustion process. In addition, the exhaust mixture contains small traces of
unreacted methane and carbon monoxide, which can be fully reacted in the combustor. If additional air and/or natural gas are needed, they are provided by the
air blower and the fuel line, respectively. The combustor output flue gas is used
as a heat source for the SMR reactor, the fuel preheater, the steam generator, and
finally the thermal storage tank.
The model includes 27 state points (nodes). In order to design and synthesize the
proposed system the energy requirements for a representative residential building

64

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

Figure 1. Initial configuration of the proposed micro-CHP system.

must be established. For the current research study, the representative residential
building is a typical Danish single-family household (130 m2 house with four persons). A micro-CHP unit of approximately 1 kWe (nominal load) is considered.
The model is simulated in the commercially available software EES (Engineering
Equation Solver), which is selected as the modeling tool for this research study
since it includes many built-in mathematical, thermo-physical property, and optimization functions. For the optimization of the HEN, the optimization software
GAMS (General Algebraic Modeling System) is used.
3. Plant model
In this section, the modeling assumptions of the main components integrated
in the proposed micro-CHP system are described and analyzed. As shown below,
the modeling characteristics are based on theoretical, empirical, and experimental
assumptions, to reflect a realistic system configuration.
3.1. Steam methane reformer reactor. The SMR reactor receives the methanesteam mixture to be converted by the steam methane reforming reaction to a
hydrogen-rich reformate gas. The highly endothermic reaction requires external
heating by means of a high temperature flue gas, provided by the combustor. For
modeling simplification purposes, both gases are assumed to behave ideally throughout the whole reactor length, assuming a plug flow pattern. All tubes in the reactor
behave independently and a uniform temperature exists throughout each catalyst
particle, which is the same as the gas temperature in that section of the catalytic
bed. Carbon deposition and bed pressure drops are neglected.
The SMR reaction is described by two reactions: the endothermic demethanation reaction and the exothermic water gas shift one. The combination of these
two partial chemical reaction mechanisms leads to the overall reaction mechanism.

APPENDIX

65

A. ARSALIS ET AL.

These three reactions are shown below,

(1)

CH4 + H2 O
CO + 3H2

(2)

CH4 + 2H2 O
CO2 + 4H2

(3)

2CH4 + 3H2 O
CO + CO2 + 7H2

The chemical equilibrium equation is given by means of a chemical equilibrium

constant of the overall reaction, which is defined as,
2
1/2
1/2
7/2 
peq
(yCO,o ) (yCO2 ,o ) (yH2 ,o )
(4)
KSM R =
3/2
p0
(yCH4 ,o )(yH2 O,o )
where yi is the chemical equilibrium mole fraction of species i, p0 is the atmospheric
pressure, and peq is the reformate gas mixture equilibrium pressure. The length of
the reactor is calculated by means of a plug flow design equation [17],
XCH4, eq

(5)

LSM R

n CH4 ,i
=
ntubes B A

Z
0

dXCH4


SM R
rCH
4

where n CH4 ,i is the inlet molar flow rate of methane (kmol/s), ntubes is the number of tubes, B is the catalyst bulk density (kg/m3 ), A is the cross-sectional area
SM R
of a single tube (m2 ), XCH4 is the kinetic methane conversion, and rCH
is the
4
demethanation reaction rate (kmol/kg-h), which is given by the following expression,


EA
SM R
SM R
(6)
rCH
=
k
exp

pCH4
0
4
SM R
RTavg
where k0SM R is the Arrhenius frequency factor (kmol/kg-h), EA is the Arrhenius
SM R
is the average reformate gas temperature (K),
activation energy (kJ/kmol), Tavg
R is the ideal gas constant (kJ/kmol-K), and pCH4 partial pressure of methane
(bar).
3.2. Water gas shift reactor. The WGS reactor receives the reformate gas from
the SMR reactor and its main purpose is to reduce the carbon monoxide content,
while simultaneously achieving a slight increase in the hydrogen-content. The carbon monoxide content should be reduced at a level below the maximum acceptable
for the fuel cell stack, which is 5% according to the fuel cell stack manufacturer
[18]. The kinetic constant is based on a power law relationship and calculated at
steady-state. The thermodynamic equilibrium of the slightly exothermic water-gas
shift reaction is calculated as follows [19],


