Id Methods

Optimum Design of
Renewable Energy
Systems:
Microgrid and Nature Grid
Methods
Shinya Obara
Kitami Institute of Technology, Japan
A volume in the Advances in

Environmental Engineering and Green
Technologies (AEEGT) Book Series
Managing Director:
Production Editor:
Development Editor:
Acquisitions Editor:
Typesetter:
Cover Design:
Lindsay Johnston
Jennifer Yoder
Allyson Gard
Kayla Wolfe
Lisandro Gonzalez
Jason Mull
Published in the United States of America by

Engineering Science Reference (an imprint of IGI Global)
701 E. Chocolate Avenue
Hershey PA 17033
Tel: 717-533-8845
Fax: 717-533-8661
E-mail: cust@igi-global.com
Web site: http://www.igi-global.com
Copyright 2014 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in
any form or by any means, electronic or mechanical, including photocopying, without written permission from the publisher.
Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or
companies does not indicate a claim of ownership by IGI Global of the trademark or registered trademark.

Library of Congress Cataloging-in-Publication Data
Obara, Shinya.
Optimum design of renewable energy systems : microgrid and nature grid methods / by Shinya Obara.
pages cm
Includes bibliographical references and index.
ISBN 978-1-4666-5796-0 (hardcover) -- ISBN 978-1-4666-5797-7 (ebook) -- ISBN 978-1-4666-5799-1 (print & perpetual
access) 1. Renewable energy sources. 2. Electric power distribution. 3. Electric power systems--Design and construction.
4. Energy storage--Equipment and supplies. I. Title.
TJ808.O23 2014
621.042--dc23
2013051299
This book is published in the IGI Global book series Advances in Environmental Engineering and Green Technologies
(AEEGT) (ISSN: 2326-9162; eISSN: 2326-9170)
British Cataloguing in Publication Data
A Cataloguing in Publication record for this book is available from the British Library.
All work contributed to this book is new, previously-unpublished material. The views expressed in this book are those of the
authors, but not necessarily of the publisher.
For electronic access to this publication, please contact: eresources@igi-global.com.
Advances in
Environmental
Engineering and Green
Technologies (AEEGT)
Book Series
ISSN: 2326-9162
EISSN: 2326-9170
Mission
Growing awareness and focus on environmental issues such as climate change, energy use, and loss of
non-renewable resources have brought about a greater need for research that provides potential solutions
to these problems. The field of environmental engineering has been brought increasingly to the forefront
of scholarly research and, alongside it, environmentally-friendly, or green, technologies as well.
Advances in Environmental Engineering & Green Technologies (AEEGT) Book Series is a
mouthpiece for this research, publishing books that discuss topics within environmental engineering or
that deal with the interdisciplinary field of green technologies.
Coverage
Alternative Power Sources
Biofilters & Biofiltration
Contaminated Site Remediation
Green Transportation
Industrial Waste Management & Minimization
Policies Involving Green Technologies & Environmental Engineering
Pollution Management
Renewable Energy
Sustainable Communities
Waste Management
IGI Global is currently accepting manuscripts

for publication within this series. To submit a proposal for a volume in this series, please contact our
Acquisition Editors at Acquisitions@igi-global.com
or visit: http://www.igi-global.com/publish/.
The Advances in Environmental Engineering and Green Technologies (AEEGT) Book Series (ISSN 2326-9162) is published by IGI
Global, 701 E. Chocolate Avenue, Hershey, PA 17033-1240, USA, www.igi-global.com. This series is composed of titles available for purchase
individually; each title is edited to be contextually exclusive from any other title within the series. For pricing and ordering information please
visit http://www.igi-global.com/book-series/advances-environmental-engineering-green-technologies/73679. Postmaster: Send all address
changes to above address. Copyright 2014 IGI Global. All rights, including translation in other languages reserved by the publisher. No part
of this series may be reproduced or used in any form or by any means graphics, electronic, or mechanical, including photocopying, recording, taping, or information and retrieval systems without written permission from the publisher, except for non commercial, educational use,
including classroom teaching purposes. The views expressed in this series are those of the authors, but not necessarily of IGI Global.
Titles in this Series
For a list of additional titles in this series, please visit: www.igi-global.com
Optimum Design of Renewable Energy Systems Microgrid and Nature Grid Methods
Shinya Obara (Kitami Institute of Technology, Japan)
Engineering Science Reference copyright 2014 303pp H/C (ISBN: 9781466657960) US $235.00 (our price)
Nuclear Power Plant Instrumentation and Control Systems for Safety and Security
Michael Yastrebenetsky (State Scientific and Technical Centre for Nuclear and Radiation Safety, Ukraine) and
Vyacheslav Kharchenko (National Aerospace University- KhAI, Ukraine, and Centre for Safety InfrastructureOriented Research and Analysis, Ukraine)
Computational Intelligence in Remanufacturing
Bo Xing (University of Pretoria, South Africa) and Wen-Jing Gao (Meiyuan Mould Design and Manufacturing
Co., Ltd, China)
Information Science Reference copyright 2014 348pp H/C (ISBN: 9781466649088) US $195.00 (our price)
Risk Analysis for Prevention of Hazardous Situations in Petroleum and Natural Gas Engineering
Davorin Matanovic (University of Zagreb, Croatia) Nediljka Gaurina-Medjimurec (University of Zagreb, Croatia)
and Katarina Simon (University of Zagreb, Croatia)
Marine Technology and Sustainable Development Green Innovations
Oladokun Sulaiman Olanrewaju (University Malaysia Terengganu, Malaysia) Abdul Hamid Saharuddin (University
Malaysia Terengganu, Malaysia) Ab Saman Ab Kader (Universiti Teknologi Malaysia, Malaysia) and Wan Mohd
Norsani Wan Nik (University Malaysia Terengganu, Malaysia)
Sustainable Technologies, Policies, and Constraints in the Green Economy
Andrei Jean-Vasile (Petroleum and Gas University of Ploiesti, Romania) Turek Rahoveanu Adrian (Institute of
Research for Agricultural Economics and Rural Development, Romania) Jonel Subic (Institute of Agricultural
Economics, Belgrade, Serbia) and Dorel Dusmanescu (Petroleum and Gas University of Ploiesti, Romania)
Energy-Aware Systems and Networking for Sustainable Initiatives
Naima Kaabouch (University of North Dakota, USA) and Wen-Chen Hu (University of North Dakota, USA)
701 E. Chocolate Ave., Hershey, PA 17033

Order online at www.igi-global.com or call 717-533-8845 x100
To place a standing order for titles released in this series, contact: cust@igi-global.com
Mon-Fri 8:00 am - 5:00 pm (est) or fax 24 hours a day 717-533-8661
Table of Contents
Preface . ...............................................................................................................................................viii
Acknowledgment . ............................................................................................................................... xv
Chapter 1
Operating Schedule of a Combined Energy Network System . .............................................................. 1
General Introduction . ..................................................................................................................... 1
Operating Schedule of a Combined Energy Network System with Fuel Cell . .............................. 2
Fuel Cell Network System Considering Reduction in Fuel Cell Capacity Using Load .
Leveling and Heat Release Loss . ................................................................................................. 18
Chapter 2
Dynamic Characteristics of a Fuel Cell System and Microgrid ........................................................... 38
General Introduction . ................................................................................................................... 38
Load Response Characteristics of a Fuel Cell Microgrid with Control of Number of Units ....... 39
Dynamic Characteristics of a PEFCL System for Individual Houses .......................................... 53
Chapter 3
Effective Improvement in Generation Efficiency of a Fuel Cell Microgrid ......................................... 72
General Introduction . ................................................................................................................... 72
Effective Improvement in Generation Efficiency due to Partition Cooperation Management .
of a Fuel Cell Microgrid ............................................................................................................... 73
Equipment Plan of Compound Interconnection Microgrid Composed from Diesel Power .
Plants and Proton-Exchange Membrane Fuel Cell . ..................................................................... 87
Chapter 4
Installation Plan of a Fuel Cell Cogeneration System ........................................................................ 103
General Introduction . ................................................................................................................. 103
Installation Plan of a Fuel Cell Microgrid System Optimized by Maximizing Power .
Generation Efficiency ................................................................................................................. 104
Fuel Cell Network with Water Electrolysis for Improving Partial Load Efficiency of a
Residential Cogeneration System ............................................................................................... 118
Chapter 5
Fuel Cell Microgrid with Wind Power Generation . ........................................................................... 136
Analysis of a Fuel Cell Microgrid with a Small-Scale Wind Turbine Generator ....................... 137
Power Characteristics of a Fuel Cell Microgrid with Wind Power Generation ......................... 154
Chapter 6
Compound Microgrid of City-Gas Engine and Proton Exchange Membrane Fuel Cell .................... 167
Amount of CO2 Discharged from Compound Microgrid of Hydrogenation City-Gas .
Engine and Proton Exchange Membrane Fuel Cell . .................................................................. 168
Capacity Optimization of PEFCL and Hydrogen Mixing Gas-Engine Compound Generator . .... 181
Chapter 7
Independent Microgrid Composed of Distributed Engine Generator . ............................................... 198
Energy Cost of an Independent Microgrid with Control of Power Output Sharing of a .
Distributed Engine Generator ..................................................................................................... 199
Improvement of Power Generation Efficiency of an Independent Microgrid Composed of
Distributed Engine Generators ................................................................................................... 216
Chapter 8
Characteristics of PEFC / Woody Biomass Engine Hybrid Microgrid and Exergy Analysis . ........... 237
Dynamic Characteristics of PEFC / Woody Biomass Engine Hybrid Microgrid . ..................... 238
Exergy Analysis of the Woody Biomass Stirling Engine and PEFC Combined System with
Exhaust Heat Reforming ............................................................................................................ 250
Exergy Analysis of a Regional Distributed PEM Fuel Cell System . ......................................... 263
Chapter 9
Design Support Using a Neural Network Algorithm . ........................................................................ 282
Dynamic Operational Scheduling Algorithm for an Independent Microgrid with .
Renewable Energy ...................................................................................................................... 283
Operation Prediction of a Bioethanol Solar Reforming System using a Neural Network ......... 300
Chapter 10
Microgrid with Numerical Weather Information ................................................................................ 321
Compound Microgrid Installation Operation Planning of a PEFC and Photovoltaics with
Prediction of Electricity Production using GA and Numerical Weather Information ................ 322
Energy Supply Characteristics of a Combined Solar Cell and Diesel Engine System with a
Prediction Algorithm for Solar Power Generation ..................................................................... 335
Chapter 11
SOFC-PEFC Combined Microgrid . ................................................................................................... 352
Power Generation Efficiency of Photovoltaics and a SOFC-PEFC Combined Microgrid .
with Time Shift Utilization of the SOFC Exhaust Heat ............................................................. 353
Power Generation Efficiency of an SOFC-PEFC Combined System with Time Shift .
Utilization of SOFC Exhaust Heat ............................................................................................. 367
Chapter 12
Bioethanol Solar Reforming System for Distributed Fuel Cell .......................................................... 385
Hydrogen Production Characteristics of a Bioethanol Solar Reforming System with Solar
Insolation Fluctuations ............................................................................................................... 386
Efficiency Analysis of a Combined PEFC and Bioethanol-Solar-Reforming System for
Individual Houses ....................................................................................................................... 399
Compilation of References................................................................................................................ 417
About the Contributors..................................................................................................................... 427
Index.................................................................................................................................................... 428
viii
Preface
Control of global warming is a common subject in the world. Therefore, the challengers of various fields
are considering methods to control global warming. Microgrid technology is expected as a next-generation energy supply system. However, since renewable energy is unstable, in many cases, it requires
support by the conventional energy equipment. We are investigating the compound energy system from
the following two viewpoints. One is the development of highly efficient energy storage equipment
represented by a battery and heat-storage tank. Another is development of the operation optimization
technology of the compound energy system including green energy. It is thought that the energy supply
method shifts from the individual operation of large-scale plant to distribution of small equipment or
microgrid. Moreover, a microgrid develops into a smart-grid by various added values with IT technology. On the other hand, it was predicted that the reduction technology of the greenhouse gas of a microgrid
progressed sharply, and we named the nature-grid. A microgrid, a smart grid, and a nature-grid require
fusion of energy technology and an information technology. For example, the operation in consideration
of the green energy change with load prediction and weather prediction of a compound energy system
can be planned. This book describes the operation optimization technology by compound utilization of
a PEFC, PEFC-SOFC combined system, bio-ethanol solar reforming, wind-power generation, woody
biomass engine, city-gas engine, diesel power plant, etc. The technology described in this book plays a
large role in the development of a small-scale power-generation system, a microgrid, a smart-grid, and
a nature-grid, which are introduced into individual houses, apartment houses or local area power supplies.
The book is organized into twelve chapters. A brief description of each of the chapters follows:
Chapter 1 has described operating schedule of a combined energy network system. In the 1st section,
the chromosome model showing system operation pattern is applied to GA (genetic algorithm), and the
method of optimization operation planning of energy system is developed. The optimization method
of this operation planning was applied to the compound system of methanol steam reforming type fuel
cell, geo-thermal heat pump and the electrolysis tank of water. The operation planning was performed
for the energy system using the energy demand pattern of the individual residence of Sapporo in Japan.
From analysis results, the amount of outputs of a solar module and the relation of the operation cost of
the system which are changed by the weather were clarified. In the 2nd section, reduction in fuel cell
capacity linked to a fuel cell network system is considered. An optimization plan is made to minimize
the quantity of heat release of the hot water piping that connects each building. Such an energy network
is analyzed assuming connection of individual houses, a hospital, a hotel, a convenience store, an office
building, and a factory. Consequently, a reduction of 46% of fuel cell capacity is expected compared
with the conventional system in the case study.
ix
Chapter 2 has described characteristics of a fuel cell system and microgrid. In the 1st section, the dynamic characteristics and generation efficiency of a microgrid structured from 17 houses were examined. A
gas engine generator with a power generation capacity of 3 kW installed in a house is made to correspond
to the base load, and a proton-exchange membrane fuel cell (PECF) with a power generation capacity of
1 kW is installed in 16 houses. Moreover, when changing the load of a microgrid, the correspondence
takes place by controlling the number of fuel cells. Using numerical analysis, the characteristics of the
power quality of a fuel cell microgrid, and the generation efficiency of the fuel cell were examined. As a
result, the relationship between the parameter of the controller and power quality and a fall in generation
efficiency by a partial load were clarified. In the 2nd section, the method of determination of the control
variables for a system controller, which controls the electric power output of a solid-polymer-membrane
fuel cell system (PEFC) during electric power load fluctuations, was considered. The power load pattern
of an individual house consists of loads usually moved up and down rapidly for a short time. This section
investigates the relation of the control variables and power generation efficiency when adding change
that simulates the load of a house to PEFC cogeneration. As a result, it was shown that an operation,
minimally influenced by load fluctuations, can be performed by changing the control variables using
the value of the electric power load of a system.
Chapter 3 has described effective improvement in generation efficiency of a fuel cell microgrid. The
fuel cell microgrid is expected as a distributed power supply with little environmental impact. In the 1st
section, a microgrid is divided into multiple and each is optimized for the purpose of maximization of
power generation efficiency. In the cooperation management of a microgrid, large fluctuations in load,
or increases and decreases in a building, can be followed with a grid using a system-interconnection
device. The system proposed in this section obtained results with high generation efficiency (from 21.1%
to 27.6%) compared with the central system (generation efficiency is 20.6% to 24.8%) of a fuel cell microgrid. In the 2nd section, an independent microgrid that compounds and connects a diesel power plant
generator (DEG) and a proton-exchange membrane fuel cell (PEFC) is proposed. The operation of DEG
is controlled to correspond to the base load of whole CIM (Compound Interconnection microgrid), and,
on the other hand, the operation of PEFC is controlled to follow the load fluctuation of CIM. A complex community model and residential area model were used for analysis. In this section, the microgrid
concerning the urban area (18 buildings) in Tokyo was investigated. From the results of analysis, it was
confirmed that CIM could be operated with a high generation efficiency of 27.1 to 29.9%.
Chapter 4 has described installation plan of a fuel cell cogeneration system. If energy-supplying
microgrids can be arranged to operate with maximal efficiency, this will have a significant influence
on the generation efficiency of the grid and will reduce greenhouse gas production. A means of optimizing the microgrid needs to be developed. In the 1st section, microgrids that use proton exchange
membrane-type fuel cells (PEFC) may significantly reduce the environmental impact when compared
with traditional power plants. The amount of power supplied to the grid divided by the heating value of
the fuel is defined as the system generation efficiency. We find that when a set of PEFCs and a natural
gas reformer are connected to the microgrid in an urban area, the annual generation efficiency of the
system slightly exceeds 20%. When a PEFC follows the electricity demand pattern of a house, it operates
at a partial load most of the time, resulting in a low efficiency of the microgrid. A method of improving the generation efficiency of a fuel cell microgrid is proposed, where a supply system of power and
heat with a high energy efficiency are constructed. In this section, a method of installing two or more
microgrids is proposed (known as the partition cooperation system). The grids can be connected in an
urban area to maximize generation efficiency. Numerical analysis shows that the system proposed in this
section (which has an annual generation efficiency of 24.6 to 27.6%) has a higher generation efficiency
than conventional PEFC systems (central generating systems have annual generation efficiencies of 20.6
to 24.8%). In the 2nd section, fuel cell energy network which connects hydrogen and oxygen gas pipes,
electric power lines and exhaust heat output lines of the fuel cell cogeneration for individual houses,
respectively is analyzed. As an analysis case, the energy demand patterns of individual houses in Tokyo
are used, and the analysis method for minimization of the operational cost using a genetic algorithm is
described. The fuel cell network system of an analysis example assumed connecting the fuel cell cogeneration of five houses. If energy is supplied to the five houses using the fuel cell energy network
proposed in this study, 9% of city gas consumption will be reduced by the maximum from the results
of analysis. 2% included to 9% is an effect of introducing water electrolysis operation of the fuel cells,
corresponding to partial load operation of fuel cell co-generation.
Chapter 5 has described fuel cell microgrid with wind power generation. Since the output of renewable energy is unstable, other energy equipment needs to cover the stability of output. Thus, in the 1st
section, the operating conditions of an independent microgrid that supplies power with natural power
sources and fuel cells are investigated. If electric power is supplied using an independent microgrid
connected to renewable energy, it can flexibly match the energy demand characteristics of a local area.
The output of wind power generation and fuel cells is controlled by proportional-integral control of an
independent microgrid for rapid power demand change. An independent microgrid that connects with
renewable energy has the potential to reduce energy costs, and reduce the amount of greenhouse gas discharge. However, the frequency and voltage of a microgrid may not be stable over a long time due to the
input of unstable renewable energy, and changes in short-period power load that are difficult to predict.
Thus, when planning the installation of a microgrid in the 2nd section, it is necessary to investigate the
dynamic characteristics of the power. About the microgrid composed from ten houses, a 2.5kW proton
exchange membrane fuel cell is installed in one building, and this fuel cell operated corresponding to a
base load is assumed. A 1kW PEFC is installed in other seven houses, in addition a 1.5kW wind turbine
generator is installed. The microgrid to investigate connects these generating equipments, and supplies
the power to each house. The dynamic characteristics of this microgrid were investigated in numerical
analysis, and the cost of fuel consumption and efficiency was also calculated. Moreover, the stabilization time of the microgrid and its dynamic characteristics accompanied by wind-power-generation and
fluctuation of the power load were clarified.
Chapter 6 has described CO2 discharged from compound microgrid of hydrogenation city-gas engine
and fuel cell. The independent microgrid is considered to be a technology in which maximum distributed
energy is realizable. However, there are many subjects, such as the stability of the dynamic characteristics of power and development of an optimal design method. If the fuel cell system of the capacity
corresponding to a load peak is installed, equipment cost will be high and energy cost will not be able
to get any profile commercially. By increasing the hydrogen concentration at the time of low load, the
power-generation efficiency of a city-gas-engine-generator (NEG) improves, and carbon dioxide emissions decrease. So, in the 1st section, a microgrid composed from a PEFC and a hydrogenation city gas
engine was investigated using numerical simulation. The system with a small load factor of NEG and
with a large load factor of PEFC system has few CO2 emissions. The system which combined baseload operation of PEFC and load fluctuation operation of hydrogenation city gas engine is the most
advantageous for the comprehensive evaluation of equipment cost, power generation efficiency, and
CO2 emissions. When the optimal system was installed into the urban area model of 20 buildings and
analyzed, power generation efficiency was 25% and CO2 emissions were 1,106 kg/Day. Distribution of
xi
the independent energy source can be optimized with regionality in mind. The 2nd section examines the
independent power supply system relating to hydrogen energy. Generally speaking, the power demand
of a house tends to fluctuate considerably over the course of a day. Therefore, when introducing fuel
cell cogeneration into an apartment house, etc., low-efficiency operations in a low-load region occur
frequently in accordance with load fluctuation. Consequently, the hybrid cogeneration system (HCGS)
that uses a proton-exchange membrane fuel cell fuel cell (PEFC) and a hydrogen mixture gas engine
(NEG) together to improve power generation efficiency during partial load of fuel cell cogeneration is
proposed. In this section, HCGS is introduced into 10 household apartments in Tokyo, and the power
generation efficiency, carbon dioxide emissions and optimal capacity of a boiler and heat storage tank
are investigated through analysis. Analysis revealed that the annual average power generation efficiency
when the capacity of PEFC and NEG is 5 kW was 27.3%. Meanwhile, the annual average power generation efficiency of HCGS is 1.37 times that of the PEFC independent system, and 1.28 times that of the
NEG independent system respectively.
Chapter 7 has described independent microgrid composed of distributed engine generator. In the 1st
section, small kerosene diesel-engine power generators are introduced into an independent microgrid
(IMG) that connects 20 houses, and power and heat are supplied to them. A 3 kW engine generator is
installed in six houses, and a boiler and a heat storage tank are also installed, and exhaust heat to make
up for insufficiency is supplied. The boiler is installed in the house that does not install the engine
generator, and heat is supplied to the demand side. Partial load operation of the engine generator has a
large influence on power generation efficiency. Therefore, this section discusses the system that controls
the power of the engine generator by the power distribution control system using the genetic algorithm
(GA), and the control system that changes the number of operations of the engine generators according to the magnitude of the power load. As a case study, the energy-demand model of the 20 houses in
Sapporo was analyzed. As a result, the annual energy cost of the number of operations system and the
power distribution control system is reducible with 16% and 8% compared with the conventional method,
respectively. However, it depends for this cutback effect on the heat demand characteristic greatly, and
when the proposed system is introduced into a community with little heat demand, effectiveness will
decrease greatly. In the 2nd section, the power generation efficiency and power cost of an independent
microgrid that distributes the power from a small diesel engine power generator was investigated using
numerical analysis. The independent microgrid built using one to six sets of 20 average houses in Sapporo and the distributed engine generators were examined using these test results. When a diesel engine
power generator is distributed, since the power generation capacity per set decreases compared with the
central system, the load factor of each engine generator rises. As a result, the operation of an engine at
partial load with low efficiency can be reduced. When the number of distributions of the engine generator increases as a result of numerical analysis, the cost of the fuel decreases.
Chapter 8 has described characteristics of PEFC / woody biomass engine hybrid microgrid and exergy
analysis. The combustion exhaust heat of woody biomass engine using Stirling cycle is high temperature.
So, in the 1st section, this exhaust heat is used for the city gas reforming reaction of a proton exchange
membrane fuel cell (PEFC) system. The woody biomass engine generator has the characteristic that
the greenhouse gas amount of emission with power generation is greatly reducible. In this study, the
microgrid system that introduces PEFC / woody biomass engine hybrid cogeneration (PWHC) is proposed. It depends on the dynamic characteristics of the grid for the power quality at the time of load
fluctuation being added to the microgrid. Especially, the dynamic characteristics of the independent
microgrid are important on security of power quality. So, in this section, the response characteristic of
xii
PEFC and woody biomass engine was investigated by the experiment and the numerical analysis. Furthermore, the response characteristic of the PWHC independent microgrid including auxiliary machinery
was investigated by the numerical simulation. Moreover, an improvement of dynamic characteristics is
proposed using the method of adding proportional-plus-integral control to PWHC. If woody biomass
engine is introduced into a house, 10.2s will be required to stabilize power quality at the maximum. On
the other hand, when woody biomass engine corresponds to a base load and PEFC corresponds to the
load exceeding the base load, settling time is less than 1.6 s. In this study, relation between the system
configuration of the PWHC microgrid and the dynamic characteristices of the power was clarified. The
woody biomass Stirling engine (WB-SEG) is an external combustion engine that outputs high-temperature
exhaust gases. It is necessary to improve the exergy efficiency of WB-SEG from the viewpoint of energy
cascade utilization. In the 2nd section, a combined system that uses the exhaust heat of WB-SEG for the
steam reforming of city gas and that supplies the produced reformed gas to a proton exchange membrane
fuel cell (PEFC) is proposed. The energy flow and the exergy flow were analyzed for each WB-SEG,
PEFC, and WB-SEG / PEFC combined system. Exhaust heat recovery to preheat fuel and combustion
air was investigated in each system. In the 3rd section, the exergy flow and exergy efficiency of a 3kW
PEFC were investigated, and the regional characteristic of the distributed energy system was considered.
In the environmental temperature range of 263K to 313K, the difference of the total efficiency of the
proposed system was 6%. On the other hand, the difference of the exergy total efficiency of the same
temperature range was 30%. Moreover, as a result of examining how to improve the exergy efficiency
of this system, certain improvement methods were proposed. (a) Preheat the city-gas and air supplied
to the system using exhaust heat, and raise the combustion temperature, (b) Preheat the water supplied
to the system using exhaust heat, (c) Change the catalyst material of each unit and reduce the amount
of cooling of the reformed gas, (d) Examination of combined cycle power generation. The exergy efficiency, in the case of introducing the proposed system into individual homes in Sapporo, Tokyo, and
Kagoshima in Japan was evaluated. Consequently, when the system was introduced into a community
with low outside air temperatures, exergy efficiency increased compared with communities with high
outside air temperatures.
Chapter 9 has described the design support using a neural network algorithm. A microgrid with
the capacity for sustainable energy is expected to be a distributed energy system that exhibits quite a
small environmental impact. In an independent microgrid, green energy, which is typically thought
of as unstable, can be utilized effectively by introducing a battery. In the past study, the production-ofelectricity prediction algorithm (PAS) of the solar cell was developed. In 1st section, a layered neural
network is made to learn based on past weather data and the operation plan of the compound system
of a solar cell and other energy systems was examined using this prediction algorithm. In this study,
a dynamic operational scheduling algorithm is developed using a neural network (PAS) and a genetic
algorithm (GA) to provide predictions for solar cell power output. We also do a case study analysis in
which we use this algorithm to plan the operation of a system that connects nine houses in Sapporo
to a microgrid composed of power equipment and a polycrystalline silicon solar cell. In this work, the
relationship between the accuracy of output prediction of the solar cell and the operation plan of the
microgrid was clarified. Moreover, we found that operating the microgrid according to the plan derived
with PAS was far superior, in terms of equipment hours of operation, to that using past average weather
data. In the 2nd section, the bioethanol reforming system (FBSR) using sunlight as a heat source is described. FBSR is a fuel production system for fuel cells with little environmental impact. An operation
prediction program of the FBSR using a layered neural network (NN) with the error-correction learning
xiii
method has been developed. The weather pattern (the amount of global solar radiation and the outside
air temperature) and energy-demand pattern for the past one year are inputted into the NN. Moreover,
training signals are calculated by a genetic algorithm. The training signals are given to the NN, and the
operation pattern of the FBSR is made to learn. As a result of analyzing using the developed algorithm,
when 20% or less of power load fluctuation occurred, the operation plan was analyzable in 14% or less
of error span. On the other hand, in operation prediction when 50% or less of fluctuation is added to
the outside temperature and global solar radiation, there was 16% or less analysis error.
Chapter 10 has described Microgrid with Numerical Weather Information. A fuel cell microgrid with
photovoltaics effectively reduces greenhouse gas emission. A system operation optimization technique
with photovoltaics and unstable power is important. In the 1st section, the optimal operation algorithm
of this compound microgrid is developed using numerical weather information (NWI) that is freely
available. A GA (genetic algorithm) was developed to minimize system fuel consumption. Furthermore,
the relation between the NWI error characteristics and the operation results of the system was clarified.
As a result, the optimized operation algorithm using NWI reduced the energy cost of the system.The
production of electricity from the solar cells continues to attract interest as a power source for distributed energy generation. It is important to be able to estimate solar cell power to optimize system energy
management. The 2nd section proposes a prediction algorithm based on a neural network (NN) to predict the electricity production from a solar cell. The operation plan for a combined solar cell and diesel
engine generator system is examined using the NN prediction algorithm. Two systems are examined in
this section: one with and one without a power storage facility. Comparisons are presented of the results
from the two systems with respect to the actual calculations of output power and the predicted electricity
production from the solar cell. The exhaust heat from the engine is used to supply the heat demand. A
back-up boiler is operated when the engine exhaust heat is insufficient to meet the heat demand. Electricity and heat are supplied to the demand side from the proposed systems, and no external sources are
used. When the NN production-of-electricity prediction was introduced, the engine generator operating
time was reduced by 12.5% in December and 16.7% for March and September. Moreover, an operation
plan for the combined system exhaust heat is proposed, and the heat output characteristics of the backup boiler are characterized.
Chapter 11 has described SOFC-PEFC Combined Microgrid. In the 1st section, the combined system of a solid-oxide fuel cell (SOFC) and a proton-exchange membrane fuel cell (PEFC) is examined.
The proposed system consists of a SOFC-PEFC combined system and a photovoltaic system (PV) as
the energy supply to a microgrid. The exhaust heat of the SOFC is used for the steam reforming of the
bio-ethanol gas with time shift utilization of the exhaust heat of the SOFC in optional time. The SOFCPEFC combined system with the PV was introduced in a microgrid of 30 residences in Sapporo, Japan.
The operation plan of the system has three cases: without solar power, with 50% and with 100% of
solar output power. Moreover, three types of system operation of using the SOFC independent operation, PEFC independent operation and SOFC-PEFC combined system are used to supply the demand
side. A comparative study between the types of system operation is presented. The power generation
efficiency is investigated for different load patterns: average load pattern, compressed load pattern and
extended load pattern. This study reported that the power generation efficiencies of the proposed system in consideration of these load patterns are 27% to 48%. In the 2nd section, a microgrid, with little
environmental impact, is developed by introducing a combined SOFC and PEFC system. This section
is investigated the operation of a SOFC-PEFC combined system, with time shift operation of reformed
gas, into a microgrid with 30 houses in Sapporo, Japan. The SOFC is designed to correspond to base
xiv
load operation, and the exhaust heat of the SOFC is used for production of reformed gas. This reformed
gas is used for the production of electricity for the PEFC, corresponding to fluctuation load of the next
day. The relation between operation method, power generation efficiency, and amount of heat storage of
the SOFC-PEFC combined system to the difference in power load pattern was investigated. The average
power generation efficiency of the system can be maintained at nearly 48% on a representative day in
February (winter season) and August (summer season).
Chapter 12 has described bioethanol solar reforming system for distributed fuel cell.
The 1st section has described hydrogen production of a bioethanol solar reforming system for distributed fuel cell. The development of a bioethanol steam reforming system (FBSR) is considered as a
means of distributing energy using PEFCs. Small-scale solar collectors (collection areas on the order
of several m2) are installed in a house as a method for applying the FBSR. However, the temperature
distribution of a reforming catalyst fluctuates under conditions of unstable solar insolation. Therefore,
heat transfer analysis applied in reforming the catalyst layer of the reactor and the temperature distribution and transient response characteristics of the gas composition of the process were investigated. In
the 2nd section, the development of a bioethanol reforming system for fuel cells (FBSR) using sunlight
as a heat source was investigated. The system was investigated using the experimental result of catalyst
performance, and numerical analysis. The overall efficiency of the production of electricity and heat power
of this system was determined by examining its thermal output characteristic. The FBSR was introduced
into standard individual houses in Sapporo, Japan for analysis. The amount of hydrogen production,
the production-of-electricity characteristic, and the thermal output characteristic were examined using
meteorological data on representative days in March and August. As a result, the overall efficiency of
the system, defined as the rate of power and heat output compared to the amount of solar heat collected,
was calculated to be 47.4% and 41.9% on the representative days in March and August, respectively.
xv
Acknowledgment
Special thanks also go to the publishing team at IGI Global, Ms. Jan Travers, Allyson Gard, Ms. Christine Smith, Ms. Erika L. Carter, Ms. Emily E. Golesh, and Mr. Mike Killian, whose contributions
throughout the whole process from inception of the initial idea to final publication have been invaluable.
In particular to Jan Travers, who assisted in keeping this project on schedule.
Shinya Obara
Kitami Institute of Technology, Japan
October 2013
Chapter 1
Operating Schedule
of a Combined Energy
Network System
ABSTRACT
This chapter consists of two sections, Operating Schedule of a Combined Energy Network System with
Fuel Cell and Fuel Cell Network System Considering Reduction in Fuel Cell Capacity Using Load
Leveling and Heat Release Loss. The chromosome model showing system operation pattern is applied
to GA (genetic algorithm), and the method of optimization operation planning of energy system is developed in the 1st section. In the case study, the operation planning was performed for the energy system
using the energy demand pattern of the individual residence of Sapporo, Japan. Reduction in fuel cell
capacity linked to a fuel cell network system is considered in the 2nd section. Such an energy network
is analyzed assuming connection of individual houses, a hospital, a hotel, a convenience store, an office
building, and a factory.
GENERAL INTRODUCTION
The summary of the 1st section is as follows. The
optimization method of this operation planning
was applied to the compound system of methanol
steam reforming type fuel cell, geo-thermal heat
pump and the electrolysis tank of water. In the case
study, the operation planning was performed for the
energy system using the energy demand pattern of
the individual residence of Sapporo, Japan. From

analysis results, the amount of outputs of a solar
module and the relation of the operation cost of
the system which are changed by the weather were
clarified. The representation day in February of
the ratio of the operation cost in case of (0% of
output rates) the rainy weather to the time of fine
weather (100% of output rates) is 1.12. And the
representation day in July is 1.71. Furthermore, the
DOI: 10.4018/978-1-4666-5796-0.ch001
Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Operating Schedule of a Combined Energy Network System
optimal capacity of accumulation-of-electricity

and thermal storage was estimated, and they are
308MJ and 23MJ, respectively.
The summary of the 2nd section is as follows.
When the power demand of the whole network is
small, some of the electric power generated by the
fuel cell is supplied to a water electrolysis device,
and hydrogen and oxygen gases are generated.
Both gases are compressed with each compressor and they are stored in cylinders. When the
electric demand of the whole network is large,
both gases are supplied to the network, and fuel
cells are operated by these hydrogen and oxygen
gases. Furthermore, an optimization plan is made
to minimize the quantity of heat release of the hot
water piping that connects each building. Such an
energy network is analyzed assuming connection
of individual houses, a hospital, a hotel, a convenience store, an office building, and a factory.
Consequently, compared with the conventional
system, a reduction of 46% of fuel cell capacity
is expected.
OPERATING SCHEDULE OF A
COMBINED ENERGY NETWORK
SYSTEM WITH FUEL CELL
Introduction
Until now, various energy devices with individual
controls have been used in buildings. However,
renewable energy and unused energy are positively
utilized from the viewpoint of global environment
problems from now on. In order to utilize renewable energy and unused energy, it is necessary to
use active energy device for stabilization of an
energy output. The object of study is to develop
the method of the operation plan and optimum
design of the combined system of active energy
device and unutilized energy. The energy network
is structured using an electric power system, a hot
water system, and a fuel system. For the operational plan of the energy network that conducts the
cooperative operation of complex energy devices,

it is necessary to solve the nonlinear problem of
many variables with objective functions provided
beforehand. In the optimization calculations of
system operational planning of a complex energy
system, linear approximation calculations based
on the mixed-integer plan-making method was
used (Ito, Shibata, & Yokoyama, 2002). However,
to analyze the operational planning of a complex
energy system with high accuracy, it is necessary to
solve the nonlinear problem with many variables.
Until now, the operation planning of an energy
system has been managed as a linear problem.
So, in this Section, the method of analyzing a
compound energy system by many variables and
nonlinearity is developed.
A genetic algorithm (Goldberg, 1989) is therefore introduced to analyze operational planning
in this Section. Previously, an analysis method
of a large-scale energy system that combined a
genetic algorithm and an annealing algorithm
(Hongmei et al., 2000) was developed (Srinivas
& Patnaik, 1994; Fujiki. et al., 1997; Yu et al.,
2000). However, an analysis method that optimizes
the operational pattern with the application of a
genetic algorithm (GA) to the compound system
built using an active energy device, a renewable
enegy device, and an unutilized energy device
has not yet been developed. In this Section, a GA
analysis method for a compound energy system is
developed as a preliminary survey of the energy
network that conducts cooperative operation. The
analysis software using GA developed in this
Section is introduced in an individual house in
Sapporo, Japan, which is a cold, snowy area, and
the operational plan is investigated. The operational planning of the compound energy system
is analyzed using the minimization of operational
costs and the maximization of renewable energy
use, and the operational planning of an active
energy device is considered. Although operation
costs and facility costs need to be considered for
a feasibility study of the system, the facility costs
of a proton exchange membrane (PEFC) fuel cell
are changing greatly. Since estimating facility

costs is difficult, the analysis in this Section only
considers operation costs. Furthermore, the device
capacity for the accumulation of electricity and
thermal storage is estimated.
System Description
Network of Distributed Energy Devices
In a dispersed arrangement of small energy
devices, a reduction in power transmission loss
and heat dissipation loss is expected. Since the
discharge of carbon dioxide is predicted, renewable
energy devices and unutilized energy devices are
connected along with established active energy
devices in an energy network, and research on supplying energy to two or more houses is required. A
network model of the fuel cell cogeneration (CGS)
installed in individual houses, as assumed for the
final target of this research, is shown in Figure
1(a). The fuel cell CGS installed in each house is
connected with hydrogen gas system piping, an
electric system power line, and hot water piping
of an exhaust heat system. The hot water system
recovers heat from fuel cells and supplies thermal
energy to individual houses. Hot water flows in
one direction, as shown by the arrows in Figure
1(a). The energy devices are connected to the
electric power and thermal energy network, and
the operational planning of a system that fulfills
the energy demands of individual buildings is
considered. The energy devices installed in each
house were controlled by autonomous distribution.
The objective of an energy network is to control
the devices linked to the network cooperatively,
and to obtain a better effect than conventional
autonomous distribution control.
Figure 1(b) shows a model of the cooperative
operational control of an energy network. The
control device of the energy network is composed
of a computer, a communication device, and a
LAN that communicates control information for
each energy device. In this system, the operational state of each energy device linked to LAN,
weather information and maintenance information
can be communicated to the outside.
The Combined Energy System

A feasibility study of the operational planning
analysis method of the energy network with
cooperative control, shown in Figure 1(a), is the
target of this Section. The analysis method in the
case of operating the compound energy system
consisting of an active energy device, a renewable energy device, and an unutilized energy
device has been developed. Figure 1(c) shows
the model of the compound energy system. The
analysis method for operational planning using
GA has been developed to minimize costs, and
the estimated device capacities and an operational
plan for a complex energy system are determined.
The Combined Energy

System to be Assumed
A block diagram of the energy system for houses
adopted in this study is shown in Figure 1(a).
In this system, methanol fuel is stored, and its
distributed power supply is also possible in the
residential areas of local cities where the city
gas piping networks are not well developed. In
individual houses and apartments, load changes
are sharp and abrupt, and partial load operation
of the energy system increases (Obara et al.,
2005). Therefore, to improve energy efficiency,
a dynamic operational plan for the energy system
is required, wherein thermal storage and electric
energy storage are introduced. A water electrolytic
bath and gas tanks are added to the electric power
storage device, and electric power generated with
a fuel cell and a solar cell is supplied to the water
electrolysis bath. Since the output of a solar cell
changes according to the weather, the operational
plan is defined by considering the amount of power
Figure 1. Combined energy network system
generation as a variable. Furthermore, when installing the fuel cell cogeneration system in houses
in cold regions, since the supply of heat energy
is insufficient, there is additional combined use
of the geothermal heat pump system.
Operation Method of Combined

Energy System
Figure 2(a) shows the compound energy system
for individual houses, and is investigated by this
study. Active energy device is a fuel cell with a
methanol reformer, and unutilized energy device
is a solar module and a geo-thermal heat pump.
This Section does not describe the equipment
cost of the system of Figure 2(a). The equipment
cost of PEFC and unutilized energy device is not
commercially realized in the present condition.
However, such equipment costs may be reduced
rapidly from now on. Therefore, it is necessary
to investigate the operation plan of a compound
energy system. Methanol fuel (mole ratio of
methanol/water = 1.0/1.4), which is contained in
the methanol tank (3), is supplied to the reformer
(2), and hydrogen and carbon dioxide are formed.
The methanol reformer is always warmed up.
The piping system of hydrogen and reformed gas
assumes use of a stainless steel tube. The specification of the reformer and the fuel cell stack was
shown in Table 1. The heat source of the reformer
drives the catalytic combustion of methanol, and
the air for combustion is supplied by the blower.
The reformed gas generated by the methanol steam
reformer is sent to the anodes, air is supplied to
the cathodes by a blower, and electricity is generated by the PEFC (1). The energy supply path for
this system is shown in Figure 2(b). The electric
power generated by the fuel cell is supplied using one of the following methods: (a) Alternating
current electric power is generated by the DC/
AC converter (8), and demand is fulfilled; (b)

Hydrogen and oxygen are generated in the water
electrolyzer (5) and stored in a hydrogen tank (6)
and an oxygen tank (7), respectively; (c) Electric
power is changed into heat by an electric heater
(9) in a thermal storage tank (10). It is possible
to drive a fuel cell at any time using the stored
hydrogen and oxygen. Selecting the appropriate
energy supply path among (a) to (c) above is also
possible for electric power generated by the solar
cell. Because the capacity of a water electrolyzer
and a thermal storage tank differs in the operation plan of a system, they are determined from
analysis output. However, electric power from the
system to the demand side is supplied only via
one of the following systems, without multiple
supply sources: (a) Methanol fuel is reformed to
generate hydrogen, which is supplied to a fuel
cell, and electric power is generated; (b) Electric
power is generated by the solar cell; (c) Stored
hydrogen and oxygen, formed by water electrolysis, are used in the fuel cell to generate electric
power. In order to reduce the discharge of carbon
dioxide, methanol fuel is not used to the extent
possible. Therefore, as many renewable energy
supply sources as possible are used with priority
set in decreasing order to be (b), (c) and (a).
A thermal storage tank has the following three
heat input sources: (a) Exhaust heat from a fuel
cell and the reformer; (b) Heat conversion of the
electric power generated by the fuel cell and the
solar cell; (c) Heat generated by the geothermal
heat pump (11). The low-temperature source of
heat pump is obtained from a bore hole with a
depth of 30m installed in soil. And maximum
output is 12 kW and COP is 3. However, when
the heat input exceeds the thermal storage capacity, some of the surplus is released. After the heat
from the thermal storage tank heats city water via
the heat medium inside the thermal storage tank,
Figure 2. Outline of the proposal system
it is supplied to the demand side. A thermal storage medium is water and the maximum temperature of thermal storage is 353 K.
For the specifications of other system component devices, we used the values shown in Table 1,
which are typical for houses in cold regions such
as Sapporo. Since a geothermal heat pump was
used, the capacity of a fuel cell was set at 4.2 kW.
With respect to device costs, installing a complex
energy system such as shown for individual houses
in Figs. 2(a) and 2(b) is difficult.
Analysis
Analysis Method Using
Genetic Algorithm (GA)
1. Indication of Device Operation by the
Chromosome Model
Figure 3(a) shows the chromosome model introduced in GA, and expresses information on electric energy output E D ,t for each time tk of
i k
device Di , heat output H D ,t , amount of electric

i k
Table 1. Energy device initial specifications
energy storage S E ,D ,t , amount of thermal storage

i k
Sst ,D ,t , and device selection switch SW ,D ,t

i k
i k
using the gene model with 0 and 1 (Obara and

Kudo, 2003). When two or more devices do not
yield a simultaneous energy supply, SW ,D ,t is
i k
introduced in order to select the device that supplies energy. As the chromosome model determined above expresses the operational pattern of
the device from tk to tk +1 . As shown in Figure
3(b), sets of the chromosome model of each sampling time of k = 0, 1, 2, , R represent all the
operational patterns for operational period R .
Although a number of chromosome models N dv'

are created as an initial generation, either 0 or 1
is selected. If the value of the random number is
less than 0.5, the gene model is set at 0, and 1 is
selected if the random number is 0.5 or more.
2. Production, Selection and Reproduction
The fitness values of the number of N dv' chromosome model groups (they indicate the patterns of
the system operation) of an initial generation are
calculated, and proliferation or selection is judged
based on the values. The combination method is
introduced in the calculation of the ranking selec-
Figure 3. Chromosome model
tion (Baker, 1985) and roulette selection (Goldberg, 1989). In the first reproduction calculation,
the chromosome models of the initial generation
are selected based on the number of N dv (here,
N dv' > N dv ), and these chromosome models are
used in subsequent calculations.
3. Crossover and Mutation of the Chromosome Model
The calculation process of crossover and mutation
is given to the chromosome model group, and the
diversity of genes is maintained. Using the calculation for the last generation, the chromosome
model with the best fit is determined as the optimal operational pattern. However, the number of
generations in the analysis is decided beforehand.
When using the chromosome model group with
the crossover process, only a specific chromosome
model evolves beyond a certain generation, and
a model with high fitness cannot be found beyond
it. In the calculation of crossover, two parent

chromosomes are chosen by probability Pcros ,
parent chromosomes are combined, and one child
chromosome is generated in the intersection position decided at random. Subsequently, the calculation process of the mutation described below is
added. In the mutation, parent chromosome models are chosen at random using probability Pmut ,
and the number and the position of the genes of
the parent chromosomes are also decided at random. If the original value of a gene is 1, it has to
change to 0, and if it is 0, it has to change to 1. In
order to progress to the next generation, the fitness
value is again evaluated with respect to all the
operational patterns of number N dv with added
crossover and mutation. Proliferation and selection
are performed using these results. The above
analysis is repeated up to the number of the last
generation, and the gene arrangement of the
model that has the highest fitness value in the
chromosome model group of the last generation

is decoded, and the optimal operational pattern
is decided.
4. Analysis Flow
The flow of calculation of the operational planning analysis of the complex energy system using
GA is shown in Figure 4(a). First, N dv' chromosome model groups described in previous Section
are generated at random. The fitness values for
each chromosome model are calculated, and the
chromosome models of N dv higher ranks are

determined by the combination of ranking selection and roulette selection. Furthermore, the
calculation of production and selection described
in previous is added to these N dv chromosome
models, and a chromosome model with a large
fitness value is obtained, maintaining diversity
by the calculation of crossover and mutation
described in previous Section. This calculation is
repeated, and the chromosome model with the
highest fitness value when reaching the number
Figure 4. Analysis flow, energy demand pattern, and output characteristics of equipment
of the last generation, decided beforehand, is

determined as an optimal model. Operational
planning of all energy devices for each sampling
time is decided by decoding the optimal model.
Cold-Region Houses
1. Characteristics of Weather in Sapporo in
Japan
Sapporo is a cold, snowy region, and the annual
average temperature for the past five years is 282
K (National Astronomical Observatory of Japan,
2003). The average temperature in February is
270 K, and the highest and the lowest temperatures on a representative day are 273 K and 266
K, respectively. Moreover, there is an average
25 days of snowfall in February. On the other
hand, the highest and the lowest temperature on
a representative July day for the past five years
are 298 K and 290 K, respectively, and the average temperature is 293 K. Since air heat-source
heat pumps cannot be used in winter, the use of a
geothermal heat pump is assumed in this Section.
Air conditioning is not needed during summer.
2. Characteristics of Individual Houses in
Sapporo
The average individual house in Sapporo is a
2-story wooden house with a 140-m2 living
area (Narita, 1996). The model of the average
electric power and thermal energy demand of
the representative February day and July day for
individual houses in Sapporo is shown in Figure
4(b). The thermoelectric ratio of representative
days is 0.90:0.1 in February and 0.5:0.5 in July.
Characteristics of Energy Devices

1. Fuel Cell Cogeneration
Figure 4(c) is the result of examining the relationship between electric power and thermal
energy output, and the fuel amount of supply.
The characteristic curve is divided into two or
more regions, and each region is approximated
10
by the least-squares method with an equation of

secondary order. The characteristics of electric
power shown in Figure 4(c) are for a model that
includes the power consumption of blowers and
electric power loss of the DC/AC converter.
Moreover, the values of the joule heat of the fuel
cells, the battery reaction heat, and the exhaust
heat of the reformer have been included in the heat
characteristics in the figure. Methanol fuel using a
burner for the heat sources of the reformer is also
included on the horizontal axis of Figure 4(c). In
addition, to start the fuel cell system, consumption
of methanol fuel equivalent to 900 kJ (250 Wh)
is considered (Takeda, 2004). In order to collect
the hydrogen and oxygen generated by water electrolysis, tanks are installed in the electric energy
storage device. For fuel cell systems using not the
gas obtained by steam reforming of methanol but
the hydrogen and oxygen in each tank, the power
generation efficiency is 0.75 and the heat output
is set at 0.05 (Obara et al., 2005).
2. Geothermal Heat Pump
Based on the examination results of hydrocarbon
binary vapor (HC-TECH Inc, 1997), we simplify
the analysis by setting the temperature
TL (= 277 K) of the low-temperature heat source
and the condensation temperatureTH (= 347 K)
to be constant, and the coefficient of pump COPt
at 3.0 .
3. Water Electrolyzer
If direct current power is supplied to a water
electrolyzer, hydrogen can be produced at a rate
of Equation (1). E EL,t indicates the amount of
k
electric energy supplied to the water electrolysis

bath, and EL expresses the efficiency of the
charge. The flow rate of hydrogen QH
,t
2 k
gener-
ated from sampling time tk to t is calculated

by the equation below. The oxygen flow rate is
also determined by the same calculation. In a
report on the water electrolysis bath for hydrogen

generation, efficiency EL of charge is given as
0.85 (Kosaka et al., 2000). Here, Ec , Fd , and EV
shows the chemical equivalent, Faraday constant
and voltage, respectively.
QH
,t
2 k
E EL,t Ec
k
Fd EV
EL
(1)
4. Thermal Storage Tank

SSt,max is the maximum thermal energy storage,
and TSt,max is the maximum temperature of the
heat medium. Equations (2) and (3) are restrictions
for thermal storage. V is the capacity of a thermal
storage medium volume (calcium chloride is assumed), C p is the specific heat and T is the air
temperature outside the thermal storage tank. The
thermal storage temperature during sampling time
tk is calculated by:
TSt ,t = SSt ,t / ( C p V ).
k
0 SSt ,t SSt ,max
(2)
T,t TSt ,t TSt ,max
(3)
The following equation is an expression of the

thermal energy storage between time tk and t .
SSt ,t SSt ,t
k 1
= {H St ,in ,t H St ,out ,t
k
St C p V (TSt ,t T,t )} t
k
(4)
H St ,in ,t and H St ,out ,t show the input and output

k
heat energies of the thermal storage tank, respectively, and the loss of thermal storage is the 3rd
term within { } on the right-hand side of Equation
(4) when it depends on open-air temperature T,t .

k
ST and shows the efficiency of thermal storage and the density of thermal storage medium,
respectively. In this Section, thermal storage loss
at time tk of the representative day will be considered as 1% of the value on the left-hand side
of Equation (4) in July and 2% in February. These
heat losses were calculated and determined from
the difference of temperature of ambient air and
a thermal storage medium.
5. Solar Modules
Figure 4(d) shows the results for a solar cell in
Sapporo in winter (representative days in February) (Obara & Kudo, 2003; Nagano, 2002). The
solar cell is a roof installation-type device installed
perpendicularly so that it does not become covered
with snow. The characteristics for the representative days in July shown in the figure are the
predicted results. Each characteristic curve is the
amount of power generated during fair weather,
and power generation falls during cloudy or rainy
weather. Using the output characteristic performance of the solar cell as 0% in snowfall, 50%
under cloudy conditions and 100% in fair weather
in Figure 4(d), the operational planning for the
representative day for each month is analyzed.
Objective Function and

Energy Equation
1. Objective Function
Only methanol is used as the fuel supplied to the
system as shown in Figure 4(b). Therefore, operational planning to minimize costs requires an
operational pattern where the methanol fuel consumption is minimum for each sampling time tk .
Di represents the energy device and the subscript
i ( i = 1 , 2 , 3 ,..., M and M is the total number
of devices) corresponds to the device number used
here. The operational costs of device Di during
sampling time tk to t are equal to fuel input
11
flow FD ,t of the device multiplied by unit fuel
and H System ,t are decided on the basis of energy
price C fuel . The operational costs of the whole

system are estimated using Equation (5). Therefore, the total operational costs of all working
periods of a system are calculated using Equation
(6) and called the best-fit solution, so that the
value of Equation (6) should be small. In the
example of the application of operational planning
in the following Section, C fuel of methanol fuel
is calculated to be 0.463 $/kg (Energy and Industrial Technology Development Organization in
Japan, 1999))
demand patterns. The left-hand side of Equation

(7) shows electric power output from the fuel cell
( E FS ,t ) and solar cell ( ESL,t ). The right-hand
i k
i =1
i k
(5)
23
C System ,day = C System ,t
(6)
tk =0 i =1
side of Equation (7) shows the electric power

consumption of the water electrolyzer ( E EL,t ),
k
electric power consumption of the heat pump (

E HP ,t ), and electric power converted into heat
k
by the electric heater ( E H ,t ), respectively. The

k
left-hand side of Equation (8) shows thermal

energy output from the fuel cell heat exhaust, heat
pump ( H HP ,t ), electric heater ( H H ,t ) and thermal
k
C System ,t = C fuel FD ,t t
k
2. Energy Balance
Equations (7) and (8) are the electric power and
thermal energy balance equations of this system,
respectively.
storage tank ( H St ,t ), respectively. FS , FFS ,t ,

k
and FS shows the calorific power of methanol

fuel, the quantity of methanol fuel mass flow and
the fuel cell stack efficiency. FS is calculated
from the relations between the amount of supply
of a methanol fuel, and a power output shown in
Figure 4(c). The right-hand side of Equation (8)
shows heat loss from the thermal storage tank (
H St ,t ) and heat release from the radiator (
k
H Rad ,t ), respectively.
k
E FS ,t + ESL,t
k
Operation Planning
= ESystem ,t + E EL,t + E HP ,t +E H ,t (7)

k
FS FFS ,t (1 FS )
k
+H HP ,t + H H ,t + H St ,t
k
= H System ,t + H St ,t + H Rad ,t
k
(8)
The left-hand sides of Equations (7) and (8)

correspond to the output energy from the system,
and the right-hand sides correspond to the amount
of consumption energy of the system. ESystem ,t
k
12
1. Analysis Method of System Operational

Planning
The mixed-integer plan-making method has
been studied to analyze the operational planning
of an energy system (Ito K, Shibata T, & Yokoyama R., 2002). In this method, the nonlinear
input-and-output characteristics of energy devices are expressed as a linear model and analyzed.
An example of the test results of electric power
and thermal energy output characteristics of a fuel
cell with a reformer is shown in Figure 4(c). If
the nonlinear characteristics of an energy device
can be made to fit a linear approximation problem,
an increase in analysis error is predicted. In the
mixed-integer plan-making method, the charac-
teristics of the electric power output are approximated by three straight lines l1 to l 3 in Figure
4(c), and heat output is approximated by four
straight lines l 4 to l7 . Generally, since the output
of small energy equipment is nonlinear, we should
use the nonlinear model for analysis. In the
analysis of the operational planning of the system
with a number of energy devices, many variables
associated with each device operation are used.
Therefore, if many variables can be calculated
simultaneously, the efficiency of the analysis will
increase. A genetic algorithm, where simultaneous
calculations of many variables and the calculation
of a nonlinear problem are possible, is introduced
in the software developed in this study. However,
neither the application of a GA to a small-scale
energy system nor a design method that optimizes the operational pattern and device capacity has been studied previously. In particular, no
research reports on the optimization of the operational plan for a compound system of an active
energy device, a renewable energy device and
unutilized energy device or their optimal capacity can be found in the literature.
2. Operation of a Chromosome Model
The chromosome model operated by the GA
calculation needs to satisfy the energy balance
in Equations (7) and (8). However, the chromosome model must also satisfy conditions (a) and
(b) described below:
1. A quantity that excludes electric energy
consumption (sum of all E ) from the
amount of electric energy output of the fuel
cell and the solar cell satisfies the electric
energy demand.
2. A quantity excluding heat loss (
H Rad ,t + H St ,t ) from the sum total of the
k
exhaust heat of the reformer and fuel cell,

heat pump, electric heater, and the heat
energy output of the thermal storage tank
should satisfy the heat energy demand.
When an operational pattern (chromosome

model) that does not fulfill one of these conditions arises, it is forced to a very low value of fit
so that it cannot proceed to the next generation.
Similarly, a low fitness value is given for an operational pattern that does not satisfy the energy
balance of Equations (7) and (8).
For the chromosome models of the initial
generation for the power generation of a fuel
cell, the total power generated by the fuel cell is
decided at random within the electric power
capacity. In addition, the total power generated
is distributed to the amounts of electric power
output ( E FS ,t ), the quantity supplied to a water
k
electrolyzer and stored as electricity ( E EL,t ),

k
and the quantity conducting heat conversion (

E H ,t ) in an electric heater at random. The total
k
amount of power generated in fine weather in

sampling time tk in a solar cell is decided as
shown in Figure 5(a). The total power generated
from the solar cell is distributed to the electric
power supplied to the electric power output (
ESL,t ), the amount of accumulated electricity (
k
E EL,t ) and amount of electric power supplied

k
to a heater ( E H ,t ) for every chromosome modk
el at random. Moreover, the power consumption

( E HP ,t ) of the geothermal heat pump for every
k
chromosome model is estimated from the amount

of heat output that was decided at random
within the limits of the device capacity and
coefficient of performance (COP).
Case Study
Analysis Conditions
Operational period R of a system is split into 23
parts, and:
tk (k = 0 , 1 , 2,....., 23)
13
Figure 5. Output characteristics of solar power generation and fuel cell, and system device cost
14
defines the sampling time. Moreover, the number

of devices M is set at five including the fuel cell
with the reformer, solar cell, geothermal heat
pump, water electrolyzer, and thermal storage
tank. In the analysis described below, the number
of initial-generation chromosome model groups
N dv' is set at 3000, and the number of chromosomes
operational after reproduction N dv is set at 2500.
The last generation is analyzed as 100 generations.
Moreover, considering the maintenance of the
diversity of gene models, the number of intersections is selected randomly. That is, with one intersection, 0.2% of the total number of chromosomes is extracted as parent chromosomes, at
maximum. The frequency of mutations governed
is about 4% of the genes in a mutated chromosome
model, at maximum. These parameters of the SGA
confirm that the value of the optimal solution is
in agreement within several percent, as a result
of trial calculations with two or more parameters.
The minimum operational costs for every
generation when performing an operational plan
by GA to minimize the operational costs with the
application of the energy demand pattern of representative days in July and February are shown
in Figure 5 (a). In this calculation, we assumed
fair weather and the electric power output of a
solar module to be 100% (same as in Figure 4(d)).
Although the fitness value of the representative
day for both months decreases rapidly, the operational costs, to almost ten generations, shows
a gradual change in successive generations. For
the best-fitness solution after 10 generations, the
mutation calculation is important. However, if
analysis conditions are changed, the number of
generations for convergence changes.
Objective of Analysis Calculation

The operational planning of a system is analyzed
using the energy demand pattern of the representative days in February and July in Sapporo. We
assume that the active energy device (fuel cell
cogeneration with a methanol reformer) is already

installed in individual houses in Sapporo in this
analysis. The renewable energy device (solar cell),
unutilized energy device (geothermal heat pump),
electricity accumulation device (water electrolyzer), and thermal storage tank are connected to an
active energy device, and the complex independent
energy system shown in Figure 4(b) is built. It is
difficult to introduce such a complex system into
individual houses because of the device costs. This
section investigates the energy network system of
the distributed energy device. This Section also
examines, the operational planning method of the
active energy device with the maximum use of
renewable energy using the information obtained
from the analysis. Furthermore, the capacity of
devices to accumulate electricity and thermal
storage is estimated.
Results and Discussion

Results of Operational Planning
Figure 5(b) shows the analytical results of operational planning assuming fair weather with
the energy demand pattern for the representative
day in each month, and it shows the device energy
output for every time period. However, the output
results of the fuel cell include the output values of
both the electric energy and the heat energy. The
breakdown of the energy output of the fuel cell is
shown in Figure 5(c). Hydrogen has been formed
by the steam reformation of methanol in the time
period with high thermal output (0:00 and 1:00
on the representative day for July and 0:00~9:00
and 19:00 and 21:00 on the representative day for
February). A thermal output is small when supplying and generating the hydrogen and oxygen
produced by water electrolysis to the fuel cell.
The reason for this is that the use of renewable
energy (solar cell) is a top priority. Therefore, if
the amount of power generated by a solar cell
exceeds the electricity demand, the fuel cell will
not operate by driving the reformer. If, from night
15
to early morning, power generation from a solar

cell is not conducted, the fuel cell operates by the
driving reformer. On a representative day, power
from a fuel cell by the operation of the reformer is
generated from 0:00 at 9:00 in February, and the
heat pump is operated using this electric power.
From the analysis results of the operational plan,
thermal storage of the heat generated by the heat
pump occurs, and the stored thermal energy is
used to conduct the time shift and to fulfill heat
demand during the daytime.
Figure 5(d) shows the analytical results of the
system operational costs for conducting operational planning in which the output proportion of
the solar cell is a variable, for the representative
day in each month. Using the characteristic output
performance of the solar cell of 0% in snowfall,
50% under cloudy conditions and 100% in fair
weather in Figure 4(d), operational planning for
the representative day in each month is analyzed.
The total value of one day of electric power and
thermal energy demand is set at 100. The representative days in February and July of the total
electric power obtained in the solar cell for fine
weather are 95 and 28, respectively. On the other
hand, from Figure 5(d), the system operation costs
during fine weather when setting the operation
costs in case of rainy weather and snowfall at 100
for February and July representative days are 88
and 29, respectively. The difference of the operation cost of every month is equal to the difference
of the consumption of a methanol fuel. Therefore,
the difference of operation cost is the difference
of a carbon-dioxide discharge.
The difference in the operational costs of the
system according to the difference in weather for
the representative days in February (100-88 = 12)
is larger than the difference in solar cell output
(100-95 = 5) due to the weather. The main reasons
for this are that fuel cell efficiency improves by
reducing operation because of low loads in Figure
16
4(c), and because the driving period to operate

a fuel cell by generating hydrogen and oxygen
using a solar cell is long. On the other hand, for
the representative July day, the generation of hydrogen and oxygen is conducted by a solar cell,
and operation of a fuel cell in most periods is
performed with high energy demand. Therefore,
the difference in the output of a solar cell due to
the weather directly affects the operational costs
of the system.
Design of the Capacity of the Water

Electrolyzer and Thermal Storage
Figure 6(a) shows the results of the operational
planning of the amount of thermal storage, and
the accumulation of electricity for a representative February day. The largest quantity of
thermal storage and electric energy storage in
the analytical results approximates the design
capacity of each device in fair weather because
it increases as the output proportion of the solar
cell increases. Figure 6(b shows the analytical
results of the thermal storage quantity and electric
energy quantity where the output proportion of
the solar cell was 100% in the energy demand
pattern for the representative day in each month.
From these results, the maximum value of the
thermal storage quantity is 308 MJ (representative
February day), and the maximum value of electric
energy storage quantity is 23 MJ (representative
July day). Thermal storage capacity is reduced
by lengthening the operational period of a fuel
cell with the reformer and the heat pump.
Conclusions
Analysis software for operational planning, when
two or more energy devices were introduced in
an individual house, was developed. A genetic
algorithm (GA), which can analyze nonlinear
Figure 6. Quantity of energy storage
problems and many variables at a time, was used

for the analysis software for the operational planning of the complex energy system developed in
this Section.
The operational planning to introduce an active
energy device (fuel cell cogeneration), a renewable
energy device (solar cell), and an unutilized energy
device (geothermal heat pump) in an individual
house in Sapporo, Japan, which is a cold, snowy
region, as an analysis example was conducted using
the developed software. Furthermore, the device

capacity of accumulation for an electricity storage
device and a thermal storage tank were estimated.
From the results of the analysis, a fuel cell with
a reformer operates from night to early morning
when a solar module is not operational. A heat
pump also operates from night to early morning.
The thermal storage of heat generated by the heat
pump is conducted, and this heat is supplied to
the heat load during the daytime.
17
FUEL CELL NETWORK SYSTEM

CONSIDERING REDUCTION
IN FUEL CELL CAPACITY
USING LOAD LEVELING AND
HEAT RELEASE LOSS
Introduction
In order for installation of the fuel cell system to
houses or a small-scale and middle-scale building
to spread, it is necessary to reduce equipment cost.
Consequently, a fuel system network (hydrogen
piping and oxygen piping) and an energy network
(a power transmission line and hot water piping)
of distribution fuel cells are proposed (Obara &
Kudo, 2005). In this system, common auxiliary
machinery is installed in a machinery room. In
this chapter, in order to reduce the capacity of the
fuel cell connected to the network, the method
of leveling the load is proposed. By this method,
hydrogen and oxygen are generated by water
electrolysis at the time of low load with little
power demand, and each gas is compressed and
stored. On the other hand, the stored gas is supplied and generated to the fuel cell in a period of
large power load. The experimental result shows
that the power generation characteristics improve
greatly compared with air supply, when supplying
oxygen to the fuel cell (Badami &Caldera, 2002).
Therefore, if the oxygen generated when load
is small can be used for a high-load period, the
installed capacity of the fuel cell can be reduced.
Moreover, the heat-energy network is hot water
piping, and supplies heat to each building. Hot
water piping distributes heat via each building.
When there is heat excess with some buildings,
it can also recover this heat through the hot water
piping. In a heat-energy network, the hot water
temperature in a building outlet changes with the
heat consumed by each building and the fuel cell
exhaust heat of each building. Therefore, the heat
release of the overall network differs according
to outside air temperature, piping distance, the
starting point of the hot water supply, and the
18
flow direction of the hot water. Consequently,

to counteract the piping heat release loss of the
heat-energy network, the minimum piping route
is examined.
In the analysis case, the capacity reduction
effect of a fuel cell when installing load leveling
using the water electrolyzer described above is
investigated regarding a local energy network that
includes houses, a hospital, a factory, an office, and
a convenience store. Furthermore, the hot water
piping route and the fuel cell capacity placed on
each building when optimizing the system with
the object of minimizing the hot water piping heat
release are considered.
Load Leveling and Arrangement