4400
(7)
KT = exp
4.063
TW GS
where KT is the equilibrium constant given in [20], and TW GS is the temperature
of the WGS reactor. The kinetic power law fit parameters are obtained from [19],


EA
W GS
(8)
KW GS = k0
exp
RT

where k0W GS is the pre-exponential factor. The carbon monoxide conversion is given
by,
dxCO
rCO
(9)
= w
dz
n CO
where is the cross-sectional area of the WGS reactor (m), w is the density of the
wire mesh catalytic material (kg/m3 ), rCO is the carbon monoxide reaction rate

66

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

(kmol/kg-s), and n CO is the molar flow rate of carbon monoxide (kmol/s). The
temperature gradient is defined as,
dT
w
(10)
=
dz
g cp,mass us (rCO dHr,1 )
where g is the density of the reformate gas (kg/m3 ), us is the superficial velocity
(m/s), and dHr,1 is the enthalpy of reaction (kJ/mol).
3.3. Fuel cell stack. The fuel cell operating temperature is an important factor in
regards of efficiency and fuel cell membrane degradation. Increased operating temperatures aid in the reduction of cooling requirements, simplify water management
and minimize contamination problems. An HT-PEMFC utilizes a PBI (Polybenzimidazole) membrane, operating at temperatures between 140 and 180o C [21].
Therefore this fuel cell type can be coupled effectively in a micro-CHP system, because waste heat can be recovered, and lower quality reformed hydrogen may be
used as fuel [22]. The model considers only the reaction of hydrogen with oxygen,
while all other species are considered inactive. The ohmic and diffusion resistances
are based on linear regressions evaluated experimentally in [23],
(11)

Rohmic = 0.0001667Tcell + 0.2289

(12)

Rdif f = 0.4306 0.0008203Tcell

where Tcell is the fuel cell operating temperature. The anode and cathode overpotentials are given by the following expressions, respectively,


i
RTcell
sinh1
(13)
a =
anode F
2keh H2




i0 + i
RTcell
i
ln
(14)
c =
+ Rdif f
4cathode F
i0
air 1
where i is the charge transfer coefficient, air is the cathode air stoichiometric
ratio, keh is the electro-oxidation rate of hydrogen, H2 is the surface coverage of
hydrogen, i is the current density, and i0 is the exchange current density. The
ohmic losses are given by,
(15)

ohmic = iRohmic

Therefore, the total cell voltage can be formulated as,

(16)

Vcell = V0 a c ohmic

where V0 is the open circuit voltage.

3.4. Heat losses. Component heat losses are calculated on the basis of each components dimensions and the insulations material properties and thickness. Heat
exchanger and piping losses are neglected since the initial HEN configuration is
not known in advance. Therefore component heat losses are only considered for the
SMR reactor, fuel cell stack and combustor. A thermal conductivity of k = 0.04 W/(m-K)
and a convective heat transfer coefficient of h = 6 W/(m2 -K) are used, as found
in the literature [24]. The total heat losses for each component are given by the
following relation,
Tc Tadj
Tc T
+
(17)
Q loss = Q loss,ca + Q loss,cc = Lca
Lcc
1
kAca + hAca
kAcc
where Q loss,ca and Q loss,cc are the component-to-atmosphere and component-tocomponent heat loss rates (W), respectively. Tc is the average temperature of the
component (K), Tadj is the temperature of the adjacent component (K), T is the
atmospheric temperature (K), Lca is the thickness of the outer insulation (m), Lcc

APPENDIX

67

A. ARSALIS ET AL.

Table 1. Fixed values for the simulation model.

Variable

Description (unit)

Value

Tcell

Fuel cell operating temperature ()

Acell

Fuel cell active area (cm2)

ncells

Number of cells in the fuel cell stack

267

diWGS

Inlet tube diameter of the WGS reactor (m)

0.13

Density of the wire mesh catalytic material of the WGS reactor (kg/m3)

1385

Catalyst bulk density of the SMR reactor (kg/m3)

1200

160
45.16

is the thickness of the component-to-component separating insulation

(m), Aca is


2
the area of the component exposed to atmospheric conditions
m
,
and
Acc is the


2
component-to-component separating surface area m .
4. Pinch analysis and process integration: HEN design and
optimization

The HEN is designed and optimized by use of pinch analysis techniques. Basic
pinch analysis and process integration leads to a HEN with a MER. This in turn
initiates the basis for an optimum HEN design and integration with the microCHP system. Since residential micro-CHP systems based on fuel cell technology
are highly integrated [3], an effective process design methodology must be applied to
allow the system to operate in an efficient and realistic manner. Such a methodology
can be summarized in the following steps:
(1) Development and optimization of a computational simulation model that
allows the calculation of the energy requirements, which can be defined as
a list of hot and cold streams;
(2) Removal of all heat exchangers from the initial configuration and application of pinch analysis to evaluate possible HEN design opportunities [1];
(3) Reevaluation of the HEN design by means of a process change optimization strategy and modification of the stream data set to minimize utility
requirements [1,2];
(4) Optimization of the HEN, with respect to its total annual cost, by use of a
simultaneous optimization algorithm as found in the literature [12,16,25].
A systematic approach of this methodology allows evaluation of different HEN
configurations, which may be integrated to the micro-CHP system as shown in
the next section. Before the pinch analysis is applied, the fixed values for the
simulation model must be defined. These values, shown in Table 1, are kept fixed
due to operational limitations imposed either by the manufacturer or defined in the
literature.
4.1. Pinch analysis. After the stream data set (including temperatures, flow rates
and heat loads) has been extracted from the system simulation model, and a minimum temperature difference Tmin is selected, the energy targets and pinch temperature can be calculated. It is predicted that the pinch analysis will yield towards

68

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

a threshold problem with a single pinch. Therefore only a cooling utility will be
required. The integration model considers cooling water as a cold utility to close
the system balance. The design can be determined by applying the plus-minus
principle (also known as the ticking-off rule), and by identifying essential matches
at the non-utility end [1,2]. The plus-minus principle suggests that a process change
will reduce the utility targets if: (a) above the pinch, the total hot stream heat load
is increased, and/or the total cold stream load is decreased, (b) below the pinch,
the total hot stream heat load is decreased, and/or the total cold stream load is
increased. Obviously some changes are not feasible due to component/process constraints, as explained in Section 3. Prior to HEN design, some processes may allow
modifications which can lead in minimization of the utilities consumption.

4.2. HEN optimization model. HEN synthesis is typically performed by means

of two major synthesis strategies: sequential and simultaneous optimization [12].
The latter approach can work towards eliminating or at least reducing the disadvantages of the former one, such as the even distribution of the total heat surface
area between the heat exchangers [26]. Minimization of match number and fixed
charges of heat exchangers are necessary in the application of binary variables [12].
Hence mixed-integer problems have to be solved. Moreover, nonlinearities in an
optimization model cause nonlinear (MINLP) problems. In general MINLP models
can optimize simultaneously energy recovery, selection of matches, and areas [12].
In this research study a superstructure model is applied to evaluate an optimum
HEN configuration [16,25]. The simplicity of the modeling structure is based on the
fact that all model constraints are linear, while only the objective function is nonlinear [26]. The superstructure is organized in a number of stages, while allowing
their corresponding stream exiting temperatures to vary as decision parameters.
Streams entering each stage are split so as to achieve all possible matches between
hot process and cold process streams. Then, the streams are mixed at the outlet
of stages. The following assumptions are imposed [12,16,26]:
(1) The heater is placed at the highest temperature region of the cold process
stream. One heater suffices for one process stream;
(2) The cooler is located at the lowest temperature region of the hot process
stream. One cooler services one hot process stream;
(3) Mixing of streams is isothermal, with each branch to a mixer having the
same temperature;
(4) Heat exchangers are modeled as matches.
The optimization identifies the least cost network embedded within the superstructure by identifying which exchangers are needed and the flow configuration of the
streams. The model allows the possibility of additional constraints if a HEN configuration with no stream splits is required. Finally, the model allows selection
of the number of stages, or alternatively to set the number of stages equal to the
maximum of the number of hot or cold process streams [12,16].
The MINLP model includes isothermal and non-isothermal streams and is based
on the superstructure formulation by [16,25]. It considers every possible combination of process streams and provides the network structure that minimizes the total
yearly cost. Stream splitting is allowed in each superstructure stage and isothermal
mixing is assumed. This requires only the inlet/outlet temperatures at each stage,
without the requirement of flow variables. The objective function is defined as the
minimization of the total yearly cost, which includes the utility cost and the fixed
and variables costs of the exchangers [12].