Plan of Fuel Cell
Fuel Cell Network System
The network model with the proton exchange
membrane fuel cell (PEFC) installed that is assumed in this chapter is shown in Figure 7 and
Figure 8. As shown in each figure, the fuel system
(hydrogen piping and oxygen piping), the power
system (power transmission line), and the heatenergy network (hot water piping) between the
fuel cells installed in each building are connected.
A heat transfer medium (hot water) is flowed in
hot water piping, the exhaust heat of a fuel cell is
recovered, and heat is distributed to each building.
The route of the hot water piping can be set up
arbitrarily, and the flow direction is one way as
the arrow in each figure shows. Figure 7 shows
the system that supplies the power to Buildings A
to G from one set of the fuel cell installed in the
machinery room, and is described as the R1 type
below. A machinery room can be installed in an
arbitrary building (Building A in Figure 7). As
shown in Figure 7 (c), the fuel cell (1: this number
corresponds to that in Figure 7), water electrolyzer
(6), city gas reformer (7), hydrogen and oxygen
compressor (9 and 11), cylinders (10 and 12), geothermal heat pump (13), heat storage tank (14),
Figure 7. Fuel cell network system model (R1 type)
Figure 8. Fuel cell network system model (R2 type)
19
etc. are installed in the machinery room. The heat

output by fuel cell exhaust heat, the heat storage
tank, and the geo-thermal heat pump is distributed
to each building through a heat transfer medium.
The piping route can be planned arbitrarily and
it is in the order of Building ABCFEDGA in the
example of Figure 7 (a). As shown in Figure 7
(b), headers (4 and 5) are set in each building at
a hot water gate. The heat of the radiator (3) and
a heat exchanger connected to the header is used
for space heating and hot water supply. Figure 8
shows the system that distributes a fuel cell in all
the buildings, and this system is described as the
R2 type below. Although the number of fuel cells
increases and the equipment cost increases for the
R2 type, heat release loss with heat transport is
small. The hot water piping route of the R2 type
and the building with a machinery room can be
planned arbitrarily. In the example of Figure 8
(a), hot water is supplied in the order of Building
ADGBFECA. The machinery room of Figure
Figure 9. Single cell performance
20
8 (c) is installed in Building A. The equipment

scheme installed in the building and machinery
room in the R2 type is shown in Figure 8 (b) and
Figure 8 (c).
Ambient air is usually supplied to the fuel cell
installed in R1 and R2 from a blower. However,
both types can also supply oxygen through piping.
Moreover, it is assumed that reformed gas of the
city gas reformer and hydrogen of the cylinder
can be supplied to the fuel cells at arbitrary times
through the network.
Power Generation Characteristics

of the Fuel Cell
Figure 9 shows the power generation characteristics when hydrogen and oxygen are supplied, and
when supplying hydrogen and air by the results
of the performance measurement of a PEFC. The
differences in these power generation characteristics are considered to be due to the difference in
oxygen partial pressure, the water balance inside

the cell, and the electrical receptivity change of
the ion exchange membrane. The power generation characteristics differ between supplying reformed gas to a fuel cell, and supplying hydrogen.
However, since there are few differences in the
power generation characteristics of reformed gas
or hydrogen, this difference is ignored.
Figure 10 shows the characteristics of the
power and heat output when supplying air or
oxygen to a cathode using the same fuel cell (the
electrode surface is 1 m2) as shown in Figure 9.
The maximum power output when supplying air
to the cathode is E fca,max =1.05 kW, and it is
E fco,max =1.9 kW in the supply of oxygen. In this
way, if oxygen is supplied to the cathode, the
power output will increase. Therefore, if oxygen
is supplied and generated to a fuel cell when there
is high power demand, the fuel cell can be miniaturized compared with the design capacity by
air supply. If the fuel cell with the characteristics

shown in Figure 10 is used with maximum output,
the fuel cell facility capacity will decrease by the
value of ( E fco,max E fca ,max ).
Load Leveling Using Water Electrolysis

Figure 11 shows the model indicating power
demand amount Eneed ,t to which is added the
power demand amount of each building in Figure
7 or Figure 8 for every sampling time t . Esep in
this figure is the threshold value of the region of
low load and high load. By using this threshold
value, load leveling is attempted using the method described below. When Eneed ,t is less than Esep ,
it generates electricity by supplying reformed gas
and air to the fuel cell. However, the production
of electricity of the fuel cell is always Esep , it
supplies power whose value is the difference
between Esep and Eneed ,t to the water electro-
Figure 10. Output characteristics of a fuel cell
21
Figure 11. Fuel cell operation
lyzer, and produces hydrogen and oxygen (the

black area in Figure 11). After compressing these
gases, they are stored in each cylinder. When
Eneed ,t exceeds Esep , it generates electricity by
supplying the hydrogen and oxygen in the cylinders to the fuel cell through the network. In the
proposed method of load leveling, it is necessary
to determine Esep , where hydrogen and oxygen
are produced at the time of low load, and the
amount is consumed at the time of high load balance.
Distribution of the Fuel Cell

Figure 12 shows the model of (a) the hot water
piping route, (b) fuel cell capacity of each building, (c) change of hot water temperature, and (d)
piping heat release per unit length of the R1 type
and R2 type. The machinery room of both types
is installed in Building A, and hot water flows in
22
the order of Building ABCDEFGA for the R1

type, and it flows in the order of ADGBFECA for
the R2 type as shown in Figure 12 (a). As shown
in Figure 12 (a) and Figure 12 (b), one fuel cell
is installed in Building A (FA) for the R1 type,
and the fuel cell of the capacity of FA to FG is
installed in Building A to G for the R2 type. Hot
water of temperature TA,in ,t is input into Building
A in the R1 type. Heat is supplied for the hot
water from the fuel cell exhaust heat (FA), heat
storage tank, and geo-thermal heat pump, and as
shown in Figure 12 (c), hot water of temperature
TA,out ,t is output from Building A. After this, there
is no heat input to for the hot water, and hot water
of temperature TA,in ,t returns to the machinery
room of Building A due to the heat consumption
of Building B to G, and piping heat release. The
temperature falls as the hot water of the R1 type
progresses to Building G from Building A. Therefore, since the difference in temperature of the
Figure 12. Arrangement plan of fuel cell units
outside air and hot water is small, as shown in

Figure 12 (d), the piping heat release per unit
length is small. On the other hand, in the R2 type,
heat is supplied to hot water from distributed fuel
cells. Therefore, the outlet hot water temperature
of each building fluctuates, as shown in Figure
12 (c). As a result, the heat release per unit length
of piping also fluctuates, as shown in Figure 12
(d).
Energy Balance Equation

At sampling time t , the water electrolyzer installed
in the machinery room and the fuel cells of M
installed in M buildings are operating ( M =1 in
the R1 type). The power balance equation in this

case is expressed with the following equation.
M
m =1
m =1
E f ,m,t = Eneed ,m,t

V
+Eel ,t + Ehp,t + Esub,v,t
(9)
v =1
The left-hand side of Equation (9) expresses

the power output in the DC-AC converter outlet
of the fuel cells of M . Moreover, the 1st term on
the right-hand side is the power demand amount
in each building, the 2nd term expresses the
power consumption of the water electrolyzer, the
23
3rd term expresses the power consumption of the

heat pump, and the 4th term expresses the power
consumption of the auxiliary machinery (the pump
of the hot water network, and the compressor of
hydrogen and oxygen).
The heat balance of the system is expressed
below.
M
H
m =1
M
H
m =1
f ,m ,t
+ H st ,t + H hp,t =
M
need ,m ,t
+ H hw ,mm ,t
(10)
Qel ,H
,t
+ Qa ,H
,t
+ Qr ,H
,t
= Q f ,m ,H ,t
m =1
Qel ,O ,t + Qa ,O ,t + Qbw ,O ,t = Q f ,m ,O ,t
2
m =1
(12)
(13)
m =1
The 1st term of the left-hand side of Equation

(10) expresses the exhaust heat of the fuel cell of
M , and the 2nd term and the 3rd term express
the heat output from the heat storage tank and the
heat pump, respectively. The right-hand side of
Equation (10) expresses heat consumption, the
1st term is the heat demand of each building connected to the network, and the 2nd term expresses the heat release of the hot water piping that
connects each building. H hw ,mm ,t expresses the
heat release of the hot water piping that connects
Building m to Building m , and is calculated
from Equation (11).
H hw ,mm ',t =

h Dp lmm ' (Tm ,out ,t Tatm ,t )
(11)
Equation (12) is the balance equation of hydrogen. The 1st term of the left-hand side of
Equation (12) expresses the quantity of hydrogen
production of the water electrolyzer, and the 2nd
term expresses the hydrogen quantity supplied to
the network from the cylinder, and the 3rd term
expresses the quantity of hydrogen production of
the reformer. Moreover, the right-hand side expresses the hydrogen consumption of the fuel cell
of M . Equation (13) is a balance equation of
oxygen. The 1st term of the left-hand side expresses the oxygen concentration of the water
24
electrolyzer, the 2nd term expresses the amount

of oxygen supplied from the cylinder, and the 3rd
term expresses the amount of oxygen in the air
supply of the blower. The right-hand side is the
amount of oxygen consumed with the fuel cell.
Operating Method of the System

The exhaust heat of each fuel cell connected to the
network is used for buildings in which a fuel cell
is installed, which is given priority. The surplus
heat of each building is recovered in the hot water
network. On the other hand, when the heat of a
certain building runs short, heat is received from
the hot water piping. Moreover, when the heat
of the overall network system runs short, heat is
supplied to the network from the heat storage tank
and the heat pump. When the network has excess
heat, surplus heat is stored in the heat storage tank.
The heat pump is operated when the heat of the
heat storage tank is insufficient.
Analysis Method
Procedure of Analysis
The analysis follows steps (1) to (3).
1. Load Leveling Using Water Electrolysis,
and Calculation of Esep
The load-leveling method using water electrolysis
is employed in the R1 type (the R2 type also uses
the same procedure). In order to determine Esep
in Figure 11, an initial value is decided at random
concerning the given power demand pattern. At
this time, the amount of production of hydrogen
and oxygen in a low-load period is calculated, and

hydrogen and oxygen consumption in a high-load
period are also calculated. The balance is calculated from the amount of production and consumption of hydrogen and oxygen. The value of Esep
is changed, and it is repeatedly calculated until
the balance of hydrogen and oxygen becomes
sufficiently small. In the analysis case in next
section, the time of less than 1% of the balance
error was adopted. Balance Equation (9) of the
power, balance Equation (12) of the hydrogen and
oxygen, and Equation (13) are iused for the calculation of the balance of hydrogen and oxygen.
Figures 10 and 13 are used as the power generation characteristics of the fuel cell and the characteristics of the water electrolyzer. When the fuel
cell capacity in the analysis exceeds that of Figure
10, it is assumed that the relation of Figure 10 can
be extrapolated.
2. Calculation of Heat Release from the Hot
Water Piping
Figure 14 shows the heat release model of the hot
water piping. The fuel cell is installed in four
houses, Building A to D. Each building is connected with piping, and hot water returns to
Building A. The machinery room is set in Build-
ing A, and the heat outputs of heat storage tank

and heat pump installed in this machinery room
are H st ,t and H hp,t . There is heat demand of
H need ,A,t to H need ,D ,t in Building A to D, respectively. In the fuel cell installed in each building,
there is exhaust heat power output of H f ,A,t to
H f ,D ,t . Therefore, the heat balance of Building A
to D is calculable from Equation (10). Moreover,
the heat release (from H hw ,AB ,t to H hw ,DA,t )
of the hot water piping that connects each building is calculated using Eq, (11). Tatm ,t in Equation
(11) employs the outside air temperature in Tokyo
as shown in Figure 15 (Inoue U., 1996).
3. Route Planning of Hot Water Piping Considering Heat Release Loss
Since a fuel cell is placed in each building for the
R2 type, it is necessary to determine the capacity
of each fuel cell. The outlet hot water temperature
of a certain building is decided by the heat balance
in the building, and the heat release of the hot
water piping is calculated from the difference in
the outlet hot water temperature and the outside
air temperature. Therefore, the heat release of the
overall network differs according to the capacity
of the fuel cell installed in each building. In this
Figure 13. Characteristics of water electrolysis device
25
Figure 14. Heat energy network model
Figure 15. Outside aie temorerure
chapter, as shown in Figure 16, information on the

capacity of the fuel cell installed in each building
and the piping route is expressed with a gene, and
these are installed into a genetic algorithm. It is
evaluated as a solution with high fitness, so that
26
there are few values in Equation (14) showing heat

release from the hot water piping. The calculation
is iterated, chromosomes are evolved and a solution
with high fitness is sought. In the last generations
chromosomes, a solution with the highest fitness
is determined as an optimal solution. From the

information on the optimum chromosome, the
capacity of the fuel cell installed in each building
and the piping route are determined.
F=
Period M
H
t =1
l =1
hw ,l ,t
(14)
Solution Parameters
As parameters of the genetic algorithm employed
in the analysis in the following section, the population is 10,000, the generation number is 20,
and the crossover probability is 0.5. The gene
manipulation of mutation is not added. Search of
the hot water piping route in the R1 type is also
analyzed using the genetic algorithm.
Case Study
Energy Demand Pattern
and Network System
In this case study, an energy network composed
of seven buildings is investigated. The energy
need pattern in winter (January), mid-term
(May), and summer (August) of each building
is shown in Figure 17(Architectural Institute of
Japan, 2002; Yamano, 2002; Ozaki & Tuziki,
1990). These energy-demand patterns are assumed to be in Tokyo. Space-cooling power
in summer is included in the power demand
shown in Figure 17, and hot water supply and
space heating are included in the heat demand.
However, the heat for convenience stores, of-
fices, and factories is supplied from an electric

heat pump. Figure 18 shows the sum of the
power demand amount of these seven buildings.
The arrangement of the buildings is shown in
Figure 19. Moreover, the broken line in Figure
19 is the hot water piping route of the shortest
distance. In these analyses, in order to make
the hot water flow rate in the piping 1 m/s or
less, the inside diameter of the piping was set at
60 mm. The hot water piping is equipped with
a 40-mm-thick polystyrene-foam system heat
insulating mould. Moreover, the overall heat
transfer coefficient on the surface of the heat
insulating mould is set at 8.0 W/m2K. Under
the conditions described above, reduction in
fuel cell capacity using load leveling, and the
route of the hot water piping for minimum heat
release are investigated.
Reduction Effect of Fuel Cell Facility

Capacity
When threshold value Esep of a low-load region
and a high-load region is calculated according to
the procedure of previous section, a representative
day in January is 109 kW and a representative
day in May and August is 125 kW. By installing
Esep into the load leveling described in previous
section the fuel cell is made to follow the power
load pattern of Figure 18. Figure 20 shows the
fuel cell exhaust heat in this case. In the heat balance on a representative day in January shown in
Figure 20, since heat runs short in the 7:00 to
17:00 period, the heat pump is operated. On the
other hand, the heat supply and demand on rep-
27
Figure 17. Energy demand patterns
resentative days in May and August show much

heat surplus. Moreover, Figure 21 shows the
calculation result of the electrode surface of the
fuel cell at the time of installing Esep and performing load leveling. The fuel cell electrode surface
in Figure 21 expresses fuel cell capacity. The data
28
enclosed within the broken line in Figure 21 are

power generation using hydrogen and oxygen.
These gases are produced using the power generated with reformed gas and air. The fuel cell
generated with reformed gas and air is operated
at time other than the broken-line region in Figure
Figure 18. Demand patterns of total electric energy
Figure 19. Arrangements of buildings
21. At the time of high load from 12:00 to 16:00

on a representation day in August, about 180-m2
electrode surface was conventionally taken. If
load leveling using water electrolysis is employed,
the fuel cell can be reduced to a 120-m2 electrode
surface, which is equivalent to 2/3 at the peak at
20:00.
Route Planning Result of Hot Water

Piping
The result of the outlet hot water temperature of
each building that composes the network is described. The outlet hot water temperature differs
according to the R1 type or the R2 type. Moreover,
29
Figure 20. Demand patterns of total heat energy
Figure 21. Results of electrode area
since the heat release of the hot water piping differs according to the outside air temperature, the
sampling time is different. The result of 4:00 and
16:00 on representative days in January and August
is shown in Figure 22. As Figure 18 shows, the sum
of the power demand of each building connected
to the network at 4:00 on representative days in
January and August is small. On the other hand,
this value is large at 16:00. The horizontal axis
in Figure 22 is the route order (No. 1 to No. 7)
30
of the hot water piping. Letters A to G in Figure

22 corresponds to the building number shown in
Figure 19. For example, in the analysis result at
4:00 and 16:00 for the R1 type on a representative day in January, hot water flows in order of
GFDCABE. The optimal path on a representative
day in January for the R1 type is GFDCABE, and
the optimal path on a representative day in August
is BEGFDCA. In this way, the starting points of
the hot water differ according to each month.
Figure 22. Hot water temperature of outlet piping of .buildings in January and August
Moreover, the route of both months representative days for the R2 type is GEBACDF. All routes
GFDCABE, BEGFDCA, and GEBACDF are the
same as a result of the shortest route shown in
Figure 19. However, GFDCABE and BEGFDCA
are clockwise rotations and GEBACDF is counter
clockwise. The outlet hot water temperature of
each building differs in the starting point of the
hot water piping, route, and flow direction, as
shown in Figure 23. Figure 23 shows the result of
the hot water temperature when setting the start-
ing point of the hot water piping as B, E, or G.

Figure 24 shows the result of the hot water piping
heat release relevant to the piping routes in Figure
23. Figure 25 shows the result of the piping heat
release in the network on a representative day
every month. Under these analysis conditions,
the difference in the heat release for the R1 type
and R2 type on a representative day is less than
3% every month. Considering the analysis error
of the genetic algorithm, these can be estimated
as the same value. Therefore, if the heat release
Figure 23. Hot water temperature of outlet piping of .buildings
31
Figure 24. Quantity of heat loss of piping between buildings
Figure 25. Waste heat of hot water supply
of the R1 type and R2 type is optimized, it will

converge on almost the same value. However,
since the R1 type in this case assumes that the
starting point of the hot water piping is movable
to arbitrary buildings according to the month, it
is not realistic.
32
Result of a Fuel Cell Arrangement Plan

Figure 26 shows the result of the fuel cell arrangement plan for the R2 type. The fuel cell capacity
installed in each building is a circle of the broken
line in Figure 26. When the electrode surface of
Figure 26. Quantity of heat loss of piping between buildings
each building shown in Figure 26 is added, it is

97 m2. The electrode surface when installing the
installed capacity reduction by load leveling is 120
m2. Furthermore, if the optimum arrangement plan
of a fuel cell is installed, the electrode surface will
be reduced to 97 m2. When load leveling using water
electrolysis and optimization of fuel cell distribution
are installed, the fuel cell electrode surface is reduced by 46% compared to the conventional system.
1. If the load-leveling method is used, the installed capacity of by fuel cell will be reduced
by 34% compared with the conventional
system.
2. Moreover, when fuel cell distribution is optimized, in accordance with the effectiveness
of (1), there is a 46% reduction compared
with the conventional system.
Conclusions
REFERENCES
For the fuel cell energy network system, the

load-leveling method that supplies air and water
electrolysis oxygen to the cathode of the fuel cell
was proposed. Furthermore, the optimum operation plan of the hot water network was proposed,
and the fuel cell capacity of each building, position of the machinery room, piping route, and
hot water flow direction were investigated. The
fuel cell energy network composed of individual
houses, a hospital, a hotel, a convenience store,
an office building, and a factory was analyzed,
and the following Conclusions were obtained.
Architectural Institute of Japan. (2002). The nationwide research study concerning the energy
consumption in the house in the 2001 fiscal year.
3, 3-6. In Japanese.
Badami, M., &Caldera, C. (2002). Dynamic model
of a load-following fuel cell vehicle: Impact of
the air system. SAE Technical Paper (SAE-200201-100), 1-10.
Baker, J. E. (1985). Adaptive selection methods
for genetic algorithms. In Proc. 1st Int. Joint Conf.
on Genetic Algorithms, ICGA85, 101-111.
33
Energy and Industrial Technology Development

Organization in Japan. (1999). The researchesand-developments trend of the low carbon
fuel for vehicle. In Japanese. Retrieved from
http://www.nedo.go.jp/kankobutsu/foreigninfo/
html9912/12119.html
Fujiki K., Akagi S., Hirokawa T., & Yoshida K.
(1997). Optimal planning method of energy plant
configurations based on a genetic algorithm.
Trans. Jpn. Soc. Mech. Eng., Series C, 64(617),
354-361. In Japanese
Nagano, K., Mochida, T., Shimakura, K.,

Murashita, K., & Takeda, S. (2002). Development of thermal-photovoltaic hybrid exterior
wallboards incorporating PV cells in and their
winter performances. Solar Energy Materials
and Solar Cells, 77, 265282. doi:10.1016/
S0927-0248(02)00348-3
Narita, K. (1996). The research on unused energy
of the cold region city and utilization for the district heat and cooling (Ph.D. thesis). Hokkaido
University, Japan.
Goldberg, D. E. (1989). Genetic algorithms in

search, optimization and machine learning. Addison Wesley.
National Astronomical Observatory of Japan.

(2003). Chronological scientific tables. Japan:
Maruzen K.K.
HC-TECH Inc. (1997). HC12a and HC22a Properties and Performance Tests Data sheets.
Obara, S., & Kudo, K. (2003). Operational optimization and scheduling of multi-kind small
capacity energy devices for cold region houses.
In Proceedings of 9th ITES, 1, 297-302.
Hongmei, Y., Haipeng, F., Pingjing, Y., & Yi,

Y. (2000). A combined genetic algorithm/simulated annealing algorithm for large scale system
energy integration. Computers & Chemical Engineering, 24, 20232035. doi:10.1016/S00981354(00)00601-3
Inoue, U. (1996). Air-conditioning handbook, 14.
Japan: Maruzen. In Japanese
Ito, K., Shibata, T., & Yokoyama, R. (2002).
Optimal operation of a cogeneration plant in
combination with electric heat pumps. Trans.
ASME J. Energy Resource Technol., 116, 5664.
doi:10.1115/1.2906010
Kosaka, K., Tani, T., & Yoshida, S., (2000). Thermal analysis of solid polymer water electrolysis
system, Trans. Jpn. Soc. Mech. Eng., 66 (642), B,
547-554. In Japanese
34
Obara, S., & Kudo, K. (2005). Study on improvement in efficiency of partial load driving of installing fuel cell network with water electrolysis
operation. Transactions of the JSME, Series B,
71(701), 237-244. In Japanese
Obara, S., Kudo, K., & Kuroda, A. (2005). Study
on small-scale fuel cell cogeneration system with
methanol steam reforming considering partial load
and load fluctuation. Transactions of the ASME.
Journal of Energy Resources Technology, 127,
265271. doi:10.1115/1.1926310
Ozaki, S., & Tuziki, I. (1990).Trial calculation of
the quantity of public electric power and city gas
to be replaced by a distributed energy system. In
Proceedings of the 9th Energy-resources seminar,
Srinivas, M., & Patnaik, L. M. (1994). Genetic

algorithms: A survey. IEEE Computer, 27(6),
1726. doi:10.1109/2.294849
Takeda, Y., Iwasaki, Y., Imada, N., & Miyata, T.
(2004). Development of fuel processor for rapid
start-up. In K. Kimura (Ed.), Proc. 20th Energy
System Economic and Environment Conference,
Tokyo, (pp. 343-344). In Japanese
Yamano, Y. (2002). Development of a load-levelling technique. [In Japanese]. Denki, 629, 5661.
Yu, H., Fang, H., Yao, P., & Yuan, Y. (2000). A
combined genetic algorithm/simulated annealing
algorithm for large scale system energy integration.
Computers & Chemical Engineering, 24(8), 2023
2035. doi:10.1016/S0098-1354(00)00601-3
35
APPENDIX
Nomenclature

C : Cost [US Dollar]

Di : Energy device
E : Electric power [W]
Dp : Heat-insulating-mould outside diameter of hot water piping [m]
E : Power [kW]
E : Power consumption [kW]
Ec : Chemical equivalent (equivalent)
E EL : Amount of electric power storage [J]
EV : Voltage [V]
E : Consumption of electric power [W]
F : Quantity of fuel mass flow [g/s]
Fd : Faraday constant [C/g]
fm : Objective function
H : Thermal energy [kW]
H : Consumption of thermal energy [W]
h : Overall heat transfer coefficient in the surface of heat insulating mould W/ m2K
l : Distance m
M : Number of energy devices, the number of buildings with a fuel cell
N dv : Number of chromosome models
P : Probability
Q : Number of select switches
R : Operation period of system [s]
S E : Amount of electric power storage [J]
SSt : Amount of thermal energy storage [J]

T : Temperature [K]
t , tk : Sampling time [s]
t : Sampling time interval [s]
V : The number of auxiliary machinery with electricity consumption
Greek Symbols

36
: Calorific power of fuel [J/g]

: Efficiency
: Density of thermal storage medium [g/m3]
Subscripts

a : Cylinder
atm : Open air
bw : Blower
el , EL : Water electrolyzer
FS : Fuel cell stack
f : Fuel cell
fca : Fuel cell generated with air
fco : Fuel cell generated with oxygen
H : Electric heater
HP : Geothermal heat pump
hp : Heat pump
hw : Heat release of hot water piping
m : The number of each building
need : Energy needs
Rad : Radiator
r : Reformer
SL : Solar module
St : Thermal storage tank
sep : Threshold value of low loading and high load
st : Heat storage tank
sub : Auxiliary machinery
37
38
Chapter 2
Dynamic Characteristics of a
Fuel Cell System and Microgrid
ABSTRACT
This chapter consists of two sections, Load Response Characteristics of a Fuel Cell Microgrid with
Control of Number of Units and Dynamic Characteristics of a PEFC System for Individual Houses. The
dynamic characteristics and generation efficiency of a microgrid structured from 17 houses are examined
in the 1st section. The characteristics of the power quality of a fuel cell microgrid, and the generation
efficiency of the fuel cell are examined by numerical analysis. In the 2nd chapter, the method of determination of the control variables for a system controller, which controls the electric power output of a
solid-polymer-membrane fuel cell system (PEFC) during electric power load fluctuations, is considered.
The summary of the 1st section is as follows. A
gas engine generator with a power generation
capacity of 3 kW installed in a house is made to
correspond to the base load, and a proton-exchange
membrane fuel cell (PEFC) with a power generation capacity of 1 kW is installed in 16 houses.
Moreover, when changing the load of a microgrid,
the correspondence takes place by controlling the
number of fuel cells. Using numerical analysis,
the characteristics of the power quality of a fuel
cell microgrid, and the generation efficiency of the

fuel cell were examined. As a result, the relationship between the parameter of the controller and
power quality and a fall in generation efficiency
by a partial load were clarified.
The operation was clarified for the response characteristics of electric power generation for setting
the control variables of proportional action and
integral action considered to be the optimal for
the system controller. The power load pattern
of an individual house consists of loads usually
DOI: 10.4018/978-1-4666-5796-0.ch002
Dynamic Characteristics of a Fuel Cell System and Microgrid
moved up and down rapidly for a short time.

Until now, there have been no examples showing the characteristics of the power generation
efficiency of a system that follows a load pattern that moves up and down rapidly. Therefore,
this study investigates the relation of the control
variables and power generation efficiency when
adding change that simulates the load of a house
to PEFC cogeneration. As a result, it was shown
that an operation, minimally influenced by load
fluctuations, can be performed by changing the
control variables using the value of the electric
power load of a system.
LOAD RESPONSE
CHARACTERISTICS OF A FUEL
CELL MICROGRID WITH CONTROL
OF NUMBER OF UNITS
Introduction
The microgrid is expected to reduce the discharge
of carbon dioxide gas, to cut the peak of an electric
power plant, and to supply backup power in an
emergency (Robert, 2004; Carlos & Hernandez,
2005; Takuma & Goda, 2005). A microgrid
technique connects energy equipment, such as
an engine generator and a fuel cell, and power
is supplied by each cooperating piece of equipment. In forming a microgrid, the coordinated
grid system with commercial power etc., and the
independent grid system should be considered. In
the coordinated grid type, the supply and demand
of power with a commercial system are possible,
and the peak cut of an electric power plant and the
buying and selling of power are possible. If the
exhaust heat of the generating equipment linked
to a microgrid needs to be conveyed only a short
distance, it can be supplied to the consumer with
small radiation loss. On the other hand, in the
exhaust heat of the large-scale conventional power
plant, long distance transport had many heat losses,
and utilization of exhaust heat was limited. One
of the problems predicted by the construction

of a microgrid is that power quality deteriorates
when the power demand and supply balance of
the grid does not balance. The deterioration of the
power quality described in this Section means the
fluctuation of the voltage and frequency because
the dynamic characteristics of an electric power
supply do not meet demand. Power quality can
be maintained comparatively easily by controlling the voltage and the frequency of a grid by a
coordinated grid type to synchronize with another
network, such as commercial power (Ohmsha,
2002). On the other hand, in an independent grid
type, the reference of the power is determined to
be any power generator linked to a grid. In addition, other voltages and frequencies of generators
are controlled to synchronize with this reference
electrode. Therefore, if the power quality of the
power generator made into a reference is not
stabilized, the power quality of the whole grid
may deteriorate. In order to stabilize the power
quality of a microgrid, there is the method of connecting a battery to a microgrid. However, when
facility cost and the maintenance cost of a facility
are taken into consideration, the introduction of
a common battery is difficult. If the number is
controlled and the distributed power generator
is made to correspond so that a rapid change of
electricity demand may be followed, the response
characteristics of each power generator will influence the power quality of a grid. In this Section,
the dynamic characteristics of a highly independent microgrid that connects a gas-engine power
generator (EG) and a proton-exchange membrane
fuel cell (PEFC) as a distributed energy system
are investigated for the possibility of realization.
Generally, in order to maintain high generation
efficiency by EG, it is necessary to extract the
area of an operation point (load and number of
rotations) as much as possible. Moreover, if large
load fluctuation is added to EG, tens of seconds
are required to stabilize the power from several
seconds (Katrasnik et al., 2005). Therefore, EG is
used for base load operation in this Section, and
39
the fuel cell distributed in each house examines

the microgrid that controls the number according
to the amount demanded.
EG and PEFC connected to a grid cooperate
according to the power demand, and supply the
power. In this case, although the exhaust heat
of EG and PEFC is supplied to each house, the
characteristics of heat supply are not dealt with
in this Section. One method of connecting a fuel
cell with the grid is to install one set of largecapacity PEFC (centralization system), and
another method is placing PEFC in each hours
(distributed system). Because the centralization
system should just install one set of PEFC, equipment cost and installation cost are low compared
with the distributed system. However, if PEFC
is operated in a time zone with little electricity
demand, low-efficiency partial-load operation
will produce a centralization system. On the other
hand, when distributing PEFC and controlling
the number, the fuel cell operated with a partial
load is one set of distributed fuel cells. Therefore,
the amount of falls in generation efficiency is
predicted to be smaller in the distributed system
than in the centralization system. Thus, an engine
generator with a generation capacity of 3 kW is
installed in a house, and it is made to correspond
to the operation of the base load as an analysis
example of this Section. In addition, the dynamic
characteristics of a microgrid and the generation
efficiency of fuel cells when a fuel cell system
with a generation capacity of 1 kW is distributed
and installed in 16 houses, and when one PEFC
with a generation capacity of 16 kW installed
(centralization system) are examined.
Microgrid Model
Power Quality of
The energy network and microgrid that are installed in an urban area are shown in Figure 1.
Figure 1 (a) shows the microgrid of the system
linked with a commercial power network. On
40
the other hand, Figure 1 (b) shows the microgrid

of the independent system that does not connect
with other grids. An energy network consists of
a hot-water piping network used for the waste
heat recovery of energy equipment, and the heat
supply to each house, a city gas network that supplies city gas to the reformer for fuel cells, and a
gas-engine power generator, and a microgrid that
supplies electric power to each house. The power
transported by the independent microgrid shown
in Figure 1 (b) needs to control the voltage and
frequency of each power generator on the basis
of the power of any one power generator. On the
other hand, it is controllable by the microgrid
connected with the other power grids shown in
Figure 1 (a) to synchronize with the network voltage and the frequency that were connected. Even
if rapid load fluctuation is added to a grid and a
difference occurs in the production of electricity
and the amount demanded, the power quality of
the coordinated grid type is stabilized by synchronizing with the connected network. With the
independent grid type, if rapid load fluctuation
is added to a grid and a difference occurs in the
production of electricity and the amount demanded
of a power generator, it is considered that a long
time is required to stabilize the power quality. If
load fluctuation is not appropriately predicted,
in an independent grid type, the power quality of
the grid is expected to be low over a long time.
System Scheme
Figure 2 (a) is a system configuration figure of
an independent microgrid that sets one set of EG
as base load operation, and connects 16 sets of
fuel cell systems to a grid. In the independent grid
type, the power quality of the power generator
operated as a base load is a standard of the power
quality of the whole grid, and is very significant.
EG connects the synchronous power generator
to a gas engine, and obtains exhaust heat from a
water jacket and an exhaust gas heat exchanger. If
rapid load fluctuation is added to EG, the power
Figure 1. Fuel cell microgrid system
quality of a grid is affected, but in this system,

EG is operated at the base load of fixed load. On
the other hand, adjustment of the power supply
when changing the load of a grid corresponds to
the number control of the fuel cells installed in
each house. The number of fuel cells installed in
House Q from House B shown in Figure 2 (a) is
expressed by F/C (1) to F/C (16).
Figure 2 (b) shows the parts of House A in
which an engine generator is installed, and Figure
2 (a) shows House B and House C in which a fuel
cell of the system is installed. The block expressing the response characteristics of an EG, a city
gas reformer, a fuel cell, an inverter, and an interconnection device with the primary-delay-
system and the secondary-delay-system is shown

in Figure 2 (b). u in this figure expresses the
power load and v2 to v4 expresses the output in
the block (from Action 1 to Action 3) that branches at the value of u. Moreover, h0 to h3 expresses
the generation capacity of the engine generator
and the fuel cell installed in House B and House
C. In this system, when the value of u exceeds
capacity h0 of the engine generator corresponding
to operation of the base load, the fuel cell (capacity h1) of House B is operated first. When the
production of electricity (h0 + h1) is less than the
value of u, the fuel cell installed in House C is
operated. Thus, the number of fuel cells is controlled by the volume of the load added to the
41
Figure 2. System block diagram
42
grid. The power of the generating equipment

installed in House C from House A can be supplied to any house through a microgrid. As shown
in Figure 2 (b), in analysis, the load pattern of
power is given to a system and Action 1-Action
3 are selected by an IF conditional branch according to the volume of load (u), and the capacity
(h0, h1, h2 ) of the generating equipment installed
in each house. In Action 1-Action 3, as shown in
Figure 2 (c), the production of electricity of each
power generator is calculated and outputted. As
shown in Figure 2 (b), the dynamic characteristics
of each power generator are expressed with a PI
controller (Controller 0-Controller 2), output
limitation equipment (Limiter 0-Limiter 2), and
a transfer function.
Capacity of Engine Generator

and Fuel Cell
In the system in Figure 2, an engine generator
with a generation capacity of 3 kW is installed in
House A, and base load operation is performed.
Furthermore, a fuel cell system with a generation
capacity of 1 kW is installed in 16 buildings of
House B to House Q, and electric power and heat
are supplied to each house. The number of installed
fuel cells in a distributed system is determined from
the electricity demand pattern shown in Figure 3
used in analysis. The electricity demand pattern of
Figure 3 is a measurement result of an individual
house for representative days in February in Sapporo, Japan (Narita K., 1996). On the horizontal
axis of this figure, the sampling time of analysis
and the time assumed (Assumed time) are written
together. In this electricity demand pattern, the
minimum load is determined as a value of the
base load (in the example in Figure 3, it could be
3 kW). In the case of a centralization system, there
is one set of installed fuel cells, and, in the case of
a distributed system, a fuel cell of 1 kW maximum
output is installed in each house. Moreover, let the
fuel cell capacity of the centralization system be a
value where the maximum load added to a grid is
satisfied. In the distributed system, the fuel cells of

F/C (1) to F/C (16) all have the same capacity and
dynamic characteristics. The electric transmission
loss of a microgrid is considered in this analysis.
Response Characteristics of System

Configuration Equipment
Generation Characteristics
of Engine Generator
The model of the generation characteristics of an
engine generator installed in the independent microgrid assumed in this Section is shown in Figure
3 (b) (Katrasnik et al., 2005). This model represents
the response characteristics when adjusting the
engine supply fuel to control the production of
electricity, and outputting power to a grid through
an interconnection device. The settling time when
converging in the range of 5% of the targeted
output value of power by adjusting the control
parameter of PI (proportionality-integration) controller can be set at about 15 seconds. The value
of the PI control parameter and transfer function
is shown in Figure 3(b).
Generation Characteristics of Fuel Cell

Figure 3 (c) shows the test results of the response
characteristics when inputting a step load of 70%
of a load factor into PEFC trial production as an
experiment (Obara & Kudo, 2005). The response
characteristics shown in Figure 3(d) are obtained
from the results of Figure 3 (c), and requires the
transfer function of a primary-delay-system so
that these characteristics may be approximated.
The equations for the transfer function are shown
in Figure 3(d). To be exact, although the transfer
function is considered depending on the load
factor, it is not taken into consideration because
this difference is small as a result of examination.
The settling time of a fuel cell when generating 1
kW maximum output is about 3 seconds (Obara
& Kudo, 2005).
43
Figure 3. Characteristics of transient response
44
Output Characteristics of
City Gas Reformer
Figure 3 (e) shows the model of the step input
given to the city gas reformer of a load factor
when it is between 100 and 80%, and between 100
and 50% (Obara & Kudo, 2005; Obara & Kudo,
2006). Figure 3 (f) shows the response result of
the load input to Figure 3(e), and the transfer
function of a primary-delay-system shows this
response characteristic in this figure. In addition to
fuel cells, although it is considered that a transfer
function influences the load factor directly, since
there is no big difference, the result of Figure 3
(g) is used in the range managed in this Section.
Inverter and Interconnection Device

An inverter uses a cheap voltage control form,
and converts and outputs input power to regular
voltage and frequency. An inverter requires 120
ms to stabilize voltage and frequency within
95% of a regular value (Obara & Kudo, 2006).
Figure 3 (g) expresses the transfer function of this
inverter with the primary-delay-system. When
switching the power of the single phase 100 V
by an interconnection device, the duration of the
change is about 10 microseconds. However, since
it is necessary to synchronize the frequency by
control, the interconnection device assumed in this
Section sets the change time to 12 microseconds.
As a result, the transfer function of the interconnection device by the primary-delay- system is
the value of Figure 3 (g).
Generation Efficiency of
Fuel Cell System
Figures 4(a)-(c) shows the output characteristics
of a fuel cell and a city gas reformer obtained in
the experiments. Figure 4 (a) shows the relation
between the load factor of a fuel cell and the
generation efficiency assumed in this Section.
Moreover, Figure (b) shows the characteristics of

the amount of hydrogen supplied to a fuel cell,
and generation and exhaust heat output. Figure
4 (c) shows the relation between the load factor
of a city gas reformer and reformer efficiency.
Here, the value that divides the calorific power
of the hydrogen contained in the reforming gas
by the calorific power of the city gas supplied to
a reformer is defined as reformer efficiency. The
city gas supplied to a reformer has the object of
producing of reforming gas, and the object of being
the heat source of a reformer. In analysis, a load
factor is calculated from the capacity of a fuel cell,
and the quantity of the load, and the generation
efficiency is calculated using the relation shown
in Figure 4(a). Moreover, the amount of exhaust
heat of a fuel cell is obtained by giving a power
load to Figure 4(b). Because the load factor of a
reformer is calculable from the load and capacity,
reformer efficiency is determined if this value is
given to Figure 4(c).
Control Variables and Analysis

Method
The response characteristics that inputted step
load into the independent microgrid shown in
Figure 2 (a) for 0.2-, 0.6-, and 1.0-kW loads are
shown in Figures 4(a)-(c) (Obara & Kudo, 2005,
2006). The response characteristics of a fuel cell
changes with the control parameter set up with a
controller, and changes and analyzes the parameters of PI control in Figures 4(d)-(f). As shown
in Figure 4(f), the result of the 1.0-kW loads does
not depend on the rise time and the settling time of
the control parameter. In the result of the 0.2-kW
loads, although the rise time of P = 12.0, I = 1.0 is
short; as for the settling time, P = 1.0 and I = 1.0
are short. When P = 12.0 and I = 1.0 is compared
with P = 1.0 and I = 1.0, as for the rise time, P =
12.0 and I = 1.0 is shorter, and the settling time
is almost the same, and the overshooting is large.
Moreover, in the result of P = 5.0 and I = 1.0, the
45
Figure 4. Output characterisics of a fuel cell system with city-gas reformer, and characteristics of output
of the system
steady state error of low load is large, and is not

suitable as a control variable. Therefore, in the
analysis example of the following Section, the
control parameter of a fuel cell is analyzed as P
= 1.0, I = 1.0 and P = 12.0, I = 1.0.
46
The dynamic characteristics of a microgrid are

analyzed using MATLAB (Ver.7.0) and Simulink
(Ver.6.0) of Math Works Cooperation. However,
in the analysis example of the following Section,
the solver to be used is made into the Runge-Kutta
method, it is calculated, and the sampling time

of the analysis is determined so that error is less
than 0.01%.
Load Response Characteristics

of Microgrid
Step Response Characteristics
The response characteristics when inputting the
step load of Figure 5(a) into the microgrid shown in
Figure 2 (a) is investigated. Figure 5 (b) to Figure
5 (f) shows the analysis results of a response of
the microgrid. However, the parameter of the PI
controller of all the fuel cell systems was set to P
= 1.0 and I = 1.0. Figure 5(b) shows the result of
a response of a microgrid, Figure 5(c) shows the
result of a response of an engine generator, Figure
5(d) to Figure 5(f) show the results of a response
of fuel cell No. 1, No. 8 and No. 12, respectively.

When the response results of an engine generator
(Figure 5(c)) is compared with the response results of a fuel cell (Figure 5(d) - (f)), there is little
output vibration and the settling time is short in a
fuel cell. If the load added to a grid increases, the
number of fuel cells to be operated is increased
corresponding to the load. According to the load
of a grid, a larger number of fuel cells is operated
such as No. 1, No. 8, and No. 12 (Figures 5(d),
(e), (f)). How the dynamic characteristics of a
microgrid would change according to the difference in the control parameter of the PI controller
of a fuel cell system was investigated.
The results of investigating the response characteristics of the grid for the control parameter
with P = 12.0 and I = 1.0 are shown in Figure 6.
Figure 6 (a) shows P = 12.0 and I = 1.0, and is a
response result when inputting the step load of
Figure 5. Step response characteristics of the system. The control parameter of the fuell cell system is
P=1.0, and I=1.0.
47
Figure 6. Analysis results of step input for microgrid system
48
Figure 5 (a) into a microgrid. When Figure 5(b)

is compared with Figure 6(a), it shows that the
overshooting in Figure 6(a) is larger. The analysis
results of the power load and settling time when
the control parameters of a fuel cell system are P
= 1.0, I = 1.0 and P = 12.0, I = 1.0 is shown is
Figure 6(b). The period until convergence of less
than 5% of the load by the power supplied to a
grid is defined as the settling time. From the result
of Figure 6(b), a load does not depend on the
settling time of a control parameter of 15 kW or
less. However, when a load exceeds 16 kW, control parameter P = 12.0, I = 1.0 of a fuel cell
system is about 5 seconds short compared with
P = 1.0, I = 1.0. Figure 6(c) shows the analysis
result concerning the difference in the power load
of a microgrid, and the power supplied to a grid
from EG and PEFC. The difference in the power
load and electric power supply is due to a time
delay in the response of EG and the fuel cell. The
result of Figure 6(c) shows that the difference in
the control parameter of a fuel cell does not affect
the difference in the power load or electric
power supply. However, the control parameter of

a fuel cell system strongly affects the settling time
and overshooting.
Application of Electric Power Demand

Pattern of House
The electric power demand pattern of an individual
house in Sapporo is inputted into the microgrid
shown in Figure 2(a), and response characteristics
are investigated. The response results of having
inputted the electric-power-demand pattern of
Figure 3 into the microgrid are shown in Figures
7 and 8. In the analysis of Figures 7 and 8, the
control parameter of the PI control device of a
fuel cell system was set up with P = 1.0, I = 1.0
and P =12.0, I = 1.0, respectively. As for overshooting, in these analysis results, P = 12.0, I =
1.0 are larger than P = 1.0, I = 1.0. Moreover,
immediately after startup (past 0:00), the reason
for many fluctuation elements is based on the
starting characteristics of an engine generator, as
shown in Figure 7 (b) and Figure 8 (b).
Figure 7. Response results analyzed using an electricity demand pattern of a house in Febryary of Sapporo. Control parameter of the fuel cell system is P=1.0 and I=1.0.
49
Figure 8. Response results analyzed using an electricity demand pattern of a house in Febryary of Sapporo. Control parameter of the fuel cell system is P=12.0 and I=1.0.
The response results of a fuel cell No. 1, No.

3, No. 5, and No. 9 are shown in (f) from (c) of
Figures 7 and 8. The analysis results of Figures
7 and 8 have large overshooting when load fluctuation is added to a fuel cell and an engine
generator. As for the analysis results of the sufficiency ratio of power supply to the power demand
amount of a microgrid, P = 1.0, I = 1.0 is shown
in Figure 9 (a), and P = 12.0, I = 1.0 is shown in
Figure 9 (b). The time when the difference in the
sufficiency ratio of power supply and demand is
large, it will be the assumed time in Figure 9 (a)
and Figure 9 (b) at 05:00 and 17:00. As Figure 3
shows, at such times, load fluctuation is large
compared with other times. A supply-and-demand
difference occurs at two or more peaks exceeding
100% in the analysis result of Figure 9 (b) compared with Figure 9 (a). The analysis result of the
sufficiency ratio of the power supply and demand
in all of the representative days is 99.63% in
Figure 9 (b) and 99.74% in Figure 9 (a). Even if
it searches by trial and error for the control parameter of a controller, the characteristics of
50
Figure 8-1 do not change. In order to change the

characteristics of Figure 9, it is necessary to improve the time constant of each piece of equipment.
Generation Efficiency of Fuel Cell

By the fuel cell microgrid of a centralization system and a distributed system, the analysis results
of generation efficiency when supplying power
to 17 houses is shown in Figure 10. The average
value of the generation efficiency of each fuel cell
under operation defines the generation efficiency
of the fuel cell in a distributed system. The generation efficiency of a centralization system and a
distributed system has the following differences.
1. Although change of the generation efficiency
of a centralization system is smooth, the
efficiency of a distributed system has large
fluctuation.
2. And, because efficiency fluctuation of a
distributed system is large, there are large
over shoots.
Figure 9. Difference in amount power demanded and amount of power supply
3. The change of the generation efficiency

through a representation day of a distributed
system is larger than a centralization system.
When the electric power demand pattern of the
representative day shown in Figure 3 is introduced
into a microgrid, the generation efficiency of a
fuel cell of a centralization system is 25.5% and of
a distributed system, it is 28.2%. The distributed
system has about 3% higher efficiency than the

centralization system. This is because the fall area
of the generation efficiency by the partial-load
operation of PEFC is small in the distributed
system with number control compared with the
centralization system. Therefore, the distributed
system with number control of the generation
efficiency of the microgrid using PEFC is more
advantageous than the centralization system.
51
Figure 10. Fuel cell efficiency results analyzed using an electricity demand pattern of 17 houses in
February in Sapporo. The control parameter of the fuel cell system is P=12.0 and I=1.0.
Conclusions
An engine generator with a power generation
capacity of 3 kW was installed in a house, corresponding to the base load, and the dynamic
characteristics of the microgrid at the time of installing a fuel cell system with a power generation
capacity of 1 kW in 16 houses was investigated
by numerical analysis.
52
As a result, the following Conclusions were

obtained.
1. The settling time (period to converge at
5% of range of the output target) and overshooting of a microgrid can be changed by
parameter setting of the controller of a fuel
cell. The settling time of this system was 10
to 15 seconds.
2. The cause of the supply-and-demand difference in the power of a microgrid is a response

delay of the generating equipment, and the
control parameter of the controller is not
related. It is necessary to improve the time
constant of each generator.
3. The fall in the generation efficiency of PEFC
by partial-load operation can be reduced
by a distributed system with the control
of a number of units compared with the
centralization system. This is because the
fall area of the generation efficiency by the
partial-load operation of PEFC is small in
a distributed system with number control
compared with a centralization system.
DYNAMIC CHARACTERISTICS
OF A PEFCL SYSTEM FOR
INDIVIDUAL HOUSES
Introduction
If small generating equipment is introduced into
individual houses and exhaust heat is recovered,
the utilization efficiency of the energy will increase
greatly compared with existing large-scale power
generation systems. In particular, the fuel cell system is attracting attention because the exhaust gas
is clean and the power generation efficiency is high
(Obara & Kudo, 2005). Compared with the load
fluctuation characteristics by the energy demand
side, the dynamic characteristics of a fuel cell system need to be rapid. If the response characteristics
of a fuel cell system are slow, the demand of the
demand person cannot be instantly satisfied. As
a result, the demand-and-supply balance becomes
unstable, and power supply becomes impossible at
worst. In order to introduce and apply small-scale
fuel cell cogeneration to a building, it is necessary
to investigate the response characteristics of the
electric power output and the heat power output

with load fluctuations. Thus far, the transient
response characteristics of the electric power and
the heat power of proton-exchange membrane
fuel cell (PEFC) cogeneration, with a town gas
reformer and an auxiliary heat source has been
investigated (Obara & Kudo, 2005). The speed of
response of power generation of a cell stack of
a PEFC system is very quick, and if the control
variables of a controller are set up suitably, it is
thought, from results of an investigation, that the
use of consumer electronics will be satisfactory
(Hamelin et al., 2001; Hatziadoniu, 2002; Kourosh,
2004; Silva et al., 2003). On the other hand, the
speed of response of a reformer is slow, compared
to a PEFC stack. The response characteristics of
the overall system at the time of providing a rapid
electric power load to the integrated system by
a reformer, a fuel cell stack, an inverter, and an
interconnection device, are not known.
Therefore, this chapter considers the determination method for the set-up of control variables
using a controller, which controls the electric
power output of the system. The response characteristics of the electric power, for setting the
optimal control variables of proportional action
and integral action for the system controller for
operating the system, are investigated (Silva et
al., 2003).
The response characteristics of systems such
as time delay and overshooting change by the
difference in control variables. Moreover, if it is
controlled to follow a system at loads that fluctuate
rapidly for a short time, a difference in the load
and the response will occur. The magnitude of
this difference depends on the fuel consumption
of a system, with the response characteristics
depending on the control variables. Therefore,
it is expected that the response characteristics of
a system will change if the control variables are
changed, and the amount of consumption of town
53
gas changes as a result. If the control variables of

the controller are changed, the transient response
characteristics, such as settling time (defined in
this chapter as the time for a value to become less
than 5% of a set value), overshoot, rise time, and
steady state error, will change (Kurata, 1999).
However, if the control variables are changed,
it is expected that the town gas consumption in
a system also changes. Therefore, if the control
variables are not set to the most optimal values,
the power generation efficiency of the system
will be low. Moreover, the power load pattern of
an individual house is a collection of loads that
usually goes up and down rapidly for short periods
of time (Kurata, 1999). Thus far, there have been
no studies on changes in electric power generation
efficiency when operating a system, wherein the
load pattern fluctuates rapidly. Therefore, in this
chapter, an electric power load pattern, fluctuating at random for a short time, is introduced into
the PEFC cogeneration, and the operation of the
system and changes in the power generation efficiency are clarified. The optimal values of the
control variables set up with a controller are also
investigated.
System Description
Outline of System
The PEFC cogeneration system outline examined
in this chapter is shown in Figure 11. The town
gas supplied to the system includes a quantity that
is changed into reforming gas by a reformer (
Qrm ,t ), and a quantity consumed by a burner as
an object for the heat source of a reforming reaction (Qcb,t ). The relationship between the supply,
the sum of the two town gas quantities (
Qrm ,t + Qcb,t ), and the electric energy output of
the system ( Esys ,t ) is shown in Figure 12 (Ibaraki
Prefecture Government Office of Education, 2002;
Mikkola, 2001). The calorific power of hydrogen
in the reforming gas generated by the reformer
divided by the supply quantity of heat of the town
gas is defined as the reformer efficiency rm ,t at
sampling time t . rm ,t is set at 73% in this system
(Obara & Kudo, 2005). The electrode areas of the
anode and the cathode in the fuel cell stack are
set to 0.5 m2, respectively, and the output characteristics of a single cell in the fuel cell assumed
Figure 11. Fuel cell cogeneration system for individual houses
54
Figure 12. The power output of the system outlet. The anode and cathode areas of the fuel cell stack are
0.5 m2 respectively. The reformer efficiency is 75%.
in this chapter is shown in Figure 13 (Ibaraki

Prefecture Government Office of Education, 2002;
Mikkola, 2001). The maximum power generation
efficiency of the fuel cell stack is 48%.
A. Efficiency of a Carbon Monoxide Oxidization System and an Inverter
A significant quantity of water is present in the
reforming gas produced by the reformer. Therefore,
the reforming gas is cooled by an air supply from
a blower with a dryer system and condensation
separation of the water is carried out. In order for
the carbon monoxide concentration of the reforming gas in the fuel cell stack inlet to be 10 ppm
or less, a carbon monoxide oxidization system is
prepared. Although carbon monoxide is burned
using a catalyst and converted into carbon dioxide
in the carbon monoxide oxidization system, it
is assumed in this chapter that nearly 2% of hydrogen in the reforming gas will also be burned.
Therefore, the efficiency of carbon monoxide
oxidization system in this chapter is 98%. Reforming gas is supplied to the fuel cell stack from the
carbon monoxide oxidization system, generating
Figure 13. Singl cell performance generated with reformed gas and air. Operating temoerature 333 K,
and reactant flow stoichiometries 2.0 both hydrogen and oxygen.
55
electricity, and the generated DC electric power

is changed into AC power of a fixed frequency
using an inverter, which is then supplied to an
interconnection device. The efficiency of the
inverter was set at 95% in this chapter.
B. The HeatSsupply Method
The exhaust heat of the reformer and the fuel cell
stack is supplied to a heat storage tank (ST), and
the heat of tap water is exchanged for the heat storage medium with a heat exchanger installed in the
heat storage tank, which then supplies hot-water
to the demand side. The thermal storage medium
is contained in the heat storage tank. When the
exhaust heat exceeds the heat demand, the heat
transfer medium is heated and the heat is stored.
The response time of the exhaust heat output in the
reformer and the fuel cell stack is long compared
with the response time of the heat output from
thermal storage (Morner & Klein, 2001). Therefore, in order to obtain a quick response, the heat
in the thermal storage is given priority and used
first. When the amount of exhaust heat exceeds
the heat demand, it is necessary to release the heat.
In this case, valves VA2 and VA3 are operated and
the excess heat is released from the system.
C. The Controller and Auxiliary Machinery
This system has two controllers, as shown in Figure 11. Controller 1 is for the power generation
system, and Controller 2 is for the heat supply
system. In each controller, the control variable
for proportional action (P action), integral action (I action) and derivative action (D action)
can be set up. In Controller 1, the temperature of
the fuel cell stack and the amount of electricity
demand are the input data, and in Controller 2,
the heat-transfer-medium temperature of the heat
storage tank and the heat amount demand are the
input data. The electric power generated by the
56
fuel cell system and commercial electric power

can be supplied to the demand side through an
interconnection device. The auxiliary machinery
of the system are operated with commercial electric power. The auxiliary machinery includes an
exhaust heat-circulating pump (PP) for the fuel cell
stack, a blower for the reformer burners (BW1),
a blower for the dryers (BW2) and a blower for
the fuel cell cathode (BW3). By using commands
from Controller 1 and Controller 2, the switches
SW1, SW2, SW3, and SW4 are operated, and the
starting and stopping of the auxiliary machinery
are performed.
System-Control Block Diagram

Figure 14 shows the control block diagram of the
system shown in Figure 11. The time constants
for each piece of equipment shown in this figure
are determined by test evaluation data and reference data, the details of which are given in reference (Obara & Kudo, 2005). The input values
used in the analysis are shown in Table 1. However, the power load and heat load are given in a
step input. Investigation of the transient response
characteristics can be performed by providing an
electric power load and a heat power load to the
system shown in Figure14. For the electric
power load, random changes can be added for
every sampling time. For each controller, limiters
are included so that extremely large overshoots
are not generated for control by Controller 1 and
Controller 2. Subsystems 1 through 8 represent
the response characteristics of the exhaust heat
output in a fuel cell stack, and details are shown
in Figure 15. The parameters a and b in this
figure express the time constant and the constant
part of a primary delay system transfer function,
respectively. The exhaust heat of the fuel cell stack
serves as a nonlinear output to the electric power
Figure 14. System control block diagram
load (Obara & Kudo, 2005). In this study, the

relationship between electric power load and the
exhaust heat output of a fuel cell is divided into
eight zones. The relationship between the electric
power load in each zone and the exhaust heat
output are approximated by the linear formulas
expressed with Subsystem 1 to Subsystem 8.
The Analysis Method

The transient response characteristics of the
electric power output in this system are analyzed
using MATLAB (Ver.7.0) / Simulink (Ver.6.0) of
Math Work. The Runge-Kutta method was used
as the numerical method (Dormand, et al.,1980,
57
Table 1. Input data used in analysis
Figure 15. Block diagram of the subsystem from 1 to 8. The parameters and represent the time constant
part of a primary delay system transfer function, representively.
Forsythe, et al.,1977, Kahaner, et al.,1989). The

sampling time of the analysis was calculated
automatically and the analysis error is less than
several percent.
System Control
The Input of the System
The operational model of the system shown in
Figure 21 is shown in Figure 16. Figures 16 (a)
and (b) are the models which show the electric
58
power load input and the heat power load input,

respectively, and both power loads are zero between the sampling times t1 and t0 . At sampling
timet1 , although the step input of the electric
power load and the heat power load is carried out
in the system, the heat power load is fixed until
t3 , and a still higher step input is given at t2 for
the electric power load. The electric power load
in individual houses fluctuates up and down
within a short period. Therefore, in the analysis
in the following section, random load fluctuations
Figure 16. System operation model
are given for every sampling time in 5% and

10% intervals of the range of the electric power
load input into the system. For the analyses shown
in Figure 17 through Figure 21 described in the
following section, the time from t0 to t1 is 5
seconds, the time from t1 to t3 is 20 seconds, and
the time from t3 to t4 is 5 seconds.
Control of Startup
Figure 16 (c) shows the model of the switches
and the valves of the system operated by the
controller of the operation. The symbols in this
figure correspond to the symbols shown in Figure
11. During the startup of the system, air is supplied to the burner, which is the heat source for
the reformer and the dryer system. For this,
switches SW1 and SW2 are turned ON by a command from Controller 1, and blowers BW1 and
BW2 are operated. At sampling time t1 , commands
are given to SW3 by Controller 1, and BW3, which
supplies air to the cathode of the fuel cell stack,
is operated. Simultaneously, a control command
is given by Controller 1 to VA1 so that valve VA1
can be opened. As shown in Figure 16 (d), operation of the fuel cell system follows electric
power load. It depends, for the response characteristics of the electric power, on transient response
characteristics, such as that of the reformer, the
fuel cell stack, and the inverter. Therefore, compared with the characteristics of the input shown
in Figure 16 (a), time delay, overshoot and steady
state error cause the response characteristics shown
in Figure 16 (d).
Control of Heat Output

Since the speed of response of the heat exhaust
output from the fuel cell stack and the reformer is
slow, heat output takes time. If the temperature T2
of the fuel cell stack rises, SW4 will be turned ON
by an order from Controller 1, and the heat exhaust
circulating pump PP will be operated. Figure 16
59
Figure 17. Characteristics of electric power output of the system
(e) shows the model for this operation. Furthermore, when the fuel cell stack and the reformer
of heat exhausts exceed the heat amount demand,
as shown in Figure 16 (f), thermal storage of the
excessive heat is carried out using a heat storage
60
tank. Finally, when the heat exhaust exceeds the

thermal storage capacity of the heat storage tank,
valves VA2 and VA3 are operated by Controller
2, and as shown in Figure 16 (g), surplus heat is
released from the system.
Figure 18. Electric power generation error for

the system
Figure 19. Electriv power generation efficiency

of the system
61
Figure 20. The settling time for electric power

generation for the system
Town Gas Consumption

Figure 16 (h) shows the model for town gas
consumption in the system. Since town gas consumption is dependent on the amount of electric
power generated by the fuel cell, the transient
characteristics of the town gas consumption are
the same as shown in Figure 16 (d). With control
variables set up by Controller 1, the output characteristics of Figure 16 (d) differ greatly. Therefore,
the town gas consumption in the system also
changes with the value of the control variables
set up by Controller 1.