APPENDIX

69

A. ARSALIS ET AL.

Table 2. Initial process stream data used in the heat integration analysis.
#

Description

Stream

Tsupply ( K )

Ttarget ( K )

H ( kW )

Reformate gas exiting the SMR reactor and entering

the WGS reactor

Hot

906.7

523.2

0.26740

Reformate gas exiting the WGS reactor and entering

the fuel cell stack anode

Hot

601.8

433.2

0.11390

Flue gas exiting the SMR reactor

Hot

1078.2

298.2

6.03600

Exhaust mixture exiting the fuel cell stack and

entering the combustor

Cold

432.0

970.6

4.00000

Methane preheating for the SMR reaction

Cold

298.2

599.2

0.04540

373.2

599.2

0.08869

373.2

374.2

0.44400

298.2

373.2

0.06183

6a
6b

Steam generation for the SMR reaction

Cold

6c

5. Results and Discussion

In the first part of this section, basic pinch analysis techniques are applied to obtain an initial HEN configuration that encompasses the need for MER. An analysis
of the composite curves and the grand composite curve (GCC) allows the study of
process modifications that will further improve the integrated system operation. In
the second part of this section, the simultaneous optimization algorithm is applied
to obtain an optimum HEN configuration, based on total cost considerations.
5.1. Pinch analysis and process integration.
5.1.1. Pinch analysis based on the initial stream data set. Before analyzing the
streams, Tmin is fixed at 20o C, which is a good approximation for the system under
study, since it involves gas-to-gas heat exchanging processes [1,2]. Table 2 shows
a list of the initially considered hot and cold streams used in the heat integration
analysis. The table includes all supply and target temperatures, identifying the
stream type (hot or cold), with a short description of the stream process. Finally
the stream heat load is given in kW. Heat loads are preferred over heat capacity
flow rates in kW/K, because although pinch analysis was originally performed using
the latter, in most practical cases it is more convenient to use heat loads. The main
reason is because internal latent heat changes, which give a high CP difference, may
be disguised. This could often lead to serious targeting errors [1].
The system composite curves are drawn in order to give a good view of the
possibilities for a HEN. They describe the total heating and cooling demands of
the system under study, based on the given inputs, as a function of temperature
intervals. Figure 2 shows the hot (upper) and cold (bottom) composite curves. The
overlap between the two curves represents the maximum amount of heat recovery
possible within the process [1]. An observation in the upper right corner indicates
no external heating is required. On the other hand, a significant amount of cooling
is required, as seen in the lower left part of the figure. Therefore, the pinch analysis
results revealed a threshold problem with a single pinch. This means that no
redesign of the original HEN configuration is required for MER design. This is
verified by the amount of minimum cold utility required (1.78 kW), which matches
the value of available heat for cogeneration found in the original simulation. A

70

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

Figure 2. Hot and cold composite curves extracted from the initial process stream data set.
more detailed observation of the composite curves shows that apart from the two
dominant cold (fuel cell stack exhaust preheating) and hot (SMR reactor flue gas)
streams, which are interdependent, the steam generation (shown in the lower left
cold composite curve) does not cause any pinch matching problems. This suggests
that no modification, such as local minimum temperature difference contribution, is
required. Finally it should be noted that some streams are very small compared to
others (e.g. #4 vs. #5), but still they cannot be removed due to their importance
in the unimpeded operation of the system.
5.1.2. Process change and system optimization. As indicated above process change
must be applied in the early stages of the pinch analysis, since the plus-minus
principle is only useful for first-stage screening [1]. In the system under study it
is vital to apply the principle simultaneously with system optimization to ensure
the feasibility of the modified processes. Specifically for the system under study,
and since this is a threshold problem, two modifications below the pinch can be
attempted: (i) decrease the total hot stream load and (ii) increase the total cold
stream load. The streams allowing modification are #3, #5, and #6 (shown in
Table 2). Streams #5 and #6, correspond to the inlet temperature of the SMR
reactor. The WGS reactor, combustor and fuel cell stack inlet/outlet temperatures
cannot be modified, because they are either fixed or dependent. The temperature
for stream #3 cannot be modified but its heat load can be reduced by reduction of
its corresponding flow rate.
The simulation model of the micro-CHP system is optimized using the genetic
algorithm method built-in in EES. The objective function is the maximization of
the net electrical efficiency of the system (el,net ) at nominal load. Four independent design parameters are selected and varied during the system optimization
process, and they are shown in Table 3, with their range of variation. The selection of the four independent variables should not be considered accidental, because
the optimization is concentrated around the SMR reactor parameters, which is an
endothermic reaction and can subsequently affect the heating utilization related to
the process change described above. The obtained value of the objective function

APPENDIX

71

10

A. ARSALIS ET AL.

Table 3. Design parameters with their range of variation used in

the process change optimization.
Variable

Description (unit)

Initial

Min

Max

Optimum

T6

Inlet temperature of the methane-steam mixture to

the SMR reactor (K)

599

400

600

587

T11

Inlet temperature of the flue gas to the methane

preheater (K)

626

500

700

597

m 17

SMR reactor input flue gas mass flow rate (kg/s)

6.82E-3

6.50E-3

6.87E-3

6.55E-3

S C

Steam-to-carbon ratio (-)

3.19

2.50

4.00

2.83

Table 4. Optimum process stream data used in the heat integration analysis.
#

Tsupply ( K )

Ttarget ( K )

H ( kW )

924.1

523.2

0.26220

603.3

433.2

0.10790

1054.4

298.2

5.59900

432.0

1021.8

3.86200

298.2

587.2

0.04325

6a

373.2

587.2

0.07464

6b

373.2

374.2

0.39500

6c

298.2

373.2

0.05500

is 35.16% for the design parameter values shown in Table 3. Therefore the optimization resulted in a significant increase of the objective function, compared to a
value of 31.86% prior to optimization.
5.1.3. Pinch analysis based on the optimum stream data set. Pinch analysis is repeated for the new stream data set obtained by the optimization strategy. Table 4
shows a list of the optimum process stream data set, which was calculated by the
optimization process. The corresponding composite curves are shown in Fig. 3.
An observation of the figure shows that the process change was successful, since
it accomplished minimization of the cold utility requirement (1.54 kW). Figure 4
shows the grand composite curve (GCC), which illustrates the net heat flow against
the shifted temperature. This graph represents the difference between the available
heat from the hot streams and the heat required by the cold streams, relative to
the pinch, at a given shifted temperature. For a threshold problem, like the one
under study, the net heat flow at the top (heating utility requirement) is zero, while
the corresponding one at the bottom (cooling utility requirement) is 1.54 kW. The
pinch temperature is 771o C, since it is the point where the net heat flow is 0 and
the GCC touches the axis.
In a threshold problem it is necessary to distinguish its characteristics. In the
analysis under study, the closest temperature approach between the hot and cold
composites is at the non-utility end and the curves diverge away from this point, as
shown in Fig. 2-4. In this case, design can be started from the non-utility end, using