Control Variables and the Response
The analysis results of the response characteristics
of control variables and electric power set up by
Controller 1 are shown in Figure 17. Figures 17
(a), (b), and (c) show the analysis results of the
response characteristics due to the difference
in the control variable of P action and I action
under each electric power load. When P action
is independently used to control this system, the
steady-state error is large. Moreover, for the use
of the D action, such as D control, PD control,
and PID control, the response characteristics of
the system become extremely unstable. Therefore,
in this analysis, independent control by P action
and control by the D action are not used. The
results in Figures 18 (a), (b), and (c) show that
the optimal values of the control variable used by
Controller 1 differ in the magnitude of the electric
power load. Therefore, if the control variables
of the system controller are changed with the
magnitude of electric power load, it is predicted
that the transient response characteristics of the
system will become favorable.
62
Figure 21. The optimal areas for PI control for different electric power losds
The Response Characteristics of the

System
1. The Amount of Electric Power Output
If a step load is provided to the fuel cell system,
until the operation of the system reaches a settling
time, the electric power outputs will go up and
down along with the load. The amount of integration of the electric power load for a definite period
of time (this is written below as the amount of
theoretical electric power loading) differs from
the electric energy output by the system, because
of the upper and lower sides of the electric power
output. Figure 18 shows the calculation results of
the difference in the amount of theoretical electric
power loading and the amount of system electric
power outputs (err), when adding electric power
load with fluctuations shown in this figure to the
system for 20 seconds. The results of 5% and
10% of the load fluctuation rate in Figures 18
(a), (b), and (c) are not shown for the zone where
P operation is large. The reason for this is that the
calculation results diverge in the zone where the
P operation is large. This tendency is strong as

there are fewer amounts of electric power output
from the system than the amount of theoretical
electric power loading, and the rate of load fluctuation becomes large by all the results shown in
Figure 18. As Figure 17 showed, the reason for
this is that each sampling time until the response
of the system achieves a settling time. Also, the
electric power outputs are less than the value of
electric power load in many cases. Moreover,
the control variable set up with a controller has a
large influence on the rising time of a response.
For this reason, the results shown in Figure 18 of
the amount of electric power output in the system
is also smaller than amount of theoretical electric power loading as the rate of load fluctuation
becomes large.
2. Electric Power Generation Efficiency
The electric power generation efficiency of the
system can be calculated from the electric energy
generated by the system by dividing it by town
gas consumption. However, the response charac-
63
teristics of the electric power output from the

system do not satisfy the amount of theoretical
electric power loading, as Figure 19 showed.
Therefore, the following equation defines the
electric power generation efficiency e of the
system in this chapter.
e = (Et / Vg ) + err
(1)
Although the cold start of a reformer takes

about 1 hour (Zhang, et al., 2004), in this analysis,
a reformer changes into the condition of already
having completed the warm-up. Therefore, the
quantity of gas used for the setup operation of the
reformer is not included in the town gas consumption, Vg . The control variables of P action and I
action, and the relationship of e are calculated
for the rate of load fluctuations of the electric
power load for different electric power loads, and
these results are shown in Figure 20. The maximum
electric power generation efficiency of this system
is 33%. For each load, except for the 1.0 kW
electric power load, the system can be operated
with a high value of e by making the control
variable of I action about 1 to 2, and by making
the control variable of P action about 5 or less.
For 1.0 kW of electric power load, the operation
of the system at the high value of e of the control
variable zones of P action and I action are narrow,
ranging from 1 to 2 for the control variable of I
action, and 1 or less for the control variable of P
action. For this reason, the maximum electric
power load which the system can respond is 1
kW, and it is possible that the zones of the optimal
control variables set up with a controller are limited.
3. Settling Time
The calculation results for the settling time are
shown in Figure 20. In the analysis results of
settling time, the reasonable zones of the control
variables P and I change greatly with the rate of
64
load fluctuations of the electric power load as well

as the electric power loads themselves. Although
a reasonable settling time is obtained for the case
where the rate of load fluctuation is 0 for 0.2 kW
electric power load, the control variable of P action
is limited by nearly 5, and the control variable of
I action is a large zone. When the power loads are
0.5 kW and 1.0 kW, the control variable zone of I
action by each load is 3 or more. For the results of
the settling time in 5% rate of load fluctuation,
any power loads have the same value and these
zones of a control variable are also the same.
However, at 10% of rate of load fluctuation,
the control variable zones of settling time differ
in the amount of electric power load. The control
variable zones of a settling time differ greatly by
electric power load and rate of load fluctuation
because the control variables to set up have a
strong influence against the transient response
characteristics of the system.
4. Selection of Control Variables
The control variables of optimal P action and I
action for every power load are chosen from each
result of the transient response characteristics of
the fuel cell system shown in Figures 18, 19 and 20.
These results are shown in Figure 21. The reason
for having determined the control variable zone
in 0.2 kW of power loads in Figure 21, the error
(err) shown in Figure 18 (a) is small in I = 0.7 or
more zones, the power generation efficiency of
the system shown in Figure 19 (a) is high in I =
1.0 or more zones, and also, the settling time of
the system shown in Figure 20 (a) is a zone with
P = 5.0 and I = 2.0 to 7.0. The control variable
zones in other electric power loads considered to
be optimal were determined similar to the example
of 0.2 kW electric power load described at the
top. Using the control variables of P action and
I action from Figure 19, the transient response
characteristics for the power output of the fuel
cell system are calculated, and the results are
shown in Figure 22. In the system controller, the
electric power output in the system over 1.0kW
Figure 22. The electric power response characteristics of the system analyzed with optimal P and I values
is restricted, the transient response characteristics

for 1.0 kW electric power load in Figure 20 do
not have an overshoot.
Operation of the System by the

Selected Control Variables
The simulation of the operation pattern at the time
of the power load input in Figure 16 (a) into the
system is carried out using the control variables
of P action and I action shown in Figure 21. In
Figure 16 (a), the time t0 to t1 is 5 seconds, time
t1 to t2 and time t2 to t3 are 100 seconds, respectively, and time t3 to t4 is 5 seconds. The
reason for the period of t1 to t3 being 200 seconds
is that the time period needs to be sufficient in
order to perform observation of the transient response characteristics of the system, and because
the differences in the responses of the amounts
of electric power cannot be compared if a period
is long. Moreover, time t1 to t2 is 0.2 kW of
electric power load and time t2 to t3 is 0.5 kW.
Figure 23 shows the power load patterns input
into the system. Figure 24 shows the calculation
results of the electric power output response of
the system corresponding to the load inputs in
Figure 23. The results of the response character-
istics of the power output of the system in Figures

24 (a), (b), and (c) are the same. Figure 25 shows
the response results of the exhaust heat output by
the fuel cell stack and the reformer. The exhaust
heat output of the fuel cell stack and the reformer shows a late speed of response. Therefore, the
heat exhaust output of the fuel cell stack and the
reformer requires progress of time from starting
of the system. As one method of improving the
speed of response of the heat, the period shift of
the heat supply-and-demand by thermal storage
is effective.
Figure 26 shows the calculation results of the
amount of town gas consumed by the system. The
consumption of town gas is hardly influenced by
the rate of load fluctuations of the electric power
like the response results of the electric power
shown in Figure 24. If the control variables set
up with the controller of the fuel cell system are
able to be set up by the results in Figure 21 according to the magnitude of electric power load,
the characteristics of transient response of a settling time, the steady state error, the overshoot,
and the rising time can be changed to the best
conditions, and the system can be operated. Moreover, by changing the control variables set up with
the controller of the fuel cell system for the magnitude of electric power loads, town gas consump-
65
Figure 23. Electric power load patterns input into the system
66
Figure 24. Response characteristics of the electric power output of the system
67
Figure 25. Heat power output of the fuel cell stack and the reformer. Electric power load fluctuations
are 0% , 5% , and 10% .
Figure 26. Town gas consumption Qrm ,t + Qcb,t as a function of time. Electric power load fluctuations
are 0% , 5% , and 10% .
tion and electric power generation efficiency are

hardly influenced by additions of fluctuations of
the electric power load to the system in a short
period.
Conclusions
The determination method of the control variables of a system controller, when introducing
and carrying out load-following operation of the
PEFC cogeneration to the electric power loads
that randomly go up and down, for individual
houses was considered. In particular, since the
electric power load pattern of an individual house
consists of loads that often rapidly rise and fall
in a short time, the difference in the load and
68
the system response is influenced by the control

variable. So, in this chapter, the relation of load
with fluctuations, control variables, and settling
time was investigated by numerical analysis. As
a result, this pchapter showed that the operation
was minimally influenced by load fluctuations and
could be performed by changing control variables
with the magnitude of the electric power loads
added to the system. If the control variables of the
system controller are changeable with the magnitude of electric power load, even if a change in
the electric power load in a short period is added
to a system, town gas consumption and the power
generation efficiency are minimally influenced
and can continue operation. Moreover, in order
to maintain the power generation efficiency of
the system, it was shown that the selected area

of the control variable of P and I operation must
be very narrow.
REFERENCES
Carlos, A., & Hernandez, A. (2005). Fuel consumption minimization of a microgrid. IEEE
Transactions on Industry Applications, 41(3),
673681. doi:10.1109/TIA.2005.847277
Dormand, J. R., & Prince, P. J. (1980). A family
of embedded Runge-Kutta formulae. Journal
of Computational and Applied Mathematics, 6,
1926. doi:10.1016/0771-050X(80)90013-3
Forsythe, G., Malcolm, M., & Moler, C. (1977).
Computer methods for mathematical computations. Upper Saddle River, NJ: Prentice-Hall.
Hamelin, J., Agbossou, K., Laperriere, A., Laurencelle, F., & Bose, T. K. (2001). Dynamic behavior
of a PEM fuel cell stack for stationary applications.
International Journal of Hydrogen Energy, 26,
625629. doi:10.1016/S0360-3199(00)00121-X
Hatziadoniu, C. J. (2002). A simplified dynamic
model of grid-connected fuel-cell generators.
IEEE Transactions on Power Delivery, 17(2),
467473. doi:10.1109/61.997919
Ibaraki Prefecture Government Office of Education. (2002). Modeling of hydrogen energy system.
High school active science project research report.
Japan: Ibaraki. In Japanese
Institute of Electrical Engineers of Japan. (2002).
Technology of a fuel cell (pp. 247292). Japan:
Ohmsha. In Japanese
Kahaner, D., Moler, C., & Nash, S. (1989). Numerical methods and software. Upper Saddle River,
NJ: Prentice-Hall.
Katrasnik, T. et al. (2005). Analysis of the dynamic response improvement of a turbocharged

diesel engine driven alternating current generating set. Energy Conversion and Management,
46(18-19), 28382855. doi:10.1016/j.enconman.2005.03.001
Kourosh, S. (2004). Dynamic and transient analysis of power distribution system with fuel cells.
Part 1: fuel cell dynamic model. IEEE Trans.
on Conversion, 19(2), 423428. doi:10.1109/
TEC.2004.827039
Kurata, S. (1999). Measurement of a domestic
energy consumption pattern. National Institute
of Advanced Industrial Science and Technology,
News, (12), 1-14. In Japanese.
Mikkola, M. (2001). Experimental studies on
polymer electrolyte membrane fuel cell stacks
(Masters thesis), Helsinki University of Technology, Finland, 58-79.
Morner, S. O., & Klein, S. A. (2001). Experimental evaluation of the dynamic behavior of
an air-breathing fuel cell stack. ASME. Journal
of Solar Energy Engineering, 123(3), 225231.
doi:10.1115/1.1385202
of the cold region city and utilization for the district heat and cooling (Ph.D. thesis), Hokkaido
University, Japan. In Japanese.
Obara, S., & Kudo, K. (2005). Study on smallscale fuel cell cogeneration system with methanol
steam reforming considering partial load and load
fluctuation. Transactions of the ASME. Journal
of Energy Resources Technology, 127, 265271.
doi:10.1115/1.1926310
Obara, S., & Kudo, K. (2005). Relation between
control variables of PEM fuel cell system and
power generation efficiency. Transaction of JSME,
Part B, 72(714), 447-454. In Japanese.
69
Obara, S., & Kudo, K. (2005). Installation planning of small-scale fuel cell cogeneration in consideration of load response characteristics (load
response characteristics of electric power output).
[In Japanese]. Transactions of the Japan Society
of Mechanical Engineers, Series B, 71(706),
16781685. doi:10.1299/kikaib.71.1678
Obara, S., & Kudo, K. (2006). Output response
characteristics of a small-scale fuel cell for a house.
[In Japanese]. Transactions of the Japan Society
162169. doi:10.1299/kikaib.72.162
Robert, H. (2004). Microgrid: A conceptual solution. In Proceedings of the 35th Annual IEEE
Power Electronics Specialists Conference, 6.
4285-4290.
70
Silva, G. J., Datta, A., & Bhattacharyya, S. P.

(2003). PID controllers for time-delay systems.
Birkhauser.
Takuma, Y., & Goda, T. (2005). Microgrid for
urban energy, transaction of the society of heating. [In Japanese]. Air-Conditioning and Sanitary
Eng. of Japan, 79(7), 573579.
Zhang, Y., Ouyang, M., Lu, Q., Luo, J., & Li, X.
(2004). A model predicting performance of proton
exchange membrane fuel cell stack thermal systems. Applied Thermal Engineering, 24, 501513.
doi:10.1016/j.applthermaleng.2003.10.013
APPENDIX
Nomenclature

a : The time constant in a primary delay system transfer function [s]

b : Constant part in a primary delay system transfer function
D: Control variable of derivative action
E : Electric power [kW]
Et : Amount of theoretical electric power loading [J]
err : Difference of amount of theoretical electric power loading and the amount of system electric
power outputs [%]
H : Heat power [kW]
H : Waste heat power [kW]
h : Capacity of generation [kW]
I: Control variable of integral action
P: Control variable of proportional action
Q: Quantity of flow [m3/s]
t : Sampling time [s]
u : Power demand [W]
Vg : Heat quantity of town gas consumption [J]
v : Electric power output [W]
e : Electric power generation efficiency [%]
rm ,t : Reformer efficiency [%]
Equipment Symbols

cb : Burner for the heat sources of the reformer

f : Fuel cell
ra : Radiator
rm : Reformer
sys : System
71
72
Chapter 3
Effective Improvement in
a Fuel Cell Microgrid
ABSTRACT
This chapter consists of two sections, Effective Improvement in Generation Efficiency due to Partition
Cooperation Management of a Fuel Cell Microgrid and Equipment Plan of Compound Interconnection
Microgrid Composed from Diesel Power. In the 1st section, the PEFC microgrid is explored as a distributed power supply with little environmental impact. The proposed system obtains results with high
generation efficiency compared with the central system of a fuel cell microgrid. An independent microgrid
that compounds and connects a diesel power plant generator and PEFC is proposed in the 2nd section.
A complex community model and residential area model were used for analysis.
The summary of the 1st sectionis as follows. The
PEFC microgrid is expected to be a distributed
power supply with little environmental impact.
However, if a microgrid is installed in an urban
area, a generation efficiency of less than 21% on
an all-year basis is expected. Generally, in planning an electric power network using a microgrid,
all the target buildings are connected and electric
power is supplied. In this study, a microgrid is
divided into multiple and each is optimized for

the purpose of maximization of power generation
efficiency. In the cooperation management of a
microgrid, large fluctuations in load, or increases
and decreases in a building, can be followed with
a grid using a system-interconnection device. The
system proposed in this study obtained results
with high generation efficiency (from 21.1% to
27.6%) compared with the central system (generation efficiency is 20.6% to 24.8%) of a fuel
cell microgrid.
DOI: 10.4018/978-1-4666-5796-0.ch003
Effective Improvement in Generation Efficiency of a Fuel Cell Microgrid

Although the microgrid using a diesel power plant
power generator (DEG) has a high possibility of
implementation, there is a problem in the emission of carbon dioxide gas. Then, an independent
microgrid that compounds and connects a diesel
power plant generator and a proton-exchange membrane fuel cell (PEFC) is proposed. The operation
of DEG is controlled to correspond to the base
load of whole CIM (Compound Interconnection
Microgrid), and, on the other hand, the operation
of PEFC is controlled to follow the load fluctuation
of CIM. In this section, the generation capacity of
DEG and PEFC installed into CIM and the grid
route of PEFC were analyzed with the object of
maximizing generation efficiency. A complex
community model and residential area model were
used for analysis. In this section, the microgrid
concerning the urban area (18 buildings) in Tokyo
was investigated. From the results of analysis, it
was confirmed that CIM could be operated with
a high generation efficiency of 27.1 to 29.9%.
EFFECTIVE IMPROVEMENT IN
GENERATION EFFICIENCY DUE
TO PARTITION COOPERATION
MANAGEMENT OF A FUEL
CELL MICROGRID
Introduction
A reduction effective of greenhouse-gas emission
is expected with the microgrid (Robert et al., 2004;
Carlos & Hernandez, 2005; Takuma & Goda,
2005). In particular, the microgrid using a fuel
cell is predicted to be a leading method of future
energy supply. In order to effectively reduce greenhouse gases to a maximum, a microgrid should be
maintained and operated at the highest possible
generation efficiency. It is necessary to optimize
the operation plan of a microgrid, as well as the

capacity of the energy equipment based on the
power demand pattern of the buildings connected
to a grid. Therefore, to produce high generation
efficiency of a microgrid, a power generator that
can maintain high efficiency is required over a
large range from high load to low load. However,
it is difficult to maintain high efficiency over a
wide operating range with the proton-exchange
membrane fuel cell (PEFC) with a reformer. Then,
the method of operating a water electrolyzer (Obara
& Kudo, 2005a), and the method of divides a fuel
cell and a reformer (Obara & Kudo, 2005b) were
examined at the time of partial-load operation.
The power load added to a microgrid is decided
by composing the power demand characteristics of
two or more buildings. However, the example of
considered the relation between the loads pattern
of a building and the power generation efficiency
of a microgrid is not reported. If the power demand
pattern of a certain building reduces the generation
efficiency of the whole microgrid, this building
sets the grid of another network with a group of
other buildings, and overall generation efficiency
may be improved. Therefore, the fuel cell microgrid (FC microgrid) is divided into multiple
grids, and this Section considers the independent
management of each grid. Two or more divided
grids can operate in cooperation with other grids
to improve generation efficiency. In this Section,
partition cooperation management of the FC microgrid and its relationship to generation efficiency
are investigated using the power demand pattern
assuming typical buildings in Tokyo. In addition
to this, the FC microgrid that is equipped with
the cooperation operation of the partition grid is
effective at Response to overload at the time of
load fluctuation, Risk abatement at the time of
accident due to distribution of power facilities
and Interruption of service not caused at the time
of maintenance.
73
System Configuration
Scheme of FC Microgrid
There is an interconnection system (Figure 1
(a)) and an independent system (Figure 1 (b))
in a microgrid. By increased power demand due
to connection of a new building and large load
fluctuations, when the grid cannot respond, an
interconnection microgrid connects to other grids.
( of Figure 1 (a) is a system-interconnection
device.) Therefore, an interconnection microgrid
is a system that can respond to the power demand
Figure 1. Microgrid model
74
pattern of a grid flexibly. On the other hand, it is

necessary to supply the entire power demand in a
grid using an independent microgrid. Therefore,
the design of an independent microgrid needs to
sufficiently consider power demand and supply
balance. The connection of a new building is
difficult for this grid type, and flexibility is poor.
However, the independent microgrid is effective
as a method of supplying electric power in areas
where the transmission line infrastructure is not
fixed. Therefore, in this Section, as shown in Figure
1 (c), one independent microgrid is divided into
multiple grids (in Figure 1 (c), these are Grid A
and Grid B). Usually, although each grid operates independently, if it is in the condition where
generation efficiency is improved, each grid will
be connected and it will perform cooperation
operation. If the method of Figure 1 (c) is used,
the independent microgrid will be improved by
the system that can respond flexibly to increases
and decreases in a building, or increases and decreases in load fluctuation. Therefore, in this study,
the relationship between the building (route of a
microgrid) connected to each of the divided grids
(in the case of Figure 1 (c), these are Grid A and
Grid B) and generation efficiency is investigated.
2 (a)) connected to one independent grid (Grid

A and Grid B, respectively) may be at one place.
Buildings other than the power plant are buildings
(Figure 2 (b)) in which a boiler is installed for heat
supply, and a system-interconnection device. An
actual FC microgrid requires a town gas distribution network and exhaust heat distribution network
other than an electric power supply network. This
Section does not describe the management of a
town gas distribution network or an exhaust heat
distribution network.
In the following, the operating method of the FC

microgrid system shown in Figures 1 (c) and 2 is
described. In Figure 2 (a), town gas is supplied to
a heat-source burner and a reformer, steam reforming of town gas is performed and reformed gas with
many hydrogen components is produced. Because
there is a lot of water in reformed gas, the water
is removed with a dryer. Furthermore, the carbon
monoxide in the reformed gas is removed with a
carbon monoxide oxidation system. Reformed gas
is supplied to the fuel cell, and power and exhaust
heat are output from the fuel cell. After storing
exhaust heat in a heat storage tank and exchanging the heat of a thermal storage medium for tap
The buildings in which power plants are installed

(Figure 2 (a)) and other buildings (Figure 2 (b))
are connected to each of Grid A and Grid B of
Figure 1 (c). A town gas reformer, PEFC, a boiler,
a heat storage tank, and a system-interconnection
device are installed in the building of Figure 2
(a). The boiler for heat supply and the systeminterconnection device for receiving electric power
from a grid are installed in the building of Figure
2 (b). Figure 2 (a) shows the model of a building
in which a power plant is installed as shown in
Figure 1 (c). In this Section, a power plant (Figure
Operating Method
Figure 2. Energy equipment model installed in a building
75
water, tap water is supplied to an auxiliary boiler,

which is output to the demand side. The power
produced with the fuel cell is supplied to a grid
through a DC-AC converter, an inverter, and a
system-interconnection device. A fuel cell is operated so that the power load may be followed. The
output control of a fuel cell adjusts and controls
the amount of town gas supplied to a reformer.
Figure 2 (b) shows the equipment model of buildings other than the power plant shown in Figure
1 (c). In these buildings, power is obtained from
a grid through a system-interconnection device.
Heat is produced with a town gas fired boiler.
Grid A and Grid B, shown in Figure 1 (c),
usually supply power to buildings independently
by each grid. However, when a load that exceeds
capacity is added to one of the grids, power can
be obtained from the other grids through the
system-interconnection device. In the partition
cooperation management of the FC microgrid
described in the following, all the grids can deliver and receive other grids and power through
the system-interconnection device.
Installation Planning of FC Microgrid

Generation Efficiency of Microgrid
Figure 3 (a) shows the cell performance curve for
an operating temperature of 333 K, with the gas
pressure at the anode and cathode being 0.1 MPa
(Takeda, Y., et al., 2004; Ibe, S., et al., 2002;
Mikkola, M., 2001; Ibaraki Prefecture Government
Office of Education, 2002). This characteristic is
the model to be used when setting the reformer
efficiency to be constant at 73%. Details of RM
are given below. Although steam reforming of
town gas at a flow rate of QRM ,t is conducted and
reformed gas is produced in the reformer, the heat
source for the reforming reaction is burned and
obtained using the town gas at a flow rate of QB ,t .
The following equation defines the reformer efficiency, and the maximum of this value is 73%
76
(Ibaraki Prefecture Government Office of Education, 2002). WH is the calorific value of the hydrogen included in reformed gas, and WB and
WR are the calorific values of town gas at a flow
rate of QBN ,t and QRM ,t , respectively. It depends
1
for the power generation efficiency of a microgrid

on the fuel cell output characteristics of Figure 3
(a). For example, the output characteristics of a
fuel cell change with the operating method of a
reformer. It needs to be careful of the characteristic of Figure 3 (a) not being the result of representing a fuel cell system.
RM =
(W
WH
+WR )
100
(1)
Figure 3 (b) shows the model indicating the

relation between the load pattern of a power demand model, and the generation efficiency of a
fuel cell. As Figure 3 (a) shows, the generation
efficiency of a fuel cell changes with the load
ratio. Next, the operation method of a fuel cell
when there are two or more power demand models is described.
In Figure 3 (c), each power demand model of
Building A, B, and C is shown. The following
three methods of supplying power to these buildings from PEFC can be considered.
1. Installing a fuel cell in all the buildings
(stand-alone system)
2. Connecting all the buildings to a microgrid,
and supplying power using one set of FC
(central system)
3. Setting two or more microgrids considering
the power demand pattern of the buildings
(partition cooperation system)
The stand-alone system determines the capacity of PEFC installed in each house so that it
may exceed the maximum of each power load of
Building A, B and C of Figure 3 (c). In the central
Figure 3. Eficiency of PEFC and power demand pattern
77
system, the power load of Building A, B and C of

Figure 3 (c) is added for every sampling time, and
the capacity of one set of PEFC is determined to
exceed the maximum. In addition, in the partition
cooperation system, Building A and Building B of
Figure 3 (c) are connected, for example, and Grid
1 operates. Building C is operated independently
(Grid 2). The capacity of PEFC installed in Grid
1 and Grid 2 is decided so that it may exceed
the maximum of the power load of Building A
and Building B, and the maximum of the power
load of Building C. The stand-alone system does
not need a grid and, moreover, does not need to
transport exhaust heat to a neighboring building. Therefore, the heat loss of exhaust heat is
small, and installation of a grid is unnecessary.
However, there are many fuel cells to install, and
it is necessary to install large-capacity fuel cells
so that load fluctuation does not have an impact.
Unlike the stand-alone system, the central system
should just install FC in one building. Therefore,
although it is advantageous in respect of equipment
cost, the transport distance for the exhaust heat
is long, and heat loss is a problem. Furthermore,
the diversification of risks at the time of accident
and the extendibility of equipment are problems.
In the central system, the power demands of the
buildings are added, and the capacity of the fuel
cell is optimized and determined to exceed the
maximum. Therefore, if the number of the buildings connected to the grid and the pattern of the
power demand model change, the load ratio will
change and generation efficiency will vary. On
the other hand, the partition cooperation system
can partially obtain each merit of the stand-alone
system and the central system. Partition of the
microgrid in the partition cooperation system is
optimized to maximize generation efficiency. For
this reason, depending on the composition of the
power demand model of the buildings in an urban
area made into an analytical object, generation
efficiency higher than that with the stand-alone
system or the central system may be applicable.
78
Power Demand Model

Figure 4 shows the power demand models of each
building, and is the mean power load of each
sampling time on a representative day in January
(winter), May (mid-term), and August (summer)
(Architectural Institute of Japan, 2002; Nagase
et al., 2003; Hatano et al., 2003; Yoshinaga et
al., 2003). However, the actual power demand
pattern is a mixture of loads that change rapidly
over a short time, such as an inrush current. In
addition, a power demand estimate for the house
actually shown in Figure 4 (a) is difficult, and
a power demand estimate for the small offices
and factories indicated in Figures 4 (d) and (e)
is comparatively easy. Although load fluctuation is not taken into consideration in the power
demand model in the analysis of this Section,
when accompanied by load fluctuations, system
interconnection between the grids is expected to
occur frequently. The power demand pattern of
a house (Figure 4 (a)) has a peak in the morning
and the afternoon. In the hotel of Figure 4 (b),
demand is stabilized when the midnight-to-earlymorning period is excluded, and power demand is
stable for 24 hours at convenience stores (Figure
4 (c)). The difference between the time zone with
small power demand from night to early morning and the time zone with great power demand
from morning to evening is clear in small offices
(Figure 4 (d)), factories (Figure 4 (e)), and small
hospitals (Figure 4 (f)).
Analysis Method
The analysis flow of the generation efficiency of
the FC microgrid is shown in Figure 5 (a). First,
a power demand model of the buildings that
compose the urban area model is prepared. In the
analysis, the power demand model shown in Figure
4 is used. Although these power demand models
are inputted into a program for every sampling
time, the input is related to all the buildings. Next,
Figure 4. Power demand models
all the routes (that is, a divided grid) of the grid

are discovered. This route planning is obtained
by calculating the permutation of the number of
buildings. The capacity of the fuel cell installed
in each route of the FC microgrid is set up, and
the power load and the load ratio of each route are
calculated for every sampling time. The generation efficiency of a route is calculated by giving
this load ratio to the approximation described in

Figure 3 (a). Furthermore, the capacity of the fuel
cell installed in each route of the FC microgrid is
changed, and the generation efficiency of a route
is calculated using the same procedure. From the
generation efficiency of all the routes obtained
by calculation, the generation efficiency (average
generation efficiency) in the entire FC microgrid
79
Figure 5. Analysis flow, power demand model, and average generation efficiency
for a representative day can be calculated. The

average generation efficiency of the stand-alone
system, the central system, and the partition cooperation system is calculable using the analysis
procedure described above.
80
Case Study
Figure 5 (b) shows the urban area model used for
this analysis, and shows the type of power demand
model of 23 buildings in this figure. In addition,
the number given in this figure is the number of

buildings, and shows the area and the maximum
load of each building assumed in Table 1. Figure
5 (b) shows a two-person family house and a
six-person family house as well as a four-person
family house (the power demand model is shown in
Figure 4 (a)). Each power demand model compares
and determines the number-of-persons rate of the
model of Figure 4 (a). Moreover, although apartment houses are shown in Figure 5 (b), these power
demand models are also relatively determined
from the number-of-persons rate of the model of
Figure 4 (a). In this analysis, Figure 5 (b) is used
as an urban area model. Moreover, Figure 4 (a)
to (f) is used as a power demand model of each
building that composes an urban area. The relation of the load ratio and generation efficiency of
PEFC installed in the model is shown in Figure
3 (a). Since the maximum generation efficiency
of the model of Figure 7 is 32%, the maximum
generation efficiency of the microgrid analyzed
here is theoretically 32%.
Analysis Results and Discussions

Stand-Alone System
Figure 5 (c) shows the analysis results of the average generation efficiency of a representative day
of every month in the case of installing the stand-
Table 1. Maximum power load of each buildings
alone system in the urban area model of Figure 5

(b). Although the average generation efficiency
differs each month, convenience stores with a small
load fluctuation range of power throughout the
year show about 30% at maximum. The average
generation efficiency of a representative day in
August with high power consumption due to airconditioners is high in hotels and hospitals. The
average generation efficiency of other buildings
is less than 20%.
The system that distributes the PEFC system
whose generation capacity is 1 kW or 2 kW to the
urban area model of Figure 5 (b) is investigated.
To install PEFC (1 kW or 2 kW) using the standalone system, it is necessary to look for buildings
that can install fuel cells of this capacity from
among buildings that compose an urban area
model. Figure 6 (a) shows the analysis results of
the number of buildings in which a fuel cell of
each capacity can be installed using the stand-alone
system, and the number of the fuel cells installing
PEFC using the partition cooperation system. If
equipment cost is taken into consideration from
the difference in the installed number of PEFC,
compared with the stand-alone system, the partition cooperation system is more advantageous.
Generation Efficiency of Central

System
Figure 6 (b) shows the analysis result of the
relationship between the capacity of PEFC and
generation efficiency in the case of installing
the central system in the urban area model. The
results of the generation capacity of PEFC using the central system is 110 kW (the average
generation efficiency of each representative day
is 25.8% and 26.1%, respectively) on representative days in January and May, and is 160 kW (the
average generation efficiency of a representative
day is 24.8%) on a representative days in August.
Therefore, it is required that a 160-kW PEFC be
81
Figure 6. Calculation results of the installed number and power generation efficiency
installed in the central system in the urban area

model, and be applied throughout the year. When
a 160-kW PEFC is installed, the average generation efficiency of representative days in January
and May is 20.6% to 20.9%. Consequently, the
generation efficiency of an FC microgrid when
installing the central system is operated at 20.6
to 24.8%.
82
Generation Efficiency of Partition

Cooperation System
Operation of FC microgrid in which is installed
PEFC with low generation capacity
Figures 7 (a) to (c) show the analysis results
of the partition cooperation system in which is
installed PEFC whose generation capacity is 2 kW.
Figure 7. Analysis results of the partition cooperation microgrid system at the time of installing 2 kW PEFC
However, because two or more buildings that have

the same power demand model are included as
Figure 5 (b) describes, when these are exchanged,
it is expressed by the route difference from Figure
7 (a) to (c). The average generation efficiency of
representative days in January, May, and August
is 23%, 22.7%, and 20.4%, respectively. Because
the space-cooling load of air-conditioners will
be included in the power demand amount of a
representative day in August, the load ratio is
high compared with other months. Therefore,
the average generation efficiency of a representative day in August is high. Because the power
demand is high on a representative day in August
compared with other months, these analysis results (the route of a grid, and the capacity of the
fuel cells installed in the grids) can be applied to
other months. Then, the analysis result of the grid

route of a representative day in August was given
to the power demand model of a representative
day in January, and the generation efficiency was
analyzed. This result is shown in Figure 8. In the
results of Figure 8, the generation efficiency of a
representative day in January is low at 2.5% compared to the generation efficiency of the whole grid
on a representative day in August, being 19.9%.
Therefore, when PEFC whose capacity is 2 kW
is installed in Figure 5 (b), the power is supplied
to the 11 buildings by the four grids (that is, four
fuel cells) of Grid A to Grid D, and the generation
efficiency is about 20% throughout the year. As
Figure 5 (c) describes, the generation efficiency
in the case of supplying power to a house (twoto six-person household) using the stand-alone
83
Figure 8. Analysis result in January when the same as analysis results of the grid in August. 2 kW PEFC
is installed in each grid. Average generation efficiency is 19.9%.
system is about 20% at maximum. Moreover,

compared with the number of PEFC installed by
the stand-alone system from the results of Figure
6 (a), the number of installations in the partition
cooperation system is about 1/3 with a generation
capacity of 2 kW. Compared with the stand-alone
system, the whole generation efficiency of the
microgrid in which is installed PEFC with small
generation capacity is improved, and equipment
cost is reduced. However, a maximum power
demand greatly over 2 kW cannot be installed in
this grid in buildings.
Operation of FC microgrid combined so that
the generation efficiency of the whole grid is
maximized.
Figures 9 (a) to (c) show each representative
day in the grid route obtained in the analysis, and
show the analysis results at the time of combining so that the generation efficiency of the whole
grid is maximized. However, because two or more
buildings that have the same power demand model
in an urban area model are included as Figure 5
(b) describes, if these are exchanged, the combination will become different from Figures 9
(a) to (c). Figure 10 shows the analysis result of
the generation efficiency when giving the route
result of a representative day in August (Figure 9
84
(c)) to the power demand model of a representative day in January. Compared with Figure 9 (c),
the result of the generation efficiency of Figure
10 falls due to each Grid A to Grid C, and the
average generation efficiency of the whole grid
falls by 23.6% to 21.1%. Therefore, as for the FC
microgrid whose average generation efficiency
in total comprises a combination of the highest
grids, generation efficiency is operated at 21.1 to
27.6%. The range of this value exceeds the average generation efficiency (from 20.6% to 24.8%)
when installing a 160-kW PEFC using the central
system shown in Figure 6 (b). Operation of a FC
microgrid that combines the grid in which average
generation efficiency exceeds 25%.
Figures 11 (a) to (c) show the result of the
combination for which the generation efficiency
of a representative day is a grid route exceeding
25%, and the generation efficiency of the whole
grid peaks every month. The grid route for which
generation efficiency does not exceed 25% does
not appear in Figures 11 (a) to (c). Consequently,
there is no guarantee that all the buildings in an
urban area model can be connected to any grid.
Figure 12 shows the analysis result of the generation efficiency when installing the power demand model of a representative day in January
Figure 9. Analysis results of the partition cooperation microgrid system which supplies the power to all
the buildings
Figure 10. Analysis result in January when the

same as analysis results of the grid in August.
Average generation efficiency is 21.1%.
into the analysis result (Figure 11 (c)) of the grid

route of a representative day in August. Compared
with Figure 11 (c), the generation efficiency of
Figure 12 falls compared to the other grids (Grid
A and Grid B), and the average generation efficiency of the whole grid falls by 19%, being 23.2%.
Therefore, as for the FC microgrid whose generation efficiency is composed from grids exceeding 25%, generation efficiency is operated at 23.2
to 28.6%. If this system is installed, the FC microgrid can be operated at the highest generation
efficiency, but buildings that lower the generation
efficiency of the whole grid are separated from
this grid.
85
Figure 11. Analysis results of the partition cooperation microgrid system when supplies the power to
the building of 25% or more of par generation efficiency
Figure 12. Analysis result in January when the

same as analysis results of the grid in August.
Average generation efficiency is 23.2%.
86
Conclusions
Even if an FC microgrid is used for urban areas
composed of buildings with various power demand
patterns and supplies power to them, generation
efficiency generally lowers. Therefore, this Section examined the generation efficiency of the
following three power supply methods: (1) Installing fuel cells in all the buildings in an urban area
model (stand-alone system); (2) Connecting all
the buildings to a microgrid, and supplying power
from one set of fuel cells (central system); and (3)
Dividing a microgrid into multiple grids consid-
ering the power demand pattern of the buildings

(partition cooperation system). The relationship
between the capacity of the fuel cell to be installed
and generation efficiency became clear from the
analysis results. Furthermore, in this Section,
generation efficiency is optimized to maximize
generation efficiency to the grid route planning
of the partition cooperation system. As a result, a
system with higher generation efficiency than the
stand-alone system and the central system could
be proposed. The average generation efficiency
of the stand-alone system is 20% or less, and
the average generation efficiency of the central
system is 20.6% to 24.8%. On the other hand, the
generation efficiency of the partition cooperation
system proposed in this Section is 21.1% to 27.6%.
EQUIPMENT PLAN OF COMPOUND

INTERCONNECTION MICROGRID
COMPOSED FROM DIESEL POWER
PLANTS AND PROTON-EXCHANGE
MEMBRANE FUEL CELL
Introduction
The installation to an urban area of the fuel cell
microgrid is the technology of spreading the utilization of hydrogen energy. For example, hydrogen
production using green energy and reforming
technology of natural gas can be introduced.
Generally, as for introduction to a city area of a
microgrid, the following points are expected: (a)
The distance for the heat supply is short and effective use of exhaust heat is possible (Obara &
Kudo, 2006; Obara, 2006); (b) Load leveling of
existing large-sized power generation equipment
is possible; and (c) Since a facility suitable for the
energy demand characteristics of a community can
be installed, it is a technology in which energy
efficiency is high and the environmental impact
is low, is expected (Robert et al. 2004; Carlos
& Hernandez, 2005, Takuma & Goda, 2005).
However, a proton-exchange membrane fuel
cell (PEFC) has an expensive electrode material

(catalyst material and solid polymer membrane).
Furthermore, since the system is complex, it is
difficult to commercialize it immediately. Then,
reducing the number of installations of the expensive fuel cell by connecting PEFC to a microgrid in
this chapter, and supplying power to two or more
buildings, is considered. However, the subject of
this system is the frequent partial-load operation
with low efficiency, when power is supplied to
two or more buildings using a large-capacity fuel
cell (FC). As technology to solve this issue, a fuel
cell is divided into small-capacity units, and there
is the method of increasing the load factor of each
unit (Obara & Kudo, 2005). However, by this
method, the number of fuel cell units increases
greatly, and facility cost increases. Consequently,
the base load of a microgrid is supplied using a
diesel engine power generator (DEG), and how to
install and interconnect two or more PEFC grids is
examined. The compound grid of DEG and PEFC
is interconnected in this chapter. This microgrid
is described as CIM (Compound Interconnection
Microgrid). There are an interconnect system and
an independent system in CIM. The interconnect
system is connected with other grids, such as
commercial power, and is operated. Although
realization of an independent system is predicted
to be difficult compared with the interconnect
system, it is considered that the effect of (b) and
(c) described above will be large. So, this chapter examines the independent microgrid system
built by two or more FC grids in which system
interconnection is possible.
There are many cases of installing DEG as
cogeneration until now, and the characteristics,
such as power generation efficiency, facility cost,
and power cost due to a number of achievements
can be estimated. Although it is expected that the
microgrid using DEG has high realizability, it is
accompanied by the problem of carbon dioxide
discharge. Consequently, DEG is introduced as
generating equipment corresponding to the base
load of the whole CIM, and operation near the
87
maximum efficiency point is examined. On the

other hand, the dynamic characteristics at the time
of load fluctuation and carbon dioxide emissions
of PEFC are good compared with DEG (Obara,
S., 2006). PEFC linked to CIM is controlled to
operate corresponding to the load fluctuation
of the grid. However, the load factor of PEFC
changes with the power-demand patterns of each
building linked to CIM in this system. As a result,
the power generation efficiency of the whole
microgrid may improve by dividing a microgrid
into two or more, and determining the buildings
connected to each grid (grid route) with the object
of maximizing the power generation efficiency.
In this chapter, an independent microgrid with
high power generation efficiency is planned by
optimizing the capacity of DEG and PEFC, and
selecting the grid routes.
Compound Interconnect Microgrid

Microgrid Model
This section examines two or more FC grids and
the independent microgrid built by DEG operated
according to a base load. However, in this section,
only the power system is investigated, and the
exhaust heat of the generating equipment is not
investigated. Figure 13 shows the CIM model that
introduced seven FC grids into 61 buildings. DEG
is installed in the buildings of any grid and outputs
constant power corresponding to the base load of
all the FC grids. The power-demand patterns of
each building in a city area differ. Therefore, the
power load pattern of each FC grid changes with
the route of the building linked to the grid. Consequently, as shown in Figure 13 (a), the building
linked to each FC grid is selected and arranged
with the object of maximizing power generation
efficiency. Figure 13 (b) shows the model of the
FC grid (FC grid A to G) in Figure 13 (a), and
power supply-and-demand is possible for each
grid through a system interconnection device
(CP1 to CP7). The system interconnection between
88
the grids is effective when supplying power from

another system for accident, maintenance, etc.,
and when there is large load exceeding a certain
grid capacity.
CIM Model
Figure 13 (c) shows the model of the FC grid and
an interconnection device shown by Figure 13 (a)
and (b). In CIM, DEG of with a capacity of PDEG
is installed and PEFC with a capacity from PFC ,A
to PFC ,G is installed from FC Grids A to G, respectively. Each grid can change over and interconnect the system interconnection device of CP1
to CP7. DEG is operated corresponding to the base
load of the city area model shown in Figure 13
(a). DEG is operated by constant load. Concerning load fluctuation, power is supplied from FC
Grids A to G. As shown in Figure 8 (c), all FC
Grids A to G are connectable with DEG.
Facility Scheme
Figure 14 (a) to (c) is a facility scheme installed
in the building linked to CIM shown in Figure
13. Figure 14 (a) shows the facility scheme of
a building of installing DEG, and Figure 14 (b)
shows the facility scheme of a building of installing
PEFC. A building in which is installed the facility
of DEG shown in Figure 14 (a) is connected to
any one grid, and a building in which is installed
the facility shown in Figure 14 (b) is connected to
all the FC grids. Figure 14 (c) shows the facility
scheme of a building in which DEG or PEFC is
not installed. Generating equipment composed
from a diesel engine, a power generator, a boiler,
a heat storage tank, a system interconnection device, etc. is installed in Figure 14 (a). Moreover,
the generating equipment composed from a town
gas reformer, PEFC, a boiler, a heat storage tank,
a system interconnection device, etc. is installed
in Figure 14 (b). In a reformer, reformed gas is
produced on a catalyst by making the combustion
Figure 13. CIM (Compound Interconnection Microgrid) model
89
gas of town gas into a heat source. Since there

is a lot of water in reformed gas generated by
steam reforming, reformed gas is cooled by the
air supply of a blower with a dryer, and the water
is condensed and is separated. In order for the
carbon monoxide concentration in the reformed
gas in a fuel cell stack entrance to be several ppm,
the carbon monoxide oxidization part is prepared.
In the carbon monoxide oxidization part, carbon
monoxide is burned on a catalyst and it changes
into carbon dioxide. Reformed gas is supplied
to a fuel cell stack from the carbon monoxide
oxidization part, and it generates electricity. The
generated DC power is changed into an alternating
current of constant frequency through an inverter,
and is supplied to a system interconnection device.
Moreover, the boiler for heat supply and the system
interconnection device for obtaining power from
a grid are installed in Figure 14 (c).
CIM Operating Method

The model of the operating method of CIM is
shown in Figure 14 (d). The power load of a
representative day is divided into the base load
of the constant load, and other loads as shown in
the figure. In operating and generating DEG at
about the base load, other loads correspond by
the power generation of FC. Although DEG is
one set, the FC grid consists of two or more sets.
Since the FC grid corresponds to load fluctuation,
it may operate at partial load with low efficiency,
but DEG can be operated by the constant load of
the maximum efficiency point.
Equipment Characteristics
Diesel Engine Power Generator
The output characteristics result of the cogeneration system using DEG is shown in Figure 15
(a). This result is the relation among the calorific
heat of the kerosene fuel supplied to DEG, the
engine-cooling-water heating value and the engine
90
exhaust gas heating value, and the production

of electricity. The engine specifications of the
cogeneration system of Figure 15 (a) are shown
in Table 2 (a). Moreover, the specifications of a
synchronous power generator are shown in Table
2 (b). The fuel of a diesel engine is kerosene and
uses 2 cylinders and 4 cycles. A power generator
is a single-phase synchronous type, and power is
transmitted through a belt from the power shaft
of the diesel engine. If the amount of kerosene
fuel is increased, the production of electricity
and exhaust gas heating value increases, but the
engine-cooling water heating value decreases.
The maximum power generation output is 3 kW,
and the kerosene supply heating value at this time
is 9.8 kW. Figure 15 (b) shows the production
of electricity of DEG and the relation of power
generation efficiency that were obtained by the examination. Although power generation efficiency
changes with the number of engine rotations, since
this difference is small, the approximated curve
shown in Figure 15 (b) is used in the analysis of
this research. Moreover, the relation between the
load factor and power generation efficiency shown
in Figure 15 (b) should be maintained even if the
capacity of DEG changes.
Proton-Exchange Membrane Fuel Cell

The output characteristics of the fuel cell stack
used in the analysis of this chapter are shown in
Figure 15 (c). The maximum power generation
efficiency of PEFC shown in Figure 15 (c) is 32%.
The supply of town gas has a system of changing it
into reformed gas, and a system of supplying it to
the burner used as a heat source of the reforming
reaction. Power generation efficiency is calculated
by dividing the production of electricity of PEFC
by the calorific power adding these two town gas
systems (Takeda et al., 2004; Ibe et al., 2002;
Mikkola, 2001; Ibaraki Prefecture Government
Office of Education, 2002). The electrode area
of an anode and cathode of the examined PEFC
is 1.0 m2, respectively. Moreover, the cost, calcu-
Figure 14. Equipment model installed in a building and an operation plan
91
Figure 15. Power generator model
92
Table 2. Specifications of equipment
efficiency total ,Day of a representative day, and is

obtained using Equation (2). In the analysis of this
chapter, the route of the FC grid and the generation
capacity of DEG and FC in case WDay is the
maximum and has been decided to be an optimal
solution.
NR
total ,t =
E DEG ,t DEG ,t + (E FC ,n ,t FC ,n ,t )
n =1
Etotal ,t
23
WDay = total ,Day = (total ,t )

t =0

(2)
(3)
When there is the same pattern among the

power-demand patterns of each building introduced into a city area model, two or more grid
routes considered to be optimal appear.
Analysis Flow
lated by dividing the calorific power of hydrogen
in the reformed gas by the calorific power of the
two town gas systems described above is defined
as reformer efficiency.
Analysis Method
Route Plan of Compound
Interconnect Grid
Equation (2) is an expression of total power generation efficiency total ,t in sampling-time t , and
calculates for all the grid routes that compose CIM.
total ,t is calculated from power generation efficiency DEG ,t
and load E DEG ,t
of DEG, and
power generation efficiency FC ,n ,t of the FC grid

of route n (n = 1, 2,..., N R ) and load E FC ,n ,t .
Equation (3) is the objective function. Objective
function WDay is equal to total power generation
The analysis flow that searches for the optimal

solution of CIM is shown in Figure 16. First, the
power-demand data of each building that composes a city area model is given to the analysis
program. Next, the base load of a microgrid is
calculated from the power demand data, and the
capacity of DEG is decided. Regarding all the FC
grid routes, the power generation efficiency for
every sampling time is calculated. However, the
FC grid route is given to a program widely assuming the number of grids and the capacity of
PEFC installed in each grid, and calculates the
power generation efficiency for all the FC grid
routes. By adding and equalizing these results,
the average generation efficiency on a representative day is obtained. The power generation efficiency of DEG and PEFC is used to calculate the
load ratio from the power demand of all the buildings connected to a grid and the capacity of DEG
and PEFC that were set up beforehand, and these
93
Figure 16. Calculation flow of generation efficiency for FC microgrid
are calculated by inserting them into Figures 15

(b) and (c). The power generation efficiency of
DEG and each route of FC grid is given to Equation (2), and the route of FC grid in case WDay
of Equation (3) is the largest value, and the capacity of DEG and PEFC are decided to be the optimal solutions.
Power Demand Model

Figure 17 shows the power demand model of
each building in Tokyo used in the analysis, and
is the mean power load of each sampling time
of the representative day in January (winter),
94
May (mid-term), and August (summer) (Architectural Institute of Japan, 2002, Nagase, O. et
al., 2003, Hatano, Y. et al., 2003). However, the
actual power demand pattern is a meeting of the
load that changes rapidly in a short time, such
as an inrush current. In Tokyo, the annual average temperature for the past five years is 289 K.
The average temperature in January is 279 K,
and the highest and the lowest temperatures on
a representative day in January are 283 K and
275 K, respectively. The average temperature
in May is 292 K, and the highest and the lowest
temperatures on a representative day in May are
296 K and 288 K, respectively. The highest and
the lowest temperatures on a representative day
in August for the past five years are 302 K and
296 K, respectively, and the average temperature
is 298 K (National Astronomical Observatory,
2003). There is a high power demand on a representative day in August compared with other
months including the space-cooling load. The
power demand estimate of the family household
shown in from (a) to (d) of Figure 17 is difficult,
and the power demand estimate of the small offices and factories indicated in Figures 17 (g) and
(h) is regular, and comparatively easy. Although
load fluctuation in a short time is not taken into
consideration for the power demand model in
the analysis of this chapter, when accompanied
by load fluctuation, it is necessary to investigate
the dynamic characteristics of the grid. The power
demand pattern of a family household (from (a)
to (d) of Figure 17) shows a peak in the morning
and the afternoon. The demand of hotels (Figure
17 (e)) stabilized when midnight to early morning was excluded, and there is continuous power
demand at convenience stores (Figure 17 (f))
with business for 24 hours. The difference in the
time zone of night to early morning with little
power demand and the time zone from morning
Figure 17. Power demand model
to evening with high power demand is clear in

offices (Figure 17 (g)), factories (Figure 17 (h)),
and hospitals (Figure 17(i)).
Case Study
Urban Area Model
Figure 18 shows the urban area model used for
analysis. The number shown in the figure is the
building number, and also shows the type of each
building. The urban area model is composed
from 20 buildings, and Table 3 shows the type
of each building. The analysis investigates each
case of Table 3 (a) and (b). Case A is a model

(complex community) assuming an urban area
that consists of various buildings, and Case B is
a model (residential area) assuming a residential
street. In addition, each power demand pattern
uses Figure 17. For example, building numbers
1 and 2 are family households (two persons), and
each power demand model used for analysis is
shown in Figure 17 (b). Therefore, the grid route
of building numbers 1 and 2 is exchangeable.
The power demand model of Figure 17 and the
urban area model of Figure 18 are installed into
the analysis program described in Figure 16,
and the efficiency of the CIM system is verified
95
Figure 18. Urban area model
Table 3. Energy demand for urban model
in analysis. However, in the analysis, the power

demand model of a representative day in May of
Figure 17 is used. The analysis using the power
demand model of the representative days of other
months is the same as that of the example of a
representative day in May, and other months are
not analyzed in this chapter.
96
Complex Community
1. Grid Route and Generation Capacity of
FC and DEG
Figure 19 (a) shows the rate of the power demand
of a representative day in May in the urban area
model of Case A of Table 3(a). A representative
day shows the greatest power demand for convenience stores (two buildings), followed by hotels,
factories, and small hospitals, in that order. As
Figure 17 describes, the difference in the power
demand for day and night is comparatively small at
convenience stores, hotels, and small hospitals and
it is large for small offices and factories. There is
a difference between family households and apartments in the amount demanded from midnight to
early morning, and daytime. In order to maintain
the high power generation efficiency of the whole
microgrid, it is necessary to plan the path of the
FC grid containing convenience stores, hotels,
factories, small hospitals, etc. with a large power
Figure 19. Analysis results in Case A
demand. The grid route shown in Figure 19 (b) was

obtained from the result of analyzing the power
generation efficiency of each grid route. Figure
19 (b) consists of three FC grids: Grid A, Grid
B, and Grid C. Figure 19 (c) shows the analysis
results of the generation capacity of FC installed in
each FC grid, and the generation capacity of DEG
corresponding to the base load. PEFC of 10 kW,
15 kW, and 47 kW is connected to each of Grid
A, Grid B, and Grid C, respectively. In addition,
57-kW DEG is installed and it corresponds to the
base load of the whole grid.
2. Result of Power Generation Efficiency
Figure 19 (d) shows the analysis result of the
electric energy to be outputted on a representative
day in May by each FC grid. In each grid, there
are base power supplied from DEG and power
corresponding to the load fluctuation supplied
from FC. The base load of each grid distributes
the power outputted by DEG. The output of DEG
is larger than FC, removing Grid B by the building
composition of Case A. In addition, the power
supply of a representative day has more DEG
than the sum total of each PEFC. Figure 19 (e)
shows the analysis results of the power generation
efficiency of FC of each FC grid, and the power
generation efficiency of the whole grid. Although
the power generation efficiency of Grid A, Grid B,
and Grid C is 19.2%, 17.4%, and 18.6%, respectively, base-load operation is added due to DEG,
and the power generation efficiency of the whole
grid improves to 27.1%. Because there are two
or more buildings with the same power demand
model in the urban area model, the grid routes
shown in Figure 19 (b) differ, but there is a case
where Figure 19 (c), (d), and (e) show the same
results. Moreover, one set of DEG or one set of
PEFC is installed into the conditions of the urban
area model of Case A, and the analysis result of
the power generation efficiency of the system that
supplies the power demand of all the buildings
(central system) is shown in Table 4 (a). The power
generation efficiency of the DEG central system
97
and the FC central system is 22.4% and 26.2%,

respectively. Therefore, the CIM system of power
generation efficiency (27.1%) proposed in this
chapter is larger. The result of the load distribution of the whole microgrid of Case A is shown
in Figure 20 (a). In this figure, allocation of the
load of DEG and the load of the FC grid (Grid
A, Grid B, and Grid C) is shown. The magnitude
of the load during the time zone from midnight
to early morning and others differs greatly, and
the power generation efficiency of the FC grid
and the total efficiency of the microgrid (equal
to CIM efficiency) are influenced.
Figure 20. Load distribution of the entire microgrid
Residential Area
1. Grid Route and Generation Capacity of
FC and DEG
Figure 21 (a) shows the rate of the power demand
of a representative day in May in the urban area
model of Case B in Table 3(b). In Case B, family
households account for 18 buildings and convenience stores account for two buildings. However,
the power demand rate of convenience stores is
84%, and the power demand rate of family households is 16%. The grid route shown in Figure 21
(b) was obtained from the analysis result of the
Table 4. Result of power generation efficiency
power generation efficiency of each grid route.

Figure 21 (b) consists of two FC grids: Grid A
and Grid B. Figure 21 (c) shows the analysis
result of the generation capacity of FC installed
in each FC grid, and the generation capacity of
DEG corresponding to the base load. PEFC of 8
kW is connected to each of Grid A and Grid B,
and DEG of 33 kW is operated as a base load of
the whole grid.
2. Result of Power Generation Efficiency
electric energy to be outputted on a representative
day in May by each FC grid. From the analysis
result of Figure 21 (d), with the composition of
the buildings of Case B, the output of DEG is
overwhelmingly larger than FC, and the power
supply of a representative day has more DEG
than the sum total of each PEFC. Figure 21 (e)
shows the analysis result of the power generation
98
Figure 21. Analysis results in Case B
efficiency of FC connected to each FC grid, and

the power generation efficiency of the whole
grid. Although the power generation efficiency
of Grid A and Grid B is 19.5% and 14.5%, respectively, base-load operation is added due to
DEG, and the power generation efficiency of the
whole grid improves to 29.9%. In addition, one
set of DEG or one set of PEFC is installed into
the conditions of the urban area model of Case
B, and the analysis result of the power generation
efficiency of the system (central system) that
supplies the power demand of all the buildings
is shown in Table 4 (b). The power generation
efficiency of the DEG central system and the FC
central system is 23.2% and 29.1%, respectively.
Therefore, the CIM system (29.9%) of power
generation efficiency proposed in this chapter is
larger. The result of the load distribution of the
whole microgrid of Case B is shown in Figure
13 (b). There is little load distribution of the FC
grid compared with Case A, and the difference
in the load at each sampling time is small. This
is the reason that the power generation efficiency
(equal to the power generation efficiency of CIM)
of the whole microgrid is high.
Conclusions
CIM (Compound Interconnection Microgrid),
which divides the grid of an independent microgrid into multiple units and interconnects
between grids, was proposed. Although there
are many examples of introducing DEG (diesel
power plant generator) as cogeneration until now,
there is the problem of carbon dioxide emission.
Therefore, in this chapter, DEG was installed as
generating equipment corresponding to the base
load of whole CIM, and the method of operating
a proton-exchange membrane fuel cell (PEFC) so
that it may correspond to the load fluctuation of
the grid was investigated by numerical analysis.
The generation capacity and the grid route of
DEG and PEFC in the case of maximum power
generation efficiency of the whole microgrid were
99
sought. From this analysis result, an independent

microgrid with high power generation efficiency
was planned. As a result, compared with the
method of installing DEG or PEFC into a microgrid
independently (central system), power generation
efficiency is confirmed to have improved. Moreover, it has been verified that the power generation
efficiency of a microgrid, the number of FC grids
introduced, and the capacity of DEG and PEFC
change with the types of building that composes
an urban area. In the complex community model
and the residential area model investigated in this
chapter, the power generation efficiency of the
whole microgrid (CIM) was 27.1% and 29.9%,
respectively. Although the microgrid that combines DEG and PEFC is advantageous concerning
power generation efficiency and carbon dioxide
emission, studies including those on the increase
in equipment cost are now required.
REFERENCES
consumption in the house in the 2001 fiscal year,
673681. doi:10.1109/TIA.2005.847277
Hatano, Y., et al. (2003). Investigation on the
energy use characteristics in the apartment house
of hot water supply and heating of an area. In
Proceedings of the Meeting of the Society of Heating, Air-Conditioning and Sanitary Engineering,
(pp.1745-1748). In Japanese.
Ibaraki Prefecture Government Office of Education. (2002). Modeling of hydrogen energy system.
High school active science project research report.
Japan: Ibaraki. In Japanese
100
Ibe, S., et al. (2002). Development of fuel processor for residential fuel cell cogeneration system.
In K. Abe (Ed.), Proc. 21th Annual Meeting of
Japan Society of Energy and Resources, Osaka,
(pp. 493-496). In Japanese National Astronomical
Observatory. (2003). Rika Nenpyo. Chronological Scientific Tables CD-ROM. Japan: Maruzen
Co., Ltd.
(Masters thesis). Helsinki University of Technology, Finland. 58-79.
Nagase, O., et al. (2003). Quantification of the
energy consumed in a house. In Proc. 19th Energy
(pp 461-466). In Japanese.
Obara, S. (2006a). The exhaust heat use plan when
connecting solar modules to a fuel cell energy
network. Transactions of the ASME. Journal of
Energy Resources Technology, 129(1), 1828.
doi:10.1115/1.2424962
Obara, S. (2006b). Load response characteristics
of a fuel cell microgrid with control of number
of units. International Journal of Hydrogen
Energy, 31(13), 18191830. doi:10.1016/j.
ijhydene.2006.05.004
Obara, S., & Kudo, K. (2005). Study on improvement in efficiency of partial load driving of installing fuel cell network with water electrolysis
operation. Transactions of the Japan Society
237244. doi:10.1299/kikaib.71.237
Obara, S., & Kudo, K. (2005). Study of efficiency improvements in a fuel-cell-powered
vehicle using water electrolysis by recovering
regeneration energy and avoiding partial load
operation. Transactions of the ASME. Journal of
Fuel Cell Science and Technology, 2, 202207.
doi:10.1115/1.1928930
Obara, S., & Kudo, K. (2006). Route planning of

heat supply piping in a fuel cell energy network.
Transactions of the JSME. Journal of Environmental Engineering, 1(1), 1728. doi:10.1299/jee.1.17

urban energy. [in Japanese]. Transaction of the
Society of Heating, Air-Conditioning and Sanitary
Eng. of Japan, 79(7), 573579.
Robert, H., et al. (2004). Microgrid: A conceptual solution. In Proceedings of the 35th Annual
IEEE Power Electronics Specialists Conference,
6, 4285-4290.
Yoshinaga, M., et al. (2003). Investigation on the

energy consumption of a house, and a residents
consciousness. In Proceedings of the Meeting of
the Society of Heating, Air-Conditioning and Sanitary Engineering, (pp. 1729-1732). In Japanese.
Takeda, Y., et al. (2004). Development of fuel processor for rapid start-up. In K. Kimura (Ed.), Proc.
20th Energy System Economic and Environment
Conference, Tokyo, (pp. 343-344). In Japanese
101
APPENDIX
Nomenclature

n : Grid route
P : Generation capacity kW
QB ,t : Town gas quantity of flow for burner [kg/s]
QRM ,t : Town gas quantity of flow for steam reforming [kg/s]

t : Sampling time [Hour]
W : Objective function
WB : Town gas calorific valve supplied to a burner [kW]
WH : Calorific valve of the hydrogen included in reformed gas [kW]

WR : Town gas calorific valve supplied to a steam reforming [kW]
WP : Purifier water producing equipment
: Power generation efficiency [%]
RM : Reformer efficiency [%]
total : Total power generation efficiency [%]
Subscripts

102
Day : Representative day

DEG : Diesel engine generator
FC : Fuel cell
103
Chapter 4
Installation Plan of a Fuel

Cell Cogeneration System
ABSTRACT
This chapter consists of two sections, Installation Plan of a Fuel Cell Microgrid System Optimized by
Maximizing Power Generation Efficiency and Fuel Cell Network with Water Electrolysis for Improving
Partial Load Efficiency of a Residential Cogeneration System. A microgrid that use PEFC may significantly reduce the environmental impact when compared with traditional power plants. The 1st section
investigates what occurs when a set of PEFCs and a natural gas reformer are connected to the microgrid
in an urban area. In the 2nd section, a fuel cell energy network which connects hydrogen and oxygen
gas pipes, electric power lines and exhaust heat output lines of the PEFC cogeneration for individual
houses is analyzed.
The summary of the 1st section is as follows. If
energy-supplying microgrids can be arranged to
operate with maximal efficiency, this will have a
significant influence on the generation efficiency
of the grid and will reduce greenhouse gas production. A means of optimizing the microgrid needs
to be developed. Moreover, microgrids that use
PEFC may significantly reduce the environmental
impact when compared with traditional power
plants. The amount of power supplied to the grid

divided by the heating value of the fuel is defined
as the system generation efficiency. The authors
find that when a set of PEFCs and a natural gas
reformer are connected to the microgrid in an
urban area, the annual generation efficiency of
the system slightly exceeds 20%. When a PEFC
follows the electricity demand pattern of a house, it
operates at a partial load most of the time, resulting
in a low efficiency of the microgrid. A method of
improving the generation efficiency of a fuel cell
DOI: 10.4018/978-1-4666-5796-0.ch004
Installation Plan of a Fuel Cell Cogeneration System
microgrid is proposed, where a supply system of

power and heat with a high energy efficiency are
constructed. In this study, a method of installing
two or more microgrids is proposed (known as
the partition cooperation system). The grids can
be connected in an urban area to maximize generation efficiency. Numerical analysis shows that
the system proposed in this study (which has an
annual generation efficiency of 24.6 to 27.6%) has
a higher generation efficiency than conventional
PEFC systems (central generating systems have
annual generation efficiencies of 20.6 to 24.8%).
The summary of the 2nd section is as follows. A fuel cell energy network which connects
hydrogen and oxygen gas pipes, electric power
lines and exhaust heat output lines of the fuel cell
cogeneration for individual houses, respectively,
is analyzed. As an analysis case, the energy demand patterns of individual houses in Tokyo are
used, and the analysis method for minimization
of the operational cost using a genetic algorithm
is described. The fuel cell network system of an
analysis example assumed connecting the fuel cell
co-generation of five houses. If energy is supplied
to the five houses using the fuel cell energy network
proposed in this study, 9% of city gas consumption will be reduced by the maximum from the
results of analysis. 2% included to 9% is an effect
of introducing water electrolysis operation of the
fuel cells, corresponding to partial load operation
of fuel cell co-generation.
INSTALLATION PLAN OF A FUEL

CELL MICROGRID SYSTEM
OPTIMIZED BY MAXIMIZING POWER
GENERATION EFFICIENCY
Introduction
Microgrid systems used in urban areas have several
notable characteristics (Robert et al., 2004; Carlos
2005; Takuma & Goda, 2005). (a) The distance
104
between the heat-supply and heat-demand sites is

small, making it possible for exhaust heat to be used
in the process. (b) The systems can also be linked
with existing large-sized electric power facilities
to perform load-leveling. A proton-exchange
membrane fuel cell (PEFC) may extend as a power
plant. These fuel cells have the advantage that they
are highly efficient and have little environmental
impact. However, the electrode material (especially the catalyst material and the solid polymer
membrane) of the PEFC is expensive, and its
system is complex. It may be possible to reduce
the number of expensive fuel cells that need to be
installed by connecting the PEFC to a microgrid
and supplying power to two or more buildings.
If the energy of the overall grid is supplied by
one set of fuel cells (central system), the facility
costs will be reduced considerably. Past work has
examined the method of supplying power to a
water electrolyzer and hydrogen and oxygen fuel
storage methods (Obara & Kudo, 2005a). Another
study looked at controlling the number of units
that divide a fuel cell and a reformer, finding that
the system efficiency falls when operated at partial
load (Obara. & Kudo, 2005b). In addition, energy
storage methods, such as batteries and flywheels,
have been considered, though this equipment is
not introduced in this chapter. Energy storage
methods must consider power fluctuations in the
microgrid affecting how the fuel cell is controlled.
Although this is an important topic, for simplicity, this chapter focuses on other issues related to
microgrids. There are no examples of the effect
of the power demand pattern of buildings linked
to a microgrid on the generation efficiency of a
fuel cell system. This chapter will examine how
the overall generation efficiency is affected by
connecting a building to another grid. In this
chapter, the fuel cell microgrid (FC microgrid) is
installed in an urban area and divided into multiple
grids. The system efficiency is improved over the
case where each grid is connected independently
(partition cooperation system). By dividing the
grid and increasing the load factor of PEFC linked

to each grid, the proposed method improves the
generation efficiency of the overall grid.
As a case study, this chapter investigates the
generation efficiency of the FC microgrid using the
power demand model assuming typical buildings
in Tokyo (individual houses, apartments, hotels,
convenience stores, small offices, factory, and
small hospitals).
Interconnected and
Independent Microgrids
Figure 1 (a) shows the interconnected microgrid
system. Figure 1 (b) shows the independent microgrid system. The system interconnection
method allows interconnections between grids
using system interconnection equipment ( mark
of Figure 1 (a)), such as a commercial power
system. Significant load fluctuations occur when
the demand of the grid cannot be satisfied; in
these situations it is possible to supply power from
other grids. Even when the number of buildings
linked to the grid increases or decreases, the
system interconnection method sees little effect
since it is inter-connected with other grids. On

the other hand, the independent microgrid must
supply power to all of the buildings linked to the
grid. When supplying power using many independent microgrids, it is necessary to deal with
the load-leveling problem. Correspondingly, the
maintenance needs for a long-distance power
network will be reduced. Moreover, if the energy
demand pattern of an urban area can be predicted,
the facility can be optimized, further reducing
facility and maintenance costs for the microgrid.
Compared with constructing an independent
microgrid, implementing an interconnection
system is relatively easy. However, many unknowns remain in assessing the generation efficiency of the independent microgrid. Therefore,
this chapter characterizes the generation efficiency of microgrids by investigating three electric power supply methods for the independent
microgrid described below.
Introduction to Fuel Cell Operation

A schematic diagram of the stand-alone system
(Figure 2(a)), the central system (Figure 2 (b))
and the partition cooperation system (Figure 2
(c)) for the independent microgrid are shown in
Figure 1. Microgrid model
105
Figure 2. Proposed independent microgrid
Figure 2. In the stand alone system, a separate fuel

cell is installed in each building. Since the power
demand pattern is different for each building,
the fuel cell capacity is determined based on the
predicted maximum value of the power demand
pattern for the proposed building. The facility
scheme for the stand-alone system is shown in
Figure 3 (a). A natural gas reformer, dryer, carbon
106
monoxide oxidization equipment, fuel cell, heat

storage tank, and boiler, etc. are installed in all
buildings. The power produced by the fuel cell
system is consumed only by the building in which
it is installed.
A schematic diagram of the central system is
shown in Figure 2 (b). In this system, power is
supplied to a single grid, which connects all the
Figure 3. Proposed system
buildings using a single set of fuel cells. In Figure

2 (b), the facility shown in Figure 3 (b) is installed
in Building 7. The power that is produced, unlike
stand-alone system described above, can be supplied to other buildings via a grid. Buildings
other than Building 7 install the facilities shown
in Figure 3 (d). In the facilities shown in Figure
3 (d), power is received from a microgrid through
an interconnection device, and heat is supplied
from a natural gas boiler.
In the partition cooperation system, as shown
in Figure 2 (c), power demand is divided into
two or more independent grids (in the case of
Figure 2 (c), these are Grids A and B). Although
installation of more fuel cells increases the partition cooperation system, the output point of each
fuel cell can be chosen to optimize generation efficiency (cooperative operation). The grid of the
partition cooperation system selects a grid route
in order to maximize the generation efficiency
of the overall microgrid. Given grid route and
fuel cell capacity may increase the load factor
of each grid, thus generation efficiency may be
higher than in the stand alone or central systems.
The partition cooperation system uses central

system shown in Figure 3 (b) in each grid. DPS,
distributed installation system of the partition
cooperation system, is one method. An alternative
model has the installation shown in Figure 3 (c) in
one building and the load distribution control in
another building (CPS: concentrated installation
method of the partition cooperation system). The
method for installing the fuel cells for the partition cooperation system shown in Figure 2 (c) is
DPS (shown in Figure 2 (e)). The CPS system
for the set up in Figure 2 (d) is shown in Figure 2
(f). In DPS, power generation equipment that is
identical to the central system is installed in each
grid. On the other hand, for CPS the power for
FC (1) and FC (2) is supplied to the building by
Grids A and B, respectively, while the power for
FC (1) and FC (2) is supplied to each building by
the two grids shown in Figure 2 (c), or the central
system of Figure 2 (d). The partition cooperation
system includes features from both the system
interconnection microgrid and the independent
microgrid. As shown in Figure 2 (c), Grids A and B
can cooperate, but the overall grid is independent.
107
System Operating Method

Figures 3 (a) to (c) show schemes for the power
plant of a microgrid. In these power plants, natural gas is supplied to the heat-source burner of a
reformer so that steam reforming of natural gas
can be performed. The sulfur content of the Japanese natural gas supply is not included. Methane,
ethane, propane, and butane comprise 90%, 6%,
3%, and 1% of the total gas, respectively. In the
reformer, reformed gas with numerous hydrogen
constituents is produced. Since there is a lot of
water in the reformed gas of a reformer outlet, the
water is removed through a drying process. The
carbon monoxide in the reformed gas is removed
using a carbon monoxide oxidation system. If
reformed gas is supplied to a fuel cell, power and
exhaust heat will be outputed. The power produced
by the fuel cell is supplied to a building through
a DC-AC converter and an inverter (stand-alone
system), or through an interconnection device
(central system and partition cooperation system),
while the exhaust heat of the reformer and fuel
cell are stored in a heat storage tank. The heat
from the thermal storage medium is exchanged
for tap water. This hot water is supplied to a
back-up boiler, where the temperature of the hot
water supply is controlled and outputted on the
Figure 4. PEFC performance
108
demand side. When the heat output of the heat

storage tank is less than the demanded amount,
the back-up boiler begins operation. When the
exhaust heat input exceeds the heat storage capacity, heat is released using a radiator. The fuel cell
of Figures 3 (a) through (c) control the amount
of natural gas supplied to the reformer to track
the power load of the grid. Figure 3 (d) shows a
model of buildings that do not have generating
equipment for the central system or the partition
cooperation system installed in them. In these
buildings power is obtained from the microgrid
via an interconnection device.
Power Model of PEFC and the

Generation Efficiency of the Grid
Figure 4 (a) shows the PEFC with the natural gas
reformer performance curve for a fuel cell stack.
The systems operating temperature is 333 K,
with gas pressure at the anode and cathode at 0.1
Mpa (Takeda, Y. et al., 2004, Ibe, S. et al., 2002,
Mikkola, M., 2001, Ibaraki Prefecture Government Office of Education, 2002). The current
density and voltage of a single cell are shown in
Figure 4 (b). The generation efficiency of a fuel
cell system changes with the load factor. In this
chapter, the curve shown in Figure 4 (a) is used
to describe the relationship between the load factor and the generation efficiency of the fuel cell
with a reformer.
Figure 5 shows the power demand and total
power models for Buildings A, B and C. In a
stand-alone system, the capacity of the PEFC
installed in each building is chosen, and it may
exceed the maximum power load of Buildings A,
B, and C from Figure 5. Since there is no other
power supply, the supply can become intermittent
when the load exceeds the design capacity.
In the central system, the power load of Buildings A, B, and C from Figure 5 are summed at
every sampling time point. The capacity of one
set of PEFCs is determined to exceed the maximum
value over the course of the simulation. In the
Central system, power generation equipment is
installed in a single building connected to a microgrid. Although this configuration is advantageous in terms of facility cost, the distance the
cost heat must be transported is considerable,
releasing heat and reducing efficiency. Moreover,
if the number of buildings linked to the grid or
the power demand model change, the capacity of
the fuel cell system will remain unchanged. As a
consequence, extending this type of facility is
problematic.
In the partition cooperation system, for example, Buildings A and B from Figure 5 are
Figure 5. Fuel cell capacity
connected by Grid 1, and Building C is connected

to Grid 2. The capacity of each fuel cell in Grids
1 and 2 is determined so that it may exceed the
maximum value of the power load sum for every
sampling time point for Buildings A and B, and
the maximum value of the power load of Building
C. Although a power plant is installed in two or
more buildings is DPS, there is one CPS.
Power Demand Models

and Analysis Methods
Power Demand Model
Figure 6 shows the power demand model for each
Tokyo-style building used in the analysis. The
plots show the mean power load at each sampling
time point for representative days in January
(winter), May (mid-term), and August (summer)
(Architectural Institute of Japan, 2002; Nagase et
al., 2003; Hatano et al., 2003; Yoshinaga et al.,
2003). However, the actual power demand pattern
is an assembly of loads that can change rapidly,
e.g. via an inrush current. In Tokyo, the annual
average temperature over the past five years has
been 289 K. The average temperature in January
is 279 K with the highs and lows of 283 K and
275 K. The average temperature in May is 292
K with high and low temperatures of 296 K and
288 K. For August, the average is 298 K, and the
high and low records are 302 K and 296 K (National Astronomical Observatory, 2003). Since
the room-cooling load of an air-conditioner is
included in the power demand for a representative
day in August, the power demand is considerably
higher than in other months.
Forecasting the power demand for the houses
shown in Figure 6 (a) to (d) is difficult. Power
demand forecasts are more straightforward for
small offices and factories like those shown in
Figures 6 (g) and (h). Although short timescale
load fluctuations are not taken into consideration
for the power demand model in this chapters
analysis, when accompanied by load fluctuations
109
Figure 6. Power consumption models
it is important to investigate the dynamic characteristics of a grid. The power demand pattern of
a house (from Figures 6 (a) to (d)) peaks in both
the morning and afternoon. A hotel (Figure 6 (e))
has a stabilized demand when the time points
from midnight to early morning are removed.
Convenience stores have a continuous power
demand (Figure 6 (f)) because their business
activities are conducted 24 hours a day. The difference between the period of night to early morning, which has little power demand, and the period of morning to evening, which has a
110
considerable power demand, is clear in offices

(Figure 6 (g)), factories (Figure 6 (h)), and hospitals (Figure 6(i)).
Figure 7 shows a model of the urban area that
was used for analysis. The type of power demand
model used for the 23 buildings is also listed. The
numbers shown in the figure correspond to Table 1,
in which the floor area and the assumed maximum
load for each building are listed. During analysis,
the urban area model from Figure 7 was used with
the power demands shown in Figures 6 (a) to (i).
The relationship between the load factor of PEFC
Figure 7. Arrangement of buildings
Table 1. Maximum power load for each building
and the generation efficiency of the model is shown

in Figure 4 (a). Since the maximum generation
efficiency of the model of Figure 4 (a) is 32%, the
maximum generation efficiency of the microgrid
analyzed is theoretically 32%.
Analysis Method
The steps used to calculate the generation efficiency of the FC microgrid are shown in Figure
8. First, the power demand model of the buildings
111
Figure 8. Analysis flow
all permutations of the number of buildings. The

power load and load factor of the fuel cell for each
grid route are calculated for every sampling time.
The capacity range of the fuel cell installed in each
route of the FC microgrid is calculated beforehand,
while the generation efficiency of this grid route
is calculated using these load factors to find the
approximate expression described in Figure 4
(a). Next, the capacity of the fuel cell installed in
each grid route of the FC microgrid is changed,
and the generation efficiency of the grid route is
calculated using the same procedure. After all the
grid routes are searched by the calculation, the
grid route with the optimal generation efficiency
is selected. The generation efficiency of the FC
microgrid is determined for a representative day
(defined as the average generation efficiency of a
representative day). Using the procedure described
above, it is possible to calculate the average generation efficiency of a representative day in a month
using the stand-alone system, central system, and
partition cooperation system models.

Stand-Alone System
in the urban area model is prepared for analysis.

These power demands are then inputted into a
program at every sampling time. The overall power
demand of all the buildings in urban areas is also
inputted into the program. Next, the program
searches for all the routes of the FC microgrid
that can be installed in the urban area model. This
route planning is accomplished by calculating
112
Figure 9 shows the analysis results for the representative average generation efficiency for the
stand-alone system for each building in the urban
area model shown in Figure 7. Although the average generation efficiency differs every month,
the efficiency for convenience stores peaks at
about 30%. In addition, the average generation
efficiency of a representative day in August, when
many room-cooling loads are high in hotels and
hospitals, finds the average generation efficiency
of other buildings to be less than 20%. Since the
distance for conveying exhaust heat is short, adding a stand-alone system to a building that has a
considerable heat demand is effective. When such
a system is installed in a house, its generation
efficiency is low.
Figure 9. Average generation efficiency of the

stand alone system
sentative days in January and May is 110 kW. On a

representative day in August, the optimal capacity
is 160 kW. The average generation efficiency of
are 25.8%, 26.1%, and 24.8%, respectively. Therefore, in order to install the central system in the
urban area model from Figure 7, a 160-kW fuel
cell system is required. When a 160-kW PEFC is
installed, the average generation efficiency of a
representative day in January and May is 20.6%
and 20.9%, respectively. Therefore, the generation efficiency of the FC microgrid is between
20.6% to 24.8%.
Partition Cooperation System

1. Distribution of Small-Capacity Fuel Cells
Central System
A. Average monthly generation efficiency
Figure 10 shows the relationship between the

capacity and the generation efficiency for the fuel
cell system when the central system is installed
in the urban area model shown in Figure 7. The
optimal capacity of the fuel cell system on repre-
PEFCs have a power generation capacity of 2 kW

or 5 kW. The way that the power is distributed to
the urban area modeled in Figure 7 is examined.
Figures 11 (a) to (c) show the analysis results for
the partition cooperation system when the PEFC is
Figure 10. Power generation efficiency of the central system
113
installed, assuming a power generation capacity of

2 kW. Figures 13 (a) to (c) show the analysis results
when the PEFC has a power generation capacity
of 5 kW. Two or more buildings with the same
power demand model are included in the urban
area model. Since these models are exchangeable,
there are other possible grid routes beyond those
shown in Figures 11(a) to (c) and Figures 13 (a)
to (c). The average generation efficiencies for
for the 2-kW PEFC are 23%, 22.7%, and 20.4%,
respectively. In the 5-kW case, they are 23.2%,
23.7%, and 21.6%, respectively.

B. Annual operation using the same grid
Since conditions in August require a considerable amount of power for room cooling, thus the
optimal grid route and fuel capacity results from
August are applicable to other months. Therefore,
the August PEFC grid route, with power generation capacities of 2 kW or 5 kW, was used in the
power demand model to calculate the generation
efficiency for January. These results are shown
in Figures 12 and 14. As shown in Figure 12, the
average generation efficiency for a representative
day in January is 19.9%, which is 2.5% lower than
in August. Therefore, if a PEFC with a power generation capacity of 2 kW is used in the urban area
model shown in Figure 7, power for 11 buildings
will be supplied by four grids, Grids A to D, and
the generation efficiency will increase from 19.9%
Figure 11. Analysis results for the partition cooperation micro-grid system using a 2kW PEFC
Figure 12. Analysis result for January, using the results from August for the grid configuration
114
Figure 13. Analysis results of the partition cooperation micro-grid system using a 5kW PEFC
Figure 14. Analysis result in January where the August grid is used (Figure 13(c)). A 5kW PEFC is
installed in each grid. Average generation efficiency is 19.2%.
to 20.4% throughout the year. In other words,

the average generation efficiency is about 20%.
Moreover, the results of Figure 14 show that the
average generation efficiency of a representative
day in January is 19.2%, which is 11.1% lower
than a representative day in August. With a 5-kW
PEFC, power for 11 buildings is supplied by three
grids, Grids A to C, and the generation efficiency
changes from 19.2% to 21.6% throughout the year.
2. Optimizing Grid Routes by Maximizing the
Generation Efficiency

A. Results of DPS
Figures 15 (a) to (c) show the grid combinations

with the highest average generation efficiency of
all the grids that were analyzed. However, since two
or more buildings have identical power demand

models, it is possible to exchange these identical
buildings to obtain the same result. Figure 16
(a) shows generation efficiency analysis results
for the grid route results shown in Figure 15 (c)
when the August power demand model is used
for a representative day in January. Moreover,
Figure 16 (b) shows the same analysis for May.
Compared with Figure 15 (c), the generation
efficiencies found in Figures 16 (a) and (b) are
lower for Grids A to C. The average generation
efficiency of the overall grid shown in Figure 16
(a) is 21.1%, and 21.3% for Figure 16 (b). Therefore, the average generation efficiency of the FC
microgrid, which is composed of a combinations
of the largest grid, varies from 21.1% to 27.6%
throughout the year. This range of this values
exceeds the average generation efficiency of each
115
Figure 15. Analysis results of the partition cooperation system (DPS)
Figure 16. Analysis result in January and May when using the grid routes from August (Figure 15(c))
months representative day when a 160-kW fuel

cell is used for the central system (from 20.6% to
24.8%) as shown in Figure 10.

B. Results of CPS
The DPS analysis results from Figure 15 (c)

show that the capacity of the fuel cell system
linked to Grids A and B is 45 kW, while for Grid
C it is 60 kW. The generation efficiency for a
microgrid with CPS is also investigated. Based on
the analysis result of Figure 15 (b), the capacity
of the fuel cell system for Grid A is 45 kW, and
the capacity of the fuel cell system for Grid B is
65 kW. Therefore, the 60 kW capacity of the Grid
C fuel cell system shown in Figure 15 (c) must be
replaced by a 65 kW system. Due to this replace-
116
ment, operation on a representative day in May

(where one grid is set at 45 kW and the other is set
to 65 kW, as shown in Figure 17 (b)) is calculated
using the fuel cell capacity of the microgrid on a
representative day in August (two sets at 45 kW
and one set at 65 kW, shown in Figure 17 (c)).
Furthermore, if Grids B and C from Figure 15
(a) are combined (25 kW + 38 kW), the power
can be supplied with the new specifications (one
set at 45 kW, and one set at 65 kW) of the grid
for a representative day in January, as shown in
Figure 17 (a). In this way, the average generation efficiency for a representative day of each
month can be achieved by replacing the capacity
of the fuel cell system as shown in Figures 17(a)
to(c). Therefore, the annual average generation
efficiency of CPS changes from 24.6% to 27.6%.
Figure 17. Analysis results of the partition cooperation system (CPS)
The value range increases from 21.1 to 27.6%

for DPS, and from 20.6 to 24.8% in the central
system. Accordingly, in the fuel cell microgrid
that connects all the buildings in the urban area
model from Figure 7, the CPS configuration has
the highest generation efficiency.
the Partition cooperation system is about 1/2 to

1/3 the number needed in the stand-alone system.
Even when small-capacity fuel cell systems are
used in urban areas, the stand alone system is at
a disadvantage because more systems are needed
than in the partition cooperation system (DPS).
3. Number of Installed Fuel Cells in the StandAlone System and the Partition Cooperation
System (DPS)
Figure 18 shows the power demand model for
a representative day each month. The analysis
results for the number of fuel cells (1 kW and
2 kW) used in the stand-alone system, and with
the partition cooperation system (DPS) are also
shown. The number of fuel cells necessary for
Conclusions
Figure 18. Method and number of units installed
The optimal grid route and fuel cell capacity for

urban microgrids was investigated in order to
maximize a power grids generation efficiency. In
this chapter, fuel cell microgrids were categorized
into stand-alone systems, central systems, and
partition cooperation systems. All three system
types were investigated. The partition cooperation
system used both the DPS (distributed installation method of the partition cooperation system)
and CPS (concentrated installation method of the
partition cooperation system) methods. After calculating the generation efficiency for each type of
microgrid by numerical analysis of urban Tokyo
settings, the following conclusions were reached:
1. When the stand-alone system is installed in a
house, the generation efficiency is less than
20% throughout the year. On the other hand,
the generation efficiencies of convenience
stores and hotels, with 24-hour business
activities and relatively constant power
demands, are high compared with houses.
August, in particular, has many room-cooling
117
loads. Thus, the load factor of the fuel cell

rises and the generation efficiency for this
month is high.
2. The generation efficiency of central system for the FC microgrid is 20.6 to 24.8%
throughout the year. Central systems, especially when used in houses, can operate with
generation efficiencies that are superior to
those obtained with the stand-alone system.
3. The generation efficiency of the DPS
microgrid, which uses a grid route that
optimizes the generation efficiency, peaks
at 21.1% to 27.6% throughout the year.
4. The generation efficiency of the CPS microgrid is 24.6% to 27.6% throughout the
year. These value are the highest of all the
methods examined in this chapter.
Thus, this chapter proposes that CPS is the
best installation method for fuel cell microgrids.
These results take into account the number of
power plants with fuel cell microgrids, generation
efficiency, and grid safety (cases where significant
load fluctuations occur are more likely to fail).
FUEL CELL NETWORK WITH

WATER ELECTROLYSIS FOR
IMPROVING PARTIAL LOAD
EFFICIENCY OF A RESIDENTIAL
COGENERATION SYSTEM
Introduction
If the distance between a power station and an
energy consuming region is large, the equipment
cost, transportation losses of electric power and
thermal energy losses become important. It is more
effective to install small cogeneration systems
(CGS) near the consumer location to overcome
these problems. The prediction of changes in
energy demand will be difficult in an individual
house, therefore partial load operation will occur
frequently for the operation of the CGS, and the
118
efficiency of the whole system will fall (Obara and

Kudo, 2000, 2003, 2005). Moreover, the distribution to the house of the fuel cell CGS with city gas
reforming device (reformer) is considered (Bauen
& Hart, 2000; Lariviere et al., 1998). Since fuel
cells differ greatly in dynamic characteristics of
temperature control from reformers, operational
plans become complicated (Yerramalla & Feliachi, 2003; Hamelin et al., 2001; Hatziadoniu
et al., 2004; He, 1995). In one example of an
examination of a fuel cell CGS, it took an hour
or more to generate stable hydrogen continuously
from a cold start (Krist, 1999; Nagata et al.,
1999). Thus, the response of the load follow-up
characteristics in city gas reformers is greatly
inferior to the response of the load follow-up
characteristics of a fuel cell. If the operation of
a reformer is started and stopped, consumption
of fuel is followed on these operations. In this
case, a hydrogen gas storage cylinder should be
introduced, the number of starts and stops of
the reformer should be reduced, and operational
planning of the fuel cell and the reformer should
be performed separately. Considering distributing
single fuel cell CGS in a residence, solutions to
the technical problems of (a) the system efficiency
falling by frequent occurrence of partial load
operation, and (b) requiring complicated operational planning due to the differences between
the dynamic characteristics of the reformer and
the fuel cell, are required.
In this chapter, we try to solve these problems
using a network of fuel cells. Operation optimization of a fuel cell network system is analyzed
using genetic algorithm. The genetic algorithm
can be used for a nonlinear problem and it is the
characteristic further that many variables are simultaneously analyzable (Obara & Kudo, 2000,
2003, 2005). In the fuel cell network system, fuel
systems (Hydrogen and oxygen gas piping) and
output systems (power lines and hot water piping) of distribution of fuel cell CGS are joined,
and carry out cooperative operation of all the
devices linked to such a network, for the primary
objective of minimizing fuel cost. For two or

more fuel cells linked to the network, if the fuel
cell group which carries out power generation is
operating in a partial load, the quantity of power
generation will be increased so that the fuel cell
group can operate at a higher efficiency. The
water electrolysis operation is performed by fuel
cells other than the fuel cell group which generates power. In this case, the hydrogen and oxygen
gases which are generated are stored in cylinders
throughout the network. These are consumed by
power generation of arbitrary fuel cells at arbitrary times. Although there is one previous case
study of a fuel cells network operation (Padull,
2000), descriptions of the measure of partial
load operation, and operational planning of fuel
cells and the reformer were not included. In this
chapter, the hypothetical effects of introduction
of fuel cell CGS network into individual houses
in Tokyo is investigated from an operational planning perspective for the objective of minimizing
operational cost, in other words, for minimizing
the consumption of city gas.
The Fuel Cell CGS Network Model,

and Energy and Substance Balance
The Network Model
The network model of the fuel cell CGS installed
in individual houses, as assumed in this chapter,
is shown in Figure 19. The fuel cell CGS installed
in each house is connected with hydrogen gas
system piping, oxygen gas system piping, an
electric system power line, and hot water piping
of an exhaust heat system. However, all the fuel
cells linked to the network are proton-exchange
membrane types. The generation of hydrogen gas
by water electrolysis and oxygen gas is possible for
all fuel cells linked to the network. The operating
temperature of the fuel cell is 330K to 345K, and
the pressure of an anode is 0.1 to 0.5 MPa, and
when supplying air to a cathode by a blower, it
is an atmospheric pressure mostly. The hot water
system performs heat recovery from fuel cells
and supplies thermal energy to each house. Hot
water flows in the one direction, as shown by the
Figure 19. Fuel cell network system model
119
arrows in Figure 19. The cylinders which store

hydrogen gas and oxygen gas are installed in the
machinery room of house 1, and both gases are
generated when carrying out water electrolysis
operation of the fuel cells linked to the network.
Both gases are assumed to be pressurized to
1.0MPa with compressors and then stored in the
cylinders. The city gas reformer is also installed
in the machinery room, and hydrogen gas can be
supplied to the network at arbitrary times. On
the other hand, it is necessary to synchronize the
operation of a fuel cell and a reformer for a system that combines a single fuel cell and a single
reformer, as in a conventional system. Moreover,
the standby time taken up by the reformer prior
to starting and stopping, and the poor follow-up
characteristics at the time of a load change, are
problems. The network system has a specific
method of storage of the hydrogen gas and the
oxygen gas, and for connecting the fuel cells to
the reformer through the network. Therefore, the
fuel cells and the reformer can be independently
employed. A thermal storage tank and a backup
boiler can also be installed in the machinery room,
and heat can be supplied to each house at arbitrary
times through the hot water system network.
The electric power used by the water electrolysis operation of fuel cells is supplied by
electric power generated by other fuel cells through
the power line. The hydrogen and oxygen gases
consumed by the fuel cells for power generation
are supplied by the water electrolysis operation
in other fuel cells, or are stored in cylinders.
However, since there is no generation method for
oxygen when generating hydrogen gas by the
reformer, air is supplied to the fuel cells. The
power generation characteristics of the fuel cells
differ greatly depending on whether oxygen gas
or air is supplied to the cathodes of the fuel cells.
This will be described further in later section.
120
The Improvement-in-Efficiency
Measure of Partial Load
Figure 20 shows the network models for the electric power line and the hot water line. The variables,
which indicate the efficiency of each electric
power and thermal energy device, are described
in Figure 2 (a) and Figure 20(b). The power generation efficiency Conv ,t of the conventional
system (combination of single fuel cell and single
reformer) in the sampling time t ( = 0, 1, 2, , R ,
operation period of the system which determined
R beforehand) is the product of the power generation efficiency fa ,m ,t of the fuel cell and the
efficiency r ,t of the reformer which supplies
hydrogen gas and air. Since the electric power
generation efficiency Net ,t for generating hydrogen and oxygen by water electrolysis of the fuel
cells follows the course of the city gas reforming,
fuel cell power generation (air supplied to the
cathode), the water electrolysis by the fuel cell,
and fuel cell power generation (oxygen supplied
to cathode), Net ,t can be calculated by the product:
r ,t fa ,m ,t el ,n ,t fo,m ,t
of each efficiency. Here, subscript t expresses
the arbitrary time containing the sampling time
t . Furthermore, to store the gases generated by
the reformer and water electrolysis operation of
fuel cells in cylinders by pressurizing with compressors, the electric power consumed by the
compression is taken into consideration in this
chapter. The water electrolysis operation of some
fuel cells linked to the network becomes advantageous from the viewpoint of the quantity of
electric power consumed by the gas compression
Figure 20. Conversion efficiency of fuel cell network
and power generation efficiency. This is the case

where the operation methods in Net ,t exceeding
er hydrogen gas consumption of the fuel cells (

M sets) for power generation thanQa ,H ,t , the
Conv ,t exists, when the value of Conv ,t falls by

partial load operation.
reformer is operated and city gas is reformed. The

quantity of city gas flow at this time is Qr ,t , and
the hydrogen gas ingredient in reforming gas is
the quantity of flow Qr ,H ,t .
Energy Flow
Figure 21 shows the energy flow in the sampling

time t of the fuel cells ( M sets) which carry out
the power generation operation and the fuel cells
( N sets) which carry out water electrolysis operation, linked to the network. The hydrogen and
oxygen gases (quantity of flow is Q f ,H ,t and
2
Q f ,O ,t ) generated by water electrolysis of fuel

2
cells are stored in separate cylinders through the

networks of hydrogen and oxygen lines, respectively. At this time, the quantity of each gas to be
stored from time period t to t + 1 is set to BH ,t
2
and BO ,t . Furthermore, quantity of hydrogen and

2
oxygen gases flowing at the time of supplying the

fuel cells, which carry out the power generation
operation from cylinders through the networks
are set to Qa ,H ,t and Qa ,O ,t . When there is great2
The electric power output from the M sets of

the fuel cells covers the total electricity demand
Et of each house, the power consumption:
Eel ,n ,t (n = 1, 2,..., N )
of N sets of the fuel cells which carry out water
electrolysis, and the power consumption (in this
chapter, the electric power EComp,H ,t and
2
EComp,O ,t consumed by the compression of

2
hydrogen and oxygen gases is taken into consideration) of the auxiliary machines of the network
through the electric power system network. With
regards to heat energy, the exhaust heat:
H f ,m ,t (m = 1, 2,..., M )
121
Figure 21. Energy flow
of M sets of fuel cells, the quantity of thermal

storage Bh ,t in a thermal storage tank, and the
heat H b,t of the backed boiler by city gas (quantity of flow is Qb,t ) provide the heat demand H t
through the hot water network.
Energy and Substance Balance

The electric power balance in the sampling time
t is shown by the following equation of the fuel
cell CGS network system, constituted from fuel
cells which perform N sets of water electrolysis,
and fuel cells which perform M sets of power
generation.
M
E
m =1
f ,m ,t
n =1
q =1
= Et + Eel ,n ,t + Esub,q ,t
(1)
The left side of this equation expresses the

gross total power output of the DC-AC converter,
of M sets of fuel cells which carry out the
power generation operation. The 1st term of the
right-hand side is the supply to houses through
122
the network based on demand, the 2nd term represents the power consumption for the operation
of the fuel cells in which N sets of water electrolysis are performed, and the 3rd term expresses the power consumption of Y sets of the auxiliary machines (pump in the hot water system
network, compressors of hydrogen and oxygen
gases, etc.) for maintaining the network.
Furthermore, the heat balance of the system
can be shown to be the following equation.
M
H
m =1
N
f ,m ,t
+ H S ,t + H b,t = H t
M
+ H el ,n ,t + H f ,i,t
n =1
(2)
i =1
The left side is the overall heat output of the

system. The 1st term expresses the heat output by
M sets of fuel cell CGS, the 2nd term expresses
the heat output of the thermal storage tank, and
the 3rd term expresses the heat output of the
backed boiler. The right-hand side expresses
consumption of heat, wherein the 1st term is the
heat output supplied to a house based on demand
through the heat system network, and the 2nd

term expresses the heat consumed by N sets of
the fuel cells which perform water electrolysis.
The 3rd term expresses heat loss from the heat
system network (i.e. the hot water piping). This
quantity is proportional to the distance li k from
the house (number i ) which carries a heat output
to the house (number k ) which receives a supply
of heat from the network, and is defined by Equation (3) shown below.
M
H
i =1
f ,i ,t
= li k
(3)
in this equation is a coefficient based on the

quantity of heat loss per unit length of hot water
piping which, in the case study in later section,
assumes that thermal insulation material is used
for keeping the general piping warm, and a temperature difference between hot water and outside
air is assumed to be 40K and is given by 15
W/m.
Equation (4) shows the hydrogen balance in
the system. The 1st term on the left side is the
hydrogen gas generated by the fuel cells while
performing N sets of water electrolysis, the 2nd
term expresses the quantity of hydrogen gas flow
supplied by the cylinder, and the 3rd term expresses the quantity of hydrogen gas generation
of the reformer. The right-hand side expresses the
amount of hydrogen gas consumed by M sets of
the fuel cells which carry out power generation.
The equation for oxygen gas balance in the system
is very similar to Equation (4).
N
Qel ,n,H
n =1
,t
+ Qa ,H
,t
+ Qr ,H
,t
= Q f ,m ,H ,t
m =1
(4)
The total quantity of the city gas Qc,t supplied

to t + 1 from the sampling time t in the system
is shown in Equation (5). The quantity of city gas
consumed by the reformer and the backed boiler

(Qr ,t and Qb,t , respectively) are added, and Qc,t
is estimated.
Qc,t = Qr ,t + Qb,t
(5)
The objective function of the system, and

calculation of the fitness value
The objective of the analysis is to search for
an operation pattern with minimum city gas consumption of the system for an operational period
from t = 0 to R . Therefore, the operational
planning represented by the value in the parenthesis of Equation (6) should be smaller for a
better solution. The genetic algorithm which is
one of the optimization techniques is introduced,
and an operational planning is performed. The
optimization analysis in the genetic algorithm
described in later section is also a solution with
a high degree of the fitness value of the genetic
algorithm.
R
minimize Qc,t
t =0
(6)
Device Characteristics and

Relational Expression
The Output Characteristics
of the Fuel Cell CGS
Figure 22 shows the results of performance measurement of the fuel cell developed at a Japanese
company (Mikkola, 2001; Ibaraki Prefecture
Government Office of Education, 2002). When
either oxygen gas or air is supplied to the cathode
of a fuel cell, a difference in catalyst reactions
arises and power generation characteristics differ
greatly. Consequently, the area (portion shown by
the straight line in Figure 22) where the electric
power output is proportional to load differs based
on whether oxygen or air is supplied to the cathode.
123
Figure 22. Cell performance
Figures 23 and 24 show the relationship between

the rate of load and power generation gross efficiency when supplying air and oxygen gas.
Water Electrolysis Operation of the Fuel

Cell
The relationship between power consumption and
hydrogen generation, when the fuel cells were
used for water electrolysis (electrode area being
approximately1m2) as seen in Figure 22, is shown
in Figure 25. Electric power consumed by the
water electrolysis operations is set to Eel , j ,t , the
quantity of hydrogen gas generated is set to
Qel , j ,H ,t , and the generation efficiency el of the
2
Figure 23. Output characteristics of a hydrogen-air fuel cell
124
Figure 24. Output characteristics of a hydrogen-oxygen fuel cell
Figure 25. Characteristics of water electrolysis of fuel cell
125
hydrogen obtained from the sampling time t to

t + 1 by water electrolysis of a fuel cell j is
calculated by the following equation.
Qel , j ,H
,t
Eel , j ,t Aj
FD Vj
el
(7)
The Output Characteristics of the City

Gas Reformer
The output characteristics of the city gas reformer
under development are shown in Figure 26 for a
Japanese manufacture company. The upper plot
in Figure 26 shows the relationship between the
ratio of load and ratio of reformation. The ratio of
reformation is low due to partial loading at about
25% or less. On the other hand, the rate of load
efficiency shows a stable rated output at 25% or
more. The lower plot of Figure 26 is the relation
between the quantity of consumption of the city gas
(methane gas is assumed in this chapter) and the
quantity of hydrogen generation of the reformer.
The Characteristics of Storage of

Hydrogen Gas and Oxygen Gas
Hydrogen and oxygen gases generated during the
water electrolysis operation of the fuel cells, and
the hydrogen gas generated by the reformer are
pressurized and stored in cylinders. The work of
Figure 26. Characteristics of city gas reformer
126
the compressor in this case is assumed to be

compressive work using an ideal gas, and is calculated by Equation (8). Here, c expresses the
whole compressor efficiency including inverter
controller loss, the power consumption in an
electric motor, power transfer loss, loss with insufficient air leaks, cooling loss, and other machine
losses. In the case study in later section, the
value of c is given as 0.5 (Kyakuno, 2003).
Moreover, both hydrogen and oxygen gases are
pressurized and stored to PComp,H = 1.0MPa
2
using compressors.
Lc,H
,t
= P U ,t ln(PComp,H / P ) / c
2
(8)
The Analysis Method of the

Operational Planning
The network is constituted from fuel cell CGS
which can choose between operations of power
generation or water electrolysis, and the operational planning of this energy system is analyzed
by the genetic algorithm.
The Chromosome Model

Figure 27 is the chromosome model introduced
into the genetic algorithm, which performs group
division of the gene model of a chromosome
model, and expresses the operation of all fuel cells

and the reformer. The following information from
the sampling time t to t + 1 is expressed with
the gene model: the information on fuel cells that
generate power and the selection of fuel cells
which carry out water electrolysis; information
on the amount of hydrogen gas generated in the
fuel cells which carry out water electrolysis; the
quantity of power generation of the fuel cells
which carry out power generation; and information on the quantity of hydrogen gas generated by
the reformer. The chromosome model of Figure
27 expresses the operation patterns of all fuel cells
and the reformer connected to the network in the
time interval of t + 1 from t . All the operational patterns of the operation period R are
decided by creating such chromosome models
with respect to each sampling time to
t = 0, 1, 2, , R . The chromosome models of all
such operation periods R of the system are called
an individual. This individual is generated in
large numbers at random, gene operations (the
reproduction, the selection, the crossover, the
mutation) are added under the objective function
shown in Equation (6), and it searches for the
optimal operational pattern with changing generations. The analysis flow using genetic algorithm
is shown in Figure 28.
Figure 28. Genetic algorithm
127
Details of the Analysis Method

Electricity demand by consumers and the quantity of electric power consumed within the fuel
cells, and the hydrogen and the oxygen gas compressor which carry out water electrolysis operation will be given by the energy balance Equation
(1), and the total quantity of power generation of
the fuel cell which carries out power generation
operation will be decided. When the fuel cells
which carry out power generation operation show
a plurality, the rate of power generated by each
fuel cell is assigned at random. City gas is consumed by the reformer(Qr ,t ) and backed boiler(
Qb,t ). The amount of hydrogen is calculated from
Qr ,H
,t
in Equation (4), and the amount of theories
of the city gas which a steam-reforming reaction

takes is calculated from this value. The amount
of theories of city gas is divided by the reformer
efficiency r ,t , and Qr ,t is obtained. Moreover,
if H b,t in Equation (2) is decided, H b,t will be
divided by boiler efficiency b,t and Qb,t will be
decided.
Case Study
System Outline
The operational planning of the fuel cell network
system for the five individual houses in Figure 19
is calculated using the energy demand model in
the individual houses in Tokyo (Nagase, 2003).
The horizontal axis of the figures under Figure
23, Figure 24, and Figure 25 is divided into two
or more domains, and the least-squares method
approximation formulas to the 4th clause are
generated for each domain, and the output characteristics of power generation operation and
the hydrogen generation characteristics of water
electrolysis operation of the fuel cell are given.
The horizontal axis of the figure under Figure 26
for the characteristics of the reformer was also
128
divided into two or more domains, to generate

least-squares method approximation equations.
The output characteristics of the backed boiler are
not based on the quantity of the hot-water supply,
but yields an efficiency of 0.75. The capacity of
the thermal storage tank and the gas cylinders are
considered to be unknown, and the hydrogen and
the oxygen gases are considered to be pressurized
to 1.0 MPa with compressors, and are stored in
cylinders.
Energy Demand Patterns

In this case study analysis, the interval of sampling
time was 1 hour, and R introduces the energy
demand pattern of an individual house in Tokyo
in every month for a representative day from 0 to
23 hours. However, if operational planning is
performed at intervals of sampling time from
several minutes to dozens of minutes, an analysis
is possible by applying the energy demand pattern
for short intervals. On the other hand, since
changes in demand such as the inrush current for
less than one second requires a special examination, it is not described in this chapter.
The analysis example in this chapter introduces
two energy demand patterns shown in Figure 29
(a) and (b). The axis of ordinate of Figure 29
expresses the quantity of the standard condition
of the town gas consumed in five houses in 24
hours. The demand pattern A (Figure 29 (a)) is
for the state in which five houses were simply
added and the energy demands of every house
was synchronized with the electric power and the
heat demand pattern of one individual house. The
demand pattern B (Figure 29 (b)) assumes the state
wherein the energy demand of an individual house
for five houses is carried out uniformly in a time
shift. In Figure 29, since the individual house is
assumed, there is much energy demanded at 17:00
to 21:00 when all families are at home. There will
be much power consumption of space heating and
cooling which uses an air-conditioner from 17:00
in 21:00. Moreover, since heat is used for hot-water
Figure 29. Energy demand patterns
supply and a bath, the demand patterns of electric

power and heat differ in Figure 11 (a). The period
of analysis stops, 24 hours after starting a system
at 0:00 of the energy demand pattern of Figure 29.
The standard energy demand pattern of individual house, the amount of demand which
shifted 1 and 2 hours before by the standard
demand pattern, and the amount of demand which
shifted 1 and 2 hours after were added, and the
demand pattern B was obtained. In the analysis,
the value of Figure 29 (b) is divided equally into
five parts, and the electric power and the quantity of heat demand of each house of the demand
pattern B are given. The electric power and the
quantity of heat demand are equalized, the demand pattern B has a small value of the maximum
load, and its demand change through the representative day is small compared with the demand
pattern A.
Analysis Parameters
In consideration of maintaining the diversity of
an individual group, the number of individuals
were 5500, the number of generations was 20, and
intersection probability and mutation probability
were set to 0.9, for the values of the parameters
introduced into the genetic algorithm, respectively.
The hot water temperature in piping is 333K. And
the fuel cell operation is following the electric
power load.
The Effect of Fuel Cell Network

Figure 30 (a) and (b) are the results of the city
gas consumption for a representative day of every
month with the energy demand patterns A and
B, respectively. The rate of annual reduction in
the quantity of fuel consumed when introducing
129
Figure 30. Comparison of operation cost
the fuel cell network system, when compared to

the quantity of city gas consumption of the conventional system is about 6% with the demand
pattern B, and about 7% with the demand pattern
A. Moreover, large reductions in the city gas
consumption were demonstrated when compared
with the conventional system for both the demand
patterns from May through October.
The analysis results of the operational planning
of thermal storage in each time interval of a representative day for every month are shown in
Figure 32. The quantity of thermal storage is
higher for each representative day from May to
October.
The ratio of heat demand to the electricity
demand of the energy demand pattern (F) introduced in Figure 31 in this analysis is shown. The
values of F from May to October when the reduc-
tion effect of the quantity of city gas consumption

of the network system is large are 1.5 or less. If
energy demand patterns, which have a high frequency of appearance of F 1.5 is applied, the
reduction effect of the driving cost of the network
system will become large. The value of F is determined by both the thermoelectric ratio of the
fuel cell CGS, and the relation of energy efficiency.
The Effects of Water Electrolysis

Operation of the Fuel Cell
From the results shown in Figure 30 (a) and (b),
the city gas consumption of the network system
in which water electrolysis operation of the fuel
cells is possible, shows a roughly 2% decrease by
the demand pattern A compared with a system
Figure 31. Ratio of electric energy demand to thermal energy demand
130
Figure 32. Quantity of thermal storage
which does not perform the water electrolysis.

The gas consumption shows about a 1% decrease
by the demand pattern B. The analysis results of
the demand pattern A is better than the demand
pattern B. The change in demand of electric power
is equalized in time shifts, and the demand pattern
B shows less number of times of appearance of
partial load. On the other hand, in the analysis of
the demand pattern A, the range of fluctuation of
load is large and there are a number of appearances of partial load. However, as Figure 23 and
24 showed, unless the load dropped to 50% or less,
the efficiency of the fuel cell did not decrease, but
in this analysis, the effect of water electrolysis
operation became small. When the capacity of a
fuel cell is large and the operation time in partial
load, for which load is much less than 50% is
long, the method of carrying out water electrolysis
operation using the fuel cells is available.
Conclusions
The analysis method of a system which connects
small fuel cell CGS installed in a house including
the piping, power lines, fuel system of hydrogen
and oxygen gases, and the output system of exhaust
heat and electric power, and carries out cooperative
operations for the purpose of minimizing operational cost, was described. This analysis method
was introduced into the energy demand pattern
of an individual house in Tokyo, and was applied
to the fuel cell network system consisting of five
houses. The following results were obtained.
1. When energy supply comes from the fuel

cell network, a 9% reduction in the quantity
of city gas consumption at the maximum
was obtained as compared to a conventional
system (the conventional system is a method
of energy supply with single fuel cell CGS
and a single reformer).
2. 2% of this is the method of storing, in cylinders, hydrogen and oxygen gases generated
by water electrolysis of the fuel cells through
the network, carrying out the time shift of
these, using at the time of power generation
of the fuel cells, and is an effect corresponding to partial load.
3. The amount of city gas consumed by a
proposal system has less ratio of electric
energy demand to thermal energy demand
in less than 1.5 fuel cell network than fuel
cells (conventional system) of distributed
installation.
REFERENCES
Architectural Institute of Japan, (2001). The nationwide research study concerning the energy
Bauen, A., & Hart, D. (2000). Opportunities
for fuel cell-based renewable energy supply in
decentralised applications. Renewable Energy,
4, 25512554.
131
673681. doi:10.1109/TIA.2005.847277
Hamelin, J., Agbossou, K., Laperriere, A., Laurencelle, F., & Bose, T. K. (2001). Dynamic behavior
of a PEM fuel cell stack stationary applications.
International Journal of Hydrogen Energy, 26(6),
625629. doi:10.1016/S0360-3199(00)00121-X
of hot water supply and heating of an area. In Proceedings of the Meeting of the Society of Heating,
Air-Conditioning and Sanitary Engineering, (pp.
1745-1748). In Japanese.
Hatziadoniu, J., Lobo, A., Pourboghrat, F., &
Daneshdoost, M. (2004). A simplified dynamic
model of grid-connected fuel-cell generators.
IEEE Transactions on Power Delivery, 17(2),
467473. doi:10.1109/61.997919
He, W. (1995). Modeling a reformer in fuel cell
power-generation systems. In Proc. ASME Heat
Transfer and Fluids Eng., 321, 317-325.
Ibaraki Prefecture Government Office of Education. (2002). Modeling of hydrogen energy system
(High school active science project research report). Ibaraki, Japan: Author. In Japanese
Ibe, S., et al. (2002). Development of fuel processor for residential fuel cell cogeneration system.
In K. Abe (Ed.), Proc. 21th Annual Meeting of
Japan Society of Energy and Resources, Osaka,
Japan (pp. 493-496). In Japanese.
Krist, K., Gleason, K., & Wright, J. (1999).
SOFC-based residential cogeneration systems.
In U. Stimming et al. (Eds.), Proc. 6th Int. Symp.
On Solid Oxide Fuel Cells VI, Pennington, NJ,
(pp.107-115).
132
Kyakuno, T., Hattori, K., Ito, K., & Onda, K.

(2003). Prediction of production power for
high-pressure hydrogen by high-pressure water
electrolysis. J. Institute of Electrical Engineers
of Japan, 124-B(4), 605.
Lariviere, E., Blanchot-Courtois, V., & Givois.
(1998). Simulation of a fuel cell in cogeneration.
Int. Gas Res. Conf., 4, 892-899.
Nagase, O., Ikaga, T., & Chikamoto, T. (2003).
Study on the effect of energy saving methods and
global warming prevention in Tokyo. In Proceedings of the 19th Energy System- Economics - Environment Conference, (pp. 461-466). i Mikkola,
M. In Japanese.
Nagata, Y., Hori, M., Sonai, A., & Shimotori,
S. (1999). Development of PEFC systems and
components at Toshiba. In I. Uchida et al. (Eds.),
Proc. 3rd Int. Fuel Cell Conference, Nagoya,
Japan, (pp. 69-72).
National Astronomical Observatory. (2003). Rika
Nenpyo, Chronological Scientific Tables CDROM. Japan: Maruzen Co., Ltd.
Obara, S., & Kudo, K. (2000). Study of small-scale
co-generation system for cold region houses. In
Proceedings of the 3-rd Int. Conf. on Cold Climate
Heating, Ventilating and Air-Conditioning, (pp.
353-358).
Obara, S., & Kudo, K. (2003). Operational optimization and scheduling of multi-kind small
capacity energy devices for cold region houses.
In Proceedings of 9th ITES, 1, 297-302.
Obara, S., & Kudo, K. (2005a). Study on improvement in efficiency of partial load driving of
installing fuel cell network with water electrolysis operation. [In Japanese]. Transactions of the
Japan Society of Mechanical Engineers, Series
B, 71(701), 237244. doi:10.1299/kikaib.71.237
Obara, S., & Kudo, K. (2005b). Study on smallscale fuel cell cogeneration system with methanol
fluctuation. Trans. ASME. Journal of Energy
Resources Technology, 2, 202207.
Padull, J., Ault, G. W., & McDonald, J. R. (2000).
An approach to the dynamic modeling of fuel cell
characteristics for distributed generation operation. IEEE Power Eng. Soc. Meeting, 1, 134-138.
6, 4285-4290.
Conference, Tokyo, (pp. 343-344). In Japanese.
urban energy. [In Japanese]. Transaction of the
Society of Heating, Air-Conditioning and Sanitary
Eng. of Japan, 79(7), 573579.
Yerramalla, S., & Feliachi, A. (2003). Modeling and simulation of the dynamic behavior of a
polymer electrolyte membrane fuel cell. Journal
of Power Sources, 124(1), 104113. doi:10.1016/
S0378-7753(03)00733-X
the Society of Heating, Air-Conditioning and Sanitary Engineering, (pp. 1729-1732). In Japanese.
133
APPENDIX
Nomenclature
Equipment Symbols

A : Electrode area [m2]

B : Quantity of storage [kg]
Bh : Quantity of thermal storage [kJ]
E : Electric power [kW]
E : Electric power consumption [kW]
F : The ratio of heat demand to the electricity demand of the energy demand pattern
FD : The number of Faraday (96500C/mol) [C/kg]
Gw : Power generation or water electrolysis selection switch

H : Heat [kW]
H : Heat consumption [kW]
L : Work [J]
l : Distance [m]
M : Sets of the fuel cell which carries out power generation operation
N : Sets of the fuel cell which carries out water electrolysis operation
P : Pressure [Pa]
P : Atmosphere pressure [Pa]
Q : Quantity of flow [kg/s]
R : Operation period of system [s]
RL : Ratio of load [-]
RR : Ratio of reformation [-]
T : Temperature [K]
t : Sampling time
t : Expresses the arbitrary time containing the sampling time t
U : Volume rate of flow [m3/s]
V : Voltage [V]
W : Number of the fuel cells linked to the network
Y : The number of the auxiliary devices which consume electric power
Greek Symbols
: The coefficient of the amount of heat loss per unit length of hot water piping (Equation (3))
[W/m]
: Efficiency
Conv ,t : Power generation efficiency of the conventional system
Net ,t : Electric power generation efficiency for generating hydrogen and oxygen by water elec-
trolysis of the fuel cells (= r ,t fa ,m ,t el ,n ,t fo,m ,t )
134
Subscripts

a : Gas cylinder
b : Backed boiler
c : City gas
el : Water electrolysis
f : Fuel cell
fa : The fuel cell which supplies air to cathode
fo : The fuel cell which supplies oxygen to cathode
m : The code of the fuel cell which carries out power generation operation
n : The code of the fuel cell which carries out water electrolysis operation
r : Reformer
s : Thermal storage tank
: Atmosphere
Equipment Symbols

C/O: CO oxidation system

FC: Fuel cell
R/M: Natural gas reformer
135
136
Chapter 5
Fuel Cell Microgrid with

Wind Power Generation
ABSTRACT
This chapter consists of two sections, Analysis of a Fuel Cell Microgrid with a Small-Scale Wind Turbine
Generator and Power Characteristics of a Fuel Cell Microgrid with Wind Power Generation. An
independent microgrid is expected to be effective in cutting greenhouse gas discharge and energy costs.
Therefore, the operating conditions of an independent microgrid that supplies power with renewable
power sources and fuel cells are investigated in the 1st section. In the 2nd section, the dynamic characteristics of fuel cell microgrid are investigated in numerical analysis, and the cost of fuel consumption
and efficiency is also calculated. In addition, the stabilization time of the microgrid and its dynamic
characteristics accompanied by wind-power-generation and fluctuation of the power load are clarified.
The summary of the 1st section is as follows. If
electric power is supplied using an independent
microgrid connected to renewable energy, it can
flexibly match the energy demand characteristics
of a local area. And an independent microgrid is
expected to be effective in cutting greenhouse gas
discharge and energy costs, as well as in eliminating the need for an emergency power supply
system. Since the output of renewable energy

is unstable, other energy equipment needs to
cover the stability of output. Thus, the operating
conditions of an independent microgrid that supplies power with natural power sources and fuel
cells are investigated. The operation conditions
of a fuel-cell independent microgrid with wind
power generation were investigated by numerical
analysis. Step loads and an apartment house power
load model were analyzed using the dynamic
DOI: 10.4018/978-1-4666-5796-0.ch005
Fuel Cell Microgrid with Wind Power Generation
characteristics of a fuel cell obtained from experiments. The output of wind power generation and
fuel cells is controlled by proportional-integral
control of an independent microgrid for rapid
power demand change.
An independent microgrid that connects with
renewable energy has the potential to reduce
energy costs, and reduce the amount of greenhouse gas discharge. However, the frequency and
voltage of a microgrid may not be stable over a
long time due to the input of unstable renewable
energy, and changes in short-period power load
that are difficult to predict. Thus, when planning
the installation of a microgrid, it is necessary
to investigate the dynamic characteristics of the
power. About the microgrid composed from ten
houses, a 2.5kW proton exchange membrane fuel
cell is installed in one building, and this fuel cell
operated corresponding to a base load is assumed.
A 1kW PEFC is installed in other seven houses,
in addition a 1.5kW wind turbine generator is installed. The microgrid to investigate connects these
generating equipments, and supplies the power to
each house. The dynamic characteristics of this
microgrid were investigated in numerical analysis,
and the cost of fuel consumption and efficiency
was also calculated. Moreover, the stabilization
time of the microgrid and its dynamic characteristics accompanied by wind-power-generation
and fluctuation of the power load were clarified.
ANALYSIS OF A FUEL CELL

MICROGRID WITH A SMALL-SCALE
WIND TURBINE GENERATOR
Introduction
As for the microgrid technique, a backup power
supply is effective in an emergency, at peak
cut of a power plant, and for the efficient use
of exhaust heat. It is predicted that an electric
power supply technique using a microgrid with
renewable energy is very effective in reducing

energy costs and cutting greenhouse gas discharge
(Robert, 2004; Abu-Sharkh et al, 2006; Carlos
& Hernandez, 2005; Takuma & Goda, 2005).
There are systems connected with other electricpower networks, such as commercial power, and
independent electric power supply systems in the
structure method of a microgrid. An independent
microgrid can reduce transport loss of electric
power and heat energy at any location, and may
serve as the future major energy supply system
from the viewpoint of energy security. Managing
an independent microgrid following local power
demand characteristics will greatly contribute to
reducing energy cost and environmental impact. It
is necessary to control the stable supply of electric
power of a microgrid so that power supply and
demand is balanced. The difference in the power
demand-and-supply balance of a grid affects the
power quality (voltage and frequency) (Carlos
& Hernandez, 2005). To connect natural power
sources, such as wind force and solar modules,
to a microgrid, it is necessary to install backup
energy equipment with quick response in the same
grid. However, the power demand of a house is
accompanied by load fluctuations over a short
time, which is often difficult to predict. An independent microgrid with natural power sources
requires a system that considers load fluctuations
and fluctuation in energy supply. As techniques of
controlling these fluctuations, there is a method
of installing accumulation-of-electricity equipment, and a method of connecting the microgrid
to other power grids (Alibhai, 2004). At present,
the method of installing large-capacity accumulation-of-power equipment is expensive. Moreover, with the method of connecting a microgrid
to a commercial power system, the installation
methods (capacity, location, equipments, etc.) of
a microgrid have limitations.
This study investigates the development of an
independent microgrid using renewable energy
(small wind power generator) with unstable output. In order to control the output fluctuation of
137
small wind power generation, a proton-exchange

membrane fuel cell (PEFC) is connected to the
grid. Dynamic characteristics when random power
load fluctuations are added to the grid for a short
time are investigated. From these results of the
investigation, the operation plan and operating
conditions at the time of operating an independent
microgrid with small wind power generation are
clarified. In the analysis, an independent microgrid
is introduced into a 4-story apartment house of 12
houses, and a 3-kW fuel cell is installed in each
floor. Furthermore, the microgrid with one small
wind power generator is analyzed. If there is little
power demand of a grid, the output fluctuation
of wind power generation will greatly affect the
power quality (voltage and frequency) of a microgrid. The transient response characteristics of
the power of the system is analyzed by MATLAB
(Ver.7.0)/Simulink (Ver.6.0) of Math Work. The
object of this study is to clarify the operation
conditions of a microgrid when it does not have
accumulation-of-electricity equipment, when it
does not connect with other grids, when load
fluctuations are taken into consideration, and
when renewable energy is used.
System Description
System Configuration and
Control Method
1. System Configuration
The model of an independent fuel cell microgrid
introduced into an apartment house of 12 houses
(House (1) -House (12)) is shown in Figure 1 (a).
A 3-kW PEFC, an inverter, and a system interconnection device are installed in a machine room on
the first floor to the fourth floor of the apartment
house. However, the dynamic characteristics of
the fuel cell, inverter, and interconnection device
(I/C) are the same for each piece of equipment.
Each fuel cell system of F/C (1) - F/C (4) consists
of a controller, fuel cell, inverter, and I/C. Moreover, the wind power system (W/P) consists of a
138
controller, an aerogenerator, an inverter, and an

I/C. An independent microgrid consists of three
lines, a power network, a town gas network, and
an exhaust heat network. In this Section, the
power network of these is called IPG (Independent
Power Grid). A town gas network is connected to
a reformer installed on the first floor. A gas tank
can be connected to a reformer with piping, and
reformed gas can be stored. The reformed gas
produced by a reformer is supplied to the fuel
cells of each floor by shifting time. Therefore,
the dynamic characteristics of the reformer do
not influence the characteristics of F/C (1) - F/C
(4). The power of a fuel cell is connected to IPG
through an inverter and a system interconnection
device. The power of IPG is assumed to be 50 Hz
in single-phase 200 V. Moreover, an aerogenerator
is installed near the apartment house, and the power
obtained by wind power generation is connected
to IPG through an inverter and a system interconnection device (W/P). The wind power generation
system composed of a generator, an inverter, and
a system interconnection device is expressed as
W/P. The capacity of wind power generation is
assumed to be 3 kW or less. Although this Section
investigates the difference in the power supply and
demand of IPG, it does not take the exhaust heat
network into consideration.
2. Control Method
Figure 1(b) is a block diagram of IPG introduced
into Figure 1 (a) composed of one wind power
generation system and four PEFC systems of F/C
(1) to F/C (4). C f 1 to C f 4 expresses the instantaneous power capacity of PEFC of F/C (1) to
F/C (4). When a short power fluctuation of second
1 or less is added to the fuel cell, the value that
can be generated with the fuel cell is defined as
instantaneous power capacity. In the case study
of following Section, the instantaneous power
capacity of each fuel cell of F/C (1) - F/C (4) is
3.3 kW, and the regular power capacity is 3 kW.
For the power supplied to IPG when the power
Figure 1. System configuration
139
demand fluctuates, there is a method of controlling the production of electricity of each fuel cell,
and a method of controlling the number of fuel
cells to be operated. If there is little power demand
of IPG, when controlling the production of electricity of each fuel cell, two or more fuel cells
may be operated at partial load with low efficiency. So, this Section examines the system that
controls the number of the fuel cells to be operated when the power demand of IPG fluctuates.
Thus, F/C (1) is a fuel cell operated corresponding to base load, and it controls the number of
operations of F/C (2) to F/C (4) by the magnitude
of the load. The power:
Ef 1 + Ef 2 + Ef 3 + Ef 4
produced by F/C (1) to F/C (4) and the power (
EW ) produced by wind power generation are
supplied to IPG, and power demand ( Ed ) is provided. The sum of the production of electricity of
F/C (1) to F/C (4) is the value except the power (
EW ) produced by wind power generation from
the power demand amount ( Ed ). Therefore, the
power balance equation of IPG is expressed with
Equation (1). The left-hand side of Equation (1)
represents each item in electric power supply, and
the right-hand side is the term for power demand.
The supply-and-demand difference (Err) of the
power is expressed with Equation (2). In this
system, F/C (1) is operated at base load, and when
the value of Err exceeds capacity C f 1 of F/C (1),
it starts operation of F/C (2). F/C (3) is operated
when the power supply is insufficient compared
with the demand using this operation method.
Thus, the number of operations of PEFC of F/C
(1) - F/C (4) is controlled by the amount demanded.
E f 1 + E f 2 + E f 3 + E f 4 + EW = Ed
140
(1)
Err = (E f 1 + E f 2 + E f 3 + E f 4 + EW ) Ed
(2)
As shown in Figure 1 (b), the operation of
each fuel cell is decided by If branch with the
magnitude of Err. The proportion-integration
(PI) controller (Controller (1) - Controller (4))
for controlling the production of electricity is
installed in F/C (4) from F/C (1). The dynamic
characteristics of the power generation of F/C (1)
to F/C (4) and W/P and the PI control parameters
beforehand set up by Controller (1) - Controller (4)
affect the dynamic characteristics of IPG. So, the
next Section describes the dynamic characteristics
of each controller and W/P.
Response Characteristics of Equipment

1. Power Generation Characteristics of Wind
Power
Figure 2 shows power output models of wind
power generation. In the case study, three maximum outputs, 1.0, 2.0, 3.0 kW, are used. Figures
2 (a) - (c) show the power output when supplying
the value decided at random between the maximum output from 0 to IPG through an inverter
and a system interconnection device at sampling
time intervals. Here, Sampling time expresses
the time scale used in analysis, and Assumption
time expresses the time on the representative day
to be simulated. The sampling time interval was
set to 0.333 s in Figure 2 and the case study. This
value corresponds to 1 minute in the Assumption
time. Following Section describes the dynamic
characteristics of the inverter and system interconnection device. The proposed model of this
study assumes an apartment house in an urban
area. As a technique for reducing greenhouse gas
discharge, planning to install a small-scale wind
turbine generator of several kilowatts in an urban
area is considered. The model of this Section
whose wind power generation capacity is 3 kW
Figure 2. Output power models of the wind power generation
is not the optimal system from the viewpoint of

economy. However, the reduction of greenhouse
gases and the development of renewable energy
are investigated.
2. Power Generation Characteristics of a Fuel
Cell
Figure 3(a) shows the test equipment of the power
transient response characteristics of a fuel cell
stack. A fuel cell stack was installed and surround-
ed with heat-insulating material, the operating

temperature was changed from 288 K to 358 K, and
the response characteristics of the cell stack were
examined. The results of the measurement of the
response characteristics when applying step input
of 45 W to PEFC for a trial are shown in Figure
3 (b). The ambient temperature was maintained
at 293 K, and reformed gas and air were supplied
and underwent experiment in a cathode and an
anode, respectively. An approximated curve of
141
Figure 3. Response characteristics of equipment here
142
the dynamic characteristics of PEFC was created

from the results of the measurement in Figure 3
(b), and the transfer function of the first-order lag
shown in the figure was obtained. Although the
transfer function is considered to be dependent on
the magnitude of the load, since this difference
was small in these test results, it is not taken into
consideration in the case study.
3. Output Characteristics of a City Gas Reformer
Figure 3 (c) shows the results of measuring the
relation between the load rate of a fuel cell, and
generation efficiency. Here, the ratio of maximum
production of electricity of a fuel cell to load is
defined as the load rate. The results of Figure 3
(c) are the characteristics of power generation efficiency when supplying reformed gas to a PEFC as
shown in Figure 3 (a), and further, taking inverter
efficiency into consideration. If the load rate of a
fuel cell is decided, the generation efficiency of
each F/C shown in Figure 1 (b) is calculable from
the characteristics of Figure 3 (c). The maximum
generation efficiency of each F/C is 32%.
4. Inverter and System Interconnection Device
The inverter is a voltage control type. A total of
120 ms is required to change the input power into
regular voltage and a regular frequency (less than
95% of a targeted value) (Kyoto Denkiki, 2001).
The value that expresses the transfer function of
this inverter with the first-order lag is shown in
Figure 3 (d). When switching the power with a
system interconnection device, the switching time
is about 10 ms (Kyoto Denkiki, 2001). However,
since it is necessary to control to synchronize the
frequency between the systems, the switching time
of the system interconnection device assumed
in this Section is set at 12 ms. As a result, the
transfer function of the system interconnection

device by the first-order lag is the value shown
in Figure 3 (d).
Analysis Method
Solver Used for Analysis
The dynamic characteristics of IPG are analyzed
using MATLAB (Ver.7.0) and Simulink (Ver.6.0)
of Math Work Co. Ltd. The solver used by numerical calculation is the positive Runge-Kutta method,
and decided with 0.333 s in Figure 2, which describes the sampling time interval. Analysis error
was managed within 0.01%.
Step Response Analysis of F/C

The response results of a step input of the power
load (1.0 kW and 2.7 kW) to F/C shown in Figure
1 (b) are shown in Figure 4. Since the response
characteristics of a fuel cell system changes by
the control parameters set up in the controller,
the PI control parameters have been changed and
analyzed in Figure 4. As for the steady-state error of 1.0 kW of loads, P =1.0 and I =1.0 are the
least. However, as for rising time, P =20.0 and I
=1.0 are the shortest. However, overshooting is
large at P =20.0 and I =1.0. When a load is 2.7
kW, compared with P =1.0 and I =1.0, the rising
time of P =20.0 and I =1.0 is short.
Determination of Step Response Error

and Control Variables
Figure 5 shows the supply-and-demand errors of
the step response of F/C obtained by the numerical simulation of Figure 4. The maximum of the
supply-and-demand error is decided by the mag-
143
Figure 4. Characteristics of F/C step response
144
Figure 5. Supply-and-demand difference of the step response of an F/C
145
nitude of a load inputted into F/C, and it is not

dependent on the control variables. However, as
for the additional value of a supply-and-demand
error, P =20.0 and I =1.0 are the least. Thus, in
the case study, considering the following characteristics and the supply-and-demand error of
load fluctuations, the PI control parameters of the
controller of F/C (1) to F/C (4) is set as P =20.0
and I =1.0, and is analyzed.
Case Study
Step Response Characteristics of IPG
In this study, the rate (percentage) of the capacity of the power generator equipment describes
load fluctuations. This is because the magnitude
of the inrush current of electric appliances has
not been widely investigated. Figure 6 shows the
analysis results of the response characteristics of
IPG when inputting a 10-kW load that has load
fluctuations into the IPG shown in Figure 1 (b).
The input model shown in Figure 6 (a) was created by adding random load fluctuations within
5%, 10%, and 20% to a 10-kW step input
at 0.5-s sampling intervals. The 5%, 10%,
and 20% load fluctuations are simulating load
fluctuations in 12 houses of an apartment house.
Figure 6 (b) shows the analysis results of a load
response of IPG. In addition, Figure 6 (c) shows
the results of the supply-and-demand differences
of IPG. Figure 6 (d) shows the analysis results
of the average efficiency of the system. The efficiency of each F/C is added, and this value that
is divided by the number of F/C operations is
defined as the average efficiency of the system.
Moreover, the efficiency of each F/C is calculable
by giving the load rate of the fuel cell to Figure 3
(c). Stable use of electric appliances requires 5%
of power in the range of fluctuation of the voltage
and frequency (standard of Japan). The microgrid
on the Response/Load shown in Figure 6 (c)
depends on the power quality (voltage, frequency).
146
Therefore, 5% of the zone is shown in Figure 6

(c). Frequency exceeding 5% of the Response/
Load increases as the range of load fluctuation
of Figure 6 (c) grows. From the result of Figure
6 (c), the width of load fluctuation shows that the
quality of the power up to about 10% is good.
The number of times that the supply-and-demand
difference in the power exceeds 95% - 100% in the
results of Figure 6 (c) is 0 to 2 times in 2 seconds
at 10% of the load fluctuation range. Similarly,
in a load fluctuation range of 20%, it is 2 to 3
times in 2 seconds. However, the variation range
of the supply-and-demand difference in the power
is 85% to 112%, when the load fluctuation range is
10%. Similarly, when the load fluctuation range
is 20%, it is 72% to 130%. The system average
efficiency of each load fluctuation range is shown
in Figure 6 (d) is less than 1% of the difference,
when the value of each sampling time is added
and compared.
Dynamic Characteristics of IPG with

Wind Power Generation
The 9-kW load with a load fluctuation range of
0%, 5%, 10%, and 20% is added to IPG. The
power is supplied to IPG by the system shown in
Figure 1 (b). The power is supplied to the grid from
the wind power generation of maximum output 1
kW or 3 kW. Figure 7 shows the analysis results
of the power supply-and-demand error of IPG at
this time. The error range is so large that the load
fluctuation range is large. Moreover, compared
with wind-generated power of 1 kW, with power
of 3 kW, the error range is large. The frequency
where the Error of Figure 7 (a) and (b) exceeds
5% increases as the load fluctuation grows. From
the result of Figure 7 (a) and (b), load fluctuation
is about 5%, and the power quality is good. The
large load fluctuation of wind power generation
affects the power quality (voltage, frequency) of
the microgrid irrespective of the power demand
amount.
Figure 6. Response results when adding 10 kW load with fluctuations to a microgrid. P=20.0, I=1.0.
147
Figure 7. Supply-and-demand error of the microgrid at the time of 9 kW of loads with load fluctuations
and wind power generation
148
Figure 8 (a) shows the relation among the load

fluctuation range, wind power generation, and
average supply-and-demand error of power. Figure 9 (a) was created from the analysis results of
Figure 6 and Figure 7, For example, wind power

generation of 3 kW, the maximum output, will be
connected to IPG, and 15% of load fluctuations
will be added to the grid. The average supply-and-
Figure 8. Characteristics of wind power generation
149
Figure 9. Outside air temperature and average power consumption
demand error of power is less than 3% in this case

from Figure 8 (a). The characteristics of the average supply-and-demand error of power in Figure
8 (a) depend on the range of the load fluctuation
rather than the amount of wind power generation.
Moreover, each characteristic of the average
supply-and-demand error of power is upward
slanting to the right in the graph of Figure 8 (a).
The main reasons are the rates of output of wind
power generation, and the load fluctuation range.
Thus, it is dependent on the rate (e f / EW ) of the
output width ( EW ) of wind power generation and
the load fluctuation range (e f ), which are shown
in Figure 8 (b). If the value of is the same when
the output width of wind power generation is large,
the average supply-and-demand error of power is
150
relatively small. When the output of wind power

generation is large (Figure 8 (b) upper), the average supply-and-demand error of power is less than
when the output of wind power generation is low
(Figure 8 (b) lower). The dynamic characteristics
of F/C can follow the output fluctuation of wind
power generation. However, it is difficult for the
output control of F/C to follow with accuracy load
fluctuations for about 0.5 seconds or fewer. An
increase in the output of wind power generation
will reduce the following error of load fluctuation.
Apartment House Model

The average temperatures in Sapporo for the
sampling time on representative days in February, May and August are shown in Figure 9 (a)
(National Astronomical Observatory, 2003).