72

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

11

Figure 3. Hot and cold composite curves extracted from the optimum process stream data set.
the pinch design rules. In this case the network design is relatively slack [1], and a
great many designs are possible as the thermodynamic constraint of the pinch does
not apply. Another aspect that can be studied is whether the endothermic SMR
reactor has been placed appropriately in terms of heat integration. Pinch analysis
can suggest minor refinements allowing improved process integration, although this
would suggest changing the reaction conditions, which may alter the overall system
performance. The possible extent of heat integration of the reactor with the rest of
the process mainly depends on the reaction temperature. The appropriate placement principle [1] suggests that for endothermic reactions, the heat should ideally
come below the pinch so that it can be driven by waste heat from the process.
This is followed for the system under study, since the pinch temperature (771o C)
is higher than the reformate outlet temperature (651o C).
The system performance is then tested at part-load operation to examine the
behavior of the proposed system at off-design conditions. Figure 5 shows the variation of efficiency at different net electrical power outputs. As the load decreases,
the net electrical efficiency increases linearly, while the thermal efficiency (and total
efficiency) decreases initially linearly, while at very low loads the decrease is higher.
This is because of the effect of the heat losses, which are temperature and area
dependent. Since component temperatures and areas remain constant at all loads,
heat losses have a more drastic effect at lower operating loads. Therefore, system
operation at low, and especially critical, loads may not be favored at periods of high
demand on cogeneration heat. The greater loss is due to the SMR reactor, which
although it is heavily insulated, it contributes 50% of the losses. This is because of
the high temperature flows (reformate & flue gas) occurring in the SMR reactor.
5.2. Simultaneous HEN optimization. The simultaneous optimization strategy analyzed in the previous section is applied to the HEN to obtain an optimized
configuration. The MINLP model with two stages is solved in GAMS, with the
DICOPT++ solver [27]. The compact nature of the micro-CHP system does not
favor the use of stream splitting, and therefore this is restricted in the code. In
addition to the potentially more complex operation, stream splitting would require

APPENDIX

73

12

A. ARSALIS ET AL.

Figure 4. Grand composite curve extracted from the optimum

process stream data set.

an additional investment cost for a control valve [12]. As in the initial configuration, the pattern of the composite curves, shown in Fig. 6, is identical to the
one shown in Fig. 3. This means that the HEN configuration still requires only
cooling (threshold problem), and therefore maintains its MER nature. This can be
explained by the significantly higher cost of running a hot utility, as compared to
the cooling one [12,16,25].
The grid diagram representation, shown in Fig. 7, shows the heat transfer operation of the optimum HEN. Hot streams are at the top running left to right,
while cold streams are located at the bottom running right to left. Heat exchange
matches are represented by vertical lines joining two arrows on the two streams
being matched. An exchanger using cold utility is represented by an arrow with
a W. The coupling of the optimized HEN configuration with the micro-CHP
system is shown in Fig. 8. The new configuration is significantly different than the
original one (shown in Fig. 1), due to the cost reduction requirements imposed in
the objective function [16]. Therefore, the optimum configuration encompasses the
need for heat exchangers with a cheaper structure, such as a minimum heat transfer

74

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

13

Figure 5. Variation of efficiency at different operating loads.

Figure 6. Hot and cold composite curves after the application of

the simultaneous HEN optimization.
area. The heat exchangers must allow a more cost efficient heat transfer distribution along the streams, although the total number of heat exchangers remains the
same (8). The results extracted from the simultaneous optimization simulation are
shown in Table 5. The HEN obtained requires a total cost of $8,147/year, with
utility and annualized investment costs of $77/year and $8,070/year, respectively.
The optimization problem is solved relatively fast due to the nature of its formulation, as explained in the previous section. The DICOPT is run under an Intel Core
2 Duo T9400 computer at 2.53 GHz. The model includes 170 continuous variables
and 187 constraints and is solved in 0.094 seconds.

APPENDIX

75

14

A. ARSALIS ET AL.

Figure 7. HEN configuration after the application of the simultaneous HEN optimization.

Figure 8. Configuration of the proposed micro-CHP system after

the application of the simultaneous HEN optimization.