There is no cooling load of summer in Sapporo.
Electricity demand includes that for household
appliances and electric lighting, and heat demand
comes from heating, hot water supply, and baths.
The average number of members in a household
is four persons. Figure 9 (b) shows the power
demand pattern at each time for the apartment
house composed of 12 houses, and used the
representative day in February and August in
Sapporo in Japan (Narita, K., 1996). The reason
for the few differences in the power demand of
both months is that the heating heat source of
winter (February) uses fossil fuel, and does not
use cooling in summer (August). Therefore, the
power demand pattern of Figure 9 (b) expresses
the power mainly consumed by electric lights and
household electric appliances. The power demand
pattern of the apartment house shown in Figure
1 (a) is used as Figure 9 (b), and the dynamic
characteristics of the power when supplying power
produced by wind power generation to IPG shown
in Figure 2 are analyzed using the block diagram
shown in Figure 1 (b). The efficiency of the fuel
cell system is calculated using the load rate of a
fuel cell as shown in Figure 3 (c).
shown in Figure 10 (b) are in good agreement, as

Figure 10 (c) shows. However, if the load of IPG
changes, overshooting and delay in rising will
appear. On the other hand, as shown in Figure
11 (c) and Figure 12 (c), there is fluctuation to
the power supply-and-demand difference in IPG.
The supply-and-demand difference in the power is
greatly influenced by the load fluctuation range of
IPG compared with the magnitude of the output
of wind power generation. This is the same reason
as explained in Figure 8 (b).
Response Characteristics of
Apartment House Model
The average whole system efficiency shown by

Figures 10 (e), 11 (e), and 12 (e) is 23 to 24%.
The average whole system efficiency does not
depend on the load fluctuation range or the output of wind power generation. Figures 10 (i), 11
(i), and 12 (i) show the change in the efficiency
of F/C (4). The operation of F/C (4) is a partial
load with low efficiency. Figure 9 (h), Figure 11
(h), and Figure 12 (h) show a change in the efficiency of F/C (3). Highly efficient operation of
F/C (3) is restricted to few periods, and most of
the time, there is partial load operation with low
efficiency. When the power generation capacity of
F/C per set is divided by increasing the installed
number of F/C and controlling the number of
Figures 10, 11, and 12 show the system response

characteristics of Figure 1 (a). The load fluctuation range of IPG and the wind power generation
output are variables. (a) in each figure is the load
model, and (b) is the response result. (c) is the
supply-and-demand difference in the power, (d)
is the number of operations of F/C, and (e) is the
whole system efficiency. (f) to (i) expresses the
generation efficiency of F/C (1) to F/C (4). Figure
10 shows analysis results when load fluctuation
and wind power generation are zero. The rate of
the load shown in Figure 10 and the response
Number of Operations of F/C

The number of operations of F/C shown in Figure
10 (d) is two sets, F/C (1) and F/C (2), constantly.
In this case, the two sets, F/C (1) and F/C (2), correspond to the base load. On the other hand, Figure
11 (d) and Figure 12 (d) need to control each F/C
at all times, in order to follow load fluctuations.
In the case of Figure 11 (d) and Figure 12 (d),
F/C (1) corresponds to the base load. Therefore,
the fuel cell capacity corresponding to the base
load of IPG differs according to the production
of electricity of wind power generation, as well
as load fluctuation range.
151
Figure 10. Analysis results of 12-house apartment load model. Wind power generator is not installed.
Load fluctuations are zero.
operations, the load rate of all the F/C increases

and the ranges of partial load operation with low
efficiency decrease. Therefore, if F/C is divided
into small capacity and the installed number is
increased, it is predicted that the whole generation
efficiency will improve. The rise in facility costs
is compared with the reduction in energy cost due
to efficiency improvement, and the capacity and
the installed number of F/C needs to be designed.
152
Conclusions
The operation conditions of an independent
microgrid connected to renewable energy (wind
power generation) with unstable output was investigated by numerical analysis. Two or more
proton-exchange membrane fuel cells were connected to the grid, the number of operations was
controlled, and a stable supply of power was at-
Figure 11. Analysis results of 12-house apartment load model. Wind power generator is 1 kW. Load
fluctuation is
tempted. As a result, it was found that the effect of

the characteristics of average supply-and-demand
error of power on the load fluctuation range is
greater than the output of wind power generation. This is because it is difficult to follow the
load fluctuation in a short time with accuracy,
although the dynamic characteristics of the fuel
cell system can follow the output fluctuation of
wind power generation. Furthermore, the average

of supply-and-demand error of power is so small
that the power supply by wind power generation is
large if the load fluctuation range of a grid is the
same. This is because the rate of the load fluctuation range will become low compared with wind
power generation if the production of electricity
of wind power generation increases. In order to
153
Figure 12. Analysis results of 12-house apartment load model. Wind power generator is 2 kW. Load
fluctuation is
improve whole generation efficiency, it is effective

to make the power generation capacity per fuel
cell system small, and to increase the number of
fuel cell systems installed. This is because the load
rate of all fuel cell systems rises and the ranges of
partial load operation with low efficiency decrease
in number. However, the improvement in generation efficiency is compared with the rise in facility
cost, and it is necessary to plan and design fuel
cell capacity and the number installed.
154
POWER CHARACTERISTICS OF
A FUEL CELL MICROGRID WITH
WIND POWER GENERATION
Introduction
It is predicted that a microgrid technique is effective about a backup power supply in an emergency,
a peak cut of power plants, and exhaust heat
utilization. Furthermore, when renewable energy
is connected to a microgrid, there is potential to

reduce the amount of greenhouse gas discharge
(Abu-Sharkh et al., 2006; Carlos & Hernandez,
2005; Robert, 2004). A microgrid has an interconnection system with commercial power etc., and
the independence supplying system of the power.
The microgrid with an interconnection system
outputs and inputs the power between other grids.
Therefore, the dynamic characteristic of the grid is
influenced by the grid of a connection destination.
When a microgrid and a large-scale grid such as
a commercial power system are interconnected,
the dynamic characteristics of the power depend
on the commercial power system. For this reason,
in the microgrid of the interconnection type, the
option of the equipment to connect is wide. On
the other hand, since microgrid can reduce transportation loss of power and heat, this technique
may become the major energy supply. The method
of connecting two or more small-scale fuel cells
and renewable energy equipment by a microgrid,
and supplying power to the demand side is effective in respect of environmental problems. So,
this section examines the independent microgrid
that connects fuel cells and wind power generation. In order to follow load fluctuation with an
independent grid system, there are a method of
installing a battery and a method of controlling
the output of power generators. Since the battery
is expensive, in this chapter, it corresponds to
load fluctuation by controlling the power output
of the fuel cell. The output adjustment of the fuel
cell has the method of controlling the production
of electricity of each fuel cell, and the method of
controlling the number of operations of the fuel
cell. However, adjustment of the production of
electricity of each fuel cell connected to the microgrid may operate some fuel cell with a partial
load with low efficiency. So, in this chapter, the
number of operations of fuel cells is controlled
to follow fluctuations in the electricity demand.
In an independent microgrid, a certain fuel
cell connected to the microgrid is chosen, and it
is considered as a power basis. The power (voltage

and frequency) of the other fuel cells is controlled
to synchronize with this base power. Therefore, if
the fuel cell that outputs base power is unstable,
the power quality of the whole grid will deteriorate. Fuel cells other than base load operation are
controlled to synchronize with the base power.
The power quality (voltage and frequency) of
the microgrid depends on the difference in the
demand-and-supply balance.
A 2.5 kW fuel cell is installed in one house of
the microgrid formed from ten houses. This fuel
cell is operated corresponding to a base load. A
1 kW fuel cell is installed in seven houses, and
a 1.5 kW wind power generator is connected to
the microgrid. According to the difference in
electricity demand of the grid and power produced by the wind power generator, the number
of operations of 1 kW fuel cells is controlled. A
city gas reformer is installed in houses in which
fuel cells are installed, and hydrogen is produced
by city gas reforming. By adding random fluctuation to an average power load pattern, the power
demand of a general residence is simulated and
it uses for analysis. The dynamic characteristics
of the microgrid and the efficiency of the system
that are assumed in this chapter are investigated
by numerical analysis.
Microgrid Model
Figure 13 shows the fuel cell independent microgrid model investigated in this chapter. There
is a network of the power and city gas in this
microgrid. Although a power network connects
all houses, a city gas network connects houses in
which a fuel cell is installed. The fuel cell installed
in each house is a proton exchange membrane type
(PEFC). The output of a 2.5kW fuel cell is decided
to be a base power of the microgrid. Moreover,
PEFC of 1 kW power is installed in seven houses.
However, the fundamental dynamic characteristics
of all the fuel cells are the same, and a fuel cell
155
Figure 13. Fuel cell microgrid system with wind power generator
and a city gas reformer are installed as a pair. One

set of wind power generator is installed, and the
power produced by wind force is supplied to a
microgrid through an inverter and an interconnection device. The power supply of the microgrid
assumes 50-Hz of the single-phase 200 V.
depending on the load factor, it is not taken into

consideration because this difference is small by
test results.
Response Characteristic of System

Configuration Equipment
Figure 14 (b) shows the output model that inputted a load of 100% load factor into the city gas
reformer stepwise (Nagano, 2002; Obara & Kudo,
2005; Lindstrom & Petterson, 2003; Oda. 1999;
Takeda, 2004; Ibe, 2002). An approximated curve
is prepared from the result of the measurement,
and the transfer function of the primary delay
of the city gas reformer is obtained. As a fuel
cell, although the transfer function of a city gas
reformer influences the magnitude of the load
significantly, since there is no large difference,
the result of Figure 14 (b) is used. Compared
with the condition of the steady operation of the
reformer, the characteristics of a startup and a
shutdown differ greatly. Cold start operation and
shutdown operation require about 20 minutes,
Power Generation
Characteristic of Fuel Cell
Figure 14 (a) shows the result of measurement
when inputting a load of 45 W into the testing
equipment of PEFC (maximum output 100 W)
stepwise. In the test, the ambient temperature
was set to 293 K, and reformed gas and air were
supplied to an anode and a cathode, respectively.
An approximated curve is prepared from the result of the measurement in Figure 14 (a), and the
transfer function of a primary delay is obtained.
Strictly, although a transfer function is considered
156
Output Characteristics of City Gas

Reformer
Figure 14. Response characteristics of system configuration equipment (Oda 1999, Takeda 2004, Ibe 2002)
respectively. In the analysis of this chapter, it is

assumed that the startup of the methanol reformer
is always a hot start.
Power Generation Characteristics

of Wind Power Generation
The model of power obtained by wind power generation is decided at random between 0 to 1.5 kW
for every sampling time, as shown in Figure 15 (a).
The power of wind power generator is supplied
to a microgrid through an inverter and a system
interconnection device. Figure 15 (b) shows the
output model of the wind power generator through
an inverter and a system-interconnection device.
Because influence is taken in the dynamic characteristic of an inverter and a system-interconnection
device, the output of wind power generation is
settled on a width of 0.75kW0.25kW range, as
shown in Figure 15 (b). The details of the transfer
function of an inverter and a system interconnection device are given with later section. The
dynamic characteristics of the inverter and system
interconnection device significantly influence
the power output characteristics of wind power
generation.
Generation Efficiency of the Fuel Cell

System
Figure 16 shows a model of the relation between
the load factor of a fuel cell, and generation efficiency (Obara and Kudo, 2005, 2005). Powergeneration efficiency is obtained by dividing the
power output of the fuel cell system by the city
gas calorific power supplied to the system. This
model was prepared from the results of the power
output when attaching the fuel cell show in Figure
14 (a) to the city gas reformer show in Figure 14
(b). If the load of a fuel cell is given to Figure
16, power generation efficiency is calculable.
The maximum efficiency of one set of a fuel cell
system is 32%.
Inverter and System Interconnection

Device
It is assumed that an inverter of a voltage control
type is used, and 120 ms is required to output
power on regular voltage and frequency (in this
chapter, it is less than 95%) (Kyoto Denkiki Co.,
Ltd. 2001). Figure 17 (a) expresses the transfer
function of such an inverter with primary delay.
157
Figure 15. Output model of wind power generator
Figure 16. Output characteristics of a PEFC with city-gas reformer
Figure 17. Transfer function of an inverter and interconnection device
158
When changing power with a system interconnection device, the change takes about 10 s
(Kyoto Denkiki Co., Ltd. 2001). However, there
is the operation of taking the synchronism of the
frequency between systems, and the model of the
system interconnection device sets the change
time to 12 ms. As a result, the transfer function
of the system interconnection device by primary
delay is shown Figure 17 (b).
Control Parameters and

Analysis Method
The response characteristics of the 1 kW fuel
cell system when inputting 0.2, 0.6, and a 1.0
kW load stepwise is shown in Figure 18. The
response characteristics of a fuel cell system
changes by the control parameters set up with
the controller. As shown in Figure 18 (c), in 1
kW step input, the rising time and settling time

(time to converge on 5% of the target output)
are not based on control parameters. In 0.2kW
step input, the rise time of P = 12.0, I = 1.0
is short, and the settling time of P = 1.0, I =
1.0 is short. In 0.6 kW step input, P = 12.0, I
= 1.0, and P = 1.0, I = 1.0 have almost the
same settling time. Moreover, overshooting is
large although the rise time of P = 12.0, I = 1.0
is short. Considering the following load fluctuations, the control parameters of the fuel cell are
analyzed by P = 12.0, I = 1.0. The dynamic
characteristics of a microgrid are analyzed using
MATLAB (Ver.7.0) and Simulink (Ver.6.0) of
Math Work Corporation. However, in analysis,
the solver to be used is the positive Runge-Kutta
system, and this determines the sampling time
from calculation converged to less than 0.01%
by error.
Figure 18. Characteristics of electric power output of the system (Obara 2005)
159
Load Response Characteristics of

the Microgrid
Step Response
The response results when applying the stepwise
input of 2, 4, 6 or 8 kW to the microgrid at intervals
of 30 seconds are shown in Figure 19 (a). The lefthand side in Figure 19 (a) shows the result of not
installing a wind power generator. The right-hand
side of the figure shows the result of a installing
wind power generator. The maximum power by a
overshooting and settling time (time to converge
on 5% of the target output) are described on
the left-hand of Figure 19 (a). Moreover, the
Figure 19. Results of step response
160
maximum power due to over shoot is described

in the right-hand side figure. The settling time
when not installing a wind power generator has
the longest period of step input of 6 kW and 8
kW for 3.9 seconds. If a wind power generator
is connected to the microgrid, many fluctuations
in the system response characteristics will occur
in a short period. If the power produced by wind
power generation is supplied to the microgrid, the
dynamic characteristics of power of the microgrid
will be influenced. Figure 19 (b) shows the analysis
result of the response error corresponding to Figure
19 (a). If wind power generator is connected to
the grid, the response error will become large as
the load of the grid becomes small. It is expected
that the power range of the fluctuation of the

microgrid will increase as the output of the wind
power generation grows. Therefore, when the load
of a microgrid is small compared with the output
of wind power generator, the power supply of the
independent microgrid becomes unstable.
Load Response Characteristics of Cold

Region Houses
Figure 20 (a) shows the power demand pattern of
a microgrid formed from ten individual houses in
Sapporo in Japan, and assumes a representative day
in February (Narita, 1996). This power demand
pattern is the average value of each hour, and the
sampling time of analyses and the assumption
time are written together on the horizontal axis.
As a base load of the power demand pattern shown

in Figure 20 (a), F/C (0) is considered as operation of 2.5 kW constant load. Figure 20 (b) and
(c) are the power demand patterns when adding
load fluctuations (1 kW and 3 kW) to Figure
20 (a) at random. The variation of the load was
decided at random within the limits of the range
of fluctuation for every sampling time.
Figure 21 shows the response results of F/C
(0) to F/C (6) when wind power generation is
connected to the microgrid and the power load
has 1 kW fluctuations. F/C (0) assumed operation with 2.5 kW constant output, with the result
that the response of F/C (0) is much less than 2.5
kW in less than the sampling time of 100 s as
shown in Figure 21 (a). This reason is because
F/C (0) was less than 2.5kW with the power of
Figure 20. 480s demand model for 10 houses in February in Sapporo
161
Figure 21. Response results of each fuel cell
wind power generation. Although the microgrid

assumed in this chapter controlled the number of
operations of F/C (1) to F/C (7) depending on the
magnitude of the load, since the power supply of
wind power generation existed, there was no
operating time of F/C (7).
Power Generation Efficiency

Figure 22 shows the analysis results of the average
power generation efficiency of fuel cell systems
for every sampling time. The average efficiency
of a fuel cell system is the value averaging the
efficiency of F/C (0) to F/C (7) operated at each
sampling time. However, the fuel cell system to
stop is not included in average power generation ef-
162
ficiency. The average power generation efficiency

of Figure 22 (a) is 13.4%, and Figure 22 (b) shows
14.3%. The difference in average efficiency occurs
in the operating point of a fuel cell system shifting
to the efficient side, when load fluctuations are
added to the microgrid. Thus, if load fluctuations
are added to the microgrid, compared with no load
fluctuations, the load factor of the fuel cell system
shown in Figure 16 will increase.
Figure 23 shows the power generation efficiency of each fuel cell in the case of connecting
wind power generation to the microgrid of 1.0kW
of load fluctuation. F/C (0) operated corresponding to a base load has maximum power generation
efficiency at all sampling times. Since the number
of operations of a fuel cell is controlled by the
Figure 22. Results of microgrid average efficiency
Figure 23. Results of efficiency for each fuel cell
163
magnitude of the load added to the microgrid, the

operating time falls in the order of F/C (1) to F/C
(6). Moreover, there is no time to operate F/C (7)
in this operating condition.
Conclusions
A 2.5 kW fuel cell was installed in a house linked to
a microgrid, operation corresponding to a base load
was conducted, and the dynamic characteristics of
the grid when installing a 1 kW fuel cell system
in seven houses were investigated by numerical
analysis. A wind power generator outputted to a
microgrid at random within 1.5 kW was installed,
and the following conclusions were obtained.
1. Although the settling time (time to converge
on 5% of the target output) of the microgrid
differs with the magnitude of the load, and
the parameters of the controller, it is about
4 seconds.
2. When connecting a wind power generator to
the microgrid, the instability of the power
of the grid due to supply-and-demand difference is an issue. This issue is remarkable
when the load of an independent microgrid
is small compared to the production of electricity of unstable wind power generation.
3. When wind power equipment is connected
to the microgrid with load fluctuation, the
operating point of the fuel cell system may
shift and power generation efficiency may
improve.
REFERENCES
Abu-Sharkh, S., Arnold, R. J., Kohler, J., Li,
R., Markvart, T., & Ross, J. N. et al. (2006).
Canmicrogrids make a major contribution to
UK energy supply? Renewable & Sustainable
Energy Reviews, 10(2), 78127. doi:10.1016/j.
rser.2004.09.013
164
Alibhai, Z., Lum, R., Huster, A., Gruver, W. A., &

Kotak, D. B. (2004). Coordination of distributed
energy resources, Systems, man and cybernetics. In H. G. Stassen (Ed.), IEEE International
Conference on Vo. 2, Hague, Netherlands, (pp.
1990-1995).
673681. doi:10.1109/TIA.2005.847277
Ibe, S., Shinke, N., Takami, S., Yasuda, Y., Asatsu,
H., & Echigo, M. (2002). Development of fuel
processor for residential fuel cell cogeneration
system. In K. Abe (Ed.), Proc. 21th Annual Meeting of Japan Society of Energy and Resources,
Osaka, (pp. 493-496).In Japanese.
Kyoto Denkiki Co., Ltd. (2001). A system connection inverter catalog and an examination
data sheet.
Lindstrom, B., & Petterson, L. (2003). Development of a methanol fuelled reformer for fuel cell
applications. Journal of Power Sources, 118,
7178.
Nagano, S. (2002). Plate-type methanol steam
reformer using new catalytic combustion for a
fuel cell. In Proceedings of SAE Technical Paper
Series (Automotive Eng. 10).
University, Japan. In Japanese.
National Astronomical Observatory. (2003). Rika
Nenpyo, chronological scientific tables CD-ROM.
Japan: Maruzen Co., Ltd.
Obara, S., & Kudo, K. (2005a). Installation
planning of small-scale fuel cell cogeneration
in consideration of load response characteristics
(Load response characteristics of electric power
output). Transactions of the JSME, Series B,
71(706), 1678-1685. In Japanese.
Obara, S., & Kudo, K. (2005b). Study on smallscale fuel cell cogeneration system with methanol
fluctuation. Transactions of the ASME. Journal
of Energy Resources Technology, 127, 265271.
doi:10.1115/1.1926310
Oda, K., Sakamoto, S., Ueda, M., Fuji, A., &
Ouki, T. (1999). A small-scale reformer for fuel
cell application. [In Japanese]. Sanyo Technical
Review, 31(2), 99106.

start-up. In K. Kimura (Ed.), Proc. 20th Energy
Tokyo, (pp. 343-344). In Japanese.
urban energy. Transaction of the Society of Heating, Air-Conditioning and Sanitary Eng. of Japan,
79(7), 573579.
Robert, H. (2004). Microgrid: A conceptual solution. In Proceedings of the 35th Annual IEEE
Power Electronics Specialists Conference, 6,
4285-4290.
165
APPENDIX
Nomenclature

Act: If action
Act_FC: Each fuel cell operation
C fn : Instantaneous power capacity of F/C (n) [kW]
CW : Instantaneous wind power capacity [kW]
Ed : Power demand [kW]
E fn : Power production by F/C (n) [kW]
Err : Error [kW]
166
EW : Wind power [kW]
e f : Load fluctuation range [kW]

F/C: fuel cell system (consists of a controller, a fuel cell, an inverter and an interconnection
device)
h : Input
h : Capacity of generation [W]
I : Integral parameter
I/C: Interconnection device
P : Proportionality parameter
PI: Proportion-integration control
u: Power load of a microgrid [W]
: Power output [W]
W/P: Wind power generation system (consists of controller, wind power generator, inverter and
interconnection device)
167
Chapter 6
Compound Microgrid of
City-Gas Engine and Proton
Exchange Membrane Fuel Cell
ABSTRACT
This chapter consists of two sections, Amount of CO2 Discharged from Compound Microgrid of
Hydrogenation City-Gas Engine and Proton Exchange Membrane Fuel Cell and Power Characteristics
of a Fuel Cell Microgrid with Wind Power Generation. In the 1st section, a microgrid composed from
a PEFC and a hydrogenation city gas engine is investigated using numerical simulation. The system
which combined base-load operation of PEFC and load fluctuation operation of hydrogenation city
gas engine is the most advantageous. The independent PEFC power supply system relating to hydrogen
energy is investigated in the 2nd section. The hybrid cogeneration system (HCGS) that uses a PEFC and
a hydrogen mixture gas engine (NEG) together to improve power generation efficiency during partial
load of fuel cell cogeneration is proposed.
The summary of the 1st section is as follows.
The independent microgrid is considered to be a
technology in which maximum distributed energy
is realizable. However, there are many subjects,
such as the stability of the dynamic characteristics
of power and development of an optimal design
method. If the fuel cell system of the capacity cor-
responding to a load peak is installed, equipment

cost will be high and energy cost will not be able
to get any profile commercially. By increasing the
hydrogen concentration at the time of low load, the
power-generation efficiency of a city-gas-enginegenerator improves, and carbon dioxide emissions
decrease. So, in this study, a microgrid composed
from a PEFC and a hydrogenation city gas engine
was investigated using numerical simulation. The
DOI: 10.4018/978-1-4666-5796-0.ch006
Compound Microgrid of City-Gas Engine and Proton Exchange Membrane
system with a small load factor of NEG and with

a large load factor of PEFC system has few CO2
emissions. The system which combined base-load
operation of PEFC and load fluctuation operation of hydrogenation city gas engine is the most
advantageous for the comprehensive evaluation
of equipment cost, power generation efficiency,
and CO2 emissions. When the optimal system was
installed into the urban area model of 20 buildings and analyzed, power generation efficiency
was 25% and CO2 emissions were 1,106 kg/Day.
Development of a small-scale power source not
dependent on commercial power may result in
various effects. For example, it may eliminate
the need for long distance power-transmission
lines, and mean that the amount of green energy
development is not restricted to the dynamic characteristics of a commercial power grid. Moreover,
the distribution of the independent energy source
can be optimized with regionality in mind. This
study examines the independent power supply
system relating to hydrogen energy. Generally
speaking, the power demand of a house tends to
fluctuate considerably over the course of a day.
Therefore, when introducing fuel cell cogeneration into an apartment house, etc., low-efficiency
operations in a low-load region occur frequently in
accordance with load fluctuation. Consequently,
the hybrid cogeneration system (HCGS) that uses
a PEFC and a hydrogen mixture gas engine (NEG)
together to improve power generation efficiency
during partial load of fuel cell cogeneration is
proposed. However, since facility costs increase,
if the HCGS energy cost is not low compared with
the conventional method, it is disadvantageous.
Therefore, in this section, HCGS is introduced into
10 household apartments in Tokyo, and the power
generation efficiency, carbon dioxide emissions
and optimal capacity of a boiler and heat storage
tank are investigated through analysis. Moreover,
the system characteristics change significantly
based on the capacity of PEFC and NEG that
compose HCGS. Therefore, in this study, the ca-
168
pacity of PEFC and that of NEG are investigated,

as well as the power generation efficiency, carbon
dioxide emissions and the optimal capacity of a
boiler and heat storage tank. Analysis revealed that
the annual average power generation efficiency
when the capacity of PEFC and NEG is 5 kW
was 27.3%. Meanwhile, the annual average power
generation efficiency of HCGS is 1.37 times that
of the PEFC independent system, and 1.28 times
that of the NEG independent system respectively.
AMOUNT OF CO2 DISCHARGED

FROM COMPOUND MICROGRID
OF HYDROGENATION CITY-GAS
ENGINE AND PROTON EXCHANGE
MEMBRANE FUEL CELL
Introduction
The introduction to a urban area of a microgrid has
the following advantages: (a) The heat transport
distance is short and effective use of the exhaust
heat of the generating equipment is possible; (b)
The optimal facility for the energy demand characteristic of a community is installed, and a system
having small environmental impact can be built;
and (c) With an independent microgrid, the scale
of equipment for distributing electricity is small
(Robert et al., 2004; Carlos & Hernandez, 2005;
Takuma & Goda, 2005). Furthermore, (d) Connecting renewable energy considering regionality
is expected as an advanced system in microgrid
technology. At present, the method of a microgrid
interconnecting with commercial power, etc. is
investigated (interconnect microgrid) (Takuma
& Goda, 2005). However, in order to achieve the
advantages of (a) to (d) described above, it is necessary to operate a microgrid independently. The
subjects of the independent microgrid are backup
in the case of overload, and securing power quality
(voltage and frequency). Furthermore, it is necessary to clarify the power-generation efficiency,
the carbon dioxide emissions, and the power
cost of an independent microgrid. An improvement in power-generation efficiency is expected

from the independent microgrid using a fuel cell
compared with conventional electric power-supply
technology. However, for the moment, fuel cells
are expensive, and whether they will spread is not
clear. As for a fuel cell independent microgrid,
power-generation efficiency and carbon dioxide emissions are expected to be advantageous
compared with existing generating equipment.
However, because the fuel cell is expensive, it
is difficult to install the capacity corresponding
to a load peak. Consequently, there is a case of
operation which limits operation of a fuel cell to
a highly efficient load region (Society of Living
Amenity, 2006). The hydrogenation technology of
a city gas engine is effective concerning efficiency
falls and increases in carbon dioxide emissions at
the time of partial load (Mohammadi et al., 2005;
Yap et al., 2004; Larsen & Wallace, 1995; Swain
& Usuf, 1993.). The power-generation system
using a city gas engine with generator (NEG) is
cheap compared with the fuel cell. So, this Section
examines the power-generation efficiency and
the carbon dioxide emissions when connecting
NEG and the proton exchange membrane fuel
cell (PEFC) to an independent microgrid. In this
Section, the load of an independent microgrid is
divided into a base-load region and a fluctuating-
load region, and the system that compounds NEG

and PEFC is examined. Below, the independent
microgrid that compounds NEG and PEFC system
is described as IMPE.
System Scheme
IMPE Model
A microgrid model is shown in Figure 1. Figure
1 (a) shows a system-interconnection microgrid.
This system is interconnected with commercial
power, etc. Power Pc is delivered and received
between other grids, and Power Pg is supplied to
a microgrid with the generating equipment installed in the machinery room of Building 5 in
the urban area model of Figure 1 (a). The power
quality (frequency, voltage) of the system-interconnection microgrid is dependent on other grids
for interconnection. Therefore, even if a large load
is added to this grid, power quality is stabilized
in a short time. On the other hand, Figure 1 (b)
shows an independent microgrid that does not
interconnect with other grid systems. The method
that supplies the power of an independent microgrid by a one-set power-generation system is
defined as a centralized system. Two sets of NEG
or PEFC system are introduced, and how to divide
Figure 1. Micro-grid model
169
into base-load operation and fluctuating-load

operation, and supply power is defined as a base
load-sharing system. However, the base loadsharing system corresponds to a base load and
fluctuating load using either FC or NEG. For
example, how to correspond base-load operation
by NEG and correspond to fluctuating-load operation by PEFC system is defined as an IMPE
system. By the IMPE system, the kinds of generating equipment of base-load operation and
fluctuating-load operation differ.
Operation Method of the Microgrid

Figure 2 shows the power load pattern of the independent microgrid shown in Figure 1 (b). The
load pattern of Figure 1 (b) is separated into a
base-load region and a fluctuating-load region in
systems other than the centralized system. As
Figure 1 (b) shows, PEFC system of Capacity PF ,l
is installed in Building 5 linked to Grid A, and
NEG of Capacity PE ,m is installed in Building 19
linked to Grid B. Grid A and Grid B can deliver
and receive the power by system interconnection
equipment CP. Therefore, PEFC system of Building 5 is made to correspond to the base-load region
of Figure 2, and NEG of Building 19 is made to
correspond to a fluctuating-load region.
Figure 2. Output share model of load power
170
Equipment Scheme
Figure 3 shows an example of equipment schemes
of the building connected to IMPE shown in
Figure 1 (b). Figure 3 (a) shows the equipment
scheme of Building m linked to an NEG central
system. The generating equipment installed with
a centralized system is any one set of FC or NEG.
Figure 3 (a) shows the equipment scheme of the
central system using NEG, where NEG, a boiler,
a heat storage tank, an interconnection device,
etc. are installed. Although city gas (QE ) is supplied to NEG, at the time of low load, hydrogen
(QEr ) is supplied through reformed gas piping.
However, equipment cost can also be reduced by
installing a city gas reformer in the same building,
m , as NEG. NEG and PEFC system are installed
in Building l , and the equipment scheme of the
IMPE system corresponding to base load or fluctuating load is shown in Figure 3 (b). The hydrogen produced by the reformer is supplied to NEG
at the time of low load, and PEFC stack. NEG,
PEFC stack, a city gas reformer, a boiler, a heat
storage tank, an interconnection device, etc. are
installed in the building shown in Figure 3 (b).
City gas (QS ) is a heat source, and city gas (QR )
produces reformed gas with the fuel for reforming.
Furthermore, in order to reduce the CO concentration in the reformed gas in a fuel cell stack entrance
to several ppm, a carbon monoxide oxidization
Section is provided. In the carbon monoxide
oxidation Section, carbon monoxide is burned on
a catalyst and it changes into carbon dioxide. The
direct current power generated by the fuel cell
stack is changed into an alternating current of
fixed frequency through an inverter, and is supplied to an interconnection device. Figure 3 (c)
shows the equipment scheme of Building n in
which NEG or PEFC system is not installed. The
power demand of Building n is received from a
microgrid through an interconnection device.
Figure 3. Energy equipment model
Moreover, heat supply is obtained by city gas (

QB ) burning of a boiler. Carbon dioxide emissions
are calculated from the city gas supplied to a
reformer (QR and QS ) and NEG (QE ).
Output Characteristics of Gas
Engine Power Generator
Figure 4 (a) shows the examination results of the
hydrogenation rate and brake thermal efficiency
of a one-cylinder city gas engine (Mohammadi et
al., 2005). The engine mean effective pressure of

hydrogenation is effective in a range that is less
than 0.8 MPa. Thermal efficiency with a mean
effective pressure large without hydrogenation
in a range exceeding 0.8 MPa can be obtained.
Figure 4 (b) shows the relation between the mean
effective pressure and brake thermal efficiency,
and the hydrogenation rate (Mohammadi et al.,
2005). The broken line shown in this figure is
the hydrogenation rate indicating the maximum
thermal efficiency. Figure 4 (c) shows the analysis
results of the production of electricity of NEG,
city gas consumption, and the amount of hydro-
171
Figure 4. Output characteristics of NEG
genation calculated from the model of Figs 4 (a)

and (b). The amount of hydrogenation of Figure
4 (c) is the result when obtaining the maximum
thermal efficiency. The specifications of a city
gas engine and a power generator are shown in
Tables 1 (a) and (b). Hydrogen consumption is
zero when the production of electricity exceeds
14 kW as shown in Figure 4 (c). This is because
high thermal efficiency can be obtained even if
there is no hydrogenation in the large range of
engine power, as Figure 4 (a) describes. Figure
4 (d) shows the relation of the production of
electricity and generation efficiency of NEG.
Although reformed gas is supplied to NEG, the
generation efficiency of Figure 4 (d) includes
reformer efficiency. The reformer is as common
172
as the PEFC system described below. Detail of

reformer efficiency is given by following Section.
Figure 5 (a) shows the relation between carbon
dioxide emissions and the production of electricity of NEG and engine hydrogenation. This
model was calculated from the characteristics of
the thermal efficiency described in Figure 4, and
the equations (from Equation (1) to Equation (3))
described below. The fuel supplied to NEG has
many hydrogen rates in a low-load region, and
there are many rates of city gas in a high-load
region. Therefore, there are many rates of carbon
dioxide discharged with a reforming reaction and
a reformer burner in a low-load region, and there
are many rates of carbon dioxide discharged by
engine burning of city gas in a high-load region.
Table 1. Specifications of NEG
Figure 5 .CO2 emission characteristics of NEG
Figure 5 (b) shows the model of a load factor and

CO2 emissions calculated from Figure 5 (a). In
Region A in this figure, NEG is mainly operated
using reforming gas. In this region, CO2 emissions
decrease slightly with the rise of a load factor. It
is because reformer efficiency will improve when
a load factor rises as described in following Section and Figure 6.
Carbon Dioxide Emissions of NEG

Equation (1) is a steam-reforming reaction equation of city gas (CH4). Since Equation (1) is an
endothermic reaction, the heat for advancing the
response of Equation (1) is produced using the
combustion reaction of CH4 shown in Equation (2).
Moreover, Equation (3) is an equation that changes
Figure 6. Output characteristics of PEM-FC system
173
the carbon monoxide of Equation (1) into carbon

dioxide and hydrogen. If the hydrogen quantity
supplied to NEG and PEFC stack is decided, the
amount of city gas supplied to a reformer and the
carbon dioxide to be discharged are calculable using Equations (1), (2), and (3). The CH4 quantity
supplied to an engine is calculable using Equations
(2), Figure 4 (a), and Figure 4 (b).
CH 4 + H 2O CO + 3H 2 206 [kJ / mol ]
(1)
CH 4 + 2O2 CO2 + 2H 2O + 802 [kJ / mol ]
(2)
CO + H 2O CO2 + H 2 + 41 [kJ / mol ]
(3)
is the power in the inverter outlet of a PEFC system. QR,F ,t expresses the calorific power of CH4
for reforming, and QS ,F ,t expresses the calorific

power of CH4 supplied to a heat-source burner.
The maximum generation efficiency of the fuel
cell shown in Figure 6 is 31%. Moreover, the
reformer efficiency in Figure 6 (a) improves with
the increase in a load factor. Figure 6 (b) shows
the CO2 emissions of the PEFC system. Figure 6
(b) shows the result of calculating based on the
power-generation efficiency and reformer efficiency in Figure 6 (a). At the time of the hydrogen
supply to PEFC stack, the amount of CO2 discharged by a reforming reaction is expressed with
GR,F ,t , and the quantity discharged by a heatsource burner is expressed with GS ,F ,t . Therefore,
the amount GF ,t of CO2 discharged by the gen-
Equation (4) expresses the amount of carbon

dioxide discharged by NEG. GE ,p,t is the carbon
eration of PEFC system is calculated by Equation

(6).
dioxide emissions when burning CH4 with Engine

p . GR,p,t is the amount of carbon dioxide dis-
F ,t = F ,t
100
+
Q
Q
(
)
R
,
F
,
t
S
,
F
,
t
(5)
GF ,t =GR,F ,t + GS ,F ,t
(6)
charged by a reforming reaction required for engine

hydrogenation. GS ,p,t is the carbon dioxide emissions of a heat-source burner. N R is the installed
number of NEG, and in the NEG centralized
system and a NEG base-load IMPE system, it is
one set, and is two sets in the NEG base loadsharing system.
NE
GNEG ,t = (GE ,p,t + GR,p,t + GS ,p,t )

p =1
(4)
As Figure 5 (b) shows, there are many CO2

emissions of NEG in a high-load zone, but there
are many CO2 emissions of a fuel cell in a low-load
zone (Figure 6 (b)). From the difference in CO2
emission characteristics, NEG is advantageous in
the operation of a partial load, and PEFC system
is advantageous in steady operation at high load.
PEFC System
Case Study
Figure 6 shows the model of the generation efficiency of PEFC system and city gas reformer
efficiency (Yasuda, 2005). Moreover, generation
efficiency F of Figure 6 (a) was calculated using
Equation (5). When the sampling time is expressed
with t , E F ,t of Equation (5), the right-hand side
Urban Area Model
174
The urban area model analyzed in this Section is

shown in Figure 7. The building number is shown
in this figure and the application for each building
is shown in Table 2. The number of buildings of an
Figure 7. Urban area model
urban area model is 20. The urban area model can

consider various patterns. This Section examines
the characteristics of the carbon dioxide emissions
of the compound grid of NEG and PEFC system
from the case of Figure 7.
Power Demand Model

Figure 8 shows the power demand model of each
building in Tokyo, and is the mean power load
of each sampling time of a representative day in
January (winter), May (mid-term), and August
(summer) (Architectural Institute of Japan, 2002;
Nagase et al., 2003; Hatano et al., 2003; Yoshinaga
et al., 2003). However, the actual power demand
pattern is an assembly of the load that changes
rapidly in a short time, such as an inrush current.
A power demand estimate of the house shown
in Figures 8 (a) to 8 (d) is difficult. On the other
Table 2. Power demand model for urban area
hand, the power demand of the small offices of

Figure8 (g) and the factories of Figure 8 (h) is
regular and easy to predict. The power demand
pattern of a house has a peak in the morning and
the afternoon. When midnight to early morning
is excluded, hotels (Figure 8 (e)) have the stable
demand, and convenience stores (Figure 8 (f)) have
24-hour power demand. In small offices (Figure
8 (g)), factories (Figure 8 (h)), and small hospitals (Figure 8 (i)), there is little power demand at
night to early morning, and there is much power
demand from morning till the evening. In Case
Study, the CO2 emissions of August representation days which are the largest power demand are
calculated. Figure 9 shows the heat demand model
in August of each building described in the top
(Architectural Institute of Japan, 2002; Nagase et
al., 2003; Hatano et al., 2003; Yoshinaga et al.,
2003). However, in a convenience store, office,
and a factory, because heat pump is introduced,
heat demand is not taken into consideration.
Analysis Flow
The analysis flow of the centralized system, base
load-sharing system, and IMPE system is shown
in Figure 10. First, the power demand data of
each building are given to the analysis program,
and the base load of the whole microgrid is calculated. Next, the power plant capacity installed
into a microgrid is given, and power generation
efficiency and carbon dioxide emissions are calculated for every sampling time concerning all
the grid routes of an urban area model. By adding
these all, the total power generation efficiency and
the total carbon dioxide emissions in the operation period, and the capacity of a power plant are
determined. The load factor is calculated from the
capacity and power load of a power plant. A load
factor is given to the approximation of Figures 4
(d) or Figure 6, and power generation efficiency
is determined. The carbon dioxide emissions of
a system are calculated by giving a load factor to
the approximation of Figures 5 or 6.
175
Figure 8. Power demand models
176
Figure 9. Heat demand models in August
Figure 10. Calculation flow

Power Load of Microgrid
Figure 11 (a) shows the result of the power load
pattern of a representative day in August of the
urban area model. As the result of time change of
power demand, Figure 11 (a) shows that the power
plant capacity of a fluctuating load is 100 kW, and
the power plant capacity of a base load is 66 kW.
Figure 11 (b) shows the result of the rate of a base
load and a fluctuating load. The base load is 1.32
times larger. Figure 11 (c) shows the composition
of the power demand added to a microgrid. The
largest load component is convenience stores
(two buildings), which takes 36% of the whole
load. Because there is 24 hour power demand
in convenience stores (Figure 8 (f)) and hotels
(Figure 8 (e)), it is a component that smoothes
the whole load added to the microgrid. Factories
and small offices of the ratio of the whole load
are large. However, the demand difference in the
day and at night is large, and is a component to
which the fluctuating load region of a microgrid
is made to increase.
Capacity of Power Plant

The analysis results of a representative day in
August are shown in Figure 12. Figure 12 (a)
shows the results of the capacity of the power
plant installed in each microgrid system. to
in Figure 7-1 expresses the power supply method
177
Figure 11. Results of the load pattern in May representative day
described in the figure. In a centralized system

( and ), one set of 166 kW power plant is connected to a microgrid. On the other hand, in the
base load-sharing system ( and ) and IMPE
system ( and ), the power plant capacity corresponding to a base load and load fluctuation is
66 kW and 100 kW, respectively.

Figure 12 (b) shows the analysis results of the
total power generation efficiency of the system
of to . Total power generation efficiency is
high at , , and . Most these systems are a
method of corresponding to base-load operation
by FC. Figure 12 (c) shows the distribution of the
178
power-generation efficiency of the system of to

except for the centralized system. In FC baseload operation, it can operate at the maximum
power-generation efficiency shown in Figure 6
(a). The maximum power-generation efficiency
of PEFC system is higher than the efficiency of
NEG shown in Figure 4 (d). Therefore, the total
power generation efficiency of the system of ,
, and using FC to base-load operation is high.
Carbon Dioxide Emissions

amount of carbon dioxide discharged from each
system. and have few CO2 emissions and these
are PEFC system base load operations. Moreover,
Figure 12. Analysis results
(NEG base load and FC load fluctuation operation) also has few CO2 emissions. When Figure
5 (b) is compared with Figure 6 (b), the change
in CO2 emissions to change of a load factor has
NEG larger than PEFC system. As Figure 5 (b)
and Figure 6 (b) showed, when the load factor of
PEFC system is large and the load factor of NEG

is small, CO2 emissions will decrease. Therefore,
there are few CO2 emissions of and . Although
is NEG base load operation, because it corresponds to load fluctuation by large capacity PEFC
system (100kW), there are few CO2 emissions
179
than the system composed only from NEG ( and

). in Figure 12 (d) is CO2 emissions of NEG
without hydrogenation. When the hydrogenation
of NEG is introduced, compared with the method
which does not add hydrogen, about 15% of CO2
emissions will reduce. After all, the order with
few CO2 emissions is , , , , , , and .
The order ( and , , , and ) of the powergeneration efficiency described in the top differs
from this order. Furthermore, when facility cost
is taken into consideration, the smallest possible
system of fuel cell capacity is advantageous.
Power-generation efficiency is high, there are few
CO2 emissions, and a system with cheap facility
cost is the best. Therefore, system is proposed
in this Section.
The energy supply by commercial power and
a kerosene boiler is defined as the conventional
method. The amount of greenhouse gas discharge
of the conventional system is able to calculate
based on the investigative commission report of
the calculation method of the amount of greenhouse gas discharge (the Ministry of Environment
in Japan, August, 2003). The commercial power
of a greenhouse gas discharge factor is 0.331 kgCO2/kWh, and a kerosene boiler is set up at 0.0685
kg-CO2/MJ, 9.5 kg-CH4/TJ, and 0.57 kg-N2O/TJ.
As a result, as shown in Figure 12 (d), the CO2
emissions of system decrease rather than the
conventional method slightly. Moreover, because
the CO2 emissions of the system of and differ

greatly, it depends for CO2 emissions to a power
load factor strongly.
Heat Demand and Exhaust Heat Output

Figure 13 shows the analysis result of the exhaust
heat output of August representation days of base
load operation and load fluctuation operation of
each system. The exhaust heat of each system of
to exceeds the heat amount demanded of the
urban area model in Figure 7, as shown in Figure
13. When an exhaust heat network is introduced
into a microgrid and exhaust heat is distributed to
each building, the boiler shown in Figure 2 will
become unnecessary.
Conclusions
Compared with the present power supply method,
power-generation efficiency may improve and
carbon dioxide emissions may decrease in independent microgrids using a fuel cell. However,
if a fuel cell microgrid is introduced into a urban
area with great load fluctuation, it becomes a very
expensive facility now. So, this Section examined
the independent microgrid of most August representation days of power demand that introduce a
hydrogenation city gas engine and PEFC system,
and are operated. The following Conclusions were
Figure 13. Heat demand and exhaust heat output in August representative day
180
obtained as a result of installing this system into

an urban area model composed from the power
load patterns of 20 buildings, such as houses, offices, and hospitals.
1. The total power generation efficiency of
centralization system, base-load sharing
system, IMPE system using PEFC and NEG
was 19% to 30% of range. Especially powergeneration efficiency has high introduction
of PEFC system base load operation.
2. The load factor of PEFC system is large
and the system with a small load factor of
NEG has few CO2 emissions. There are few
CO2 emissions of PEFC base-load sharing
system as a result of analysis. Next, there
is little IMPE system of PEFC and NEG.
Moreover, when hydrogen is added to NEG,
CO2 emissions will be reduced 15%.
3. A PEFC base load and the system which
operates a hydrogenation city gas engine in a
load fluctuation region are the most advantageous under the overall evaluation of facility
cost, power generation efficiency, and CO2
emissions. When the urban area model was
analyzed using the highest system of an
overall evaluation, 25% of power-generation
efficiency and CO2 emissions were 1,106 kg/
Day. However, it depends for CO2 emissions
on power-demand fluctuation (load factor)
of the microgrid which introduces a system
greatly.
The maximum effect is expected by making an
energy-demand characteristic reflect in planning
of a microgrid. The relation of the locality of an
energy demand characteristic and the optimal design of a microgrid will be investigated hereafter.
CAPACITY OPTIMIZATION OF PEFCL

AND HYDROGEN MIXING GASENGINE COMPOUND GENERATOR
Introduction
Cogeneration using a proton-exchange membrane
fuel cell (PEFC) has a high maximum power
generation efficiency compared with other power
plants. However, reduced power generation efficiency during partial-load operation of a system
with a reformer remains an issue. In particular,
in general houses, partial-load operations with
significant fluctuation of power load and low
efficiency occur frequently. On the other hand,
hydrogen mixing technology is studied, concerning exhaust gas cleanup and improved efficiency
at the time of partial load of gas-engine cogeneration (Mohammadi et al., 2005; Yap et al., 2004;
Larsen & Wallace, 1995; Swain & Usuf, 1993).
If the hydrogen rate of fuel is increased when
the load is small, studies confirm that exhaust
cleanup and brake thermal efficiency improve
(Mohammadi et al., 2005; Yap et al., 2004;
Larsen & Wallace, 1995; Swain & Usuf, 1993).
Although the facility cost of a hydrogen/gas hybrid engine (NEG) cogeneration is low compared
with PEFC, the maximum power generation efficiency is inferior. However, if the partial-load
characteristics of NEG are better compared with
PEFC, the introduction of PEFC to a house may
be disadvantageous. Therefore, in this chapter,
a PEFC and NEG hybrid cogeneration system
(HCGS) having the maximum power generation
efficiency of PEFC and the partial-load characteristics of NEG is proposed. In HCGS, PEFC is
operated corresponding to a base load, and NEG
is operated corresponding to a fluctuating load.
181
In this case, since PEFC operates at a constant

load, the capacity of the fuel cell is designed to be
optimal, and can always be operated at high power
generation efficiency. On the other hand, NEG is
operated corresponding to fluctuating loads except
for the base load. Therefore, a low-load operating range, where PEFC is disadvantageous, can
correspond to NEG. The facility cost of HCGS
increases compared with the case where PEFC or
NEG is operated independently. Therefore, in order
for the introduction of HCGS to city areas to be
effective, compared with the individual operation
of PEFC or NEG, the power cost and carbon dioxide emissions must be advantageous. However,
the power generation efficiency and the carbon
dioxide emissions of PEFC, NEG, and HCGS
are dependent on the load pattern of a building.
Moreover, since the exhaust heat characteristics
differ according to PEFC, NEG, or HCGS, the
optimal capacity of the heat storage tank and boiler
of each system also vary, while the power generation efficiency and exhaust heat characteristics
of HCGS change with the capacity of PEFC and
NEG to be introduced. Therefore, in this chapter,
the relations among the capacity of PEFC and
NEG, power generation efficiency, carbon
dioxide emissions, and the optimal capacity of
Figure 14. HCGS block diagram
182
a boiler and a heat storage tank are investigated

with reference to the introduction of HCGS into
10 household apartment houses in Tokyo.
HCGS Scheme
A block diagram of HCGS proposed in this
chapter is shown in Figure 14. City gas includes
supply systems to NEG, to a reformer, to a heat
source burner, and to a boiler respectively, where
the Japanese conditions apply to the composition
of City gas. Accordingly, City gas consists
mainly of methane. The extent of the HCGS city
gas consumption is calculated by Equation (7).
U total ,t = U N ,t + U R,t + U B ,t + U S ,t
(7)
The operation model of HCGS is shown in

Figure 15 (a). In the operation of HCGS, PEFC
is used for a base load and NEG is used for other
fluctuating loads. Therefore, a high load of PEFC
is always highly efficient, and it can be operated.
C N' and C F' in this figure express the maximum
electric load of NEG and PEFC, respectively. It
is necessary to determine the capacity of NEG
and PEFC (C N ,C F ) introduced into HCGS with
Figure 15. HCGS operation method
a value exceeding C N' and C F' of the figure. Since

the load factor (load/capacity) changes according
to the capacity setup of C N and C F , the fuel
consumption and carbon dioxide emissions of
HCGS also change. So, in this chapter, the fuel
consumption and carbon dioxide emissions when
changing the value (relatively) of C N and C F are
investigated. Figure 15 (b) shows a model showing the relation of the load factor and power
generation efficiency of NEG and PEFC. In the
low-load operating range of NEG, an improvement
in brake thermal efficiency is expected by increasing the hydrogen concentration of fuel (Mohammadi et al., 2005). However, there is no effect of
hydrogen mixing in the high-load operating range
of NEG. Moreover, as shown in Figure 15 (b),
compared with PEFC, the maximum efficiency
point of NEG is low. Therefore, operation in a
low-load zone of NEG is advantageous, and the
same applies to operation in a high-load zone of
PEFC. Since HCGS has the characteristics of the
power generation efficiency of PEFC and NEG,
it is expected to be comparable to the individual
operation of PEFC or NEG, in which power generation efficiency is advantageous.
Output Characteristics of NEG
Past examination results are used for the output
characteristics of a city gas engine with hydrogen
mixing (Mohammadi et al., 2005). The examined
hydrogen mixing engine is 857 cc in a single
cylinder, and lean burn is enabled by injecting
reformed gas into an inlet pipe. Figure 16 (a)
shows the model of the city gas consumption
(CH4) of NEG, the amount of hydrogen mixing,
and the production of electricity. These characteristics were calculated from the hydrogen mixing rate, based on the reference examination result
(Mohammadi et al., 2005). In Figure 16 (a), when
the electricity production exceeds 14 kW, the
amount of hydrogen mixing is zero, because high
thermal efficiency can be obtained, even if there
is no hydrogen mixing within the large operating
range of engine power. Figure 16 (b), meanwhile,
shows the carbon dioxide emission characteristics
of NEG, calculated based on the reference examination result (Mohammadi et al., 2005). In
this figure, the approximate expression showing
the relation between load factor R and CO2 emis-
183
sions is shown. These approximate expressions

differ bordering on 60% of the load factor, because
the fuel supplied to NEG has a high hydrogen rate
in a low-load range and the rate of city gas is high
within a high-load range. In a low-load range, the
carbon dioxide ratio discharged by city gas reforming for hydrogen manufacture and the burner for
reformers is high. On the other hand, in a highload range, the rate of carbon dioxide generated
in engine city gas burning is high. Figure 16 (c)
shows the relationship between the load factor
and the generation efficiency of NEG. Although
reformed gas is supplied to NEG, reformer effi-
ciency is included in the power generation efficiency shown in this figure. Equation (8) defines
reformer efficiency.
q
100
R = H 2
qCH
4
q H in Equation (8) is the calorific power of

2
hydrogen in reformed gas, and qCH expresses the

4
calorific power of the city gas supplied to a reformer, while the calorific power of the city gas
Figure 16. Output characteristics of the hydrogen mixing gas engine cogenerator
184
(8)
[%]
for reforming and the city gas for heat-source

burners is included in qCH . In this chapter, R
4
was set at 73% (T. Nakamura and M. Sei., (2002),

K. Oda, et al., (1999), Takeda, Y. et al., (2004)).
Generally, engine thermal efficiency increases,
so that the capacity expands. Consequently, as
shown in Figure 16 (c), the NEG power generation
efficiency characteristics are separated into each
capacity of NEG, which are then introduced into
the analysis.
Figure 16 (d) shows the power and heat output
of 10-kW NEG, and a model of overall efficiency.
The heat outputs are engine exhaust, cooling
water, and reformer exhaust heat. However, the
overall efficiency in this figure assumes a case
where all the power and heat outputted by NEG
are consumed. Introduced in the Case Study are
3- to 10-kW NEG. The difference between the
generation efficiency and the carbon dioxide emissions of 3-kW NEG and 10-kW NEG is about 6%.
Figure 18 (b) described in later Section explains
the exhaust heat of the NEG.
Output Characteristics of PEFC

Figure 17 (a) shows a model of the output characteristics and overall efficiency of 5-kW PEFC
with a city gas reformer (Mikkola, 2001; Ibaraki
Prefecture Government Office of Education,
2002). The heat output includes the exhaust heat
of the reformer and cell stack. Moreover, the
power output is the value of the inverter outlet.
Overall efficiency assumes the case where all of
the power and heat to output are consumed. Figure 17 (b) shows the model of the carbon dioxide
emissions and generation efficiency of PEFC of
Figure 17 (a). The carbon dioxide discharged by
the operation of PEFC is based on a reforming
reaction (Equations (9) and (10)) and city gas
burning of a reformer burner (Equation (11)). The
power generation efficiency F of Figure 17 (b)
was calculated from Equation (12). E F ,t , on the
right-hand side of Equation (12), shows the
power of the inverter outlet, q R,CH ,t shows the
4
calorific power of CH4 supplied to the reformer,
Figure 17. Output characteristics of the PEFC
185
Figure 18. Operation map
186
and q B ,CH
,t
shows the calorific power of CH4
supplied to the heat-source burner of the reformer. The maximum generation efficiency of
the PEFC model shown in Figure 17 (b) is 32%.
CH 4 + H2O CO + 3H2 206 [kJ/mol]
(9)
CO + H2O CO2 + H2 + 41 [kJ/mol]
(10)
CH 4 + 2O2 CO2 + 2H2O + 802 [kJ/mol]
(11)
F ,t
E
= F ,t
q R,CH + q B ,CH ,t
4
4
100
(12)
System Operation Map

In this chapter, the chart showing the city gas
calorific power (consumption) supplied to HCGS
and the relationship between the production of
electricity and the heat output of a system is defined
as an operation map. The fuel calorific power
in Figure 18 has a lower calorific value. Figure 18
(a) shows the operation map of a 10-kW PEFC,
and Figure 18 (b) shows the operation map of
the 10-kW NEG respectively. Overall efficiency
is calculable because the calorific power of the
supply fuel is shown in Figure 18. The Operation
Area A of each figure is the production range of
electricity and exhaust heat output when operating
PEFC or NEG independently. In Area A, when the
production of electricity is decided, heat output
as shown in the figure occurs. Operation area B,
meanwhile, indicates the range of production of
electricity and the heat output (the amount of
exhaust heat, and boiler output) when operating a
boiler in addition to PEFC or NEG. However, the
boiler efficiency of Operation Area B was set at
0.9. The operation map describes primary input
power (based on city gas LHV, described as city

gas consumption below) consumed by a system.
When the operation maps of PEFC (Figure 18
(a)) and NEG (Figure 18 (b)) are compared, the
NEG of Operation Area A is wider. Furthermore,
in the Operation Area A of NEG, if the electricity
production increases, exhaust heat output will also
increase rapidly, but in PEFC, even if the production of electricity increases, the increase in exhaust
heat output is lost. Figure 18 (c) shows the operation map when introducing 5-kW PEFC and 5-kW
NEG into HCGS. In HCGS, PEFC corresponds
to the base load and NEG is operated corresponding to a fluctuating load. The maximum power
generation efficiency point of PEFC approaches
the maximum load rate, as shown in Figure 17 (b).
On the other hand, as shown in Figure 16, NEG
shows positive power generation efficiency and
exhaust gas characteristics at low load. As for the
installation of low-load zone operation of NEG,
high effectiveness is expected if the generation
efficiency and the load factor of PEFC and NEG
have the relationship as shown in the model in
Figure 15 (b). Therefore, the power generation
efficiency of HCGS is predicted to improve when
compared to circumstances where PEFC or NEG
is introduced independently.
Case Study
Figure 19 shows a power and heat demand
model for the representative days each month
of 10 residence apartment houses in Tokyo, and
indicates the average load of each sampling time
(Architectural Institute of Japan, 2002). However,
the actual power-demand pattern is a set of loads
that change rapidly within a short time, such as
an inrush current. Since a cooling load is included
in the electric power demand, there is considerable power demand in the summer season. The
annual power demand amount of ten houses is
57.6MWh, while the heat demand is 86GJ and
the load factor of the power every month is shown
in Figure 19. However, when the supply- demand
187
balance is calculated by averaging power demand

with inrush current, the result will be estimated
to be far smaller (Andrew & Steven, 2007; Adam
& Matthew, 2005). The amount of city gas consumption is obtained using the operation map of
the system shown in Figure 18 for every time the
power and heat demand model shown in Figure
19 is sampled. However, when the exhaust heat
output of a system exceeds heat demand, surplus
heat is stored in a heat storage tank. On the other
hand, when the exhaust heat is less than the heat
demand, heat is supplied from a heat storage tank
and when the heat still remains insufficient, it is
output by a boiler. When exhaust heat exceeds
the capacity of the heat storage tank, the excess
amount is released using a radiator. In the analysis
of this chapter, the daily radiation loss of the heat
storage tank becomes 5%. Generation efficiency is
calculated by dividing the power demand amount
by the calorific power of the consumed city gas
fuel. Moreover, the value that divides the value

adding the production of electricity and the amount
of heat output by the calorific power of the city
gas fuel supplied to a system is defined as overall
efficiency. The carbon dioxide emissions of the
system for every sampling time are obtained by
giving a load factor to Figures 16 (b) and 17 (b),
while total carbon dioxide emissions are calculated
by adding these values and the amount of carbon
dioxide discharged by the boiler.
Analysis Results and Discussions

Generation Efficiency and
Fuel Consumption
Figure 20 (a) shows the analysis output of the annual average generation efficiency, and Figure 20
(b) shows the result of the calorific power of the
city gas fuel consumed each year. However, the
Figure 19. Power and heat demand for ten houses apartment in Tokyo
188
Figure 20. Result of annual average power-generation efficiency and fuel consumption
fuel quantity consumed by a boiler is not included

in Figure 20 (b). The average generation efficiency
in the case of PEFC and NEG of 5 kW peaks in
this case, as shown in Figure 20 (a). The generation
efficiency in this case is 1.37 compared with the
PEFC individual system, and is 1.28 compared
with the NEG individual system. This difference is
equivalent to the difference in the fuel consumption
shown in Figure 20 (b). The conventional method
illustrated is the value where the presumed value
in a thermal power station is power generation
efficiency of 40%, while transmission loss is not
included. Since the power generation efficiency of
the proposed power system is lower than a thermal
power station, there is a considerable fuel amount
of heat for the proposed system and it is difficult
to improve the power generation efficiency of
the independent distributed power supply rather
than that of a conventional system. An independent distributed power supply has the economic
advantages of low power transmission loss and
waste heat loss. Therefore, when exhaust heat can
be used effectively, or when the distance between
a power plant and demand area is considerable,
the merit of an independent source becomes clear.

However, it is difficult for the proposed system to
be economically feasible. Generation efficiency
is higher with the installation of 10-kW NEG
rather than that of 10-kW PEFC. If the energydemand pattern shown in Figure 19 is installed
into 10-kW PEFC or 10-kW NEG, the generation efficiency of PEFC will fall compared with
NEG. Therefore, a system (NEG) with sufficient
generation efficiency at low load has a stronger
influence than one (PEFC) with high generation
efficiency at high load.
Heat Storage Tank and Boiler Capacity

Figure 21 shows the analysis result of the exhaust
heat of a representative day every month. The
system with the most exhaust heat is 10-kW NEG
(Figure 21 (a)), while 10-kW PEFC produces the
least, due to the difference in the Operation Area
A of the system operation map, as shown in Figure
18. In other words, although the increase in exhaust
heat is lost due to the increase in the production
of electricity of PEFC, NEG changes rapidly. For
189
Figure 21. Result of the exhaust heat output of an every month representation day
190
this reason, when the power load alters sharply,

the exhaust heat characteristics of PEFC and NEG
differ considerably. Based on the result of Figure
21, the amount of exhaust heat increases, so that
the NEG capacity does the same. Moreover, since
there is little heat demand in the summer season,
there is considerable surplus exhaust heat.
Figure 22 shows the analysis result of the investigation into the relation between heat storage
tank capacity and boiler capacity. Since exhaust
heat characteristics also vary according to the
capacity of PEFC and NEG, the optimum heat
storage and boiler capacity differ for each system.
As shown in the example of Figure 18 (c), there
is little exhaust heat of PEFC and NEG that composes HCGS compared with Figures 18 (a) and
(b). This is attributable to the low capacity of
PEFC and NEG. Consequently, periods of time
when the exhaust heat of HCGS and the heat
supply of a time shift of the heat storage tank do
not satisfy heat demand appear. The boiler capacity installed into HCGS is smallest in the order
of 5-kW PEFC and 5-kW NEG, 3-kW PEFC and
7-kW NEG, and 7-kW PEFC and 3-kW NEG
respectively. This order is the same as that of the
average generation efficiency described in Figure
20 (a). In other words, if generation efficiency is

high, the amount of exhaust heat will decrease,
and periods of heat shortage emerge.
Carbon Dioxide Emissions

Figure 23 shows the analysis result of the carbon
dioxide emissions of a representative day every
month. The carbon-dioxide emissions of the
boiler are included in Figures 23 (b), (c), and
(d). On the other hand, in Figures 23 (a) and (e),
heat demand can be fulfilled by storing the exhaust heat of HCGS. Since a large cooling load
is added to the system in the summer season,
carbon dioxide emissions are considerable when
compared with other seasons. Although HCGS
(Figures 23 (b), (c) and (d)) is accompanied by
the boiler operation, there is a greater amount
of carbon dioxide discharged by power generation than that discharged by the boiler operation.
Figure 24 shows the carbon-dioxide emissions of
the conventional method (commercial power and
boiler). Moreover, Figure 25 shows the result of
the annual carbon-dioxide emissions using the
conventional method and the proposed system.
The amount of carbon-dioxide-emissions using
Figure 22. Analysis result of the heating storage capacity and boiler capacity when satisfying a heat
balance
191
Figure 23. Result of the carbon-dioxide emissions of an every month representation day
192
Figure 24. Result of the carbon-dioxide emissions for conventional method
Figure 25. Result of the annual carbon-dioxide emissions
the conventional method is calculated based

on the investigative commission report of the
calculation method of the amount of greenhouse
gas discharge (Ministry of the Environment in
Japan, 2003). There is a peak of 10 kW PEFC
in the result of Figure 25. This is because of the
considerable load fluctuation of power and the
considerable operation frequency of a low-load
region in a general house, while the reduced carbon
dioxide emissions of a system depend more on a
system (NEG) with better operation characteristics
under partial load than one (PEFC) with a large
maximum efficiency point.
Conclusions
The PEFC and NEG hybrid cogeneration system
(HCGS) equipped with a proton-exchange membrane fuel cell (PEFC) at base load operation and
the hydrogen mixing gas engine (NEG) of a load
that follows operation were examined. In this
chapter, generation efficiency, carbon dioxide
emissions, and the optimal capacity of the boiler

and heat storage tank were investigated regarding the installation of HCGS into 10 household
apartment houses in Tokyo. Consequently, the
following conclusions were obtained:
1. When the capacity of PEFC and NEG that
composes HCGS was changed and the
annual average generation efficiency was
investigated, the case of each remaining at 5
kW was good at 27.3%. This value was 1.37
times that of the PEFC individual system,
and 1.28 times that of the NEG individual
system respectively.
2. The capacity of PEFC and NEG that constitute HCGS was changed, and the relationship between the heating storage capacity
and boiler capacity was clarified. Since the
exhaust heat decreases to enhance the average
generation efficiency, a large boiler capacity
is required.
193
3. NEG independent systems produce a small

annual amount of carbon dioxide emission.
When inputting the energy-demand pattern
of a general house into HCGS, the carbon
dioxide emission characteristics in a partialload region at low load have a significant
influence.
Larsen, J. F., & Wallace, J. S. (1995). Comparison

of emissions and efficiency of a turbocharged
lean-burn natural gas and hythane-fuelled engine.
ASME ICE, 24, 3140.
REFERENCES
Mohammadi A., et al. (2005). Development of

highly efficient and clean engine system using
natural-gas and hydrogen mixture fuel obtained from onboard reforming (NEDO report
ID:03B71006c). In Japanese
Adam, H., & Matthew, L. (2005). Impacts of

temporal precision in optimization modeling of
micro-combined heat and power. Energy, 30(10),
17591779. doi:10.1016/j.energy.2004.11.012
Andrew, W., & Steven, F. (2007). The nature of
domestic electricity-loads and effects of time
averaging on statistics and on-site generation
calculations. Applied Energy, 84(4), 389403.
doi:10.1016/j.apenergy.2006.09.008
3, 3-6. In Japanese
673681. doi:10.1109/TIA.2005.847277
of hot water supply and heating of an area. In
Proceedings of the Meeting of the Society of Heating, Air-Conditioning and Sanitary Engineering,
(pp. 1745-1748).
Japanese Ibaraki Prefecture Government Office of
Education. (2002). Modeling of hydrogen energy
system (High school active science project research
report). Ibaraki, Japan: Author. In Japanese.
194

Nagase, O., et al. (2003). Quantification of the

energy consumed in a house. In Proc. 19th Energy
461-466. In Japanese.
Nakamura, T., & Sei, M. (2002). Energy related
technology. High-efficiency fuel processor for
fuel cell system (Technical Report 77 4-9). Japan:
Matsushita Electric Works, Ltd. In Japanese.
Oda, K. etal. (1999). A small-scale reformer for
fuel cell application. [In Japanese]. Sanyo Technical Review, 31(2), 99106.
6, 4285-4290.
Society of Living Amenity. (2006). Family-size
fuel cell cogeneration systems development of
Tokyo Gas. ALIA News, 89. In Japanese.
Swain M. R., & Usuf M. J. (1993). The effects of
hydrogen addition on natural gas engine operation.
SAE Tech Pap Ser, SAE-932775.
urban energy. [In Japanese]. Transaction of the
Society of Heating, Air-Conditioning and Sanitery
Eng. of Japan, 79(7), 573579.
Yasuda I. (2005). Development of hydrogen production technology for fuel cell. Energy Synthesis
Engineering, 28(2). In Japanese
the Society of Heating, Air-Conditioning and Sanitary Engineering, (pp. 1729-1732). In Japanese
Yap D., et al. (2004). Effect of hydrogen addition

on natural gas HCCI combustion. SAE Tech Pap
Ser, SAE-2004-01-1972.
195
APPENDIX
Nomenclature

C : Capacity kW
C : Maximum load kW
E : Power kW
E F : Power of an inverter outlet [kW]
E g : Production of electricity [kW]

E : Power load [kW]
F : Fuel cell
G : CO2 emission [g/s]
G : CO2 emission [g/(skW)]
N : Gas engine generator
N E : Installed number of NEG
P : Capacity of generator [kW]
Pc : Interconnection power with other grids [kW]
Pg : Production of electricity of the generating equipment of a system interconnection grid [kW]

Q : Amount of fuel supply [kg/s]
q : Primary power input [kW]
R : Load factor [%]
S : Boiler
t : Sampling time
U : Flow rate [m3/s]
Greek Symbols

: Efficiency [%]
Subscripts

196
B : Burner installed in a reformer

E : City-gas engine generator (NEG)
Er : Reforming gas
F , FC: Fuel cell
l : Building model with PEFC system and NEG
m : Building model with NEG
n : Building model without PEFC system and NEG
p : City-gas engine
R : Reformer
S : Reformer burner
Equipment

B: Boiler
C/O: Carbon monoxide oxidation equipment
CGS: Cogeneration
DC/AC: DC-AC converter
FC: Fuel cell
G/T: Generator
HCGS: Hybrid cogeneration
I/T: Inverter
NEG: Gas engine cogenerator
R: Reformer
PEFC: Proton-exchange membrane fuel cell
Vb: Valve
197
198
Chapter 7
Independent Microgrid
Composed of Distributed
Engine Generator
ABSTRACT
This chapter consists of two sections, Energy Cost of an Independent Microgrid with Control of Power
Output Sharing of a Distributed Engine Generator and Improvement of Power Generation Efficiency
of an Independent Microgrid Composed of Distributed Engine Generators. In the 1st section, small
kerosene diesel-engine power generators are introduced into an independent microgrid, and power and
heat are supplied to 20 houses. If the proposed system is introduced into a community with little heat
demand, effectiveness will decrease greatly. The 2nd section investigates the power generation efficiency
and power cost of an independent microgrid that distributes the power from a small diesel engine power
generator. When the number of distributions of the engine generator is installed, the cost of the fuel
decreases.
In the 1st section, small kerosene diesel-engine
power generators are introduced into an independent microgrid (IMG) that connects 20 houses,
and power and heat are supplied to them. A 3 kW
engine generator is installed in six houses, and a
boiler and a heat storage tank are also installed,
and exhaust heat to make up for insufficiency is

supplied. The boiler is installed in the house that
does not install the engine generator, and heat is
supplied to the demand side. Partial load operation
of the engine generator has a large influence on
power generation efficiency. Therefore, this study
discusses the system that controls the power of the
engine generator by the power distribution control
DOI: 10.4018/978-1-4666-5796-0.ch007
Independent Microgrid Composed of Distributed Engine Generator
system using the genetic algorithm (GA), and the

control system that changes the number of operations of the engine generators according to the
magnitude of the power load. As a case study, the
energy-demand model of the 20 houses in Sapporo
was analyzed. As a result, the annual energy cost
of the number of operations system and the power
distribution control system is reducible with 16%
and 8% compared with the conventional method,
respectively. However, it depends for this cutback
effect on the heat demand characteristic greatly,
and when the proposed system is introduced into a
community with little heat demand, effectiveness
will decrease greatly.
The power generation efficiency and power cost
of an independent microgrid that distributes the
power from a small diesel engine power generator
was investigated using numerical analysis. The
fuel consumption of a small diesel engine and
the relation between power generation and heat
power were obtained in experiments using a prototype. The independent microgrid built using one
to six sets of 20 average houses in Sapporo and
the distributed engine generators were examined
using these test results. However, the operation
of a diesel engine power generator controls the
number of operations according to the magnitude
of the power load of the microgrid. When a diesel engine power generator is distributed, since
the power generation capacity per set decreases
compared with the central system, the load factor
of each engine generator rises. As a result, the
operation of an engine at partial load with low
efficiency can be reduced. When the number of
distributions of the engine generator increases as
a result of numerical analysis, the cost of the fuel
decreases. However, when the rise in facility cost
is taken into consideration, the number of engine
generators for distribution is in fact 3 or 4.
ENERGY COST OF AN
INDEPENDENT MICROGRID WITH
CONTROL OF POWER OUTPUT
SHARING OF A DISTRIBUTED
ENGINE GENERATOR
Introduction
The independent microgrid (IMG) that accommodates power and heat independently without
connecting with other power systems is expected
to back up power supply in an emergency, and
at peak cuts of a power plant, and the effective
use of exhaust heat is anticipated (Alibhai et al.,
2004; Carlos & Hernandez, 2005; Abu-Sharkh et
al., 2006; Robert 2004). Usually, since the energy
transport distance of a microgrid is short, effective use of exhaust heat is possible. The change
factor of the power load changes over a short
period, such as with an inrush current, and there
is a change over a long period in the demand. In
IMG with large load fluctuation, the dynamic
characteristics of the grid and fall in generation
efficiency due to partial load are subjects. On
the other hand, because the power is supplied to
various demand sides, the load of a large-scale
power generation system is leveled. Load fluctuations can be leveled by installing a battery in
IMG (Gouda et al., 2004). However, installation
of a battery is disadvantageous when facility cost
and maintenance are taken into consideration. In
IMG that does not have a battery installed, the
power generator connected to the grid is expected
to have frequent partial-load operations with low
efficiency. As a result, it is expected that power
generation efficiency in the entirety of distributed
power generators is inferior to large-scale existing
power facilities. Therefore, in this Section, regarding IMG introduced into city areas, the operation
plan of each engine generator is investigated for
199
the efficiency fall of the distributed small engine

generator by numerical analysis, and improvement
is attempted.
The distributed engine generator supplies the
power and heat to the house that installed the
facility first. Furthermore, the surplus power of
the engine generator is supplied to other houses
through IMG. The heat demand of the houses in
which an engine generator is installed is supplied
by the heat power of the exhaust heat of the engine
generator, backup boiler, and heat storage tank.
In houses in which an engine generator is not installed, a boiler is operated and heat is supplied.
Therefore, the cost of the heat in each house differs
according to the operation method of the engine
generator. Then, a controller adjusts the power
distribution of the engine generator connected to
IMG, and the operation method aiming at high
power generation efficiency and effective use of
exhaust heat is examined. Therefore, this Section
discuss the system that controls the power of engine generators by the power distribution control
system using the genetic algorithm (GA) (Baker,
1985; Goldberg, 1989), and the control system
that changes the number of operations of engine
generators according to the magnitude of power
load. GA can apply to a nonlinear many-variables
problem, and there are many introduction cases
until now (Baker, 1985; Goldberg, 1989). Since
GA is a random search, there are few constrain
conditions, and judgment of a designer is hard to
be taken into consideration. The microgrid can
use exhaust heat with few heat radiation compared
with large-sized power facilities. For that reason,

if the microgrid is used in an area with much heat
demand, it will be predicted that the reduction effective of energy cost is large. Therefore, in this
Section, the energy-demand model of the house
in Sapporo with much heat demand is analyzed.
In this Section, the energy efficiency, the energy
cost, and the operation method when using IMG
by these two control methods are investigated.
System Description
Independent Microgrid Configuration
A schematic figure of IMG constituted from the
engine generator assumed in this Section is shown
in Figure 1. The engine generator is installed in
20 houses. As houses connected to the IMG,
there are houses that installs the engine generator (Figure 2), and houses that is not installed. A
boiler is installed in all 20 houses and the heat of
the house with that the engine generator is not
introduced is supplied from the boiler. On the
other hand, in the house that installs the engine
generator, the heat storage tank for exhaust heat is
installed. Among the 20 houses connected to the
IMG, engine generators are installed in 6 houses.
In addition, the power generation capacity of the
engine generators is 3 kW, and the output characteristics of all the engine generators are set to
be the same. The engine generator consists of a
kerosene diesel-engine and a synchronous power
generator. The power generated with the engine
Figure 1. Diesel engine independent micro-grid system
200
Figure 2. Individual house with engine-generator connected to an independent micro-grid system
generators are supplied to each house through a

system interconnection device and the IMG. The
system interconnection device also intermediates
the connection of the load and the IMG. The synchronization of the generated alternating current
power and the power of the IMG is controlled by
the system interconnection device.
The system configuration of the engine generator installed in each house is shown in Figure
2. The exhaust heat outputted with an engine
generator is engine-cooling water and exhausts
gas, and this heat is stored in a heat storage tank
through a heat exchanger. In addition, the heattransfer-medium heat exchanger for supplying
heat to a house is installed in the heat storage tank.
After exchanging this heat transfer medium for
exhaust heat in the heat storage tank, it is supplied
to a backup boiler. When there is a large amount
of exhaust heat compared with the heat load, the
excess exhaust heat is stored in the heat storage
tank. On the other hand, when the exhaust heat
exceeds the heating storage capacity of the heat
storage tank, the exhaust heat is released from a
radiator. In houses that the engine generator is not
installed, power generated with the engine generator of other houses through IMG is used.
Operating Method
Figure 3 (a) shows the power generation efficiency
model of the central system of the engine generator,
and Figures 3 (b-1), (b-2) and (b-3) show the power
generation efficiency model of the distribution
system. In the central system, the power is supplied
with the one-set engine generator. The capacity
of the engine generator shown in Figure 3 (a) is
9kW. Here, load factor is the value that divided
the amount of power generation by the capacity
of the power generator. As shown in Figure 3 (a),
the power generation efficiency in the case of corresponding to 3.0, 4.5, 6.0 and 9.0 kW load with
one set of engine generator (E/G (0)) are 18, 24,
27, 30%. Figures 3 (b-1), (b-2), and (b-3) are the
models of the power generation efficiency of the
distributed system. In this distributed system, it
is assumed that a three-set engine generator, E/G
(1), E/G (2), and E/G (3) with a power generation
capacity of 3 kW, are installed. When 3.0, 4.5, 6.0,
and 9.0 kW power load are added to IMG with
the distributed system, each power generation
efficiency is 30, 29, 30, and 30%, respectively.
It is because a three-set engine generator can be
operated at the maximum efficiency point to 3,
201
Figure 3. Installed numbers of an engine generator, and efficiency model
6, and 9kW power load. For example, when 6kW

load is added to IMG, E/G (1) and E/G (2) will
be operated at the maximum efficiency point, and
E/G (3) will stop. Furthermore, when 4.5kW load
is added to IMG, for example E/G (1) and E/G (2)
are operated with a 2.25kW output, it can operate
at efficiency higher than Figure 3 (a). Therefore,
this Section investigates the operation method
of IMG regarding the power distribution control
method of each engine generator, and the control
of the number of operations method of the engine
generators.
Control System of Engine Generators

The engine generators connected to the IMG are
controlled by the power distribution control system and the control of the number of operations
system. In this Section, a power distribution control system is analyzed using the genetic algorithm
(GA). The chromosome model (Figure 4 (a))
introduced into GA expresses the load ratio of six
202
engine generators, and describes each gene group

by 8 bits of 0 and 1. The production of electricity in sampling time t of each engine generator is
calculable from the amount of power demand and
load rate of the IMG. An objective function is the
minimization of the fuel cost of one year, as Equation (1) shows. It is evaluated that the value with
a smaller fuel cost of one year shown by Equation
(1) of the adaptive value of GA is higher. The fuel
cost of one year is the sum total of the fuel cost
of an engine generator (Ye,t ,n ), the fuel cost of the
boiler introduced into a house that the engine
generator is installed (Ybe,t ,n ), and the fuel cost of
the boiler introduced into a house that the engine
generator is not installed (Ybn ,t ). Figure 5 shows
the analysis flow of the GA. In the GA, many
chromosome models (load ratio of engine generators) are prepared at random, and let these be
initial generations. The number of individuals
with a high adaptive value of objective function
Equation (1), initial-generation chromosome
Figure 4. The chromosome model used for genetic algorithm
Figure 5. GA analysis flow
model group, is increased, and it is operated so

that a low adaptive value may be selected. Mutation is added to the chromosome model group,
and the generation is changed, maintaining as
much collective variety as possible. The mutation
operation of a chromosome model group and
selection / increase operation are repeatedly calculated until it results in the last generation that
has been determined beforehand. In addition, the
model with the highest adaptive value is determined to be the optimal load ratio of the engine
generators in the last-generation chromosome
model group.
Yday =
Year 6

minimization (Ye,t ,n + Ybe,t ,n ) + Ybn ,t

t =0 n =1
(1)
203
Ye,t ,n = C fuel Fe,t t
(2)
Ybe,t ,n = C fuel Fbe,t t
(3)
Ybn ,t = C fuel Fbn ,t t
(4)
In the operation of mutation, a parent chromosome model is first selected at random under the
mutation probability given beforehand. Moreover,
the position of the varying gene is decided at
random. If the original value of a gene is 1, it
changes to 0; if it is 0, it changes to 1.
Diesel-Engine Generator System

Engine Generator Specification
The test equipment of the cogeneration system
using a small-scale diesel engine is shown in
Figure 6 (Obara et al., 2005). The analysis of this
Figure 6. Test diesel engine co-generation system
204
Section uses the experimental results obtained by

the operation of the test equipment of Figure 6. The
engine specifications of the cogeneration system
of Figure 6 are shown in Table 1. In addition, the
specifications of a synchronous power generator
are shown in Table 2. The fuel of a diesel engine
is kerosene, and the engine has 2 cylinders with
4 cycles. The power generator is a single-phase
synchronous type, and the power is transmitted
from the power axis of the diesel engine.
Output Characteristics of a Small-Scale

Diesel-Engine Co-Generation (Obara et
al., 2005)
The examination results of the calorific power
of the kerosene fuel supplied to the small dieselengine power generator are shown in Figure 6,
and the heating value of the engine-cooling water,
the heating value of engine exhaust gas, and the
production of electricity are shown in Figure 7. If
Table 1. Engine specifications
Table 2. Generator specifications
Figure 7. Characteristics of diesel engine generator output. Engine number of revolutions is

1600rpm
the kerosene fuel supplied to an engine is increased,

the production of electricity and the heating value
of the exhaust gas will increase, but the heating
value of the engine-cooling water will decrease.
The kerosene supply heating value at the time of
3 kW maximum output is 9.8 kW. Figure 8 shows
the examination results of the production of elec-
tricity and the power generation efficiency of a

diesel-engine power generator, as well as the load
factor and power generation efficiency. Although
the engine number of rotations was changed and
examined from 1440 rpm to 2000 rpm, the influence on power generation efficiency was small.
An approximate expression of each examination
result is shown in the figure. These approximate
expressions are used in the calculation of the
power generation efficiency of the case study.
Generally, in an engine generator of the same
form, even if the maximum engine production of
electricity differs, the characteristics of the load
factor and power generation efficiency can be
managed similarly. Then, similar to the relation
between the load factor and the power generation
efficiency of the engine generator used in the case
study, it is managed without being dependent on
the power generation capacity. The relation of the
load factor and exhaust heat output of the 3kW
engine generator shown in Figure 9 was prepared
based on the result of the experimental device
shown in Figure 6. The exhaust heat output characteristic of the 18kW engine generator in Figure
9 was calculated from the experimental result of
a 3kW engine generator. However, the maximum
power-generation efficiency of the 18kW engine
generator increases compared with 3kW so that
following Section may describe. For this reason,
the exhaust heat characteristic of the 18kW engine
generator is not 6 times the 3kW engine. In the
case study described in the following Section, the
exhaust heat output characteristic of the engine
generator shown in Figure 9 is introduced.
Case Study
Analysis Program
The calculation of power cost takes the unit price
of kerosene fuel. In this Section, the kerosene
cost is set at 466 dollar/m3 (Fuji Futures, 2005).
Moreover, the conversion rate of the dollar to the
205
Figure 8. Diesel engine generator efficiency
yen is 1 dollar = 118 yen. For the operation characteristics of the power distribution control method
of an engine generator, software for the analysis
flow of Figure 5 is prepared and investigated.
Analysis Conditions of GA
In the analysis of the power distribution control
method of the engine generator using GA, the
number of generations is 5000, the number of
chromosome models is 3000. Here, the model used
for analysis is the chromosome model shown in
Figure 4 described in previous Section. Operation
of mutation is added to 10-6% of the gene models
shown in Figure 4 among all the chromosome
models. The generation number is the number of
the rough indications to convergence achievement
Figure 9. Exhaust heat characteristics
206
of the analysis. The generation number when

converging is still smaller. In addition, in each
generation, the adaptive value increases in the
50 highest chromosome models, and selects 50
chromosome models with a low order of adaptive
value. Each of these parameters was decided by
trial and error. The chromosome model introduced
in this Section cannot express the method of controlling the number of operations of the engine
generator. Although this is the reason for using
the output characteristics of the engine generator
shown in Figure 7 for the analysis program of GA,
even when the power is zero, the heat power of
the engine-cooling water and exhaust gas is not
zero. The chromosome model cannot express the
stop of an engine generator.
Energy Demand Models

The power demand pattern of Figure 10 (a) is the
measurement result of an individual house for a
representative day in February in Sapporo, Japan
(Narita, 1996). Moreover, Figure 10 (b) shows the
heat demand pattern of each month of an average house in Sapporo (Narita, 1996). The power
demand pattern does not change significantly by
each month. This is because there is no cooling
load in the summer, and the exhaust heat and the
heat of the auxiliary boiler are supplied to a house
assuming the heating load of winter. In the case
study, the load factor, the heating value of the
cooling water, the heating value of the exhaust
Figure 10. Average daily power and heat needs of a cold region house with traditional heating system space
gas, the fuel consumption, and the power generation efficiency of each engine generator are
calculated using the power demand pattern of the
representative day shown in Figure 10 (a). The
analysis model of IMG connects and constitutes
20 houses of the power pattern of Figure 10 (a).

Power Distribution Control of
Each Engine Generator
1. Load Factor of Each Engine Generator
Figure 11 shows the result of analyzing the load
factor of six engine generators placed in the
power distribution control system. The power
demand model used for analysis is February
representation day shown in Figure 10 (a). In the
analysis using GA, the value of the power load
distribution rate with six engine generators can
be obtained. From the power load distribution
rate allocated to each engine, load factor of each
engine shown in Figure 11 was calculated. However, the determined the power load distribution
rate is given to six engines at random. If it is the
same sampling time, planning of each load factor
of EG (1) to EG (6) shown in Figure 11 replaces

for other engine generators. Figure 12 shows the
result of the power generation efficiency of the
central system of one set of 18kW engine generators, and the power distribution control system
of six sets of 3kW engine generators shown in
Figure 11. Distribution of the engine generator
will increase power generation efficiency clearly
compared with the central system, as shown in
Figure 12. Figure 13 shows the operation plan
that replaces the result of the load factor of each
engine generator shown in Figure 11, and gives
a load to descending order from E/G (1) to E/G
(6). In the operation plan shown in Figure 13, the
power generation efficiency is the same as power
distribution control system shown in Figure 12.
Thus, the result of power generation efficiency of
the operation plan shown in Figure 11 and Figure
13 is equivalent.
2. Fuel Cost
Figure 14 shows the analysis results of the calorific
fuel power consumed by the engine generator
installed in 6 houses on a representative day in
August and February. The power demand pattern shown in Figure 9 and the analysis results
207
Figure 11. Analysis result of a load factor with the power distribution control of six engine generators
Figure 12. Fuel consumption of all the engine generators with power distribution control
208
of Figure 14 are similar figures, and the analysis

using GA shows the operation characteristics that
follow the power demand. Figure 15 shows the
results of the cost of the fuel consumed by the
boiler installed in 14 houses in which an engine
generator is not installed.
Figure 16 shows the results of the fuel cost

when supplying energy to 20 houses connected
to IMG by the power distribution control of the
engine generators in February and August on a
representative day. The sum totals of the cost of
the fuel consumed by 6 engine generators and the
Figure 15. Boiler cost analysis results of individual houses without engine generator
209
Figure 16. Cost analysis results of IMG with the power distribution control of six engine generators
cost of the fuel consumed by 6 boilers are shown

in Figure 16 (a) and Figure 16 (b). When the results
of February and August are compared, there is
no large difference in the fuel cost of the engine
generators, but the cost of the fuel for the boilers
is significantly different. Compared with the
representative day in August, the results of the
representative day in February show that the fuel
cost of the engine generator is 1.02 times, the cost
of the boiler fuel is 12.7 times, and the total is 2.6
times higher.
Control of the Number of Engine

Generators
1. Relation of the Capacity and the Maximum
Efficiency of an Engine Generator
Generally, if the capacity of an engine generator
increases, efficiency will also increase. Therefore,
Figure 17 is used for the relation of the capacity

and the maximum efficiency of an engine generator in analysis. The relation shown in Figure 17
was obtained from cogeneration products using
a small-scale diesel power plant (YANMAR Co.,
Ltd., 2006).
2. Output Characteristics of Each Engine
Generator
Figures 18 show the analysis results of the heat of
the engine-cooling water, the heat of the engine
exhaust gas, and the power generation of each
engine generator for distribution with six engine
generators while the number of operations is
controlled. The heat of the engine-cooling water
changes little, even if the power load changes
sharply according to the output characteristics of
the engine generator shown in Figure 7. Therefore,
if the demand characteristics of the power and the
Figure 17. Rate of power generator capacity and the maximum power generation efficiency
210
Figure 18. Energy output characteristics when controlling the number of operations of six engine generators
heat in Sapporo shown in Figure 9 are taken into

consideration and the engine generator operates
from 23:00 to 05:00 with little power demand, a
significant amount of heat from the engine-cooling
water can be obtained.
3. Power Generation Efficiency

Figures 19 show the analysis results of the power
generation efficiency and the load factor of each
sampling time of an engine generator with arranged
distribution. The load factor of the central system
211
is 28 to 89% of the range, and the power generation

efficiency changes from 13% to 28% of the range.
On the other hand, E/G (1) of the analysis results
of the number of engine generators is 6 as shown
in Figure 19, and is 100% of the load factor at all

sampling times. From 02:00 to 05:00, E/G (2)
has a load factor of 70% or more. The load factor
is 100% in other time zones. E/G (3) has a load
Figure 19. Load factor and generation efficiency characteristics when controlling the number of operations of six engine generators
212
factor of 100% in the 06:00 to 09:00 time zone

and the 16:00 to 23:00 time zone. E/G (1) to E/G
(3) all show an operation pattern with a high load
factor compared with the central system. On the
other hand, E/G (4) and E/G (5) have a high load
factor from 16:00 to nighttime, but in other time
zones, the load factor shows a low operation pattern. Regarding E/G (6), the operation of the load
factor at a maximum of 33% takes place only from
16:00 to 22:00, and the hours worked are short. If
an engine generator is distributed and arranged,
the operating ratio of the engine generator will be
distributed. If the efficiency of an engine generator
that has stopped is not taken into consideration, the
power generation efficiency of the whole system
of the distribution system of engine generators
is improved compared with the central system.
However, if the engine generator stops, an engine
generator with short working hours arises. If the
life of the whole engine generator linked to IMG
is taken into consideration, periodically replacing
the starting order of the engine generator of E/G
(1) to E/G (6) is required, for example.
4. Power Costs
Figure 20 shows the analysis results of expressing
the number of distributions of an engine generator,
and the amount of consumption of kerosene fuel
on a representative day with the calorific power.
However, the result shown in Figure 20 is only the

result of the fuel used for power generation, and the
fuel used for heat supply is not included. When the
number of engine installations is one, it shall be
the central system, and the capacity of the power
generator shall be 18 kW. This value can cover
20 houses with the maximum amount demanded
of the power shown in Figure 9. Moreover, 2, 4,
and 6 engine installations show 9 kW, 4.5 kW,
and 3 kW in capacity, respectively. The relation
between engine capacity and power generation
efficiency was calculated using Figure 17. If the
number of distributions of an engine generator
increases as shown in the figure, fuel consumption
will decrease. However, when four distributions
of an engine generator is compared with six, the
difference is 6%, and even if the number of installations is further increased, the reduction effect
of fuel consumption shows slight improvement.
Figure 21 shows the analysis results of the
years of operation of IMG in which engine generators are distributed, and the cost of the kerosene
fuel used for power generation. However, the
result shown in Figure 21 is only the result of the
fuel cost for power generation; the fuel cost for
heat supply is not included. The years of operation
were analyzed as 3 or 10 years. An increase in the
number of distributions of an engine generator
will reduce the cost of kerosene consumption as
Figure 20. Result of the relation between the installed numbers of an engine generator, and the fuel
consumption
213
Figure 21. Analysis results of the number of installations of an engine generator, and kerosene fuel cost
shown in Figure 20. In particular, when the number of distributions of an engine generator is 2,
compared with the central system, the decline is
large, and there is a 27% reduction in kerosene
cost. If the number of engine generators for distribution is increased, the cost of the kerosene
fuel used for power generation will decrease, but
the ratio of the effect becomes so insignificant
that the number of engine generators increases.
If the rise in facility costs due to the increase in
the number of installations of an engine generator
is taken into consideration, even if the number of
distributions of an engine generator is determined
to be 2, a reduction in kerosene fuel cost is fully
attained. Because the transportation distance of
exhaust heat will increase when the number of
installation of an engine generator is reduced, the
boiler cost in winter increases.
Relation between the Control Method

and Fuel Cost
Figure 22 shows the analysis results of all the
engine generators of the 20 houses connected to
IMG, and the cost of fuel consumption by all the
boilers. The heating storage capacities connected
to IMG are a total of 1620 MJ. The following
Section describes the comparison results of the
fuel cost of the control of the number of operations method and the power distribution control
method. Compared with the power distribution
control system, the analysis results of the control
of the number of operations system show the fuel
cost of an engine generator to be 74%, the cost
of boiler fuel to be 108%, and the total cost to
be 96%. The power distribution control system
has much exhaust heat from the engine genera-
Figure 22. Cost analysis results of IMG with the number control operations of an engine generator
214
tors compared with the control of the number of

operations system, and the cost of the boiler fuel
is low. On the other hand, in the control of the
number of operations method, the fuel cost of
the engine generators is low compared with the
power distribution control method, and total cost
falls by 4%. Compared with the power distribution control system, the control of the number of
operations system can be performed at low cost
of power generation fuel.
Figure 23 shows the result of the annual energy cost of each operating method. As shown in
Figure 23, the annual energy cost due to the microgrids is cheaper than the conventional method.
Compared with the conventional system, the reduction rate of the fuel cost of the control of the
number of operations system is 16%. And, compared with the conventional system, the reduction
rate of the fuel cost of the power distribution
control system is 8%. The generation efficiency
of the engine generator with load fluctuation does
not exceed large compared with commercial
power. Therefore, the reason the energy cost of
the microgrid is advantageous is that a power unit
price and heat supply cost differ from the conventional method. In a microgrid, heat cost is cheap
by a waste heat recovery. From this, the energy
supply cost of the microgrid changes in the energy demand characteristic to introduce. Only by
generating cost, the control of the number of
operations system is reduced 22% compared with
the conventional method. On the other hand, the

power distribution control system increases
slightly. When a system is introduced into communities with few heat demand amounts, the
difference of the heat cost shown in Figure 23
will become small. And the advantage of the
energy cost of the microgrid decreases. Here, the
basic monthly charge for commercial power is
8.28 dollars, and the unit price is 0.205 dollar/
kWh. Compared with the energy total cost of the
conventional method, the reduction of the number
of control operations and of the power distribution
control operations is 38% and 33%, respectively.
The engine microgrid with improvement of partialload operation falls in energy cost compared to
the conventional method.
Conclusions
The relation between two control systems and the
energy costs of the independent microgrid (IMG)
in which small diesel-engine power generators
are distributed were investigated by numerical
analysis. As a result, the following Conclusions
were obtained. Three systems of the conventional
system (commercial power, heat storage tank and
kerosene boiler),the power distribution control
system,and the control of the number of operations
system were evaluated using the energy demand
model of the 20 houses in Sapporo. Compared
with the conventional system, the reduction rate
Figure 23. Cost analysis results of IMG with the number control operations of an engine generator
215
of the fuel cost of the control of the number of

operations system is 16%. And, compared with
the conventional system, the reduction rate of the
fuel cost of the power distribution control system
is 8%. In the microgrid, since engine exhaust
heat is used, heat cost is cheap. The energy supply cost of the system takes influence in a heat
demand characteristic greatly. When the proposal
microgrid is introduced into a community with
little heat demand, the effectiveness over the
conventional system will decrease. When only
the generating cost of the microgrid and the
conventional method is compared, the control of
the number of operations system will be reduced
22%. On the other hand, the power distribution
control system increases slightly.
IMPROVEMENT OF POWER
GENERATION EFFICIENCY OF
AN INDEPENDENT MICROGRID
COMPOSED OF DISTRIBUTED
ENGINE GENERATORS
Introduction
Application of microgrid technology provides
a backup power supply in an emergency, peak
cut of an electric power plant, and effective use
of exhaust heat (Alibhai et al., 2004; Carlos &
Hernandez, 2005; Abu-Sharkh et al., 2006). In
addition, an independent microgrid system (IMG:
Independent Microgrid) that supplies electric
power and heat without interconnecting with
other power systems realizes the advantages of a
distributed power source. It allows the construction
of a power supply system with low environmental
impact that uses renewable energy including green
energy and unused energy. Furthermore, since the
energy transport distance of a microgrid is short,
it uses exhaust heat effectively. Therefore, overall
efficiency is improved compared with conventional power generation systems. However, IMG
is required for the rapid dynamic characteristics
216
that follow changes in power load compared with

a grid interconnected for other systems (Robert,
2004). Although the independent microgrid realizes maximum distributed energy, there are issues
to tackle such as the stability of the dynamic
characteristics of power and the development of
an optimal design method (Tanrioven, 2005). In
addition, potential drops in efficiency at partial
load operation of the generating equipment occurring in the load fluctuations of a grid need to
be improved. The change factor of the power load
changes rapidly, such as during an inrush current,
or a change in a long period for the demand. This
chapter describes changes in a short period as
load fluctuation, and changes in a long period as
demand fluctuation.
When a building linked to a grid is a house,
in IMG, both load fluctuation and demand fluctuation are expected to be large. In a large-scale
power plant, since power is supplied to various
demands, demand fluctuation is smoothed. In
IMG with big load fluctuation, if no electricity
storage system is installed, the operating point
of the power generator will change significantly.
However, if facility cost and maintenance cost are
taken into consideration, installation of an electricity storage system can be avoided. When not
installing electricity storage equipment in IMG,
a power generator with partial-load operations is
expected to frequently operate at low efficiency,
and the power generation efficiency in total is
expected to be lower than that in existing largescale electric power facilities. In this chapter, the
power generation efficiency when constituting
IMG from a distributed small diesel-engine power
generator is investigated by numerical analysis.
The aim is to control the number of operations
of the distributed power generator, and raise the
load per power generator and decrease operation
with a partial load, which has low efficiency.
The fall in the power generation efficiency is
due to partial load and the transportation loss of
exhaust heat decreases, so that the capacity of the
power generator is reduced and the number of
distributions is increased. However, if the number

of power generators increases, the facility cost
and the maintenance cost rise. Therefore, in this
chapter, the relationship among the number of
diesel-engine power generators for distribution,
power generation efficiency and power cost is
investigated by numerical analysis.
System Description
Independent Microgrid Configuration
The schematic figure of IMG constituted from the
engine generator assumed in this chapter is shown
in Figure 24. An engine generator can be installed
in arbitrary buildings. An engine generator consists of a diesel engine and a synchronous power
generator. The power generated with the engine
generator is supplied to each building through a
system interconnection device and IMG. An engine generator can be installed in the machinery
room of another building besides the machinery
room of each building. A system interconnection
device also intermediates the connection of the
load and IMG. The synchronization of the generated alternating current power and the power of
IMG is controlled by a system interconnection
device.
The system configuration of the engine generator installed in each building is shown in
Figure 25. The exhaust heat outputted with an

engine generator is engine-cooling water and
exhaust gas, and this heat is stored in a heat storage tank through a heat exchanger. In addition,
the heat-transfer-medium heat exchanger for supplying heat to a building is installed in the heat
storage tank. After exchanging this heat transfer
medium for exhaust heat in the heat storage tank,
it is supplied to a backup boiler. When there is a
large amount of exhaust heat compared with the
heat load, the excess exhaust heat is stored in the
heat storage tank. On the other hand, when the
exhaust heat exceeds the heating storage capacity of the heat storage tank, the exhaust heat is
released from a radiator.
Control of the Number of Engine

Generators
Figure 26 (a) shows the power generation efficiency model of the central system of an engine
generator, and Figure 26 (b) shows the power
generation efficiency model of distribution. The
model of Figure 26 (a) expresses the relation
between the power of an engine generator with a
power generation capacity of 9 kW, and the load
factor and efficiency. Here, the ratio of production
of electricity to maximum production of electricity
is defined as the load factor. As shown in Figure
26 (a), the power generation efficiency in the
Figure 24. Diesel engine independent micro-grid system
217
Figure 25. Diesel engine co-generation system
Figure 26. Installed numbers of an engine generator, and efficiency model
218
case of corresponding to a 7.5 kW load with one

set (E/G (0)) of engine generators is 29%. The
load factor of the engine generator at this time
is 82%. The model of Figure 26 (b) is a model
of the power, the efficiency, and the load factor
when installing three sets of engine generators
(E/G(1), E/G(2), E/G(3)) with a power generation
capacity of 3 kW. The number of operations of a
three-set engine generator is controlled, and it is
operated so that the power load may be followed.
When the same load (7.5 kW) as Figure 26 (a)
is added, E/G (1) and E/G (2) are operated with
a full load with high efficiency (respectively 3
kW), and E/G (3) corresponds to the remaining
loads (1.5 kW). Since the load of E/G (3) is small,
E/G (3) is the operation of a partial load with
low efficiency. E/G (1) and E/G (2) of the load
factor of an engine generator are 100%, and E/G
(3) is 50%. If the power generation efficiency of
the centralized system of Figure 26 (a) and the
distributed system of Figure 26 (b) are compared,
the distributed system may have a high load factor
per set, and the power generation efficiency in
total may become advantageous. Therefore, this
chapter investigates the difference in the power
generation efficiency of the centralized system of
IMG and the distributed system. The maximum
efficiency of a 9 kW commercial engine generator
is high about 2% compared with a 3 kW engine
generator (Yanmar, 2006). And, although a 3 to
20 kW engine generator is installed in the example
reported in the case study below, the difference of
the maximum efficiency of an engine generator
is about 3% (Yanmar, 2006). In this chapter, the
difference in the efficiency due to engine generator
size is not taken into consideration.
Diesel Engine Generator System

Engine Generator Specifications
The test equipment of the cogeneration system
using a small-scale diesel engine is shown in
Figure 27. The analysis of this chapter uses the
experimental results obtained by operation of the

test equipment of Figure 27. The engine specification of the cogeneration system of Figure 27 is
shown in Table 3. In addition, the specifications
of a synchronous power generator are shown in
Table 4. The fuel of the diesel engine is kerosene,
and the engine has 2 cylinders with 4 cycles. The
power generator is a single-phase synchronous
type, and the power is transmitted from the power
axis of the diesel engine.
Output Characteristics of a Small-Scale

Diesel Engine Cogeneration System
The heating value of the engine-cooling water,
the heating value of engine exhaust gas, and the
production of electricity are shown in Figure
28. If the kerosene fuel supplied to the engine is
increased, the production of electricity and the
heating value of the exhaust gas increases, but
the heating value of the engine-cooling water
decreases. The kerosene supply heating value
at the time of 3-kW maximum output is 9.8 kW.
Figure 29 shows the examination results of the
relation between the amount of kerosene fuel
supplied to the engine cogeneration shown in
Figure 27, and the power and the exhaust heat
output power of the system. The hatched area
in Figure 29 (Area A) is the area of the power
where the exhaust heat is outputted only with the
engine generator, and the other areas (Area B)
are areas that include the heat output power in a
kerosene boiler in addition to the output power
of Area A. Figure 30 shows the examination
results of the production of electricity and the
power generation efficiency of a diesel-engine
power generator, and the load factor and power
generation efficiency. Although the engine
number of rotations was changed and examined
from 1440 rpm to 2000 rpm, the influence on
the power generation efficiency was small. An
approximate expression of each examination
result is shown in the figure. These approximate
expressions are used in the calculation of the
219
Figure 27. Test diesel engine co-generation system
Table 3. Engine specifications
220
Table 4. Generator specifications
Figure 28. Characteristics of diesel engine generator output. Engine number of revolutions is 1600rpm.
power generation efficiency of the case study.

In an engine generator of the same form, even
if the maximum engine production of electricity differs, the characteristics of load factor and
power generation efficiency can be managed
similarly. Then, similar to the relation between

the load factor and the power generation efficiency of the engine generator used in the case
study is managed without being dependent on
the power generation capacity.
221
Figure 29. Fuel supply and energy output of engine generator. Engine number of revolutions is 1600rpm.
Area A corresponds to the condition in which only the engine-generator is working. Operating Area B
corresponds to the condition along which boiler is added to the engine-generator.
Figure 30. Diesel engine generator efficiency
222
Case Study
Analysis Method
Figure 31 shows the analysis program of the model
that distributes two sets of engine generators, E/G
(1) and E/G (2). Each block in Figure 31 expresses
a calculation component. Although E/G (1) is first
operated in the analysis program, when the electricity demand exceeds the maximum production
of electricity of E/G (1), E/G (2) is started further.
E/G (2) is started on the basis of the If branch in
Figure 31. Moreover, the production of electricity
in all the engine generators in operation is equal
to the electricity demand, and an engine generator
that follows the power load is in operation. The
power generation capacity of each engine generator
is inputted beforehand into the analysis program of
Figure 31. Calculation of the fuel consumption, the
heating value of the cooling water, and the heating
value of the exhaust gas is performed using the
engine power characteristics shown in Figure 28.
If an electricity demand pattern is inputted into an
analysis program and the number of operations
of an engine generator is controlled, the power
generation efficiency can be obtained by intro-
ducing the load factor of each engine generator

into Figure 30. The characteristics of the heat of
the cooling water and the heat of the exhaust gas
of each engine generator are calculated as being
relatively the same as the experimental results of
Figure 28, even if the power generation capacity
differs. In this chapter, the number of arrangements
of an engine generator is analyzed regarding the
model of 1 (central system: 18 kW, one set), 2
(9 kW, two sets), 4 (4.5 kW, four sets), and 6 (3
kW, six sets). The calculation of power cost takes
the unit price of kerosene fuel. In this chapter,
the kerosene cost in the U.S. is set as 409 $/m3,
and the kerosene cost in Japan is set as 466 $/ m3
(Fuji Futures, 2005). Moreover, the conversion
rate of the dollar to the yen was set at 1 dollar =
118 yen. The input-output characteristics of the
power of IMG are analyzed using MATLAB (Ver.
7.0) and Simulink (Ver. 6.0) of Math Work Co.
Ltd. Analysis error was managed within 0.01%.
Weather Conditions in Sapporo

The average temperatures in Sapporo for the
sampling time on representative days in February,
May, and August are shown in Figure 32 (National
Figure 31. Block diagram of analysis. Number of diesel-engine generator to be distributed is two.
223
Figure 32. External temperature model in Sapporo
Astronomical Observatory, (2003)). There is no

cooling load in the summer in Sapporo. Electricity demand includes household appliances and
electric lighting, and heat demand comes from
heating, hot water supply, and baths. The area
of an average individual house (a 3- or 4-person
household) in Sapporo is 140 m2 with 2 stories,
and the houses are made of wood.
Energy Demand Models

The power demand pattern of Figure 33 (a) is
the measurement result of an individual house
for representative days in February in Sapporo,
Japan (Narita, K., (1996)). Moreover, Figure
33 (b) shows the heat demand pattern of each
month of an average house in Sapporo (Narita,

K., (1996)). The electricity demand pattern
does not change significantly each month.
This is because there is no cooling load in the
summer and the exhaust heat and the heat of
the auxiliary boiler are supplied to a house assuming the heating load of winter. In the case
study, the load factor, the heating value of the
cooling water, the heating value of the exhaust
gas, the fuel consumption, and the power generation efficiency of each engine generator are
calculated using the electricity demand pattern
of the representative day shown in Figure 33
(a). The analysis model of IMG connects and
constitutes 20 houses of the electricity demand
pattern of Figure 33 (a).
Figure 33. Average daily power and heat needs of a cold region house with traditional heating system space
224

Load Distribution of the
Engine Generator
Figure 34 shows the characteristics of the load
added to each engine generator when distributing
and arranging an engine generator and controlling
the number of operations. In the central system,
one set of engine generators with a power generation capacity of 18 kW corresponds to the
load of all sampling times. On the other hand,
in Figures 34 (b), (c), and (d), it corresponds to
the load by controlling the number of operations
of two sets, four sets, and six sets, respectively,
of engine generators. As a result of analysis, the
load of each E/G (1) of Figures 34 (b), (c), and
(d) is the largest, and the load of E/G (2) is the
next largest. For example, in Figure 34 (c), E/G
(1) is the maximum power at all sampling times,
and the power of E/G (2) has an operating time
that is not the maximum output in the time zone
of midnight to 6 oclock in the morning, or the
time zone of 8:00 to 16:00. E/G (4) is operated
only in the time zone of 16:00 to 21:00. Using the
method of controlling the number of operations as
in this case, E/G (1) shows the highest production
of electricity, and this decreases in the order of
E/G (2), E/G (3), and E/G (4).
Number of Distributions, and Full Force

Power
Figure 35 shows the number of engine generators for distribution, as well as the results of the
breakdown of the energy outputted at each sampling time with each engine generator. Compared
with the central system, fuel consumption on a
representative day decreases in total by the system
shown in Figures 35 (b) to (d). However, in the
7:00 to 9:00 time zone, and 17:00 to 21:00 time
zone, there is little reduction in the fuel consumption of the system of Figures 35 (b) to (d). The
reason is that Figure 34 (b) is E/G (2), Figure 34
(c) is E/G (3), and Figure 34 (d) is E/G (4) and

E/G(6), operating in the 7:00 to 9:00 time zone,
and the 17:00 to 21:00 time zone, with a partial
load with low efficiency as shown in Figures 34
(b), (c), and (d). Although E/G (5) of Figure 34
(d) is operated in the 17:00 to 22:00 time zone,
the load factor of E/G at this time (5) is 50% to
100%. Moreover, there is no operation in the other
time zones. The 7:00 to 9:00 time zone and the
17:00 to 21:00 time zone show much higher fuel
consumption compared with the time zones before and after each figure. The fuel consumption
of an engine generator decreases in the order of
Figures 35 (a), (b), (c), and (d). Therefore, the
fuel consumption resulted in a decrease, so that
the number of distributions of the engine generator increased.
Output Characteristics of Each Engine

Generator
Figure 36 shows the analysis results of the heat of
the engine-cooling water, the heat of the engine
exhaust gas, and the power generation of each
engine generator for distribution. The heat of
the engine-cooling water changes little, even if
it changes the power load sharply according to
the output characteristics of the engine generator
shown in Figure 28. In the 7:00 to 9:00 time zone
of Figure 36 (b), the power generation of E/G
(2) is 2 kW or less, and the load factor is 22% or
less. As a result, in the 7:00 to 9:00 time zone of
Figure 36 (b), the heat of the engine-cooling water
shows a large value as the output characteristics
of Figure 28 show. Therefore, if the demand characteristics of the power and the heat in Sapporo
shown in Figure 33 are taken into consideration
and the engine generator operates from 23:00 to
5:00 with little electricity demand, a significant
amount of heat from the engine-cooling water
can be obtained. It can prepare for the large heat
demand at 6 oclock in the morning in winter
(January, March, and November) shown in Figure 33 (b) by storing this exhaust heat. Since the
225
Figure 34. Load of each engine generator
226
Figure 35. Installed numbers of an engine generator, and total input-output of power
227
Figure 36. Analysis results of energy output
228
relation between the power load and exhaust heat

output is nonlinear, if the load distribution of each
engine generator is controlled, an operation plan
fit for the demand pattern of power and heat may
be realizable. However, this chapter does not deal
with the supply and demand of heat. As for Figure
36 (d-1), in the results of E/G (1) to E/G (3) of 6
system, (d-2) is the results of E/G (4) to E/G (6).

Figure 37 shows the analysis results of the power
generation efficiency and the load factor of each
sampling time of an engine generator with arranged
distribution. The load factor of the central system
is 28 to 89% of the range, and power generation
efficiency changes from 13% in 28% of the range.
On the other hand, E/G(1) of the analysis result
of the number of engine generators is 6 as shown
in Figures 37 (d-1) and (d-2), is 100% of the load
factor at all sampling times. From 2 oclock to 5
oclock in the morning, E/G (2) is a load factor
of 70% or more. The load factor is 100% in other
time zones. E/G (3) is 100% of the load factor in
the 6:00 to 9:00 time zone and the 16:00 to 23:00
time zone. E/G (1) to E/G(3) all show an operation
pattern with a high load factor compared with the
central system. On the other hand, E/G (4) and E/G
(5) have a high load factor from 16:00 to night as
shown in Figure 37 (d-2), but in other time zones,
the load factor is a low operation pattern. Regarding E/G (6), the operation of the load factor at a
maximum of 33% is only from 16:00 to 22:00, and
the hours worked are short. If an engine generator
is distributed and arranged, as Figures 37 (d-1)
and (d-2) show, the operating ratio of an engine
generator will be distributed. If the efficiency of
an engine generator that has stopped is not taken
into consideration, the power generation efficiency
of the whole distribution system of an engine
generator is improved compared with the central
system. However, the engine generator stops, and
an engine generator with short working hours
appears. If the life of the whole engine generator
linked to IMG is taken into consideration, periodically replacing the starting order of the engine
generator of E/G(1) to E/G(6) in Figures 37 (d-1)
and (d-2) is required, for example.
Power Cost
Figure 38 shows the analysis results of expressing
the number of distributions of an engine generator,
and the consumption of fuel for power generation
on a representative day with the calorific power
of fuel. If the number of distributions of an engine generator increases as shown in the figure,
the fuel consumption will decrease. However,
when four distributions of an engine generator
are compared with six, the difference is 6%, and
even if the number of installations is further increased, the reduction effect of fuel consumption
shows slight improvement. If the facility cost and
maintenance cost of an engine generator can be
estimated, the number of years required to recover
costs is calculable from the reduction result of
fuel consumption shown in Figure 38. However,
in order to determine the facility cost of an engine
generator, it is necessary to assume a particular
heat supply method, which is beyond the scope
of this study. This chapter considers the number
of engine generators for distribution, and the relationship between the reduction of fuel cost and
power generation.
Figure 39 shows the analysis results of the
years of operation of IMG that distributes engine
generators, and the cost of the kerosene fuel used
for power generation. The years of operation were
analyzed as 3, 5, 8 or 10 years. Moreover, since
the kerosene prices in the U.S. and Japan differed
as described in previous section, they were analyzed using the average kerosene price of each
country. An increase in the number of distributions
of an engine generator will reduce the cost of
kerosene consumption as shown in Figure 38. In
particular, when the number of distributions of
an engine generator is 2, compared with the central system, the decline is large, and there is a 27%
229
Figure 37. Analysis results of efficiency and load factor
230
Figure 38. Result of the relation between the installed number of an engine generator, and the fuel
consumption
Figure 39. Analysis results of the number of installations of an engine generator, and kerosene fuel cost
231
reduction in kerosene cost. If the number of engine

generators for distribution is increased, the cost
of the kerosene fuel used for power generation
will decrease, but the ratio of the effect becomes
so insignificant that the number of engine generators increases. If the rise in facility costs due
to the increase in the number of installations of
an engine generator is taken into consideration,
even if it determines the number of distributions
of an engine generator to be 2, a reduction in
kerosene fuel cost is attained.
Figure 40 shows the result of the reduction
value when comparing the cost of the kerosene
fuel when operating IMG for 1 year with the cost
of kerosene fuel when operating IMG for 3, 5, 8,
and 10 years. For example, when IMG comprises
four engine generators and it operates for 8 years
in the U.S., a cost reduction of the kerosene fuel
of 43,000 dollars is predicted. Therefore, the price
per engine generator is 10,750 dollars (43,000
dollars / four sets). However, considering the
facility cost of IMG, the maintenance cost, the

cost of equipment of the heat system, etc. will
further reduce facility costs and maintenance costs
per engine generator. Under the same conditions,
if the number of arrangements of an engine generator is two sets, a cost reduction of the kerosene
fuel of 33,400 dollars is predicted assuming operation for 8 years. The price per engine generator is 16,700 dollars (33,400 dollars / two sets).
Table 5 is an example of the cost comparison
when installing the proposed system in Japan.
However, equipment cost, installation cost, and a
subsidy are included in the initial cost. The equipment cost included in the initial cost was obtained
by referring to the data of business-use cogeneration and gas-engine cogeneration (Yanmar, 2006;
Osaka Gas, 2006). Maintenance cost is decided
by the installed number and capacity of an engine
generator. Initial cost and Maintenance cost in
Table 5 differ according to the product and the
usage. If the number of installations of an engine
Figure 40. Analysis results of the number of installations of an engine generator, and the reduction value
of kerosene cost. Comparison with the operation cost for one year.
232
Table 5. Cost comparison
generator changes, the power generation capacity

of an engine generator will also change. Therefore,
although it is necessary to take the difference in
power generation capacity and the relationship of
the facility cost into consideration, the number of
engine generators for distribution is expected to
be 3 or 4 sets; the analysis results of Figure 39
and Table 5 provide the same prediction.
The relation between the number of installations

of an engine generator and the power generation
efficiency and the power generation cost was
investigated by numerical analysis regarding the
independent microgrid (IMG) that distributes
small diesel-engine power generators. The following conclusions were obtained.
appears. If the life of the engine generator

is taken into consideration, it is necessary
to replace the starting order of the engine
generator within a definite period.
2. The relationship between power generation
and the exhaust heat of an engine generator is nonlinear. Therefore, operation fit for
the power and heat demand pattern may be
realizable by controlling the distribution of
the load of the distributed engine generator.
3. There is a reduction in the fuel for power
generation, so that there are many engine generators distributed and arranged. However,
if the facility cost and maintenance of an
engine generator by distribution are taken
into consideration, the number of installations of an engine generator would be 3 or
4 sets.
1. If the engine generator linked to IMG is distributed and arranged, the power generation
efficiency of the whole system will improve
compared with the central system. However,
an engine generator with short hours worked
The fall in power generation efficiency at

the time of partial load operation of IMG was a
problem. This chapter proposes a more efficient
method of IMG with load fluctuation. Application
of IMG using renewable energy is also considered.
Conclusions
233
REFERENCES
Abu-Sharkh, etal. (2006). Can microgrids make
a major contribution to UK Energy supply? Renewable & Sustainable Energy Reviews, 10(2),
78127. doi:10.1016/j.rser.2004.09.013
Hussain, M. M., Baschuk, J. J., Dincer, X., & Li,

I. (2005). Thermodynamic analysis of a PEM
fuel cell power system. International Journal of
Thermal Sciences, 44, 903911. doi:10.1016/j.
ijthermalsci.2005.02.009
Alibhai, Z., et al. (2004). Coordination of distributed energy resources. In H. G. Stassen (Ed.),
Systems, Man and Cybernetics, IEEE International
Conference on Volume 2, (pp. 1990-1995).
Ishihara, A., Mitsushima, S., Kamiya, N., & Ota,

K. (2004). Exergy analysis of polymer electrolyte fuel cell systems using methanol. Journal
of Power Sources, 126, 3440. doi:10.1016/j.
jpowsour.2003.08.029
Architectural Institute of Japan. (2002). The

national research study concerning the energy
consumption in the house, 3, 3-6. In Japanese.
Japanese National Astronomical Observatory.

(2003). Rika Nenpyo, chronological scientific
tables CD-ROM. Japan: Maruzen Co., Ltd.
Ay, M., Midilli, A., & Dincer, I. (2006). Exergetic

performance analysis of a PEM fuel cell. International Journal of Energy Research, 30, 307321.
doi:10.1002/er.1150
Kazim, A. M. (2005). Exergoeconomic analysis

of a PEM electrolyser at various operating temperatures and pressures. International Journal
of Energy Research, 29, 539548. doi:10.1002/
er.1073
Baker, J. E. (1985). Adaptive selection methods

for genetic algorithms. In Proc. 1st Int. Joint Conf.
on Genetic Algorithms: ICGA85, (pp. 101-111).
673681. doi:10.1109/TIA.2005.847277
Fuji Futures Co. Ltd. (2005). Commodity futures
market total information. Retrieved from http://
www.fuji-ft.co.jp/selection/toyu/index.htm
Goldberg, D. E. (1989). Genetic algorithms in
search, optimization and machine learning. Addison Wesley.
Gouda, T. etal. (2004). Microgrid. Japan: Denki
Shimbun. In Japanese
Hsin-Sen, C., Fanghei, T., Yi-Yie, Y., Kan-Lin,
H., & Fa-Lin, C. (2008). The development of a
small PEMFC combined heat and power system.
Journal of Power Sources, 176(2), 499514.
doi:10.1016/j.jpowsour.2007.08.072
234
Kopac, M., & Kokturk, L. (2005). Determination

of optimum speed of an internal combustion engine
by exergy analysis. Int. J. Exergy, 2(1), 4054.
doi:10.1504/IJEX.2005.006432
Kuzgunkaya, E. H., & Hepbasli, Arif. (2006). Exergetic performance assessment of a ground-source
heat pump drying system. International Journal
of Energy Research, 31, 760777. doi:10.1002/
er.1268
Mathiak, J., Heinzel, A., Roes, J., Kalk, T., Kraus,
H., & Brandt, H. (2004). Coupling of a 2.5 kW
steam reformer with a 1 kW PEM fuel cell.
Journal of Power Sources, 131(1-2), 112119.
doi:10.1016/j.jpowsour.2004.01.024
Meteorological Agency. (2007). Meteorological
observation in Japan. Retrieved from http://www.
data.kishou.go.jp/index.htm
Mohamad, J. E., & Mofid, G. B. (2005). Exergetic
analysis of gas turbine plants. Int. J. Exergy, 2(1),
3139. doi:10.1504/IJEX.2005.006431

University, Japan.
Obara, S. etal. (2005). Study on small-scale cogeneration system for domestic house considering
partial load and load fluctuation. [In Japanese].
Journal of the Japan Society of Mechanical Engineers, 71(704), 11691176.
Obara, S., & Kudo, K. (2005). Relation between
control variables of PEM fuel cell system and
power generation efficiency. Transaction of JSME,
Part B, 72(714), 447-454. In Japanese.
Osaka Gas Co. Ltd. (2006). Osaka Gas home use
gas appliance synthesis information site. Retrieved
from http://www.g-life.jp/html/scene/cogeneration/ecowill/point/point09.html
Ozturk, A., Senel, A., & Onbastoglu, S. U. (2005).
Thermodynamic optimization of combined cycles.
International Journal of Energy Research, 29,
657670. doi:10.1002/er.1098
Robert, H. (2004). Microgrid: A conceptual solution. In Proc. 35th Annual IEEE Power Electronics
Specialists Conference, 6, 4285-4290.
Shukuya, M. (2006). The way of looking and way
of thinking of exergy. In order to develop familiar
green engineering. [In Japanese]. IEEJ Journal,
126(4), 196197. doi:10.1541/ieejjournal.126.196
Tanrioven, M. (2005). Reliability and costbenefits of adding alternate power sources to
an independent microgrid community. Journal
of Power Sources, 150, 136149. doi:10.1016/j.
YANMAR Co. Ltd. (2006). Cogeneration package. Retrieved from http://www.yanmar.co.jp/
products/mgasc/cogene04.htm
Yasuda, I. (2005). Development of hydrogen production technology for fuel cell. Energy Synthesis
Engineering, 28(2). In Japanese.
Yoshida, K. (1999). Exergy engineering, as actual
as the theory. Japan: Kyoritsu Shuppann Ltd. In
Japanese
235
APPENDIX
Nomenclature

E/G : Engine generator

IMG : Independent microgrid
Eeg : Power of an engine generator [kW]
Reg : Load factor [%]
eg : Generation efficiency [%]
236
237
Chapter 8
Characteristics of PEFC /
Woody Biomass Engine Hybrid
Microgrid and Exergy Analysis
ABSTRACT
This chapter consists of three sections, Dynamic Characteristics of PEFC / Woody Biomass Engine
Hybrid Microgrid, Exergy Analysis of the Woody Biomass Stirling Engine and PEFC Combined System
with Exhaust Heat Reforming and Exergy Analysis of A Regional Distributed PEM Fuel Cell System.
The chapter describes the exhaust heat of the combustion of woody biomass engine using a Stirling
cycle that was used for the city gas reforming reaction of a PEFC system. The response characteristic
of PEFC and woody biomass engine is investigated by the experiment and numerical analysis. Finally, a
combined system that uses the exhaust heat of the woody biomass Stirling engine for the steam reforming
of city gas and that supplies the produced reformed gas to a PEFC is proposed.
The summary of the 1st section is as follows. The
combustion exhaust heat of woody biomass engine
using Stirling cycle is a high temperature. This
exhaust heat is used for the city gas reforming
reaction of a PEFC system. The woody biomass
engine generator has the characteristic that the
greenhouse gas amount of emission with power
generation is greatly reducible. In this study, the
microgrid system that introduces PEFC / woody

biomass engine hybrid cogeneration (PWHC) is
proposed. It depends on the dynamic characteristics of the grid for the power quality at the time
of load fluctuation being added to the microgrid.
Especially, the dynamic characteristics of the
independent microgrid are important on security
of power quality. So, in this study, the response
characteristic of PEFC and woody biomass engine was investigated by the experiment and the
DOI: 10.4018/978-1-4666-5796-0.ch008
Characteristics of PEFC / Woody Biomass Engine Hybrid Microgrid
numerical analysis. Furthermore, the response

characteristic of the PWHC independent microgrid
including auxiliary machinery was investigated by
the numerical simulation. Moreover, an improvement of dynamic characteristics is proposed using
the method of adding proportional-plus-integral
control to PWHC. If woody biomass engine is
introduced into a house, 10.2s will be required
to stabilize power quality at the maximum. On
the other hand, when woody biomass engine corresponds to a base load and PEFC corresponds
to the load exceeding the base load, settling time
is less than 1.6 s. In this study, relation between
the system configuration of the PWHC microgrid
and the dynamic characteristics of the power was
clarified.
The woody biomass Stirling engine (WB-SEG)
is an external combustion engine that outputs
high-temperature exhaust gases. It is necessary
to improve the exergy efficiency of WB-SEG
from the viewpoint of energy cascade utilization.
So, a combined system that uses the exhaust heat
of WB-SEG for the steam reforming of city gas
and that supplies the produced reformed gas to a
PEFC is proposed. The energy flow and the exergy
flow were analyzed for each WB-SEG, PEFC,
and WB-SEG / PEFC combined system. Exhaust
heat recovery to preheat fuel and combustion air
was investigated in each system. As a result, (a)
improvement of the heat exchange performance
of the woody biomass combustion gas and engine
is observed, (b) Reduction in difference in the
reaction temperature of each unit, and (c) removal
of rapid temperature change of reformed gas are
required in order to reduce exergy loss of the
system. The exergy efficiency of the WB-SEG
/ PEFC combined system is superior to PEFC.
The summary of the 3rd chapter is as follows.
The exergy flow and exergy efficiency of a 3kW
PEFC were investigated, and the regional characteristic of the distributed energy system was considered. In the environmental temperature range of
238
263K to 313K, the difference of the total efficiency

of the proposed system was 6%. On the other hand,
the difference of the exergy total efficiency of the
same temperature range was 30%. Moreover, as
a result of examining how to improve the exergy
efficiency of this system, certain improvement
methods were proposed. (a) Preheat the city-gas
and air supplied to the system using exhaust heat,
and raise the combustion temperature, (b) Preheat
the water supplied to the system using exhaust
heat, (c) Change the catalyst material of each unit
and reduce the amount of cooling of the reformed
gas, (d) Examination of combined cycle power
generation. The exergy efficiency, in the case of
introducing the proposed system into individual
homes in Sapporo, Tokyo, and Kagoshima, was
evaluated. Consequently, when the system was
introduced into a community with low outside
air temperatures, exergy efficiency increased
compared with communities with high outside
air temperatures.
DYNAMIC CHARACTERISTICS
OF PEFC / WOODY BIOMASS
ENGINE HYBRID MICROGRID
Introduction
If the microgrid is introduced into an urban area, it
will be expected that the energy cost of a distributed
power supply and emission of greenhouse gas can
be reduced. Authors investigated until now about
the operating method that connects distributed
proton exchange membrane fuel cells (PEFC) in
a power network, and cooperates (Obara, 2006a;
Obara, 2006b). Although generation efficiency
of a PEFC is high, greenhouse gas discharges by
the reforming reaction of city gas. On the other
hand, micro combined heat and power (microCHP) using a small-scale Stirling engine generator
(SEG) is examined in U.K. as an energy system
for individual houses (Peacock & Newborough,
2007; Hawkes & Leach, 2007). By using woody

biomass so that carbon dioxide may circulate, the
greenhouse gas amount of emission of a power
generation system can be decreased. Therefore,
introduction of SEG using woody biomass is
effective in emission control of greenhouse gas
(De Ruyck et al., 2007; Bonnet, Alaphilippe, &
Stouffs, 2005; Obara, 2006c). However, compared
with an internal combustion engine or a fuel cell,
the conventional SEG has subjects in generation
efficiency, volume efficiency, equipment cost, etc.
The energy supply system using the microgrid can
reduce equipment cost compared with the method
of introducing generating equipment into each
house. Moreover, the energy equipment linked to
the microgrid can cooperate, for example on the
minimization object of a greenhouse gas amount
of emission. So, in this section, the dynamic
characteristics of the power of the independent
microgrid using hybrid cogeneration (PWHC)
of PEFC and SEG using woody biomass are
investigated. It depends for the control response
characteristic of SEG on engine structure, the
configuration of the combustion chamber, the
heat transmission characteristic of the heat source,
etc. Until now, optimization of the combustion
chamber configuration and the heat transmission
characteristic of combustion gas are investigated
(Ozeki, 1991; Eguchi, & Kurosawa, 1988). Commonly, the power-demand pattern of a house
or an apartment house consists of many peaks
changed for a short time. Since such power load
is followed, a rapid control response characteristic
is required of generating equipment. In order to
manage the power quality of the microgrid, it is
necessary to clarify the dynamic characteristics
of the power with load fluctuation. So, in this
section, the dynamic characteristics of the PWHC
microgrid are clarified by a numerical analysis
using the results of an investigation of the SEG
test machine and PEFC.
System Scheme
Hybrid Microgrid
Figure 1 (a) shows the model of the independent
microgrid that introduces two-set PWHC (PWHC
(1) in House (1), and PWHC (2) in House (5)).
The microgrid of this model consists of eight
buildings of House (1) to House (8). Heat supply
of the exhaust heat of PWHC, a heat storage tank,
a boiler is separated into the group of House (1) to
House (4), and the group of House (5) to House
(8). The power of two-set PWHC is supplied to
each building through the power grid. The system
interconnection device is installed in the contact
point of PWHC and a power grid. Moreover, the
power of PWHC is changed into 100V and 50Hz
with an inverter. On the other hand, the exhaust
heat of PWHC, the heat of a heat storage tank, and
a boiler is supplied to each building through hot
water pipings (1) and (2). However, examination
of this study is limited to the dynamic characteristics of the power for the microgrid, and is not
managed about the heat system.
Figure 1 (b) shows the energy flow and
chemical reaction of each component of the proposed system. Chip fuel is supplied to woody
biomass engine (SEG), and power is transmitted
to an alternating current synchronous power generator. The heat output of SEG is the high-temperature exhaust gas of the combustion chamber,
and engine-cooling hot water. Moreover, as the
heat output of PEFC, there is fuel cell stack exhaust
heat and reformer exhaust heat. In the proposed
system, the combustion chamber high-temperature
exhaust gas of SEG is supplied to the heat exchanger of the reformer. With a catalyst in the
reformer, city gas is changed into reformed gas
with a high hydrogen concentration with a reaction
temperature of 970 K to 1070 K using this exhaust
heat. Reformer exhaust heat is the remaining heat
239
Figure 1. Outline of the proposed system and operation method
240
after providing heat to the catalyst through the

heat exchanger. In the case study, exhaust heat
that can be supplied to the demand side is taken
as the reformer exhaust heat and SEG cooling
water. Moreover, the demand side is supplied
after changing the power of SEG and PEFC into
an alternating current of constant frequency.
Outline of Testing SEG

Table 1 and Table 2 are the operating conditions
and specifications of SEG and the power generator that are examined in this section. Although
the maximum output of SEG is 3.7 kW, the
maximum power load examined according to
restrictions of combustion chamber capacity, etc.
is 1.6 kW. Figure 1 (c) is a general view of the
test equipment. Chip fuel (woody biomass) is fed
into the hopper of the combustion chamber. Chips
are mixed with air preheated before entering the
combustion chamber. The rate of feed of chip fuel
is controllable by the fuel feed system installed
in the lower part of the hopper. Power is transmitted to the power generator shown in Table 2 by a
belt from the power shaft of SEG. Since the test
SEG is a single cylinder, its vibration is large.
Table 1. SEG specification
Consequently, the combustion chamber is connected with the engine by a buffer duct so that
the vibration of the engine does not spread to
the combustion chamber. The exhaust gas of the
combustion chamber is discharged from the
system through a duct. The quantity of heat of
the exhaust gas QEx and cooling water QEy is
obtained from the value of the temperature sensor and the flow meter by calculating the transport volume of enthalpy. Moreover, the amount
of heat radiation on the combustion chamber
surface (QEz ) is measured by heat flow rate sensor q , and the heat-medium pressure is measured
using sensor Pg .
Figure 1 (d) shows the experimental results of
the energy flow of the test SEG. The energy flow
is separated into auxiliary machinery loss, cooling
water quantity of heat, exhaust gas quantity of
heat, production of electricity and other losses.
Other losses of the energy flow are mechanical
loss of radiation of heat and friction of SEG, vibration, etc. Other losses decrease, so that the
production of electricity of SEG is large. The
power generation efficiency of SEG improves by
reducing other losses that hold a large part of the
energy flow at the time of low load. The quantity
of exhaust gas heat holds the largest part in the
energy flow, and it is always large compared to
the cooling water quantity of heat. Since there is
large exhaust gas heat, the development of a
compound cycle of operating a steam turbine using the high-temperature exhaust gas of SEG, for
example, is possible. Auxiliary machinery loss
holds very few parts in the whole energy flow.
Microgrid System Operating Method

Table 2. Power generator specification
Figure 1 (e) shows the PWHC operation model

in a representation day. In this operation pattern,
SEG is operated in a range smaller than the base
load set up beforehand. In addition to SEG, PEFC
is operated in a larger load range than the base
load. When a load exceeds the base load, SEG
241
can be operated at a maximum efficiency point.