76

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

15

Table 5. HEN configuration results after the application of the

simultaneous HEN optimization.
Variable

Description (unit)

Value

Q util

Total heat load of utilities (kW)

N HEx

Total number of heat exchangers (incl. utilities) (-)

AHEx

Total area of heat exchangers (m2)

0.13

Autil

Total area of utilities (m2)

0.05

Total area of heat exchangers (incl. utilities) (m2)

0.18

AHEx ,tot

1.53
8

Cutil

Total cost of utilities ($/year)

Cinv

Investment cost ($/year)

8070

Total cost of the heat exchanger network ($/year)

8147

CHEN

77

6. Conclusions
An optimization methodology, based on process integration techniques, is applied
on an HT-PEMFC-based micro-CHP simulation model to improve its performance,
in terms of net electrical efficiency and total HEN cost. This is accomplished by
solving two objective functions: (1) maximization of the net electrical efficiency
of the system, (2) minimization of the total HEN cost. The first optimization
strategy includes a pinch analysis and process change procedure. Pinch analysis is
applied on the initial set of stream data, and then by use of the plus-minus principle
(process change), the system is optimized with a genetic algorithm optimization
methodology. This approach is diverse, because it allows studying and evaluation
of the optimum system parameters (e.g. steam-to-carbon ratio), based on the
knowledge input suggested by the pinch analysis technique. The proposed system
exhibits high electrical efficiencies, which vary from 35.2 to 43.6% (based on LHV).
The total system efficiencies vary from 65.2 to 91.1%. The system total efficiencies
are significantly decreased at low operating loads, because of the higher impact of
heat losses at those loads.
In the second optimization strategy the optimum stream data set, calculated
in the previous optimization strategy, is used as the input in an MINLP problem
formulation. This optimization attempts to minimize the total annual HEN cost,
by suggesting a HEN configuration fulfilling the targets with a reduced cost. This
is successfully accomplished, and most importantly not in the expense of the net
electrical efficiency, since the energy input is not forced to change. Therefore the
heat integration is accomplished requiring only cold utilities (threshold problem).
A HEN, significantly different than the initial one, is generated by the optimization
strategy, suggesting the effectiveness of the optimization algorithm. The minimized
total annual cost of the HEN is $8,147/year. The calculation is rapid because of the
nature of the MINLP formulation, which consists of a nonlinear objective function,
but with linear constraints.

APPENDIX

77

16

A. ARSALIS ET AL.

Acknowledgment
The authors would like to acknowledge the support of Danfoss and Dantherm
Power throughout the realization of this research study.
References
[1] I.C. Kemp, Pinch Analysis and Process Integration: A User Guide on Process
Integration for the Efficient Use of Energy, 2nd ed., Butterworth-Heinemann,
2007.
[2] R. Smith, Chemical Process Design and Integration, Wiley, Chichester, UK,
2005.
[3] J. Godat, F. Marechal, Optimization of a fuel cell system using process integration techniques, Journal of Power Sources. 118 (2003) 411-423.
[4] R. OHayre, W. Colella, S.-W. Cha, F.B. Prinz, Fuel Cell Fundamentals, 2nd
ed., Wiley, 2009.
[5] J. Larminie, A.L. Dicks, Fuel Cell Systems Explained, 2nd ed., Wiley, Chichester, UK, 2003.
[6] F. Marechal, F. Palazzi, J. Godat, D. Favrat, Thermo-Economic Modelling and
Optimisation of Fuel Cell Systems, Fuel Cells. 5 (2005) 5-24.
[7] B. Linnhoff, J.R. Flower, Synthesis of heat exchanger networks: I. Systematic
generation of energy optimal networks, AIChE Journal. 24 (1978) 633-642.
[8] B. Linnhoff, J.R. Flower, Synthesis of heat exchanger networks: II. Evolutionary
Generation of Networks with Various Criteria of Optimality, AIChE Journal.
24 (1978) 642-654.
[9] B. Linnhoff, E. Hindmarch, The pinch design method for heat exchanger networks, Chemical Engineering Science. 38 (1983) 745-763.
[10] N.C. Monanteras, C.A. Frangopoulos, Towards synthesis optimization of a fuelcell based plant, Energy Conversion and Management. 40 (1999) 1733-1742.
[11] C. Wallmark, P. Alvfors, Design of stationary PEFC system configurations to
meet heat and power demands, Journal of Power Sources. 106 (2002) 83-92.
[12] L.T. Biegler, I.E. Grossmann, A.W. Westerberg, Systematic Methods of Chemical Process Design, 1st ed., Prentice Hall, 1997.
[13] V. Verda, F. Nicolin, Thermodynamic and economic optimization of a MCFCbased hybrid system for the combined production of electricity and hydrogen,
International Journal of Hydrogen Energy. 35 (2010) 794-806.
[14] F. Palazzi, N. Autissier, F. Marechal, D. Favrat, A methodology for thermoeconomic modeling and optimization of solid oxide fuel cell systems, Applied
Thermal Engineering. 27 (2007) 2703-2712.
[15] N. Autissier, F. Palazzi, F. Marechal, J. van Herle, D. Favrat, Thermo-Economic
Optimization of a Solid Oxide Fuel Cell, Gas Turbine Hybrid System, Journal
of Fuel Cell Science and Technology. 4 (2007) 123.
[16] T.F. Yee, I.E. Grossmann, Simultaneous optimization models for heat integrationII. Heat exchanger network synthesis, Computers & Chemical Engineering. 14
(1990) 1165-1184.
[17] N.G. Georgopoulos, Application of a Decomposition Strategy to the Optimal
Synthesis/Design and Operation of a Fuel Cell Based Total Energy System,
Virginia Tech, 2002.
[18] Serenergy, Available at: http://www.serenergy.dk/ [Accessed February 2, 2011]
[19] R.L. Keiski, T. Salmi, P. Niemisto, J. Ainassaari, V.J. Pohjola, Stationary and
transient kinetics of the high temperature water-gas shift reaction, Applied
Catalysis A: General. 137 (1996) 349-370.
[20] J. Davies, D. Lihou, Optimal design of methane steam reformer, Chem. Proc.
Eng. 52 (1971) 71-80.

78

APPENDIX

OPTIMIZATION OF A HIGH TEMPERATURE PEMFC-BASED MICRO-CHP SYSTEM

17

[21] D.F. Cheddie, N. Munroe, Mathematical model of a PEMFC using a PBI membrane, Energy Conversion and Management. 47 (2006) 1490-1504.
[22] J. Zhang, Z. Xie, J. Zhang, Y. Tang, C. Song, T. Navessin, et al., High temperature PEM fuel cells, Journal of Power Sources. 160 (2006) 872-891.
[23] A.R. Korsgaard, R. Refshauge, M.P. Nielsen, M. Bang, S.K. Kr, Experimental
characterization and modeling of commercial polybenzimidazole-based MEA
performance, Journal of Power Sources. 162 (2006) 239-245.
[24] K. Kim, M.R. von Spakovsky, M. Wang, D.J. Nelson, A hybrid multi-level
optimization approach for the dynamic synthesis/design and operation/control
under uncertainty of a fuel cell system, Energy. (2010).
[25] J. Ponce-Ortega, A. Jimenez-Gutierrez, I.E. Grossmann, Optimal synthesis of
heat exchanger networks involving isothermal process streams, Computers $
Chemical Engineering. 32 (2008) 1918-1942.
[26] S. Sieniutycz, J. Jezowski, Energy Optimization in Process Systems, 1st ed.,
Elsevier, Amsterdam, 2009.
[27] J. Viswanathan, I.E. Grossmann, A combined penalty function and outer approximation method for MINLP optimization, Computers & Chemical Engineering. 14 (1990) 769-782.
Department of Energy Technology, Aalborg University, Pontoppidanstrde 101,
9220, Aalborg ., Denmark
E-mail address: aar@et.aau.dk

APPENDIX

79