However, when a load is less than the base load,
load following operation is required of SEG.
Control Response Characteristics

of PEFC and SEG
1. Control Block Diagram
Figures 2 (a) and (b) are the block diagram of the
feedback control on the microgrid by SEG and
PEFC, respectively. Proportional-plus-integral
control (PI control) is introduced into control of
each system. PEFC and SEG are controlled by
the controller. Each controller is controlled based
on the PI control parameters ( P and I ) set up
beforehand. The power generated by SEG and
PEFC is supplied to the demand side through an
inverter and a system interconnection device. The
transfer functions of each equipment shown in
Figures 2 (a) and (b) describes the determination
method in following sections.
The control block diagram in the case of oneset SEG operating corresponding to a base load,
and corresponding to the load exceeding the base
load by one-set PEFC is Figure 2 (c). SEG supplies the power to the load below the base load
set up beforehand. PEFC is also operated when
the load of the microgrid exceeds the base load.
The control block diagram in the case of one-set
PEFC operating corresponding to the base load,
and corresponding to the load exceeding the base
load by one-set SEG is Figure 2 (d).
The control block diagram of the PWHC
microgrid in the case of one-set SEG operation
corresponding to the base load, and corresponding
to the load exceeding the base load with multiple
generators is Figure 2 (e). In Figure 2 (e), SEG
(1) operates corresponding to the base load, and
operates SEG (2), PEFC (1), and PEFC (2) according to the magnitude of a load. When supplying
the power to the microgrid from the combined
cycle system, the dynamic characteristics of the
microgrid are decided with the transfer functions
242
of each equipment. So, in this section, the transfer

function and control parameters of PEFC, SEG, an
inverter, and a system interconnection device in
Figure 2 are determined by the method described
to following sections.
The transient response characteristics of the
power output of the SEG, PEFC, auxiliary machine
and PWHC microgrid are analyzed by MATLAB
(Ver.7.0) / Simulink (Ver.6.0) of Math Work. In
the solver to be used, the Runge-Kutta method
is installed, and the sampling time of analysis is
calculated automatically and decided so that error
may be less than 0.1%.
2. Response Characteristic of PEFC
Table 3 is a result of the transfer function investigated by the last study about the fuel cell
stack, the reformer, the inverter, and the system
interconnection device (Obara, 2006; Zhang et al.,
2004). The transfer function of the fuel cell stack
was determined from the experimental result, and
the transfer function of other equipment is decided
from references (Zhang et al., 2004; Sedghisigarchi, 2004; Hamelin et al., 2001; Yerramalla, et al.,
2003; Oda et al., 1999; Takeda et al., 2004; Ibe,
2002; Koike et al., 1999; Kyoto Denkiki, 2001).
In the further last study, it is investigating also
about the optimal value of the parameters of the
PI control introduced into the controller of PEFC.
The transfer function and control parameters on
PEFC of the control block diagram shown in
Figure 2 introduce each value of Table 3.
Figure 3 (a) shows the result of the step response
of 0.2kW, 0.4kW, 0.6kW, 0.8kW, 1.0kW of PEFC
with a reformer (Obara, 2006; Obara & Kudo,
2007). In the analysis of Figure 3 (a), the control
block diagram of Figure 2 (b) was used. The
control parameter with short settling time and
small overshooting was investigated by numerical
analysis, and P = 12.0 and I = 1.0 were decided. The response time of a system when converging on 5% of a target value is defined as
settling time.
Figure 2. Control block diagram of power supply
243
Table 3. Transfer function of a power output
Figure 3. Step response of PEFC and SEG
244
3. Response Characteristics of SEG

Figure 3 (b) shows the experimental result of the
step response of 0.2kW, 0.4kW, 0.6kW, 0.8kW,
1.0kW of Testing SEG. As shown in Figure 3 (b),
the step response of Testing SEG has large overshooting, and its settling time is long compared
with PEFC. The heat transmission characteristics
between the combustion gas of a chip and the
heat exchanger of SEG is considered to influence
settling time greatly. However, it is difficult to
improve the rate of heat transfer of the combustion gas of a chip, so that the load fluctuation of
the power can be followed. So, in order to shorten
the settling time of SEG as much as possible and
to reduce overshooting, PI control is added to
operating of SEG.
Figure 3 (c) shows the example as a result of
a step response obtained in the operating experiment of SEG (Figure 1 (d)). The model of the
transfer function that simulated this step response
is shown in Figure 3 (c). The settling time of
Testing SEG exceeds 10 s. Therefore, when SEG
is operated so that fluctuating load may be followed, the unstable time of voltage and a frequency is long. Figure 4 shows the analysis result
of a step response when adding PI control to the
system using the transfer function in Figure 3 (c).
The control block diagram used in this analysis
is Figure 2 (a), and the control parameters of SEG
introduced P = 0.1 and I = 0.001 . Moreover,
a response result in case there is no PI control is
also shown in Figure 4. Settling time becomes
short by adding PI control to SEG, and an overshoot does not appear. For example, the settling
time of the 2kW step response that does not use
PI control is about 16 s. However, if PI control is
added, it will improve at about 6 s.
Result of Dynamic Characteristics

Analysis of PWHC Microgrid
Power Response
Characteristic of PWHC
1 kW PWHC microgrid consists of 0.5 kW SEG
and 0.5 kW PEFC. Figure 5 shows the analysis
result of the step response of 0.2 kW, 0.4 kW, 0.6
kW, 0.8 kW, 1.0 kW of this system. Results in
case SEG corresponds to the base load and PEFC
follows the fluctuating load exceeding the base
load are Figures 5 (a) and (b). The control block
diagram used in the analysis in Figures 5 (a) and
(b) is Figure 2 (c). However, the value in Figure
3 (a) ( P = 12.0 and I = 1.0 ) and Figure 4 (
P = 0.1 and I = 0.001 ) was used for the control
parameter of the analysis in Figure 5 (a). The
speed of response of PEFC shown in Figure 3 (a)
is quick compared with the speed of response of
SEG shown in Figure 4. From the difference in
this speed of response, as shown in the step response of 0.8kW and 1.0kW in Figure 5 (a), the
response of a quick response part and a late response part appears. Consequently, the control
parameters of PEFC with a quick speed of response
are changed, and an improvement of the response
Figure 4. Step response results of 2 kW SEG model
245
Figure 5. Step response results of 1 kW PWHC
characteristics of the PWHC microgrid is tried.

Figure 5 (b) shows the response characteristics at
the time of changing the control parameters of
PEFC into P = 0.95 and I = 1.1 . These control
parameters were decided by trial and error. Two
response parts, 0.8kW and 1.0kW in Figure 5 (a),
have improved.
Step response results in case PEFC corresponds
to the base load and SEG follows the fluctuating
load exceeding the base load are Figures 5 (c) and
(d). In the analysis in Figures 5 (c) and (d), the
control block diagram shown in Figure 2 (d) was
used. In Figure 5 (c), the control parameters of
PEFC are P = 12.0 and I = 1.0 , and the control
parameters of SEG are P = 0.1 and I = 0.001 .
Since the over shoot of the response shown in
Figure 5 (c) is large, the control parameters of
PEFC is changed and an improvement is tried.
246
Figure 5 (d) shows the response characteristics at

the time of changing the control parameters of
PEFC into P = 0.95 and I = 1.1 . These control
parameters were decided by trial and error. Compared with the response of Figure 5 (c), the response of Figure 5 (d) has small overshooting,
and its settling time is short.
Response Characteristics of
SEG and PEFC Microgrid Using
Power Load Pattern for Houses
1. Response Result of SEG
The response characteristics in the case of supplying the power to the microgrid from SEG or
PEFC are investigated. However, the power load
pattern added to the microgrid assumes two houses
on February representative day in Sapporo. The
power load pattern consists of time average values

of the load consumed by the household appliances
and electric lights. Space cooling and heating
loads are not included in this power load pattern.
Therefore, the power load pattern does not have
a large difference at every month.
Figure 6 shows the analysis result of a load
response at the time of supplying the power to the
microgrid using 2 kW SEG. The control block
diagram used in the analysis of Figure 6 is Figure
2 (a). Moreover, the control parameters of SEG
are P = 0.1 and I = 0.001 as well as Figure 4.
The horizontal axis of Figure 6 is the representative time of analysis. Real time is also displayed
on the horizontal axis of Figure 6. Since the calculation time is enormously long, real-time
analysis is performed by shortening real time to
1/180 in this section. Figure 6 (a) shows the result
of a load input and the system response, and
Figure 6 (b) shows the result of the error of a load
input and a response. As for the broken-line part
shown in Figure 6 (b), the error of the load and
the response is over 5%. A large rising error
occurs immediately after 0:00 in Figure 6 (b).
Actually, since the system is operated continu-
ously, this rising error does not exist. Figure 6 (c)

shows the analysis result of time period for the
error of the load and the response to exceed 5%.
Accordingly, the results of Figure 6 (c) express
settling times. The settling time when installing
SEG into the microgrid from the result of Figure
6 (c) is 10.2 s at the maximum. When the microgrid
is composed from SEG, the unstable period of
voltage and a frequency is 10.2 s at the maximum.
2. Response Result of PEFC
Figure 7 shows the analysis result of a load response at the time of installing 2 kW PEFC into
the microgrid. The control block diagram used in
the analysis of Figure 7 is Figure 2 (b). The control parameters set up with the controller are
P = 12.0 and I = 1.0 as well as Figure 3 (a).
The settling time in case PEFC composes the
microgrid from the result of Figure 7 (c) is 1.6 s
or less. However, rising parts are excluded. The
power supply due to PEFC has a short settling
time compared with SEG. Therefore, the dynamic characteristic of the power of PEFC microgrid is good compared with SEG microgrid.
Figure 6. Dynamic characteristics analysis results of the microgrid at the time of installing the power
demand model of two houses in Sapporo. The power is supplied to the grid from 2 kW SEG of one. SEG
of P=0.1 and I=0.001.
247
demand model of two houses in Sapporo. The power is supplied to the grid from 2 kW PEFC of one.
PEFC of P=12.0 and I=1.0.
Response Characteristics of PWHC

Microgrid Using Power Load Pattern for
Houses
Figure 8 shows the analysis result of a load response of the microgrid composed from 8 kW
PWHC. Eight houses are connected to the microgrid. Woody biomass engine generators installed into the microgrid are 2 kW SEG (1) and
2 kW SEG (2), in addition install 2 kW PEFC (1)
and 2 kW PEFC (2). Moreover, the control block
diagram used in the analysis of Figure 8 is Figure
2 (e). The control parameters set up with the
controller of PEFC are P = 0.95 and I = 1.1
as well as Figure 5 (d), and SEG are P = 0.1 and
I = 0.001 . Since the speed of response of SEG
is slow, the dynamic characteristics of SEG (2)
have large influence on the microgrid. It is because
SEG (2) is followed and operated to load fluctuations. As a result, a settling time becomes long
as shown in Figure 8 (c). Consequently, installation of SEG shall be one set corresponding to the
base load. And about the load exceeding the base
load, it corresponds by installing two-set PEFC.
Figure 9 shows the analysis results of the load
248
response of the microgrid composed from one-set

of 2 kW SEG and two-set of 2.5 kW PEFC. Eight
houses are connected to the microgrid. This system was analyzed by modifying the control block
shown in Figure 2 (e). The control parameters set
up with the controller of PEFC are P = 12.0 and
I = 1.0 , and SEG are P = 0.1 and I = 0.001 .
The error-analysis result shown in Figure 8 (b)
and Figure 9 (b) shows the well alike profile.
However, as shown in Figure 9 (c), the settling
time of the microgrid becomes very short compared with Figure 8 (c). The system of Figure 9
is the PWHC microgrid stabilized dynamically.
Conclusions
The load response characteristics were investigated using the testing Stirling engine power
generator (SEG) that uses woody biomass as a
fuel. The transfer function was determined from
these results and the dynamic characteristics of
the power of the microgrid composed from SEG
were investigated. Moreover, hybrid cogeneration
(PWHC) that uses the combustion exhaust heat of
SEG for the heat source of the reformer of PEFC
demand model of teight houses in Sapporo. The power is supplied to the grid from 2 kW SEG of two sets
and 2 kW PEFC of two sets. PEFC of P=0.95 and I=1.1, and SEG of P=0.1 and I=0.001.
demand model of teight houses in Sapporo. The power is supplied to the grid from 2 kW SEG of one set
and 5 kW PEFC of two sets. PEFC of P=12.0 and I=1.0, and SEG of P=0.1 and I=0.001.
249
was proposed. The dynamic characteristics of the

power when composing the microgrid from this
PWHC were investigated by numerical analysis.
And, the control parameters installed into each
controller of SEG and PEFC were examined
from the results of numerical analysis. The load
response characteristics of the microgrid using
PWHC were investigated, and the following
Conclusions were obtained.
addition, the detail of the greenhouse gas emission characteristic and economic evaluation of
the proposed system is reported independently.
1. The settling time of the SEG microgrid for

the power supply to houses exceeds 10 s at
the maximum. Therefore, in SEG independent microgrid, voltage and a frequency may
often be instability in a long time.
2. The microgrid composed from PEFC has a
short settling time at the time of load fluctuation compared with the power supply due to
SEG. Therefore, the dynamic characteristics
of the power of the PEFC microgrid are good
compared with SEG.
3. The microgrid system that combined baseload operation of SEG and the load following operation of PEFC was proposed. The
settling time of the proposal microgrid with
eight houses is 1.6 s or less. The microgrid
that installs the proposed system is stable,
and there are few amounts of emission of
greenhouse gas.
Introduction
There are waveform distortion of a higher harmonic wave (about 10-4 to 10-2 seconds), voltage
fluctuation (about 10-2 to 100 second), frequency
change (about 100 to 102 seconds) and an overvoltage / undervoltage (about 102 seconds or more)
as dynamic characteristics that should be secured
concerning the power supply of the independent
microgrid. In this section, the influence of a period
about longer than a voltage fluctuation grade (10-2
seconds) was investigated. Waveform distortion of
a higher harmonic wave is an important subject,
and reports a detail by the following report. In
250
EXERGY ANALYSIS OF THE WOODY

BIOMASS STIRLING ENGINE AND
PEFC COMBINED SYSTEM WITH
EXHAUST HEAT REFORMING
Utilization of woody biomass is discussed from

the view of the environmental problem (Qian et
al., 2007; Rhen et al., 2007; Jianbang & Smith,
2006). In Japan, the utilization as a woody biomass fuel of thinned wood or building scrap
wood is investigated. Authors are evaluating the
test Stirling engine (WB-SEG) that burns these
woody biomasses directly (Obara et al., 2007).
The theoretical thermal efficiency of a Stirling
cycle is higher than a Diesel cycle at heat for the
same input temperature level. However, in actual
WB-SEG, the heat transmission control between
combustion gas and the engine is difficult and heat
supply temperature are limited (Obara et al., 2007).
In the experiment, the heat released down stream
of the engine represents at least 60% in total loss
(Obara et al., 2007). Concerning the exergy efficiency of an energy system, loss in the combustor
is large (Taniguchi & Ishida, 2005). Therefore,
the high-temperature heat released when woody
biomass needs to be used effectively for utilization of WB-SEG. However, when such release
is used for space heating and hot water supply, it
is not effective from the view of energy cascade
utilization. So, in this study, the combined system
of WB-SEG and PEFC is proposed. There are
very few reports of research regarding the WBSEG and PEFC combined system (Obara & Itaru
Tanno, 2007). The study on the complex system
of a fuel cell and an engine is uncivilizedness,
and the example of the system especially using a

Stirling cycle is few. The burning exhaust heat of
WB-SEG is used for the heat source of a steam
generator (vaporizer), and a reforming unit for
steam reforming of city gas in this system. The
exhaust heat of each unit can be used for preheating of fuel and combustion air in this system.
However, low-temperature exhaust heat, such as
engine cooling water of WB-SEG and cell stack
exhaust heat of PEFC, is used for space heating
and hot water supply. There are many methods
to optimize the leading the different efficiency
level. In the WB-SEG / PEFC combined system,
because PEFC is operated using the exhaust heat
of WB-SEG, exergy efficiency should improve.
Therefore, in this study, the energy flow and the
exergy flow are investigated in both of WB-SEG,
PEFC, and the WB-SEG / PEFC combined system
including preheating of fuel and combustion air.
Stirling Engine Generator
Figure 10 (a) is the block diagram of WB-SEG.
Woody biomass fuel (chips) and combustion air
are supplied to the combustor. A heat exchanger is
introduced and this is exchangeable for the heat of
the exhaust gas. The heat generated in the combustor is supplied to the engine heat exchanger. As
exhaust heat, there is exhaust gas from burning
and engine-cooling water. The heat of the enginecooling water is supplied to a heat storage tank,
and can be supplied to the demand side with a
time shift. On the other hand, the combustion
gas can select preheating of the biomass fuel and
the combustion air, or supply to the heat storage
tank. As shown in Figure 10 (b), engine power is
transmitted to the power generator using a belt.
The power outputted with the power generator is
converted into a regular frequency by the inverter,
and supplies stable voltage power and frequency
to the grid. The output characteristics of WB-SEG
shown in Figure 10 (c) are introduced into this

analysis. These output characteristics were taken
by examining an engine of the specification shown
in Figure 10 (b). The heat of each test WB-SEG
was calculated using the following equations.
Equation (1) is the heat of the woody biomass
fuel supplied to the engine. Equations (2) and
(3) are the heat of the engine-cooling water and
combustion gas, respectively. Equation (4) is the
heat discharge of the combustor.
qb = Lb Vf
(1)
qw = w C w tw Vw
(2)
qa = a C a ta Va
(3)
qrm = K Arm trm
(4)
PEFC with a City Gas Reformer

1. Reformed Gas System
Figure 11 (a) is the block diagram of PEFC with
a city gas reformer. The composition of city gas
is considered to be CH4 in this analysis. Figure
11 (b) shows the reaction temperature of each unit
of the reforming system and the gas concentration.
City gas and combustion air are supplied to the
combustor. Heat is supplied to the vaporizer and
the reforming unit using the combustion gas of
the combustor. The city gas and water for reforming are supplied to the vaporizer. In the reforming
unit, there is H2, CO, and the steam occurring by
the steam reforming of city gas. In the shifter unit,
CO2 and H2 are generated from CO and steam.
Some CO contained in the reformed gas is reduced
by 10 ppm or less by CO oxidation reaction. This
reformed gas is supplied to the anode of the cell
stack. Among H2 in the reformed gas supplied to
the fuel cell, utilization rate rce is used for power
251
Figure 10. Block diagram of WB-SEG
252
Figure 11. Proposed PEFC system
253
generation. H2 that is not used for power generation is supplied to the combustor, and is used for
vaporization of water and the heat source of the
reforming reaction.
2. Heat System
Exhaust heat is outputted from the vaporizer, the
reforming unit, the shifter unit, the CO oxidation
unit, and the cell stack. The high-temperature exhaust heat of the reforming unit, the shifter unit,
and the CO oxidation unit is used for preheating
the fuel (CH4 for burning and CH4 for reforming),
the water for reforming, and the combustion air.
Whether the exhaust heat of the reforming unit,
the shifter unit, and CO oxidation unit is used for
preheating of the fuels, etc., or whether a heat storage tank is supplied can be selected. In fact, the
distribution rate of the exhaust heat of preheating
and thermal storage can be determined arbitrarily.
However, in this chapter, all the high-temperature
exhaust heat is used for preheating, or all of it is
stored in the heat storage tank. On the other hand,
the exhaust heat of the vaporizer and the cell stack
is always supplied to the heat storage tank.
3. Output Characteristics of the Cell Stack
Figure 11 (c) shows the output characteristics
taken from a single cell experiment (Mikkola M.,
(2001), Ibaraki Prefecture Government Office of
Education, (2002)). The single cell shown in Figure
11 (c) is built up, and the cell stack is composed.
The number of cells of the cell stack decides that
the generation efficiency of the maximum output
point becomes the maximum. Figure 11 (d) shows
the relation of the load factor and efficiency of the
city gas reformer and PEFC to the city gas reformer
(Mikkola M., (2001), Ibaraki Prefecture Government Office of Education, (2002), Oda K, et al.,
(1999)). The reformer efficiency and generating
efficiency of the PEFC system were calculated
using Equations (5) and (6), respectively.
r =
254
qlhv,H
((m
gc
+ m gv ) LCH
(5)
sc =
E s
((m
gc
+ m gv ) LCH
(6)
WB-SEG / PEFC Combined System

1. Block Diagram
Figure 12 (a) is a block diagram of the combined
system of WB-SEG and PEFC. Moreover, the
chemical reaction and efficiency of each unit
are shown in Figure 12 (b). Woody biomass and
combustion air are supplied to the combustor unit.
Moreover, city gas and water are supplied to the
vaporizer unit. The combustion gas of the woody
biomass is first given to the Stirling engine, and
the exhaust heat is supplied to the vaporizer unit
and the reforming unit. The exhaust heat of the
reforming unit, the shifter unit, and the CO oxidation unit can select supplying preheating of the
biomass fuel, combustion air, CH4 for reforming
and water, or storing in the heat storage tank. The
power of the engine generator and the fuel cell is
converted into alternating current power of constant frequency, and is supplied to the power grid.
2. Chemical Reaction and Efficiency
The chemical reaction, the efficiency, and the
molar flow rate of each unit of the WB-SEG /
PEFC combined system are shown in Figure 12
(a). In the next section, the energy flow and the
exergy flow are calculated from the input-output
and efficiency of each unit.
Analysis Method
Energy Flow of the WB-SEG /
PEFC Combined System
The conditions used for analysis of the system are
shown in Table 4. Analysis of the energy flow
and the exergy flow is calculated under standard
conditions (25 degrees Celsius, 0.1 MPa). A
lower calorific value is used for fuel heat. Equations (7) to (12) are the energy balance equations
of each unit. h in each equation expresses en-
Figure 12. WB-SEG/PEFC combined system
255
Table 4. Analysis parameters
thalpy. H and E express heat and power, respectively. The left-hand side of each equation expresses power input, and the right-hand side
expresses output.

Reformer Unit:
m ve hve H ra + H r = m re hre + H ro
256
(8)
(9)
(10)
CO Oxidation Equipment Unit:
m se hse + m ca hca + H cr = m ce hce + H c

(11)

Vaporizer Unit:
m gv hgv + m rw hrw H va

+H v = m ve hve + H vo

(7)
Shifter Unit:
m re hre + H sa = m se hse + H s

Combustor Unit:
m wb hwb + m ga hga + m fe h fe

+H cb = m br hbr + H rb + H si
Cell Stack:
m ce hce + m fa h fa + H fr =

m fe h fe + H fe + f E f
(12)
Equation (13) is the energy balance equation

of the system. The left-hand side expresses
power input terms, and the right-hand side expresses output terms. n and m in this equation
are the number of units inputted and outputted,
respectively. H sys and E sys of the right-hand side

express the heat output and the power output of
the system, respectively. Moreover, the generation
efficiency of the system is calculated using Equation (14). qlhv,wb in this equation expresses the
calorific value of the woody biomass.
n
m
i =1
in ,i
hin ,i = m out , j hout , j

j =1
E sys ,1 + E sys ,2
m wb qlhv .wb + m gv qlhv,CH
100
4
[%]
(14)
The quantity of exergy is the sum of chemical

exergy, thermomechanical exergy, kinetic exergy,
and potential exergy. In this chapter, kinetic exergy and potential exergy are not calculated, because they are very small values compared with
chemical exergy (ech ) and thermomechanical
exergy (etc ). So, the value of each exergy is calculated using Equation (15).
(15)
The exergy balance of the system is calculated by Equation (16). ein and eout show the
input and output of exergy, and ediss and edest show
exergy dissipation and exergy destruction, respectively.
ein = eout + ediss + edest
m
i =1
m
in ,i
ein ,i + 1 T0 Tin H in =
m out, j eout, j + 1 T0 Tout H out + E + I
Equations (18) to (23) are exergy balance expressions of each unit. The left-hand side of each
equation expresses input, and the right-hand side
expresses output.
Combustor Unit:
m wb ewb + m ga ega + m fe e fe =

1 T0 Trb H rb + 1 T0 Tsi H si + Isi
(18)
Vaporizer Unit:
m gv egv + m rw erw

+ 1 T0 Tv H v = m ve eve + Iv
(19)
Reformer Unit:
(
(
)
)
m ve eve + 1 T0 Tr H r =

m re ere + 1 T0 Tro H ro + Ir
(20)
Shifter Unit:
m re ere = m se ese + 1 T0 Ts H s + Is
(16)
Equation (17) is an exergy balance expression.

The left-hand side is the input term, and there are
substance inputs and heat inputs. The right-hand
side is the output term. The first term of the righthand side expresses the output of substance, the
(17)
(13)
Exergy Flow of the WB-SEG /

PEFC Combined System
e = ech + etc
j =1
+H sys + E sys ,1 + E sys ,2 + H loss + E loss
en ,sys =
second term expresses the heat output, the third

term expresses the power output, and the last term
expresses exergy loss.
(21)
m se ese + m ca eca = m ce ece

+ 1 T T H + I
(22)
257
Cell Stack:
[%]
and thermomechanical exergy (etc ) of the entrance

of each unit are calculated, and they give Equations (18) to (23). As a result, the exergy balance
in each unit can be taken. When the exergy balance in each unit is decided, the exergy balance
in the overall system will be given by giving this
result to Equation (17). The exergy efficiency of
the system can calculate the exergy supply and
the production of electricity of the fuel taken from
the result of the exergy balance described above
by giving Equation (24).
(24)
Analysis Result
rce m ce ece + rfa m fa e fa =

m fe e fe + 1 T0 Tfe H fe + f E f + If
(23)
The exergy efficiency of the system is calculated using Equation (24).

ex ,sys =
eout
ein
E sys ,1 + E sys ,2
m wb ewb + m gv egv
100
Analysis Procedure
Energy Balance and Exergy Flow
The analysis procedure of the WB-SEG / PEFC

combined system shown in Figure 12 is described.
The analysis procedure of each system of WBSEG and PEFC shown in Figure 10 and Figure
11 is the same as that of the analysis method of
the WB-SEG / PEFC combined system.
Figures 13, 14, and 15 show the analysis result

of the energy and exergy flow of the WB-SEG,
the PEFC, and the WB-SEG / PEFC combined
system, respectively. The numbers in brackets in
these figures show exergy. Figure (a) in each figure is the calculation result in the case of heating
the fuel previously using the exhaust heat of the
system. Each exhaust gas of the combustion gas
of WB-SEG, the reformer unit of PEFC, shifter
unit, and CO oxidation unit is used for preheating
the fuel. The exhaust heat of the engine-cooling
water of WB-SEG, the cell stack of PEFC, and
the vaporizer unit is always supplied to the heat
storage tank, and is supplied to the heat demand
side. On the other hand, Figure (b) in each figure
is the calculation result in the case of storing all
the exhaust heat of the system in the heat storage
tank, without preheating the fuel.
1. Energy Flow and Generation Efficiency

First, the generating end output ( E sys,1 and E sys,2 )
of WB-SEG and PEFC is determined. The molar
flow rate and the amount of input-output of heat
of each unit are calculable using the efficiency
and the chemical reaction shown in Figure 12 (b).
The energy flow of the system calculates and
obtains the energy balance in each unit using
Equation (7) to Equation (12). The whole energy
balance is calculated by giving the amount of
energy input-output in each unit to Equation (13).
The generation efficiency of the system is calculated by introducing into Equation (14) the amount
of fuel supply and the production of electricity
taken from the energy balance.
2. Exergy Flow and Exergy Efficiency
First, the fluid temperature in each unit outlet is
calculated from the chemical reaction shown in
Figure 12 (b), and the energy flow described in
the section above. Next, chemical exergy (ech )
258
1. WB-SEG
As shown in Figure 13 (b), the exhaust heat of
the exhaust gas system is given to preheating the
woody biomass fuel and the combustion air at
50%, respectively. Exergy loss of the combustor
and the Stirling engine is large in Figures 13 (a)
and 13 (b) (3440 W and 3430 W). It is expected
that the cause of this is the heat transmission characteristic of the high-temperature combustion gas
Figure 13. Energy flow and exergy flow of WB-SEG
in the combustor and the engine heat exchanger.

The combustion gas of the woody biomass is
considered to be exhausted without sufficiently
exchanging heat for the engine heat exchanger.
In order to reduce exergy loss of the system, the
heat exchange quantity of combustion gas and the
Stirling engine needs to be increased. The Stirling
engine using woody biomass is attended by solid
burning with difficult heat transmission control.
2. PEFC
As shown in Figure 14 (b), 30% of the exhaust heat
of the reformer unit, shifter unit, and CO oxidation unit is given to the city gas for burning. 20%
of heat of the exhaust gas is given to the air for
burning. Moreover, 50% of exhaust heat is given
to the city gas and water for reforming.
Exergy loss of the combustor of PEFC is 1805
W when heating the fuel previously. The exergy
259
Figure 14. Energy flow and exergy flow of PEFC with reformer
loss when not accompanied by preheating the

fuel is 1830 W. These values are small compared
with the combustor of WB-SEG described in
Figure 13. Moreover, exergy loss at the inlet and
outlet of the shifter unit of PEFC is large. This
is because the reformed gas of the reformer unit
260
is supplied to the shifter unit after cooling. The

reformed gas temperature of the reformer unit
outlet is 800 degrees Celsius (1073 K), and the
reformed gas temperature of the shifter unit inlet is
180 degrees Celsius (about 450 K). As one of the
improvement means of exergy loss, it is feasible to
Figure 15. Energy flow and exergy flow of WB-SEG/PEFC combined system
261
smooth the difference in the reaction temperature

in each unit. Accordingly, the rapid temperature
change of reformed gas can be avoided. For
example, development of a shift catalyst with
high-temperature reaction is effective. On the
other hand, development of a reforming catalyst
with low-temperature reaction is also effective.
3. WB-SEG / PEFC Combined System
When previously heating the fuel for reforming,
the fuel for burning (woody biomass), and combustion air of the WB-SEG / PEFC combined
system, exhaust heat is distributed as shown in
Figure 15 (b). This distribution ratio is the same
as PEFC shown in Figure 14 (b). Exergy loss of
the combustor as well as of WB-SEG shown in
Figure 13 is large. Exergy loss of the combined
system with fuel preheating is 5310 W, and loss
if there is no preheating is 5100 W. Exergy loss
in front of and behind the shifter unit shown in
Figure 14 is also large.
Generation Efficiency and

Exergy Efficiency
Figure 16 shows the analysis result of the generation efficiency based on energy and exergy.
Figure 10 (a) shows the result in the case of
supplying room-temperature fuel to the system.
Figure 16 (b) shows the result of the system with

preheating of the fuel. The energy generation
efficiency of PEFC is the highest. Moreover,
although the energy generation efficiency of
the WB-SEG / PEFC combined system is lower
than that of PEFC, it has improved more than
that of WB-SEG. Energy generation efficiency
and exergy generation efficiency will increase
by all the systems, when a fuel is previously
heated using exhaust heat. WB-SEG and the
WB-SEG / PEFC combined system with fuel
preheating improve energy efficiency and exergy
efficiency greatly. This result is because the
high-temperature exhaust heat of the combustor
of the Stirling engine was used effectively. The
energy generation efficiency of the WB-SEG
/ PEFC combined system with fuel preheating
has improved greatly compared with WB-SEG.
Moreover, the exergy efficiency of the WB-SEG
/ PEFC combined system exceeds that of PEFC.
The WB-SEG / PEFC combined system realizes
energy cascade utilization, and possible effective
use of energy.
Conclusion
A combined system that uses the high-temperature
exhaust heat of WB-SEG for city gas reforming and
that supplies the produced reformed gas to PEFC
Figure 16. Analysis results of system generating efficiency
262
was proposed. The energy flow and exergy flow

of this system were analyzed, and the following
conclusions were obtained.
1. Exergy loss of the combustor of the WBSEG is large, due to difference between the
high-temperature combustion gas and the
heat exchanger of the Stirling engine. In
order to reduce exergy loss of the system, it
is necessary to improve the heat-exchange
characteristic described in the top.
2. The exergy loss in front of and behind the
shifter unit of PEFC is large. This is because
the reformed gas outputted from the reformer
unit is cooled greatly. In order to improve
exergy loss, the temperature change of each
unit should be smoothed. For example, it
is effective to develop a shift catalyst with
high- temperature reaction and a reforming
catalyst with low- temperature reaction.
3. Although the energy generation efficiency
of the WB-SEG / PEFC combined system
with preheating of the fuel is lower than that
of PEFC, it is greatly improved compared
with the WB-SEG system. Moreover, the
exergy efficiency of the WB-SEG / PEFC
combined system exceeds that of the PEFC
system. Therefore, the proposed system in
this chapter can realize cascade use of energy,
and can use energy effectively.
EXERGY ANALYSIS OF A
REGIONAL DISTRIBUTED
PEM FUEL CELL SYSTEM
Introduction
When introducing a distributed energy system,
consideration of regional characteristics is required. However, no cases of examination using
exergy in relation to the regional characteristics
of a distributed energy system have been found.
The aspect of this study that differs from others
is when evaluating the regionality of a distributed

fuel cell using exergy analysis. This investigates the
relation between the regionality (environmental
temperature, load characteristic) and the generation efficiency of the PEM fuel cell. If the exergy
efficiency of system components becomes clear,
specification of a unit with considerable exergy
loss can be considered (Shukuya, 2006). The
exergy flow of the energy system influences the
environmental temperature (standard reference
temperature) (Yoshida, 1999; Kopac & Kokturk,
2005). Therefore, installation of the same energy
system as an area where weather differs will
change the exergy flow. If the exergy flow of the
energy system differs for every community, the
system configuration and operation method will
be considered to have an optimal solution for every
community. Therefore, in this study, the exergy
of the fuel cell cogeneration for houses where the
environmental temperature differs is investigated,
and the regional characteristics of the system are
compared. The method of improving the exergy
efficiency of the system is not necessarily in agreement with that of improving energy efficiency. To
date, the exergy of power systems, such as fuel
cells, cogeneration, and combined cycle power
generation systems, has been investigated (Kuzgunkaya & Hepbasli, 2006; Kazim, 2005; Senel et
al., 2005; Mohamad & Mofid, 2005; Ay, Midilli,
& Dincer, 2006). However, no cases investigating the exergy efficiency of residential fuel cell
systems in Japan due to differences in reference
temperature have emerged. Moreover, neither
were any details found of reports investigating
the regional characteristics of the distributed fuel
cell system using exergy analysis.
The fuel cell system used for the analysis in
this study is a proton-exchange-membrane fuel
cell (PEM-FC) with a city-gas (with components
almost equivalent to those of natural gas) reformer
(Hsin-Sen et al., 2008; Obara & Kudo, 2005;
Mathiak et al., 2004). The power demand of the
household electric appliances in an individual
house is an average of 1kW or less at any one
263
time. When the electric power demand for cooling in the summer season is added to this load,
the instant power load will increase to between 2
and 3 kW. Therefore, in this study, the capacity
of a fuel cell is set up for 3kW. Based on numerical analysis, the relation between the reference
temperature, and energy and exergy efficiency
is investigated. Moreover, the characteristic of
energy and exergy efficiency when setting the
outside air temperatures, with Sapporo, Tokyo,
and Kagoshima taken as reference temperatures,
is investigated and the regional characteristics
of the small-scale PEM-FC cogeneration were
considered from these results.
Figure 17. PEFC system model for house
264
Scheme of the Distributed

Energy System
Fuel Cell Cogeneration for Houses
Figure 17 shows the basic system of the small-scale
fuel cell cogeneration for houses as assumed in
this study. Figure 17 (a) shows the system scheme
while Figure 17 (b) shows the flow of the energy
and the system substance. The proposed system
consists of a city-gas reformer, a fuel cell, and
other auxiliary machinery. The city-gas reformer
consists of a combustion chamber (burner) unit,
a vaporizer unit, and a reforming unit. Moreover,
the fuel cell consists of a cell stack, a shifter unit,

and carbon monoxide combustion equipment. The
main auxiliary machinery is a heat storage tank, a
backed boiler, a DC-AC converter, an inverter, and
a high-speed changeover switch. Although city-gas
is supplied to the reformer and the backed boiler,
as shown in Figure 17 (b), it is separated into the
reforming system and the combustion equipment
system in the reformer unit. Heat generation by the
electrochemical reaction and overvoltage of the
cell stack and the heat generation of the reformer
unit, the shifter unit, and the carbon monoxide
combustion unit are recovered. Exhaust heat is
stored in the heat storage tank, which can subsequently be supplied to the demand side when
the period is shifted. Moreover, the production
of electricity (I/T outlet in Figure 17 (b)) of the
system is a maximum of 3kW.
Operating Condition of Each Unit

The reaction and exhaust heat temperatures of each
unit are shown in Figure 18 (a), which also shows
the component of reformed gas. The combustion
gas of the combustor unit is supplied to both the
vaporizer and reformer units, following which
combustion gas is supplied to the heat exchanger
in the heat storage tank. The steam used for steam
reforming is generated in the vaporizer unit. In
the reformer unit, reformed gas is produced from
city-gas and steam and many kinds of catalyst
are used in the reformer unit, the shifter unit,
the CO oxidation unit, and the cell stack, with
proper temperature ranges for each reaction. The
temperature ranges of the reformed gas system
shown in Figure 18 (a) are the optimal range of
each catalytic reaction. Cooling air is supplied to
the shifter unit, the CO oxidation equipment unit,
and the cell stack by a blower. All such exhaust
heat is supplied to the heat storage tank, and stored.
The stored heat shifts a period and can supply the
heat to the demand side, while the backed boiler is
operated when the heat storage temperature is low.
Energy and Substance Flow

The reaction formulas, mass transfer, efficiency,
and the relation between the power and heat output
of each unit are shown in Figure 18 (b). Although
both city-gas and air are supplied to the combustor,
the air-fuel ratio is set to 3.0 based on all operating
conditions. Moreover, the fuel utilization rates
of the cell stack anode and cathode are set to 0.5
and 0.3, respectively, and both the efficiencies of
the DC-AC converter and the inverter are fixed at
0.95. In the CO oxidation unit, a little hydrogen
also burns at the time of CO oxidization. In this
study, the hydrogen-burning rate in reformed gas
is set up with 3 vol%. The oxygen is supplied to
the combustor unit, the CO oxidation equipment
unit, the cell stack, and the boiler. Moreover, water
is supplied to the vaporizer unit using a pump. As
described above, in the proposed system, power
is consumed to provide a supply of air and water,
but the demand required is not included in the
analysis of this study.
Analysis Method
Performance of the Fuel Cell
Figure 19 (a) shows the output performance of
the cell stack used for analysis (Hussain et al.,
2005). The current density of the maximum output point is about 1.5 A/cm2 and ohmic polarization ( Vop ), concentration polarization ( Vcp ),
and activation polarization ( Vap ) exist as resistance components of the fuel cell (Hussain et al.,
2005). The voltage Vvt of the fuel cell is the
value excluding Vop , Vcp , Vap from the
theoretical voltage Vrt , as expressed in Equation
(25). From these resistance components, the relation between the load factor of the cell stack and
generation efficiency is decided. Here, (production-of-electricity / power generation capacity of
the facility)100 [%] is defined as the load factor.
265
Figure 18. System configuration
266
Figure 19. Performance of the PEFC with natural gas reformer
As shown in Figure 19 (b), the reformer efficiency also changes with the load factor (Yasuda,
2005). Here, (calorific heat of the hydrogen in
reformed gas) / (calorific heat of the city-gas supplied to the reformer) 100 [%] is defined as the
reformer efficiency. The generation efficiency of
the system is the value multiplied by the reformer efficiency, and the generation efficiency of the
fuel cell as shown in Figure 19 (b).
Vvt = Vrt Vap Vop Vcp
m gv hgv + m rw hrw

+m bv hbv Q v = m ve hve

Equations (26) to (31) show the energy-balance

equation of each unit, while m and h in each
equation express the molar flow rate and enthalpy, respectively, while Q and W express the
heat and power, respectively. The left-hand side
of each equation expresses inputs, and the right
hand side expresses outputs.
Combustor Unit:
rb
bv
bv
br
br
(28)
(25)
Energy Balance and Energy Efficiency
m gc hgc + m ga hga + m fe h fe

+Q = m h + m h
(27)
Reformer Unit:
m ve hve + m br hbr Q r =

r m re hre + m rx hrx

Vaporizer Unit:
(26)
Shifter Unit:
m re hre + Qs = m se hse + m sx hsx

(29)
CO Oxidation Unit:
m se hse + m ca hca + Qc = m ce hce + m cx hcx

(30)

Cell Stack:
m ce hce + m fa h fa = m fe h fe + Q fe + f W f
(31)
Equation (32) is the energy balance equation
of the system. The left-hand side expresses the
267
input terms and the right hand side the output

terms, while Qsys and W sys of the right hand side
express the heat and power outputs of the system,
respectively. Moreover, the total energy efficiency of the system is calculated using Equation
(33). qlhv in this equation expresses the calorific
heat of the city-gas.
n
m
i =1
ech = ni ei0
in ,i
en ,sys =
hin ,i = m out , j hout , j
(32)
100
(33)
j =1
W sys + Q f + Qe
(m gc + m gv ) qlhv
Exergy Analysis
(etc ), kinetic exergy (eki ), and potential exergy (

ep ), as shown in Equation (35). However, in this
study, kinetic exergy and potential exergy are not
calculated from very small values compared with
chemical exergy and thermomechanical exergy.
e = (h h0 ) T0 (s s 0 )
T
n

etc = ni C p,i T T0 T0 ln

i =1
T0
(37)
2. Exergy Balance and Exergy Efficiency
The exergy balance of the system is described in
Equation (38) (Ishihara, A.,et al., (2004)). ein and
eout express the input and output of exergy, while
1. Exergy Calculation
By environmental temperature T0 , the exergy of
the reversible change of the open system is calculated by Equation (34). It means that T0 (s s 0 )
is not convertible for mechanical energy among
the enthalpy change (h h0 ) . Here, h , s , and
T express enthalpy, entropy, and temperature
respectively, while subscript 0 indicates a standard
condition. Exergy is calculated by the sum of
chemical exergy (ech ), thermomechanical exergy
e = ech + etc + eki + ep
(36)
i =1
+Qsys +W sys + Qloss
(34)
(35)
ech and etc in Equation (35) are calculated

using Equations (36) and (37), respectively (Yo-
268
shida, K., (1999)). Here, the number of components of the fluid is set to n , and ni and ei0 express
the number of moles and the standard chemistry
exergy of component i . C p,i expresses the specific heat of component i .
ediss and edest express the exergy dissipation and

exergy destruction, respectively. Moreover, Equation (39) expresses exergy loss.
ein = eout + ediss + edest
(38)
eloss = ein eout = ediss + edest
(39)
Equation (40) is the exergy balance equation.

The left-hand side is the input terms, comprising
inputs of substance and heat, while the right hand
side is output terms. The 1st term of the right hand
side expresses the output of substance, the 2nd
expresses the output of heat, the 3rd expresses the
power output and the final term expresses the
exergy loss rate I .
n
m
i =1
m
in ,i
ein ,i + 1 T0 Tin Q in =
m out, j eout, j + 1 T0 Tout Qout +W + I

j =1
(40)
Equations (41) to (46) represent the exergy

balance equation in each unit. The left-hand side
of each equation expresses the input, and the right
hand side expresses the output.
m gc egc + m ga ega + m fe e fe =
m bv ebv + m br ebr + 1 T0 Tcb Qcb + Icb

Vaporizer Unit:
m gv egv + m rw erw + m bv ebv =

m ve eve + 1 T0 Tv Q v + Iv
(41)
(42)
Reformer Unit:
m ve eve + m br ebr + 1 T0 Tr Q r =
(43)
m e + m e + I
r
re
re
rx
rx
Shifter Unit:
m re ere = m se ese + m sx esx

+ 1 T T Q + I
(44)
m se ese + m ca eca = m ce ece

+m e + 1 T T Q + I
cx
cx
(45)
Cell Stack:
rce m ce ece + rfa m fa e fa = m fe e fe

+ 1 T T Q + W + I
fe
fe
W sys
+m cx ecx + 1 T0 Tfe Q fe

m gc egc + m gv egv
Combustor Unit:
eout
=
ein
+ m rx erx + m sx esx
ex ,sys =
(47)
100
Analysis Procedure
Each equation described by previous sections is
calculated using the newly developed computer
program, with the analysis procedure described
below. The energy balance of each unit is calculated using Equations (26) to (31), and the energy
balance of the whole system is calculated by
Equation (32). The relation of the load factor,
efficiency of the reformer and fuel cell is determined from the relation of Figure 19 (b). On the
other hand, calculation of the exergy balance of
each unit is calculated using Equations (41) to
(46). Moreover, the exergy balance of the system
is calculated by Equation (40) and the energy and
exergy flowcharts are prepared from these results.
Next, reference temperature (T0 ), used for the
calculation of energy and exergy, is changed, and
the relation between the environmental temperature, energy efficiency, and exergy efficiency of
the system is investigated. These relations are
introduced into the outside-air-temperature profile
of each city, and the energy and exergy efficiencies
of the system are calculated. However, the operation of the system follows power load fluctuations.
The typical parameters used in analysis are
shown in Table 5.
Analysis Result
(46)
The total exergy efficiency of the system is

calculated by Equation (47).
Energy Balance and Exergy Balance

Figure 20 shows the results of the energy balance
and exergy balance at reference temperature 298
K and 0.1013 MPa. However, the production of
269
Table 5. Analysis parameters
Figure 20. Analysis results of energy flow and exergy flow. In the case of the power generation output
of 3 kW, and standard reference temperature 298 K.
270
electricity (W sys ) of the system is 3kW. Moreover,

Figure 21 (a) shows the analysis result of the
exergy loss of each unit when changing the reference temperature. As shown in Figures 20 (b) and
21 (a), exergy loss in the combustor unit is considerable, the vaporizer unit, and the reformer
unit. In order to reduce the exergy loss of the
combustor unit, it is necessary to preheat city-gas
and air using the exhaust heat of the system, and
to raise the combustion temperature. For example,
in order to reduce the exergy loss Icb in Equation
(41), it is necessary to boost the value of
1 T T Q of the 3rd term of the right hand
cb
Figure 21. Analysis results
cb
side. Therefore, the increase in the combustion

temperature Tcb is effective in reducing the exergy loss. Likewise, preheating the water supplied
to the vaporizer also decreases exergy loss. When
the proposed system is introduced into each city
as will be described in the latter section, exhaust
heat will often exceed heat demand, meaning
preheating of the fluid using exhaust heat is a
sensible move.
On the other hand, the reason for the considerable exergy loss of the reformer unit is the need
to cool the hot-reformed gas according to the
catalytic reaction (See Figure 18 (a)). Although
the reformed gas temperature of the reformer unit
is 1073K, it is cooled and supplied to the shifter
unit at 450K (See Figure 18 (a)). From the
shifter unit to the CO oxidation equipment unit,
and the CO oxidation unit to the cell stack, the
temperature change of the reformed gas is modest,
with little exergy loss. In order to decrease the
exergy loss, it is necessary to minimize any drop
in temperature due to the cooling of reformed gas.
For example, the catalyst material of each unit
may be modified and improvements to reduce the
temperature variation between units can be considered.
The total energy efficiency shown in Figure
20 (a) is 72%, and the total exergy efficiency
shown in Figure 20 (b) is 40%. The total exergy
271
efficiency is low because of the small heat output

efficiency and the usage of heat from reformed gas
in particular requires modification. As a method
for improving exergy efficiency, the combined
cycle power generation using the heat of reformed
gas is effective.
Influence of the Environmental

Temperature
Figure 21 (b) shows the analysis result of the
energy and exergy efficiencies when changing
the reference temperature of the proposed system.
The difference of the exergy generation efficiency of 263K (-10 C ) and 313K (40 C ) is
1.5%, while the difference of the efficiency of a
heat output is 42% and that of exergy total efficiency is about 13%. Conversely, the difference
of the energy generation efficiency within the
same temperature range is 1.5%, and the difference
of the heat output efficiency of energy is 3%,
while that of energy total efficiency is about 3%.
Although the difference in environmental temperature has a significant influence on exergy
efficiency, its influence on energy efficiency is
small. Figure 21 (c) shows the relation between
the load factor of the system, and energy total
efficiency and exergy total efficiency. In the case
of load factor 4% to 100% and an environmental
temperature range of 263K (-10 C ) to 313K (40
C ), the difference of energy total efficiency is
6%. On the other hand, the difference in exergy
total efficiency within the same range is 30%. In
this way, the exergy total efficiency changes significantly with the environmental temperature.
Evaluation of Regionality
1. Analysis System
In this study, the energy and exergy efficiencies
at the time of introducing the proposed system
into individual homes of each city in Sapporo,
Tokyo, and Kagoshima are investigated. Figure
21 (d) shows the temperature of the representative
272
days for every month in each city. The temperature of these representative days is a value that
is averaged based on the outside air temperature
at the same time every month 12 to 18 in 2005.
The meteorological data for each city, released by
the Meteorological Agency, was used to obtain
these outside air temperature data (Meteorological Agency meteorological observation in Japan,
(2007)). Figures 22 (a) and 22 (b) show the power
demand and heat demand models of an individual
house in each city (Narita, K., (1996), The National Research Study Concerning the Energy
Consumption in the House, (2002)), with the
load of household appliances, electric lights, and
space cooling included in the power load and that
for space heating, hot water supply, and the bath
included in the heat load. Because there is no space
cooling in Sapporo, there is no significant difference in the power load pattern every month. On
the other hand, because the cooling load increases
compared with Tokyo, the power load in summer
is considerable in Kagoshima.
2. Regional System Characteristics
Figure 23 shows the analysis result of the energy
and exergy efficiencies in winter (February), midterm (May), and summer (August) representative
days. Figures 24 (a) and 24 (b) show the difference
of energy and exergy efficiencies in each city in
comparison with Tokyo. Except for the generation
efficiency of the summer in Kagoshima, there is
no difference in each city in terms of considerable
energy efficiency. There are many cooling loads
and the load factor increases in summer in Kagoshima, meaning generation efficiency follows
suit. On the other hand, in the analysis result of
the exergy efficiency shown in Figure 23 (b), the
heat output efficiency differs greatly in each city.
The difference shown in Figure 24 (c) is between
the exhaust heat output of the system, and the
heat demand amount. When there is insufficient
heat, operation of the backed boiler is required,
and in the case of a heat surplus, heat release is
required. Figure 25 shows the analysis result of
Figure 22. Energy demand
273
Figure 23. Energy and exergy efficiency of the proposal system
Figure 24. Regional difference of the analysis results
274
Figure 25. System efficiency when taking waste heat into consideration
the energy and exergy efficiencies when releasing surplus heat. Moreover, Figure 26 shows the
difference of the energy and exergy efficiencies
in each city in comparison with Tokyo. The fall
in the energy and exergy efficiencies due to the
release of surplus heat is remarkable in the city
with a high outside air temperature. Because the
heat release tends to increase when the outside air
temperature is high and there is little heat demand,
total energy efficiency falls. However, in the city
with a high outside air temperature, both the space
cooling load and generation efficiency increase.
Conclusions
The exergy flow and exergy efficiency of the
distributed energy system using small-scale PEMFC were investigated. As a result, when the same
energy system was introduced into a community
where climates differ, the exergy flow was shown
to vary. The exergy efficiency when the proposed
system was introduced into individual houses in
Sapporo, Tokyo, and Kagoshima was evaluated.
At the case of a load factor of 4% to 100% and
environmental temperature of 263K to 313K, the
Figure 26. Difference with Tokyo of the energy and exergy efficiency when taking release into consideration
275
difference of energy total efficiency is 6%. On

the other hand, the difference in the exergy total
efficiency within the same range is 30%. The
following methods were suggested for improving
exergy efficiency:
1. Preheat the city-gas and air supplied to the
system using exhaust heat, and raise the
combustion temperature.
2. Preheat the water supplied to the system
using exhaust heat.
3. In order to reduce the cooling of the reformed
gas as much as possible, modify the catalyst
material of each unit.
4. The examination of combined cycle power
generation.
When the proposed system is introduced into
a community with a low outside air temperature,
exergy efficiency will increase compared with a
community with a high outside air temperature.
Moreover, exhaust heat must be released into
communities with a high outside air temperature,
meaning a considerable drop in energy and exergy
efficiencies. On the other hand, because the space
cooling load of the system introduced into this
community increases, the generation efficiency
does the same.
REFERENCES
Bonnet, S., Alaphilippe, M., & Stouffs, P.
(2005). Energy, exergy and cost analysis of a
micro-cogeneration system based on an ericsson engine. International Journal of Thermal
Sciences, 44, 11611168. doi:10.1016/j.ijthermalsci.2005.09.005
De Ruyck, J. et al. (2007). Co-utilization of
biomass and natural gas in combined cycles
through primary steam reforming of the natural
gas. Energy, 32(4), 371377. doi:10.1016/j.energy.2006.07.010
276
Eguchi, K., & Kurosawa, Y. (1988). Study on high

load swirl burners for Stirling engines. Reduction in NOx and heat adsorption characteristics
in exhaust gas recirculation. In Proc. of the 26th
Japanese Symp. on Combustion, (pp. 131-133).
Gan, J., & Smith, C. T. (2006). A comparative
analysis of woody biomass and coal for electricity
generation under various CO2 emission reductions and taxes. Biomass and Bioenergy, 30(4),
296308. doi:10.1016/j.biombioe.2005.07.006
Hamelin, K. etal. (2001). Dynamic behavior of
a PEM fuel cell stack for stationary applications.
International Journal of Hydrogen Energy, 26,
625629. doi:10.1016/S0360-3199(00)00121-X
Hawkes, A. D., & Leach, M. A. (2007). Costeffective operating strategy for residential microcombined heat and power. Energy, 32(5), 711723.
doi:10.1016/j.energy.2006.06.001
Hussain, M. M., Baschuk, J. J., Dincer, X., & Li,
I. (2005). Thermodynamic Analysis of a PEM
Fuel Cell Power System. International Journal
of Thermal Sciences, 44, 903911. doi:10.1016/j.
ijthermalsci.2005.02.009
(High School Active Science Project Research
Report). Ibaraki, Japan: Author. In Japanese
Ibe, S. (2002). Development of fuel processor for
residential fuel cell cogeneration system. In K.
Abe (Ed.), Proc. 21th Annual Meeting of Japan
Society of Energy and Resources, (pp. 493-496).
In Japanese.
Koike, S., et al. (1999). Demonstration program
of 1kW-class PEMFC system for residential use
by the Japan gas association. In Proc. 3rd Int. Fuel
Cell Conference, (pp. 497-498).
Kyoto Denkiki Co., Ltd. (2001). A system connection inverter catalog and an examination
data sheet.
Mikkola, M. (2001). Experimental Studies on

Polymer Electrolyte Membrane Fuel Cell Stacks
University, Japan. In Japanese
Obara, S. (2006a). Effective improvement in
generation efficiency due to partition cooperation
management of a fuel cell microgrid. Journal of
Thermal Science and Technology, 1(2), 4253.
doi:10.1299/jtst.1.42
Obara, S. (2006b). Load response characteristics
Energy, 31(13), 18191830. doi:10.1016/j.
Obara, S., Tanno, I., Kito, S., Hoshi, A., & Sasaki, S. (2007). Design of Environmental Energy
System Using Stirling Engine. In Proceedings of
the 5th International Symposium on Environmentally Conscious Design & Inverse Manufacturing
(EcoDesign), B2-1-4S, 1-8, Tokyo, Japan.
Oda, K. etal. (1999). Small-scale reformer for fuel
Review, 31(2), 99106.
Ozeki, H. (1991). Research on a combustor for
Stirling engines. 4. Effects of a combustor engine
operation parameter on exhaust characteristics. In
Proc. of the 29th Japanese Symp. on Combustion,
(pp. 208-210).
Peacock, A. D., & Newborough, M. (2007).
Controlling micro-CHP systems to modulate
electrical load profiles. Energy, 32(7), 10931103.
doi:10.1016/j.energy.2006.07.018
Obara, S. (2006c). Load response characteristics

of the PEM fuel cell for individual cold-region
houses. Transactions of the ASME. Journal of Fuel
Cell Science and Technology, 5(1), 011005-111.
Qian, Y., Zuo, C., Tan, J., & He, J. (2007). Structural analysis of bio-oils from sub-and supercritical
water liquefaction of woody biomass. Energy,
32(3), 96202. doi:10.1016/j.energy.2006.03.027
Obara, S., Kito, S., Hoshi, A., & Sasaki, S. (2007a).

Dynamic characteristics of PEFC / woody biomass
engine hybrid microgrid. Journal of Thermal Science and Technology, 2(1), 111122. doi:10.1299/
jtst.2.111
Reduction in NOx and heat adsorption characteristics in exhaust gas recirculation. In Proc. of
the 26th Japanese Symp. on Combustion, (pp.
131-133).
Obara, S., Kito, S., Hoshi, A., & Sasaki, S. (2007b).

Energy balance of Stirling engine cogeneration
and installation potentiality to cold region houses.
Journal of Environmental Engineering, 2(4),
696707. doi:10.1299/jee.2.696
Obara, S., & Kudo, K. (2007). Installation planning
of small-scale fuel cell cogeneration considering
transient response characteristics (load response
characteristics of electric power output). Journal
of Environmental Engineering, 2(3), 621631.
doi:10.1299/jee.2.621
Rhn, C., hman, M., Gref, R., & Wsterlund, I.

(2007). Influence of fuel ash composition on high
temperature aerosol formation in fixed bed combustion of woody biomass pellets. Fuel, 86(1-2),
181193. doi:10.1016/j.fuel.2006.07.001
Sedghisigarchi, K. (2004). Dynamic and transient
analysis of power distribution system with fuel
cells-Part 1: Fuel-cell dynamic model. IEEE Transactions on Energy Conversion, 19(2), 423428.
doi:10.1109/TEC.2004.827039
277
Taniguchi, Y., & Ishida, M. (2005). Passive load
following method for PEFC system with reformer
and its efficiency improvement evaluated by using the concept of exergy. Electrical Engineering
in Japan, 152(4), 1726. doi:10.1002/eej.20148
278
Yerramalla, S. et al. (2003). Modeling and

simulation of the dynamic behavior of a polymer
electrolyte membrane fuel cell. Journal of Power
Sources, 124, 104113. doi:10.1016/S03787753(03)00733-X
Zhang, Y. (2004). A model predicting performance of proton exchange membrane fuel cell
stack thermal systems. Applied Thermal Engineering, 24, 501513. doi:10.1016/j.applthermaleng.2003.10.013
APPENDIX
Nomenclature

Ar : Surface area of the combustion chamber [m2]
C a : Mean specific heat of exhaust gas [J/(kgK)], [J/molK]

C w : Specific heat of engine-cooling-water [J/(kgK)]
E : Power [W]
EC : Capacity of power generation [kW]
Es : Generating end output [W]

e : Exergy [W], [J/mol]
e 0 : Standard chemical exergy [J/mol]
H : Heat [W]
h : Enthalpy [J/mol]
I : Integration parameter
I : Exergy loss rate [W]
K : The rate of overall heat transfer [W/(m2K)]
Lb : Lower calorific value of the woody biomass [J/g]
LCH : Lower calorific value of city-gas [J/kg]
m : Molar flow rate [mol/s]

P : Proportionality parameter
Pg : Pressure gauge
Q : Heat [W]
q : Heat flux sensor

q : Heat quantity [W]
qa : Exhaust gas quantity of heat [W]
qb : The amount of combustion heat of the woody biomass [W]
qlhv : Calorific value of city gas [J/mol]
qr : The amount of heat release from the combustion chamber surface [W]
qw : Engine-cooling-water quantity of heat [W]

r : Fuel utilization rate
R : Power load factor [%]
rce : Utilization rate of anode gas [%]
rfa : Utilization rate of cathode gas [%]

s : Entropy [J/molK]
T : Temperature [K]
ta : Difference of exhaust gas temperature and room temperature [K]
tr : Difference of combustion-chamber temperature and room temperature [K]

tw : Difference of engine-cooling-water and room temperature [K]
279
u : Power input [kW]

V : Voltage [V]
V : Overpotential [V]
Va : Exhaust gas flow rate [m3/s]
Vf : Supply of the woody biomass [g/s]
Vw : Engine-cooling-water flow rate [g/s]

W : Power [W]
Greek Letters

: Efficiency
a : Mean density of emission
w : Density of engine cooling water
Subscript

280
a : Exhaust gas of the combustion chamber

ap : Activation polarization
br : Combustion gas or combustion gas supplied to the reformer unit
bv : Combustion gas supplied to the combustor unit
c : CO oxidation unit
ca : Air supplied to the combustor unit
cb : Burning or combustor unit
ce : Reformed gas of the CO oxidation equipment unit
ch : Chemical
cp : Concentration polarization
cr : CO oxidation heat
cx : Cooling air of the CO oxidation equipment unit
D : DC-AC converter
dest : Destruction
diss : Dissipation
Ex : Exhaust gas of SEG
Ey : Cooling water of SEG
Ez : Heat radiation of SEG
e : Exhaust heat (Figure 15 (b))
en : Energy
ex : Exergy
FC : Proton exchange membrane fuel cell
f : Cell stack
fa : Air supplied to the cell stack
fe : Exhaust gas of the cell stack
fr : Cell stack heating valve
ga : Air supplied to the combustor unit

gc : City-gas supplied to the combustor unit
gv : City-gas supplied to the vaporizer unit
I : Inverter
ki : Kinetic
m : The number of composition of the fluid
n : The number of composition of the fluid
op : Ohmic polarization
p : Potential
r : Reformer unit
ra : Reforming heat
rb : Combustion gas
re : Reformed gas of the reformer unit
rm : Heat release of the combustion chamber
rw : Water supplied to the vaporizer unit
rx : Exhaust gas of the reformer unit
SEG : Stirling engine
s : Shifter unit
sa : Shifting heat
sc : Cell stack
se : Reformed gas of the shifter unit
si : Output to the stirling engine
sx : Cooling air of shifter unit
sys : System
tc : Thermomechanical
tc : Thermomechanical
v : Vaporizer unit
va : Amount of heat of vaporization
ve : Reformed gas of the vaporizer unit
vo : Exhaust heat of the vaporizer unit
vt : Power voltage output
w : Cooling-water of the Stirling engine
wb : Woody biomass
0 : Standard reference
281
282
Chapter 9
Design Support Using a

Neural Network Algorithm
ABSTRACT
This chapter consists of two sections, Dynamic Operational Scheduling Algorithm for an Independent
Microgrid with Renewable Energy and Operation Prediction of a Bioethanol Solar Reforming System
Using a Neural Network. In the 1st section, a dynamic operational scheduling algorithm is developed
using a neural network and a genetic algorithm to provide predictions for solar cell power output (PAS).
The section shows that operating the microgrid according to the plan derived with PAS was far superior, in terms of equipment hours of operation, to that using past average weather data. Because solar
radiation and outside air temperature are unstable, it is difficult to predict operation of the system with
accuracy. Therefore, the 2nd section developes an operation prediction program of the FBSR (bioethanol reforming system) using a layered neural network (NN) with the error-correction learning method.
microgrid with the capacity for sustainable energy
is expected to be a distributed energy system that
exhibits quite a small environmental impact. In an
independent microgrid, green energy, which is
typically thought of as unstable, can be utilized
effectively by introducing a battery. In the past
study, the production-of-electricity prediction

algorithm (PAS) of the solar cell was developed.
In PAS, a layered neural network is made to learn
based on past weather data and the operation plan
of the compound system of a solar cell and other
energy systems was examined using this prediction algorithm. In this study, a dynamic operational scheduling algorithm is developed using
a neural network (PAS) and a genetic algorithm
DOI: 10.4018/978-1-4666-5796-0.ch009
Design Support Using a Neural Network Algorithm
(GA) to provide predictions for solar cell power

output. We also do a case study analysis in which
we use this algorithm to plan the operation of a
system that connects nine houses in Sapporo to
a microgrid composed of power equipment and
a polycrystalline silicon solar cell. In this work,
the relationship between the accuracy of output
prediction of the solar cell and the operation plan
of the microgrid was clarified. Moreover, we found
that operating the microgrid according to the plan
derived with PAS was far superior, in terms of
equipment hours of operation, to that using past
average weather data.
The bioethanol reforming system (FBSR) using
sunlight as a heat source is a fuel production system for fuel cells with little environmental impact.
However, because solar radiation and outside air
temperature are unstable, it is difficult to predict
operation of the system with accuracy. Therefore,
an operation prediction program of the FBSR using a layered neural network (NN) with the errorcorrection learning method has been developed.
We developed a method of analyzing the operation
of a natural energy system with sufficient accuracy. The weather pattern (the amount of global
solar radiation and the outside air temperature)
and energy-demand pattern for the past one year
are inputted into the NN. Moreover, training signals are calculated by a genetic algorithm (GA).
The training signals are given to the NN, and the
operation pattern of the FBSR is made to learn.
Operation of the FBSR on arbitrary days can be
predicted by inputting the weather pattern and the
energy-demand pattern into this learning NN. In
this study, the operation prediction program of
the FBSR is developed, and details of the analytic
accuracy are clarified. As a result of analyzing using the developed algorithm, when 20% or less
of power load fluctuation occurred, the operation
plan was analyzable in 14% or less of error span.
On the other hand, in operation prediction when
50% or less of fluctuation is added to the outside

temperature and global solar radiation, there was
16% or less analysis error.
DYNAMIC OPERATIONAL
SCHEDULING ALGORITHM FOR
AN INDEPENDENT MICROGRID
WITH RENEWABLE ENERGY
Introduction
Microgrid technology with the capacity for sustainable energy operation has been widely discussed recently from the point of view reducing the
environmental impact of society (Abu-Sharkh et
al., 2006; Muselli, Notton, & Louche, 1999; Ismail,
Kemmoku, Takikawa, & Sakakibara, 2002). In
these setups, the operation optimization program
installed in the controller of a combined system
is the most important aspect of the technology
for determining the performance of the system (
Obara, 2006). However, because an output prediction for the green energy contribution to the
system is required, the dynamic operation plan of
a system that combines conventional energy equipment (for example, a diesel engine, a gas engine,
a fuel cell, etc.) and green-energy equipment can
be very difficult to design. In this work, we use
a neural network (NN) to obtain output predictions for a solar cell. Weather data from the past
14 years (amount of solar radiation and outside
temperature) is fed into the learning process of the
NN. This NN production-of-electricity prediction
algorithm was developed by the author and is
described to be PAS in (Obara & Tanno, 2008).
Power fluctuations are known to occur in
systems that utilize green energy on an independent microgrid and that experience large or rapid
changes in load (Obara, 2007). Given this, power
storage equipment must be introduced and the
dynamic characteristics of the microgrid must be
283
improved. Due largely to the proliferation of hybrid

vehicles and the like, the cost and performance
of batteries have recently improved remarkably
(Jorgensen, 2008). With this in mind, this Section
investigates algorithms for the operation planning
of a microgrid that combines conventional type
energy equipment, a solar cell and a battery. Since
a microgrid is typically built up of two or more
energy systems, we have to solve a nonlinear
problem with many variables. In this Section, the
operation condition of the generating equipment
is expressed with chromosome codes, these are
introduced into a GA, and the optimal operation
planning is determined.
Independent Microgrid with

Renewable Energy and Battery
Figure 1 shows the case of an independent microgrid that distributes power to a small collection
of generating equipment (here we use gas engine
generators as an example, but our analysis is not
limited to this case). The microgrid is controlled
by a system controller on which we have installed
dynamic operation planning software. In this Section, we investigate the power supply-and-demand
characteristics of the microgrid; the examination

concerning the supply-and-demand of heat in a
similar system will be reported independently.
Dynamic Operation Planning

1. Operation Planning Based on a Solar Cell
Output Power Model
Figure 2 shows the operation method of a microgrid
with a solar cell. Dynamic operation of the microgrid is planned based on the output power
model of a solar cell. In this Section, the solar cell
output model is based on PAS predictive data and
past average weather data (Homepage of Japan
Meteorological Agency, 2007; NEDO Technical
information data base, 2008). The PAS is a production-of-electricity prediction algorithm that
was designed specifically for the solar cell in the
work (Obara & Tanno, 2008). Figure 2 (a) shows
the method of planned dynamic operation based
on the output power model of the solar cell described in the introduction. From time t0 to time
t2 , there is no output from the solar cell and so
in this interval, more electric energy is stored in
the battery than is required by the load. In light
of this, generating equipment is operated during
this period (Generator output (1)). Each piece of
generating equipment is operated by the fixed
Figure 1. Independent microgrid system using gas engines
284
Figure 2. Operation planning of a microgrid with a solar power
load near the maximum efficiency point. The

supply of electric power in the microgrid is adjusted by controlling the number of pieces of
power equipment in operation. When the supply
of electric power exceeds the load in the period
from t0 to t2 , the amount of surplus charge is
moved into the battery. There is an output of the
solar cell in the period t3 to t6 . In the period t2
to t4 , the system responds to a load by discharg-
ing the battery, and supplying charge from output

of the solar cell. Because the output of the solar
cell exceeds the load in the period t5 to t4 , surplus
power can again be stored in the battery. If the
output characteristics of the solar cell and the
characteristic of power load are able to be predicted in the early morning, one can minimize the
number of equipment hours of operation and the
number of charge and discharge cycles on the
285
battery throughout the day. It is known, however,

that the load following operation of power equipment is disadvantageous when it comes to installed
capacity and the hours of operation required of
power equipment when compared with the output
predictive model of a solar cell (Mohammadi et
al., 2005).
2. Error of the Output Predictive Model of a
Solar Cell, and the Influence on an Operation
Plan
Figure 2 (b) shows the method of an operation
plan when a difference occurs between the output
model of the solar cell shown in Figure 2 (a), and
an actual solar cell output. This system starts out
(during t0 to t2 ) by following the operation plan
shown by the command of the controller in Figure
2 (a). However, as shown in Figure 2 (b), since
there is so little power being produced in the
solar cell, the period of t4 to t5 shown in Figure
2 (b) requires additional operation of power equipment (Generator output (2)). The system is continuously controlled by the controller to follow
the operation shown in Figure 2 (a). However,
there is little storage of electricity by the solar cell
in Figure 2 (b) compared with Figure 2 (a). For
this reason, when discharged from a battery in the
period t7 to t8 , the system will change to load
following operation (Generator output (3) and
(4)). As a result, the number of hours of operation
of power equipment increases, and the additional
capacity of the battery cannot be sufficiently
utilized further.
Solar Cell System

The area of power-generation of the solar cell
introduced into the microgrid will be called Ssol
and we are assuming a polycrystalline silicon type
solar cell. The production of electricity of the
solar cell Psol is calculated using Equation (1).
The power-generation efficiency changes as the
286
temperature of the solar cell Tref changes (the

efficiency falls as the temperature increases). The
temperature coefficient in this case is called RT .
T is a reference temperature and sol is the
power-generation efficiency of the solar cell at
T . H D in Equation (1) expresses the solar radiation intensity for direct delivery (the intensity
of radiation which enters into the acceptance
surface). Moreover, H S in Equation (1) expresses the solar radiation intensity of the dispersion
component.
Psol = Ssol sol (H D + H S )

R
1 Tref T T
100
(1)
Power Balance and

Objective Function
Power Balance
Equation (2) expresses the power balance equation
in the proposed microgrid. The left-hand side of
the equation is the power outputted by the composition of equipment that makes up the system,
and the right-hand side expresses the power consumed by the microgrid. E gen ,i,t , Ebt ,t and Esol ,t
of the left-hand side express the output of the
generating equipment, battery, and solar cell
between time t and t + 1 , respectively. Moreover,
N eng is the number of pieces of generating equipment introduced into the microgrid. Eneed , j ,t is the
power demand at time t of the house j . N house
is the number of the houses connected to the
microgrid. The last term on the right-hand side
of Equation (2) ( Eloss ,t ) expresses the power
loss in the system. The charge-and-discharge efficiency of a battery, power transmission loss,
etc., are included in this term. In the analysis of
this Section, the charge-and-discharge efficiency
of a battery is the only effect included in this term.

The sampling interval for each piece of equipment,
such as the generating equipment, a solar cell, or
a battery, is set to be one hour in this Section.
Neng
i =1
N house
gen ,i ,t
E
j =1
+ Ebt ,t + Esol ,t =
need , j ,t
value. The solution that more closely satisfies

Equation (4) is described as having a large adaptive value.
EOT t =
(2)
+ Eloss ,t
Objective Function
The number of hours of operation of the generating equipment EOT t between time t to t + 1
is obtained by calculating Equation (3). However,
ngen ,i,t expresses the operational status of the
generating equipment i in the time between t
and t + 1 (1 indicates operation, 0 indicates
idle). N gen is the number of pieces of generating
equipment installed in the microgrid. The total
number of hours of operation for all pieces of
generating equipment that were in operation during the periods t = 1,2,.....,Psys is calculated by
Equation (4). The optimization of the dynamic
operation plan of the microgrid is examined using
a genetic algorithm (GA). In the GA, the objective
function shown by Eq (4) is defined as an adaptive
Ngen
n
i =1
gen ,i ,t
Engine operation time =

Psys
minimize EOT t
t =1
(3)
(4)
Analysis Method
Production-of-Electricity Prediction
Algorithm of Solar Cell (PAS)
A layered neural network (NN) is introduced
and the production of electricity of a solar cell is
predicted according to the following procedures.
1. Input-and-Output Data of NN Used by PAS
Figure 3 expresses the input-and-output data of
the NN used for PAS. dw expresses the present
date and t expresses the present time. In the
learning and analysis process of the NN, the average amount of solar radiation and average outdoor
air temperature are inputted as data for each time
of the present date. The input data described in
Figure 3. Input data introduced into proposal NN
287
the introduction is fed into Equation (1), and the

production of electricity Psol of the solar cell is
obtained. This Psol used as the teaching data in
the learning process of the NN.
2. Input Data
The data inputted into the NN the learning process
includes the average amount of solar radiation
and average outdoor air temperature for each time
of the day (from time 0 to time t ). On the other
hand, the average amount of solar radiation and
average outdoor air temperature from t to 24 of
a given day give the values measured on the same
time and the same day of previous years, as obtained from the standard weather and the solar
radiation database on weather government office
and AMEDAS (1990 to 2003) (Homepage of
Japan Meteorological Agency, 2007), and NEDO
technical information data base (METPV-3)
(NEDO Technical information data base, 2008).
The data inputted into the NN during the analysis
process are the same as those inputted during the

learning process from 0 to t . The average amount
of solar radiation and average outdoor air temperature from t to 24 of a given day give the data
measured at same time the previous day.
3. Structure of the NN, and Output Data
Figure 4 shows the structure of the NN introduced
into PAS. Figure 4 (a) shows the learning process
and Figure 4 (b) shows the analysis process. The
NN used in this proposal has three layers. In the
learning process, the input data ( x 1 to x 48 ) (described in Section (2)) are fed into the first layer
(the power-input layer) and the teaching data ( y1
to y24 ) are fed into the third layer (the output
layer). The weight of each network connection
between neurons is decided using back propagation (Obara, S., & Tanno, I., 2007) so that the
relationship between each input data point and
each teaching data point may be realized. Input
data can be given to a learned NN (in the analysis
Figure 4. Layered neural network of the proposal system
288
process), and the solar cell output power ( y1 to

y24 ) in each time of dw can be obtained from the
output layer.
number of individuals of high adaptive value

(gotten by evaluating of the objective function)
may be multiplied at a fixed rate.
Optimization of Dynamic Operation

using a Genetic Algorithm (GA)
Crossover
Chromosome Model
Dynamic operation planning of the microgrid is
optimized using a GA based on the solar cell
output power model (the PAS predictive data and
the past average weather data). Figure 5 expresses the chromosome code used by the proposed
GA. One chromosome (individual) consists of 24
genes. Each gene shows the operational status of
the various pieces of generating equipment that
are on the grid during each time step in the date
dw . With the number of the pieces of generating
equipment introduced into the grid being called
N gen , the value of each gene is an integer between
0 and N gen . The genes of the initial generations
chromosome are decided using random numbers.
Multiplication and Selection

By decoding the genes in a chromosome, the
operation condition ngen ,i,t of the generating
equipment in Time t can be obtained. By giving
ngen ,i,t to Equation (3), EOT t is calculable.
Furthermore, the value of the objective function
of each chromosome is calculated from Equation
(4). The software implementation is such that the
Change of generation is repeated, adding the

genetic manipulation of crossover to chromosomes
to maintain diversity (an evaluative process). In
the chromosome code, when arriving at the last
generation, a fitness value decides the individual
most suited to be the optimal operating method.
In the crossover calculation, two parent chromosomes are selected by the crossover probability
Pcros given beforehand, and the crossover position
common to both of parent chromosomes is decided at random. The genes of both of the parent
chromosomes are rearranged bordering on the
crossover position, and the chromosome with a
new gene is generated.
Analysis Flow of Operation Planning

Analysis flow of the operation planned optimization of the microgrid by the proposal GA is shown
in Figure 6. In the Calculation part (a) of the
figure, an electricity demand pattern, a solar cell
output power model (PAS or output power pattern
of the solar cell based on the past average weather data), and the parameter of GA are inputted.
The initial generations chromosome group is
generated in Calculation part (b). In Calculation
part (c), the fitness value of all the chromosomes
is calculated and the order of chromosomes is
decided at Calculation part (d) according to the
Figure 5. Chromosome code used for genetic algorithm
289
Figure 6. Calculaton flow of the optimal operation plan using GA
magnitude of their fitness values. Chromosomes

of large fitness value are made to increase in
number at a fixed rate, while chromosomes of
small fitness value are deleted. In Calculation part
(e), a parent chromosome is chosen at random
290
under crossover probability, a crossover position

is decided at random, and genes are exchanged.
In Calculation part (c), (d), (e), the calculation of
the fitness value, the operation of arranging chromosomes in order of fitness value, and exchange
of the gene by crossover is done. These operations

are calculated and repeated until we reach the
final generation number. In the last generations
chromosomes, an individual with the highest fitness value is decided to be the optimal solution.
By decoding the chromosome code of the optimal
solution, operation of the generating equipment
in each time step of date dw can be planned.
Case Analysis
Analysis System
In this case analysis, we assume the introduction
of the microgrid into nine houses in Sapporo. In
this Section, the difference of the operation plan
when introducing the proposal method or the past
average weather data is investigated. The generating equipment introduced into the microgrid shall
be structured in five sets, and the power output
of each piece of generating equipment shall be
1kW. As described in next Section, the power load
introduced into this case analysis has a maximum
of about 7.5 kW (Figure 7).
The solar cell is a flat plate type of polycrystalline silicon, and installs the cell of area 150 m2
facing south by sloping 30 degrees. Since the area
of the solar cell installed in an average house is
20 to 40 m2, the capacity of the solar cell linked
to the proposal microgrid is equivalent to 4 to 7
houses. The battery introduced into the system is
a nickel hydrogen type. The performance and
specification were decided from the reference

(Obara & Tanno, 2007). The analysis conditions
of the solar cell, battery, each converter used in
the simulation, and an inverter are shown in Table
1. The loss of charge efficiency, discharge efficiency, and natural discharge are all included in
the table.
Analysis Conditions
1. Power Load
The time average of the electricity demand pattern of nine houses in the representation day of
the every month of Sapporo is shown in Figure
7 (Narita, 1996). Air conditioning load is not included in the electricity demand pattern of Figure
7 and it is assumed that there are four residents
per house on the average. Space heating load is
supplied with engine exhaust heat, and cooling
load is unnecessary. Although the actual electricTable 1. Specifications of equipment
Figure 7. Power demand of the nine-houses microgrid in Sapporo
291
ity demand pattern changes sharply on short time

scales, a gradually varying time-averaged value
is used throughout this Section.
2. Amount of Solar Radiation and Outdoor
Air Temperature
In this case analysis we will investigate about
six days, from the 2nd day to 7th day in each of
March, June, September, and December (the 1st
day of every month is used to check the prediction
of the PAS analysis). Figures 8 and 9 show the
measured amount of solar radiation and outside
temperature for seven days in every month from
1990 to 2004 in Sapporo. The daily fluctuations
in the amount of solar radiation are typically large
when compared to the daily fluctuations of the
outdoor air temperature in every month.
3. GA Parameters
In this case analysis, the number of an initial
generations chromosome codes is 8000. In the
genetic manipulation of multiplication and selec-
tion, the 20 chromosome codes (that is, 5% of the

population) with the largest fitness values are
made to increase their number in the next generation. The crossover probability is set to 0.2. The
generation number was set to 10 because of the
large total number of chromosome codes. These
parameters of GA were arrived at by a trial-anderror method, and the final values decided using
analysis accuracy as a reference.
Analysis Results
Prediction of Solar Cell
Output Power Via PAS
1. Relationship between Prediction Start Time
and Analysis Accuracy
Figure 10 shows the predicted amount of solar
radiation on June 6, 2007 in a south-facing set
of 30 distinct angular orientations, calculated
using PAS. In Figure 10, it will be the present
time in 5, 8, and 11, and will be each prediction
Figure 8. The slope-face amount of solar radiation in Sapporo in 1990 to 2004. 30 degrees of the angle,
for south.
292
Figure 9. Outside temperature data in Sapporo in 1990 to 2004
start time in 6, 9, and 12. Actual data refers to the

weather data measured on June 6, 2007. We see
that the difference between past average weather
data and actual data is larger than the difference
between the actual data and the predicted results.
Moreover, the data that came from using a start
time of 9 or 12 are closer to the actual data than
that which came from using a start time of 6. In
the result of Figure 10, if there is a large power
input when using the weather data measured at
that day (accordingly, the prediction start time is
late), the analysis accuracy will have improved.
In the analysis using PAS or later, prediction start
time is 6 oclock.
2. Prediction Result in Every Month
The result based on the prediction result of solar
cell output power, actual value, and past average
weather data (Homepage of Japan Meteorological
Agency, 2007; NEDO Technical information data
base, 2008) of having used PAS in representation
days every month is shown in Figure 11. One can
see from Figure 11 that in many cases one gets
closer to the actual value by using the PAS prediction than one would get by using past average
weather data.
Prediction Error of PAS, and Operation

Method of Generating Equipment
1. Operation Planning of Generating Equipment
Figure 12 shows the result of the dynamic operations analysis on March 2, 2007. In order to
investigate the influence of operation planning
on battery efficiency, we here set the battery
efficiency to 100%. The battery efficiency in
Table 1 means charge-and-discharge efficiency.
In analyses other than Figure 12, the battery efficiency shown in Table 2 is used. Accordingly,
the loss based on battery efficiency is not taken
into consideration in Figure 12. Figure 12 shows
the calculation result except charge-and-discharge
loss of the battery, in order to clarify the power
balance about consumption, the solar cell, and
the generator equipment. Figure 12 (a) shows the
result of operation planning of generating equipment using the solar cell output power prediction
generated by PAS. On the other hand, Figure
12 (b) shows the result of operation planning of
generating equipment using the actual amount of
solar radiation and outdoor air temperature. If the
PAS generated solar cell output power is the same
293
Figure 10. Solar radiation in 30-degree slope of the south sense using PAS, and actual solar radiation.
Prediction start time of PAS, and the characteristic of prediction solar radiation.
Figure 11. PAS prediction results and the actual value of the solar cell output in Sapporo
294
Figure 12. Results of the dynamic operation plan of the microgrid system (March 2, 2007)
295
Table 2. Conditions of the system components
as that coming from actual data then the optimal

operation planning method is shown in Figure
12 (a). However, one must also note that errors
are introduced in the PAS predictions and actual
operation may more closely resemble the result
of Figure 12 (c). Accordingly, the output power
of the generating equipment as controlled by PAS
and that controlled by actual weather conditions
are compared, and the system is operated in such a
way as to accommodate the worst conditions (this
is a typical way of doing things in similar systems).
2. Operation Planning of Battery
Figure 12 (d) shows the results that come from
the operation planning of a battery when using
either PAS predictions or actual data to calculate
the solar cell output power. One sees that the
operating characteristics of the battery greatly
influence the method and duration of operation of
the generating equipment. Looking at Figure 12
(d), one sees that, in at least two different places,
there is a significant difference in the battery
capacity calculated in the two different schemes.
3. Output Characteristics of the Solar Cell,
and Operation Planning
Figure 12 (e) shows the operation planning of the
generating equipment when using either past average weather data or actual weather data. Figure 12
(f) shows the resultant battery capacity planning
296
under the same conditions. Figure 12 (g) shows

the result of operation planning of generating
equipment. When Figure 12 (c) is compared with
Figure 12 (g), one sees that distinct methods of
planning for generating equipment begin to deliver noticeably divergent results after about time
15. This happens as the output power models of
the solar cell introduced into operation planning
begin to produce diverging results. Accordingly,
the accuracy of the output power prediction generated by PAS has a large influence on operation
planning of the generating equipment the capacity
planning of the battery.
Result of Dynamic Operation Planning

1. Operation Planning of Microgrid based on
the PAS and the Past Average Weather Data
Figure 13, shows the result of dynamic operation
planning of the system (including generating
equipment and batteries) when using either past
average weather data or PAS prediction data as
the solar cell output power model. The error in
the solar cell output power in Figure 13 is the
difference between the output determined by
a given simulation and that determined from
actual weather conditions. In operation of an
actual microgrid, a difference is in the solar cell
output power model described in the introduction,
and the solar cell output power operating under
Figure 13. Dynamic operation planning based on the past average weather data and the PAS predicted
data. In the analysis, the influence on operation planning by the error of the past average weather
data and the PAS predictive data to actual weather is not taking into condition.
297
actual weather conditions. As previous Section

described, in dynamic operations planning using the solar cell output power model, additional
operation of generating equipment is expected in
an unfavorable condition. For example we see, in
Figure 13 (b), that the partial output power of the
generating equipment on December 5 (calculated
in the past average weather data scheme) exceeds
5 kW, the maximum power of the proposed system. We can make sense of this as the smallest
amount of solar radiation comes in the month of
December. Because of this, the solar cell output
power grows small and the discrepancy with past
average weather data grows large. Thus, if the
error between the model output power and the
actual output power is large, the working time of
the generating equipment is expected to increase.
In comparison, for operations planning using the
PAS predictive data (see Figure 13), the generating equipment is never asked to exceed 5 kW in
the entire month of December. We see here that
operations planning by PAS is at an advantage
over planning with past average weather data.
2. Actual System Operation using the PAS

If a difference occurs in the solar cell output
power based on the PAS predictive value and
actual weather conditions, the operation of generating equipment will follow the method described below. Power balance (Equation (2)) of
the microgrid in the sampling time t is calculated, and when electricity demand exceeds supply, the generating equipment starts operation
immediately. The amount of pieces of generating
equipment sent into operation at this time is decided to be the minimum number plausible to
avoid the case where the power supply exceeds
the demand. However, it is not this limitation
when large electricity demand is predicted at
later time from the analysis of dynamic operation
planning in that case. Accordingly, when demand
is expected to exceed the maximum power supply
(5kW), we can increase the number of pieces of
equipment in operation beforehand. Figure 14
shows the operation of the generating equipments
and battery as a result of adding the modification
(the operation method described in the above)
Figure 14. Results of the dynamic operation planning of the power generator with PAS prediction. The
error of the PAS predictive data to the actual weather is taking into consideration.
298
concerning a prediction error to operation planning using PAS shown in Figure 13. Figure 14
shows operation planning of the generating equipment using PAS predictive data, as well as operation planning with adjustment of PAS prediction error (here called Adjustment value). In
actual operation of a microgrid, operation planning with PAS and with operation planning of
the generating equipment by PAS predictive data
and operation plans by adjustment value in each
sampling time are compared, while a large
power output is being performed.
3. Hours of Operation of Generating Equipment
Figure 15 shows the result of operating the generating equipment in the microgrid according to the
method described in previous Section. When the
hours of operation using past average weather data
and PAS prediction data are compared, there is few
hours worked method when introducing PAS to all
the month. The advantageous operation method
can be obtain from this rather than the operation
planning method using the past average weather
data. Moreover, Figure 16 shows the reduction in

the total equipment working hours due to using
the PAS prediction data rather than using the past
average weather data. By introducing PAS into
dynamic operation planning of the microgrid, the
working time of generating equipment is reduced,
as compared to using past average weather data,
from 3% to 30%.
Conclusions
The operations planning of the microgrid was
investigated here using a proposed algorithm.
Operation of the proposed microgrid was analyzed
using actual weather data (amount of solar radiation and outside temperature) collected from the 1st
to the 7th in the months March, June, September,
and December of 2007 in Sapporo. However, it
is thought that the following results change by
the introductory region of the microgrid. If the
analysis method described in this Section is introduced into various places, the microgrid system in
consideration of regionality will be constructed.
The following Conclusions have been obtained:
Figure 15. Result of the generator hours of operation and analyzed under the past average weather data
and the PAS predictive data. Consideration with error to the actual value is added.
299
Figure 16. Reduction rate of the power generator operation hours in the case of introducing sult of the
generator hours of operation and analyzed under the past average weather data. Consideration with
error to the actual value is added.
1. If the PAS predictive value is introduced as

the predictive value of solar cell output power
compared with the past average weather data,
the working time of the generating equipment
can be reduced from 30% to 3%. However,
working time of generating equipment is
under the influenced of battery capacity.
2. However, there is the possibility that in
seasons with only a small amount of solar
radiation, if the prediction error of PAS is
large, the original operation planning will
change greatly. In this case, a storage capacity over and above the capability of the
generating equipment introduced into the
microgrid is predicted. When introducing
the proposal analysis method into areas other
than Sapporo, it is necessary to investigate
the relation between following important
factors and system operation plan.
a. Influence of space cooling load and
space heating load.
b. Influence of a rainy season with a small
amount of solar radiation.
300
OPERATION PREDICTION
OF A BIOETHANOL SOLAR
REFORMING SYSTEM USING
A NEURAL NETWORK
Introduction
Hydrogen supplied to a fuel cell is produced from
fossil fuel, fuel alcohol, water electrolysis, etc.
The environmental impact of the fuel cell changes
with these hydrogen production methods. When
distributing small-scale fuel cells to an urban area,
installation of steam reforming of natural gas is
influential (Nogare et al., 2007; Radulescu, 2006).
Hydrogen production methods not dependent
on fossil fuel have been investigated (Schenk,
2007; Pengmei, 2007). We are investigating the
technique of bioethanol steam reforming using
daylight as a heat source (Obara & Tanno, 2007).
In the proposed system, hydrogen is produced
by bioethanol steam reforming, and a protonexchange-membrane fuel cell (PEFC) is operated. The heat of a Condensing Solar Collector is
used for evaporation and steam reforming of the

bioethanol fuel. In the utilization of bioethanol,
CO2 is hardly discharged. As renewable energy
is used as the heat source, the proposed system
is very clean equipment. However, the amounts
of hydrogen production of this system differ according to the weather, time, season, installation
location, etc. There are many examples of the
relation of meteorological data to the energy demand of buildings (Ueno, 2006; Laouadi, 2004),
In this chapter, when the weather pattern and the
energy-demand pattern are inputted, a program that
outputs the operation of the system is developed.
We have been developing an analysis program
of the dynamic operation planning of an energy
system using a genetic algorithm (referred to
as GA) (Obara & Kudo, 2005). However, a GA
cannot predict the operation of an energy system.
Moreover, the analysis time becomes very long
with the increase in design variables. The optimal
solution analyzed with a GA is used as the training
signal of the neural network (referred to as NN)
(Hastie, 2001). If the NN is made to learn based on
a training signal, an operation result similar to the
optimal solution can be predicted in a short time
under arbitrary conditions (weather condition and
energy-demand pattern). There have been many
cases of operation planning of energy systems
using a NN until now (Al-Alawi et al., 2007;
Sanino & Reischel, 2007). However, operation
prediction of a dynamic system containing green
energy has not been achieved. If this prediction
method is realized, the optimum configuration for
the area of introduction of the bioethanol solar
reforming system can be proposed. The cost of
equipment can be reduced because equipment
capacity is optimized. Moreover, if the dynamic
optimum operation of the system is calculable
at high speed, installation in the controller will

be possible. As a result, it is expected that the
results of this study will contribute to the spread
of distributed installation of fuel cells.
Fuel Cell System with Bioethanol
Solar Reforming System (FBSR)
Figure 17 is a block diagram of the fuel cell system with the bioethanol solar reforming system
(referred to as FBSR) (Obara & Tanno, 2007).
In the FBSR, two paraboloid rotating mirrors
(described as Condensing Solar Collectors) that
have a solar tracking system are used. The highdensity solar energy in Condensing Solar Collector
A is used for vaporization of the bioethanol fuel.
The solar energy in Condensing Solar Collector
B is used as the heat source to reform the fuel
steam. The type and the condensing ratio of the
Condensing Solar Collectors A and B differ. We
are also evaluating the basic property of a condensing solar collector (Sun and Ice BmbH, n. d.)
which will be reported in another study. When
setting the condensing ratio from 1:120 to 1:150
and examining in Tomakomai at a time of clear
sky in summertime, the condensing temperature
exceeds 673K. When solar energy cannot be used,
it generates electricity using the reformed gas
stored in the cylinder, or commercial power is
supplied. In order to reduce CO, a shifter and CO
oxidation equipment are installed. As reformed
gas contains a lot of steam, when compressing
and storing reformed gas, moisture is removed
by the cooler. Reformed gas is supplied to the
fuel cell and the power is converted into regular
voltage and a regular frequency with a DC-DC
301
Figure 17. Fuel cell system with bioethanol solar reforming system (FBSR)
converter and an inverter. This power is supplied

to an individual house or a power grid through
an interconnection device. The exhaust heat of
the cooler, the fuel cell, and the CO oxidization
equipment shown respectively in Figure 17 is
supplied to the heat storage tank, and this heat is
supplied to the demand side.
Installation Method of FBSR

Introducing the FBSR into a microgrid is considered. Expected benefits from the microgrid
are backup power supply in case of emergency,
reduction of environmental impact, and peak cut
off for the power plant (Obara, 2007). Moreover,
effective use of exhaust heat is possible and total
efficiency is improved compared with the conventional power generation system. However,
in the case study of this chapter, the example of
introducing FBSR into one house is investigated.
This case is the simplest system.
302
Control of Reformed Fuel

The S/C (mole ratio of steam to ethanol) of the
steady-state value of the reformed fuel supplied to
the vaporizer equipment is 3.0. Moreover, the SV
value (space velocity) in the catalyst layer in the
solar reformer unit is 3000 hour-1, the conversion
ratio of the reformed gas is 95%, the methanization
rate is 5%, and the CO generation percentage is
10% (Urasaki, U., (2005)). When there is little solar
radiation, the SV value is controlled to maintain
the reaction temperature of the reforming unit at
673K or more.
Storage of Reformed Gas

The water content in the reformed gas is removed
using the dryer. The pressure of this gas is P0 and
the quantity of flow is U 0,t . This gas is pressurized
( Pcp = 10MPa ) and stored with a compressor.
The work of compressor Wp,t is supposed to be
the work of compression of the ideal gas, and is

calculated using Equation (5). cp is the overall
using the mass balance, and the hydrogen quantity of flow (Q SR ) in the solar reforming unit
compressor efficiency. In the case study, cp is

set at 50% assuming a real machine. The power
consumed with the compressor is supplied from
the fuel cell.
outlet. C in the equation expresses the conversion ratio.
Wp,t = P0 U 0,t ln Pcp P0
cp
(5)
Equations (6) and (7) express the balance of the

power and heat, respectively. The left-hand side
of each equation expresses inputs, and the righthand side expresses outputs. IT and CT in
Equation (6) are the efficiency of the inverter and
the DC-DC converter, respectively. Moreover,
E AUX expresses the demand of auxiliary machinery. H
on the right-hand side of Equation (7)
Loss
is the heat loss of the system.
Power Balance:
(6)
Heat Balance:
H CL + H CS + H OU + H ST

+H BL = H SYS + H Loss
(8)
Steam Reforming of Bioethanol:
C2 H5OH + 3H2O 2CO2 + 6H2 173kJ/mol

(9)
Energy Balance
IT CT E CS + E CC = E SYS + E AUX
Hydrogen Balance:
Q SR + Q HC + Q EX = Q OU + Q OU Q SU

Energy and Mass Balance
(7)
Mass Balance
Equation (8) expresses the mass balance of hydrogen, and Equation (9) expresses the reaction
formula of the steam reforming of ethanol. Equation (10) expresses the relation between the
quantity of flow of bioethanol (Q BE ) calculated
Q SR = 0.2746 C Q BE
(10)
Operation Prediction
Program of the SRF
Operation Prediction Algorithm
Changing the equipment will not be easy if an
actual system is built. Therefore, optimization
considering annual operation is required. Figure
18 shows the preparation procedure of the operation prediction algorithm of the SRF developed
in this chapter. In the operation prediction using
the NN, as shown in Figure 18, it is necessary to
perform the learning process first. The training
signal used for the learning calculation is previously calculated by the GA. The training signal is
the optimal solution of the dynamic operation plan
on a representative day. Moreover, the input data
given to the NN are the past weather pattern (the
global solar radiation and the outside air temperature at each time on January 1, 2006 to December
31) and the past energy-demand pattern (average
value of every month for every time of power and
heat demand). In the learning calculation by the
NN, as described in following sections, the weights
of the network of each neuron are determined. If
303
Figure 18. Procedure of the SRF prediction algorithm
the learning calculation of the NN finishes, the

weather pattern and energy-demand pattern of
arbitrary days is given to the NN. As a result, the
collecting areas of Solar Collectors A and B, the
cylinder capacity of the reformed gas, and the
capacity of the heat storage tank are obtained.
Structure of the Neural Network

The structure of the layered neural network introduced in this chapter is shown in Figure 19. This
neural network consists of three layers, the input,
the medium, and the output. All the neurons between each layer of the NN are connected with
networks. Each neuron is outputted to output
layer So,k , j according to the magnitude of the
input. Input-output between the neurons of n 1
layer and n layer is shown in Figure 20. Input
x nj of neuron j in n layer is calculable using
n , n 1
output outkn1 and weight w j , k
of neuron k of
n 1 layer, as shown in Equation (11):

Where j = 1, , Ln and k = 1, , Ln 1 .
304
Ln 1
x nj = w j , k
n , n 1
k =1
outkn 1
(11)
Output out jn of neuron j in n layer is given

with the sigmoid function of input x nj , as shown
in Equation (12).
out jn =
1 + exp x nj
(12)
Learning Calculation
1. The Learning Method
n , n 1
All weight w j ,
in the NN shown in Figure 19
using the error-correction learning method is

determined. So, the past weather pattern and the
energy-demand pattern in a building are given to
the NN. Training signals are solutions of the
operation plan on the representative day previn , n 1
ously calculated by the GA. When w j , k
is de-
cided by learning of the NN, the error of the
Figure 19. Layered neural network of the proposed system
training signal and the data of the output layer

can be evaluated from Equation (13). This error
is expressed as ErrN , and the NN is made to learn,
n , n 1
changing weight w j ,
so that ErrN may approach
zero.
L
ErrN =
1 n
y j out jN
2 j =1
(13)
n , n 1
k
w j,k
n , n 1
= w nj ,,kn 1 + w j , k
n , n 1
w j , k
is expressed with Equa-
ErrN
n , n 1
w j , k
n , n 1
(14)
ErrN
x nj
out jn 1
(15)
tion (14) using the weight before modification

w nj ,,kn 1 and amount of modifications w j , k .
in Equation
(14) is expressed with Equation (15). The term

of the partial differential of the right-hand side of
Equation (15) is calculable by Equations (16) and
(13) (Murakami and Izumida Lab, 2007).
n , n 1
2. Modification of Weights
Corrected weight w j ,
n , n 1
Amount of modifications w j , k
At the case of n = N
305
Figure 20. Input and output of the neuron
3. Analysis Flow of the Learning Process

Figure 21 shows the analysis flow of the learning
n , n 1
process of the NN. All weights w j ,
are first
determined at random, and learning rate in

Equation (15) is given to the program. Next, input
data x nj described in previous sections and training signal y j described at the bottom are inputted
into the program, and input x nj and output out jn
of each neuron are calculated. Equation (16) is
calculated sing out jn and y j , and Equation (17)
is calculated further. These results are introduced
n , n 1
into Equation (15), and w j , k
are calculated.
When this value is given to Equation (14), the

weight of each neuron can be updated. The
analysis error is calculated using Equation (13),
and calculation will finish if this value is smaller
than the value set up previously. On the other
hand, when the analysis error is larger than the
value set up previously, as shown in Figure 21,
the process is returned and calculated repeatedly.
Preparation of the Training Signal using

a GA
ErrN
x nj
= y j x jN out jN 1 out jN
(16)
At the case of n < N

ErrN
x nj
306
Ln Err
n +1, n
n
n
N
=
w
out j 1 out j
l, j
l =1 x n +1
(17)
1. Dynamic Operation Plan of a Representative Day

A GA can investigate optimization of the nonlinear
problem of many variables. However, determination of the solution parameter (the population of
chromosomes, selection, mutation probability,
crossover probability, generation number) concerning genetic manipulation requires many trials.
Moreover, because a GA has the characteristic of
random search, the analysis time will be long when
the design variables increase in number. Therefore,
when the operation plan of the FBSR is analyzed
by a GA, a very long time to try will be required.
Figure 21. Analysis flow of the NN learning peocess
On the other hand, with the NN, although based

on the network structure of the neurons, the time
to try is short compared with GA calculation. So,
in this chapter, the training signal of the NN is
prepared using a GA and actual operation prediction is analyzed only by the NN.
2. Start Time of System Operation
Figure 22 shows the case of the power-demand
pattern in a building and the operation pattern of
the system. In this figure, solar radiation is obtained
with a Condensing Solar Collector in period Rdh
from 06:00 to 18:00. In this case, start time tst of
the system operation is 06:00. In insufficient time
zones, the system generates electricity using the
reformed gas stored during the daytime. In period
R of time tgs to time tge shown in Figure 22,
energy is supplied from the fuel cell. Although

tgs and tge can be determined arbitrarily, in the
case study in following section, it is set at 23:00
and 07:00. This time zone is set at midnight
power in many domestic electric power companies.
3. Chromosome Model and Analysis Flow
The collecting area of Solar Collector B of the
FBSR is expressed with the chromosome code
shown in Figure 23. The chromosome code is
13-bit genetic code of 0 or 1. Figure 24 shows
the analysis flow for obtaining training signals
using the GA. Many chromosome codes are
generated, and the adaptive values (objective
function) described in the following section for
all the chromosomes are calculated. Chromosome
groups with a low adaptive value are exchanged
307
Figure 22. Operation plan of a representative day
Figure 23. Chromosome code.here
for chromosome groups newly generated at random. Moreover, handling of the crossover and
the mutation are added to chromosomes and
diversity is maintained. These calculations are
308
repeated by the number of generation numbers set

up previously. In the last generation, it is decided
that the chromosome with the highest adaptive
value is the optimal solution. The collecting area
Figure 24. Analysis flow of the training signal using GA
of Solar Collector B is decided by decoding this

chromosome code. Using this result, the cylinder
capacity of the reformed gas, the capacity of the
heat storage tank, and the collecting area of Solar
Collector A are calculated. The maximum value
of the quantity of reformed gas and exhaust heat
to be stored at each time on a representative day
is decided as the capacity of the reformed gas
cylinder and the heat storage tank. The collecting
area of Solar Collector A is calculated from the
heat quantity required to reform bioethanol fuel
by Solar Collector B.
4. Objective Function and Adaptive Value
Equation (18) expresses the sum total of the difference in the power of the system and demand
on a representative day. When the cylinder capacity of the reformed gas and the capacity of the
heat storage tank are expressed as VH and Vhw ,

2
respectively, objective function fo of the system

is given in Equation (19). wo,1 , wo,2 , and wo,3 in
Equation (19) express the weight of each term. In
this chapter, it is decided that the adaptive value
in the GA is a high solution, so that the value of
fo is small. Moreover, the reformed gas cylinder
and the heat storage tank are designed to be as
small as possible. In this objective function, the
supply-and-demand balance error of power is
minimized. Furthermore, the reforming gas cylinder and the heat storage tank are made small,
and the whole system is miniaturized.
tge
Errday = E N E SYS
(18)
tgs
309
fo = wo,1 Errday + wo,2 VH + wo,3 Vhw

2
(19)
Case Study
Analysis System
Operation of the system when the FBSR shown
in Figure 17 is introduced into an individual house
is predicted. The energy demanded of the individual house used as input data introduces the
data of the reference 15). Moreover, the database
of the Japan Meteorological Agency is used concerning outside air temperature and global solar
radiation. The reference (Obara & Tanno, 2007)
has described the calculation method of the amount
of direct solar radiation obtained by the Condensing Solar Collector. Operation of the system is
taken as the pattern shown in Figure 22 where,
tgs = 08:00 and tge = 21:00. The data inputted into
the NN by learning calculation are the weather
pattern in 2006 and the energy-demand pattern
Figure 25. Cell stack power output
310
of a house (Japan Meteorological Agency, 2007).

Operation prediction of the FBSR on arbitrary
days is calculable by giving the learning NN the
weather pattern and the energy-demand pattern.
Characteristics of the System

The single-cell performance of the PEFC is shown
in Figure 25. This performance is obtained from
the experiments in the references (Mikkola, M.,
(2001), Ibaraki Prefecture Government Office of
Education, (2002)). The maximum output point of
the single cell is decided as 100% of a load factor in
this chapter. Figure 26 shows the relation between
the load factor and the power density prepared
from single-cell performance. The characteristics
shown in Figure 26 are divided into three areas, and
approximate expressions are shown in this figure.
When the production of electricity of the PEFC
is decided, the generation efficiency, the power
density, and the amount of exhaust heat will be
decided from Figures 25 and 26. The conditions
of the system components used in analysis are
shown in Table 3.
Figure 26. Relation between load factor and power output of the cell stack
Table 3. Parameters of the GA
Analysis Condition
The parameters introduced into the analysis of
the GA and the NN are shown in Tables 3 and
4, respectively. These values were determined
by applying a trial-and-error method for many
analyses. When operation of the FBSR is analyzed by the NN using the conditions in Table
4, the computation time is 90 to 110 times faster
compared with the GA.
The analytical object is an individual house in

Sapporo with the weather pattern and the energydemand pattern shown in Figure 27. There is no
cooling load in the summer season, and hot water
supply, baths, and space heating load are included in the heat demand. Moreover, the load of
electric lights and household appliances is included in the power demand, and there is no large
difference in each month.
311
Table 4. Parameters of the NN
Figure 27. Meteororogical pattern and energy demand pattern in Sapporo
In the operation plan in this chapter, the priority of design variables is in order of the collecting
area of the solar collector, the cylinder capacity
of the reformed gas, and the capacity of the heat
storage tank. Then, weights wo,1 , wo,2 and wo,3 ,
of the objective function shown in Equation (19)
is set at 0.7, 0.2, and 0.1, respectively.
the collecting area of Solar Collectors A and B, the

cylinder capacity of the reformed gas (expressed
with hydrogen quantity), and the capacity of the
heat storage tank, respectively. The rate of energy
supplied every month from the FBSR to the powerand heat-demand amount of a representative day
is shown in Figure 28 (c).
Results of Operation Prediction
1. Difference for Every Year

In order to examine the difference for every year,
the collecting areas of the solar collector in January,
November, and December in 2004 to 2006 were
analyzed. Since January, November, and December have small solar radiation, their difference in
the collecting area of the solar collector is larger
Preparation of Training Signal

Figure 28 shows the result of the operation plan of
the FBSR by the GA that introduced the weather
pattern in 2006. Figures 28 (a) and 28 (b) express
312
Figure 28. Analysis results of the operation plan using GA
than in other months. These results are shown in

Figure 28 (a). The difference of every month of
these years was 12% at maximum.
tions (18) and (19)) of the FBSR. Therefore, the

power supply rate from the FBSR is high compared
to demand.
2. The Cylinder Capacity of the Reformed

Gas and Capacity of the Heat Storage Tank
As shown in Figure 28 (b), the change of the
cylinder capacity of the reformed gas is less than
12%, which is small compared with the heat storage
capacity. The reason is the objective function of
the system described in the following section. On
the other hand, because the heat-demand amount
changes greatly with the season, the change in the
heating storage capacity every month is large.
Operation Prediction by the NN
3. Amount of Energy Supplied from the FBSR

As shown in Figure 28 (c), the power supplied
from the FBSR is 67% to 76% of the demand.
However, the heat supplied from the FBSR is 3%
to 47%, differing greatly according to the season.
Supplying energy for the power demand from tgs
to tge is included in the objective function (Equa-
1. Highest Analytic Accuracy of the NN

Operation of the system is predicted by the learning
NN using the same input data introduced by the
learning process of the NN. These analysis results
have the highest accuracy in the proposed NN.
Figure 29 shows the analysis result in this case.
Except for the amount of hydrogen storage, there
is less than 1% error. The largest error in hydrogen
storage is 1.7%. From this result, with the difference of input data and the data of the learning
process, system operation can be predicted with
an accuracy of less than several percent.
2. Analytic Accuracy of Operation Prediction
Figure 30 shows the analysis error when adding
5%, 10%, and 20% of fluctuation at random
to the power-demand pattern introduced into
313
Figure 29. Calculation error of Neural network to GA
Figure 30. Analysis results of the neural network in the case of with power change
314
learning of the NN. Figure 31 shows the analysis

error when adding 10%, 30%, and 50% of
fluctuation at random to the weather pattern (the
outside air temperature and the global solar radiation) used by the learning process of the NN.
However, when loads are added to input data at
random, the pattern of the analysis error of the NN
has diversity. But the maximum value with error

converges on the constant range. When operation
of the system is predicted from the input data with
20% of load fluctuation added to the power load
of the learning process, the analysis errors of the
solar collecting area, the cylinder capacity of the
reformed gas, and the heating storage capacity
Figure 31. Analysis results of the neural network in the case of with outside temperature and solar radiation change
315
are 6%, 14%, and 11%, respectively. On the other

hand, when operation of the system is predicted
from the input data with 50% of load fluctuation
added to the outside air temperature and the global
solar radiation of the learning process, the analysis
errors of the solar collecting area, the cylinder
capacity of the reformed gas, and the heating storage capacity are 16%, 13%, and 14%, respectively.
In the operation prediction program of the FBSR
developed in this chapter, accurate analysis within
the limits described above is possible.
Conclusions
The operation prediction program of a fuel cell
system with a bioethanol solar reforming system
using a neural network was developed. In this
program, the operation plan of a building analyzed
using a genetic algorithm is introduced as a training signal. When input data (weather pattern and
energy-demand pattern) are given to the learning
program, the operation method of the system on
arbitrary days can be obtained. The operation
method obtained in analysis is the collecting
area of the solar collector, the cylinder capacity
of the reformed gas, and the capacity of the heat
storage tank. Operation prediction of individual
houses was analyzed using the program developed.
Moreover, the analytic accuracy of the operation
prediction by the proposal NN program in case
random fluctuation is added to the weather pattern
and the energy-demand pattern is shown. When
operation of the system is predicted from the
input data with 20% of load fluctuation added
to the power load of the learning process, there
were analysis errors of 14% or less. On the other
hand, when operation of the system is predicted
from the input data with 50% of load fluctuation
added to the outside air temperature and the global
solar radiation of the learning process, there were
analysis errors of 16% or less. Even if power load
fluctuation (20% or less) or weather fluctuation (50% or less, outdoor air temperature and
the amount of global solar radiation) occurs, the
316
operation analysis error of the proposal algorithm

is 16% at the maximum. The fluctuating range of
the power, outside air temperature and amount of
solar radiation was determined in consideration of
the actual design. Especially prediction of outdoor
air temperature and the amount of global solar
radiation is difficult. But the analytic accuracy
realized in this chapter is available enough to a
design and operation plan of the system.
REFERENCES
Abdelaziz, L. (2004). Development of a radiant
heating and cooling model for building energy simulation software. Building and Environment, 39(4),
421431. doi:10.1016/j.buildenv.2003.09.016
Abu-Sharkh, etal. (2006). Canmicrogrids make
a major contribution to UK energy supply? Renewable & Sustainable Energy Reviews, 10(2),
78127. doi:10.1016/j.rser.2004.09.013
Al-Alawi, A., Al-Alawi, S. M., & Islam, S. M.
(2007). Predictive control of an integrated PVdiesel water and power supply system using an artificial neural network. Renewable Energy, 32(8),
14261439. doi:10.1016/j.renene.2006.05.003
Hastie, T., Tibshireni, R., & Friedman, J. (2001).
The elements of statistical learning. Germany:
Springer. doi:10.1007/978-0-387-21606-5
(High School Active Science Project Research
Report). Ibaraki, Japan: Author.
Ismail, Y., Kemmoku, Y., Takikawa, H., &
Sakakibara, T. (2002). An operating method for
fuel savings in a stand-alone wind/diesel/battery
system. Journal of Japan Solar Energy Society,
28(2), 3138.
Japan Meteorological Agency. (2007). Homepage.
Retrieved from http://www.data.jma.go.jp/obd/
stats/etrn/index.php
Jorgensen, K. (2008). Technologies for electric,

hybrid and hydrogen vehicles: Electricity from renewable energy sources in transport. Utilities Policy, 16(2), 7279. doi:10.1016/j.jup.2007.11.005
Laouadi, A. (2004). Development of a radiant heating and cooling model for building energy simulation software. Building and Environment, 39(4),
421431. doi:10.1016/j.buildenv.2003.09.016
Murakami and Izumida Lab. (2007). Homepage.
Retrieved from http://ipr20.cs.ehime-u.ac.jp/
column/neural/index.html
Muselli, M., Notton, G., & Louche, A. (1999).
Design of hybrid-photovoltaic power generator,
with optimization of energy management. Solar
Energy, 65(3), 143157. doi:10.1016/S0038092X(98)00139-X
University, Japan.
NEDO Technical Information Data Base. (2008).
Standard meteorology and solar radiation data
(METPV-3). Retrieved from http://www.nedo.
go.jp/database/index.html
Nogare, D. D., Baggio, P., Tomasi, C., Mutri, L.,
& Canu, P. (2007). A thermodynamic analysis of
natural gas reforming processes for fuel cell application. Chemical Engineering Science, 62(1820), 54185424. doi:10.1016/j.ces.2006.12.065
Obara, S. (2006). Operating schedule of a combined energy network system with fuel cell. International Journal of Energy Research, 30(13),
10551073. doi:10.1002/er.1203
Obara, S. (2007). Dynamic operation plan of a

combined fuel cell cogeneration, solar module,
and geo-thermal heat pump system using genetic
algorithm. International Journal of Energy Research, 31(13), 12751291. doi:10.1002/er.1300
Obara, S. (2007). Improvement of power generation efficiency of an independent microgrid
composed of distributed engine generators.
Transactions of the ASME. Journal of Energy Resources Technology, 129(3), 190199.
doi:10.1115/1.2748812
Obara, S., & Kudo, K. (2005). Study on improvement of efficiency of partial load driving of
installing fuel cell network with geothermal heat
pump and water electrolysis operation. In Proc.
of the 8th Int.Energy Agency Heat Pump Conf.,
P5-1, Las Vegas (pp. 1-9).
Obara, S., & Tanno, I. (2007). Development of
distributed energy system due to bio-ethanol PEM
fuel cell with solar reforming, Part 1Evaluation
of basic performance. Trans. of the Society of Heating. Air-Conditioning and Sanitary Engineers of
Japan, 123, 2332.
Obara, S., & Tanno, I. (2007). Operation prediction of a bioethanol solar reforming system using
a neural network. Journal of Thermal Science and
Technology, 2(2), 256267. doi:10.1299/jtst.2.256
Obara, S., & Tanno, I. (2008). Fuel reduction
effect of the solar cell and diesel engine hybrid
system with a prediction algorithm of solar power
generation. Journal of Power and Energy Systems,
2(4), 11661177. doi:10.1299/jpes.2.1166
Pengmei, L., Zhenhong, Y., Longlong, M., Wu,
C., Chen, Y., & Zhu, J. (2007). Hydrogen-rich
gas production from biomass air and oxygen/
steam gasification in a downdraft gasifier. Renewable Energy, 32(13), 21732185. doi:10.1016/j.
renene.2006.11.010
317
Sanino, L. A. M., & Reischel, R. A. R. (2007).

Modeling and identification of solar energy water heating system incorporating nonlinearities.
Solar Energy, 81(5), 570580. doi:10.1016/j.
solener.2007.01.012
Ueno, T., Sano, F., Saeki, O., & Tsuji, K. (2006).

Effectiveness of an energy-consumption information system on energy savings in residential houses
based on monitored data. Applied Energy, 83(2),
166183. doi:10.1016/j.apenergy.2005.02.002
Schenk, N. J., Moll, H. C., Potting, J., & Benders, R. M. J. (2007). Wind energy, electricity,
and hydrogen in the Netherlands. Energy, 32(10),
19601971. doi:10.1016/j.energy.2007.02.002
Urasaki, K. (2005). Application of lattice oxygen in

metal oxides to catalysts for hydrogen production
(Ph.D. thesis). Waseda University, Tokyo, Japan.
Sun and Ice BmbH, Germany. (n. d.). Retrieved

from http://www.sun-and-ice.de
318
APPENDIX
Nomenclature

E : Power [W]
E N : Power demand [W]
E : Production of electricity of the system [W]
SYS
E AUX : Consumption power of auxiliary the machinery [W]

ELF : Load factor
Errday : Difference of the power generation of the system and demand in a representativeday [W]
ErrN : Error of the output data and training signal of the NN
fo : Objective function
H : Heat [W]
H : Heat loss [W]
i , j , k : The number of a neuron

Ln : The number of neurons of n layer
N , n : Layer number
outyz : The output of the neuron y of layer z
Pcp : Compression pressure of reformed gas [Pa]
P0 : Reformed gas pressure at the reforming unit outlet [Pa]

Q : Quantity of flow [g/s]
Q OU : Hydrogen quantity lost in CO oxidation unit [g/s]
q : Amount of global solar radiation [W/m2]
R : Operating period of the fuel cell [s]
Rdh : Period of the day light [s]
Loss
Sm , j ,i : The intermediate layers neuron connected to j from neuron i
So,k , j : The output layers neuron connected to k from neuron j

T : Outside air temperature [K]
tge : Operation finish time of a fuel cell
tgs : Start-up time of a fuel cell
tst : Sampling start time [oclock]
U 0,t : Reformed gas quantity of flow at the reforming unit outlet [m3/s]
VH : The amount of hydrogen storage [g]
Vhw : Exhaust heat quantity to be stored [J]
Wp : Work of the compressor [W]
w j , i : Weight of the neuron i of layer m , and the network of neuron j of layer n
n, m
319
w : Weight before modification

w : Amount of modifications of weight
wo : Weight of the component of the objective function
x yz : Input to the neuron y of layer z
y j : Training signal
Greek Symbols

C : Conversion ratio [%]
CT : Inverter efficiency [%]
IT : DC-DC converter efficiency [%]
cp : Overall compressor efficiency [%]

: Learning rate in the NN
Subscripts

320
BE: Bioethanol
BL: Boiler
CC: Commercial power
CL: Reformed gas cooler
CS: Cell stack
OU: CO oxidation unit
SR: Solar reforming unit
ST: Heat storage tank
SU: Shift unit
SYS: System
321
Chapter 10
Microgrid with Numerical

Weather Information
ABSTRACT
This chapter consists of two sections, Compound Microgrid Installation Operation Planning of a
PEFC and Photovoltaics with Prediction of Electricity Production Using GA and Numerical Weather
Information and Energy Supply Characteristics of a Combined Solar Cell and Diesel Engine System with
a Prediction Algorithm for Solar Power Generation. The optimal operation algorithm of a photovoltaics
compound microgrid is developed using numerical weather information (NWI) in the 1st section. The
relation between the NWI error characteristics and the operation results of the system is clarified. The
2nd section proposes a prediction algorithm based on a neural network to predict the electricity production from a solar cell. The operation plan for a combined photovoltaics and diesel engine generator is
examined using the NN prediction algorithm.
fuel cell microgrid with photovoltaics effectively
reduces greenhouse gas emission. A system operation optimization technique with photovoltaics
and unstable power is important. In this study, the
optimal operation algorithm of this compound
microgrid is developed using numerical weather
information (NWI) that is freely available. A GA

(genetic algorithm) was developed to minimize
system fuel consumption. Furthermore, the relation between the NWI error characteristics and
the operation results of the system was clarified.
As a result, the optimized operation algorithm
using NWI reduced the energy cost of the system.
The production of electricity from the solar cells
DOI: 10.4018/978-1-4666-5796-0.ch010
Microgrid with Numerical Weather Information
continues to attract interest as a power source for

distributed energy generation. It is important to
be able to estimate solar cell power to optimize
system energy management. This section proposes
a prediction algorithm based on a neural network
(NN) to predict the electricity production from
a solar cell. The operation plan for a combined
solar cell and diesel engine generator system is
examined using the NN prediction algorithm. Two
systems are examined in this study: one with and
one without a power storage facility. Comparisons
are presented of the results from the two systems
with respect to the actual calculations of output
power and the predicted electricity production
from the solar cell. The exhaust heat from the engine is used to supply the heat demand. A back-up
boiler is operated when the engine exhaust heat is
insufficient to meet the heat demand. Electricity
and heat are supplied to the demand side from
the proposed systems, and no external sources
are used. When the NN production-of-electricity
prediction was introduced, the engine generator
operating time was reduced by2.5% in December
and 16.7% for March and September. Moreover,
an operation plan for the combined system exhaust
heat is proposed, and the heat output characteristics
of the back-up boiler are characterized.
COMPOUND MICROGRID
INSTALLATION OPERATION
PLANNING OF A PEFC AND
PHOTOVOLTAICS WITH PREDICTION
OF ELECTRICITY PRODUCTION
USING GA AND NUMERICAL
WEATHER INFORMATION
Introduction
An energy supply system using a microgrid constitutes the optimal system for energy demand.
Therefore, its use as a clean energy supply technique is expected to spread (Obara, 2008a; Aki
322
et al., 2006; Jiayi, Chuanwen, & Rong, 2008). A

microgrid using a PEFC (proton exchange membrane fuel cell) may become the mainstream of
future distributed energy. In addition, the application of green energy to a microgrid is desired.
Accordingly, this Section examines the PEFC
and a photovoltaics compound system. Power
can be supplied to a grid from each PEFC and
photovoltaic component in this system. The hydrogen supply method to the PEFC assumes that
the steam reforms the LPG (liquefied petroleum
gas). However, the power generation output characteristics and PEFC exhaust heat with a steam
reformer are nonlinear with a load factor (Obara,
2006a). Furthermore, although the power and
exhaust heat of the proposed system are utilized
effectively, battery installation and a heat storage
tank are planned. Consequently, the operation plan
of the proposed microgrid must be optimized as
a nonlinear system considering electricity and
heat storage. Concerning operation optimization
of a nonlinear system with heat storage, we have
summarized the use of a GA (genetic algorithm)
(Obara, 2006b, 2007a, 2007b). In addition, it is
necessary to predict unstable photovoltaic electricity production for every sample time while
optimizing operation of a compound microgrid
with a PEFC and photovoltaics. Accordingly,
numerical weather information (NWI) is used to
predict photovoltaic electricity production (Online
data service, 2009; Data of Japan Meteorological
Agency, 2009). Anyone can obtain NWI in Japan
through the Internet. However, there is an error
in the photovoltaic electricity production calculated using NWI compared to using the actual
meteorological data. Consequently, the operation
plan of the system using the NWI differs from
operation under actual weather. The cause of this
difference in operation is not addressed in this
Section. Instead, the relation between the NWI
error and the operation results of the system is
clarified. It is shown that installing the operation
optimization algorithm using NWI is important
for operation of a PEFC microgrid with photovoltaics. The objective of this study is to develop
an analysis algorithm to optimize operation of a
PEFC microgrid with green energy.
System Configurations
PEFC and Photovoltaics
Compound Microgrid
Figure 1 shows a scheme of a compound microgrid
with PEFC and photovoltaics. The compound
microgrid consists of a power system and a heat
system. Here, the power system is not connected
with a commercial power system. The power from
a PEFC and a solar cell can be supplied simultaneously to a microgrid. Moreover, these power
sources can accumulate electricity using a battery.
The hydrogen (reformed gas) supplied to a PEFC
is produced from LPG (liquefied petroleum gas)
using a steam reformer.
Each piece of equipment of the power system
and the heat system is operated by a system con-
troller. The photovoltaic electricity production for

every sample time in a target day is predicted
using the NWI (the amount of solar radiation and
outdoor air temperature) obtained by the system
controller at 0:00 on the target day. Based on this
prediction result, the optimal system operation on
the target day is planned by the system controller.
The objective given to the system controller is to
minimize fuel (LPG) consumption. As Figure 1
shows, fuel is consumed by a PEFC and a boiler
in the proposed system. The optimization analysis of the operation plan in this Section considers
operation of a power system and a heat system.
The NWI used for analysis is the information
obtained at 0:00 on a target day. Therefore, the
NWI does not match actual meteorological data.
If a system is employed according to the first
optimization plan (the plan is made at 0:00 on the
target day), then depending on the magnitude of
this error, the fuel consumption may get worse.
For example, the operation hours of a PEFC and
a boiler may be extended under the actual weather conditions. Investigating the relation between
Figure 1. System scheme
323
the NWI error and system fuel consumption

evaluates the operation optimization algorithm
using NWI.
System Operation
Figure 2 shows power demand (a) on a representative day, heat demand (d), electricity production
and exhaust heat of PEFC (b), (e), the operation
model of a battery (c) and the operation model
of heat storage and the boiler (f). Predicted photovoltaic electricity production based on the
NWI obtained at 23:00 on a representative day
is shown in Figures 2 (a) and (d). Furthermore,
the photovoltaic electricity production obtained
under actual weather conditions at each time is
shown in this figure.
The relation between the load factor and

power generation efficiency of a PEFC with a
reformer, load factor and heat generation efficiency is nonlinear. Figure 3 shows the relation
of the load factor and power generation efficiency of a home fuel cell with an LPG reformer
released by Tokyo Gas Co. Ltd. (Development of
home fuel cell cogeneration system in Tokyo Gas,
2009). As shown in Figure 3 (a), power generation
efficiency also falls according to load decreases.
Therefore, the PEFC operates well under high
load. Accordingly, as shown in Figure 2 (b), it is
desirable to operate the PEFC near its maximum
efficiency point. When the PEFC is operated near
the maximum efficiency point, exhaust heat of
the PEFC is an output characteristic, shown in
Figure 2 (e). Operation of the PEFC in the periods
Figure 2. System operation planning of the compound microgrid installation of a PEFC and photovoltics
with prediction of electricity
324
Figure 3. Efficiency characteristics of the PEFC with steam reformer
from t0 to t1 and t8 to t9 in Figures 2 (b) and (e)

are the optimal operation plan based on predicted
solar power shown in Figures 2 (a) and (d). On
the other hand, operation of PEFC in the periods
from t2 to t4 , t7 to t8 and t10 to t11 in Figures 2
(b) and (e) covers the photovoltaic shortage compared to the actual solar power (the actual solar
power is smaller than the predicted solar power),
as shown in Figures 2 (a) and (d). When there is
little actual solar power compared to the predicted solar power, additional PEFC operation is
required. In this case, as shown in Figures 2 (c)
and (f), battery, heat storage tank and boiler operations change. Accordingly, the relation between
the magnitude of the difference of the predicted
solar power and the actual solar power, and the
fuel consumption of the system is investigated.
By considering this result, the influence of the
NWI error on the system operation plan can be
identified.
sentative day is calculated as shown in Figure 1.

RT in Eq (1) is the temperature coefficient, and
Analysis Method
I D ,t can be determined from the NWI. Moreover,
Power System
1. Photovoltaics
In this Section, installation of the polycrystalline
silicon solar module of area Ss is assumed. The
average production of electricity Ps ,t of the solar
module from sample time t to t + 1 on a repre-
when the temperature Tc,t of the solar cell rises,

power generation efficiency will fall. To is a
reference temperature, and s is the power generation efficiency under To . The temperature Tc,t
of the solar cell is calculated from the specific
heat of the polycrystalline silicon and the amount
of solar radiation at sampling time t . When the
intensities of direct solar and sky solar radiation
are expressed by H D ,t and H M ,t , respectively,
among the solar radiation input into the acceptance
surface, Ps ,t will be calculated by Equation (1).
Direct solar insolations and sky solar radiation
are used for power generation in a flat solar cell.
Global-solar-radiation intensity, direct solar radiation intensity, and horizontal sky solar radiation
intensity at time t (t =0.2, , 23) are expressed
with I H ,t , I D ,t and I M ,t , respectively. I H ,t and
I M ,t can be calculated using I H ,t and I D ,t . The
incidence angle to the acceptance surface of
sunlight is calculated using Equation (2). Here,
, , and show the latitude of a setting point,
the solar celestial declination, and hour angle,
respectively, while Equation (3) is a calculation
formula for the sky solar radiation component
H D ,t .
325
Ps ,t = Ss s (H D ,t + H M ,t )

R
1 (Tc,t To ) T
100
sin = cos sin sin cos cos

H D ,t = I D ,t cos
(1)
(2)
(3)
Equation (4) calculates the incidence sky solar

radiation component H M ,t of the solar cell. The
first term on the right-hand side of Equation (4)
is the air solar radiation component; the second
term is the reflective solar radiation component;
is the angle of gradient of the acceptance surface by Equation (5); and is the reflection
factor of the ground.
H M ,t = I M ,t
1 + cos
1 cos

+ I H ,t
2
2
(4)
cot = cos cot + sin cosec tan

(5)
2. Power Balance
Equation (6) is a power balance equation. Pfc,t ,
Ppv,t and Pbt ,t on the left-hand-side in the equation
are the PEFC power, photovoltaic power, and
battery power, respectively. Pneed ,t , Pbtc,t , Ploss ,t
on the right-hand side in the equation represent
power demand, the amount of battery charge, and
loss of power, respectively. Charge-and-discharge
loss of a battery is included in the power loss
Ploss ,t .
Pfc,t + Ppv,t + Pbt ,t = Pneed ,t + Pbtc,t + Ploss ,t
326
(6)
3. Heat Balance
Equation (7) is a heat balance equation. H fc,t ,
H bl ,t and H st ,t on the left-hand-side in the equation are the heat power of a fuel cell, a boiler, and
a heat storage tank, respectively. H need ,t , H sts ,t
and H loss ,t on the right-hand side of the equation
are heat demand, the amount of heat storage, and
the heat loss, respectively. Heat storage loss is
included in the heat loss H loss ,t on the right-handside of the equation.
H fc,t + H bl ,t + H st ,t = H need ,t + H sts ,t + H loss ,t
(7)
Optimal Analysis Using GA

1. Objective Function
If Pfc,t in Equation (6) and H bl ,t in Equation (7)
are determined, the heating value of LPG Q fuel ,t
consumed by a compound microgrid is calculable.
Here, the amount of fuel with the output power
of Pfc,t and H bl ,t is decided by the PEFC power
generation efficiency and the thermal efficiency
of the boiler. Equation (8) defines the objective
function in this study. The objective function
minimizes the system fuel consumption Qsystem ,day
on one day. The fuel consumption Q fuel ,t of the
system from sample time t to t + 1 is the sum
of the fuel consumption Q fc,t of a fuel cell, and
the fuel consumption Qbl ,t of a boiler.
23
23
t =0
t =0
Qsystem ,day = Q fuel ,t = (Q fc,t + Qbl ,t )
(8)
2. Optimal Operation Planning Algorithm

In this study, the optimal operation plan of the
proposal compound microgrid is analyzed using
a GA. Figure 4 shows the operation optimization algorithm developed in this Section, and the
analysis flow is explained below.
1. The energy demand pattern data, equipment
specifications, GA parameters, numerical
weather data, efficiencies, initial conditions
and system loss are used as input into a
computer (system controller) in Calculation
(A) in Figure 4.
2. In Calculation (B), many initial generation
chromosome models are generated at random. One individual of the chromosome
model expresses PEFC operation and power.
The PEFC operation is represented with a 1
bit binary number and the PEFC power is
represented by a 14 bit binary number.
3. In Calculation (F), the PEFC power is
determined by decoding the chromosome
model. Furthermore, in Calculation (G), the
production of electricity of photovoltaics is
calculated using NWI.
4. In Calculations (H) through (K), battery,
heat storage tank, and boiler operations are
planned based on the power balance and
heat balance equations.
5. The fuel consumption is calculable from the
amount of PEFC and boiler power. In
Calculation (L), these values are totaled, and
the fuel consumption of the system in the
sampling time t is determined.
6. Calculations (E) through (M) are repeated
from sampling time 0 to 23 for one chromosome model. In Calculation (N), the adaptive
value (namely, the objective function shown
in Equation (8)) of the chromosome model
is obtained from this result.
7. The adaptive value of all the chromosome
models is decided by repeating Calculations
(D) through (O). The ranking of the chromosome models is decided according to
the magnitude of the adaptive value of each
chromosome (Calculation (P)).
Figure 4. Optimal operation algorithm
8. The chromosome models with low adaptive

value are selected, and they are exchanged
for the new randomly generated model.
Moreover, the genetic manipulation of
crossover and mutation is added based on
the probability given in Calculation (A)
concerning the chromosome models with
high adaptive value (Calculation (Q)).
9. Calculations (D) through (Q) (repeated
calculation of Calculations (C) to (R)) are
327
repeated for a defined number of generations.

In the last generations chromosome group,
the solution with the highest adaptive value
is chosen to be the optimal system operation
plan (Calculation (S)).
Case Analysis
Equipment Specifications
The equipment specifications for the case analysis of the PEFC and photovoltaics compound
microgrid are shown in Table 1. The microgrid
assumes that the equipment is installed in Sapporo,
Japan (latitude 43.062-degree north and longitude
141.354-degree east, a cold and snowy area).
1. PEFC with a Reformer
The maximum PEFC power with a reformer is
3 kW, and this performance is shown Figure 3.
2. Photovoltaics
The maximum efficiency and photovoltaic
temperature coefficients are 16.4% and 0.4%/K,
respectively. These values are general facility values used in Japan. The solar panel is installed in
the roof, with a slope of 30-degrees facing south.
Moreover, the solar cell area is set to 60.0 m2.
The area of the general solar cell installed into
individual houses in Japan is usually 25 m2 to 40
m2 (for a solar cell with a 3 kW to 5 kW capacity).
Table 1. Specifications of equipment
3. Battery, Converter and Inverter

The self-discharge of a battery is set to 10% per
hour. The converter and inverter efficiencies are
both set to 95%.
4. Heat Storage Tank and Boiler
Heat dissipation loss of a heat storage tank is set to
5% per hour, and the boiler efficiency is set to 90%.
GA Parameters
The GA parameters in the proposed algorithm are
shown in Table 2. These values were chosen by
repeating trial and error so that the convergence
solution was as stable as possible. Since the convergence solution (analysis result) has dispersion
for every analysis, the optimal solution is obtained
by repeating the same analysis.
Energy Demand Pattern

Power and heat are supplied to three individual
houses in Sapporo, Japan using the proposed
microgrid. Figure 5 shows the power and heat
demand on a representative day every month
(Narita, 1996). There is no cooling load and heating
is included in the heat load. Therefore, the power
load pattern on a representative day of every month
does not vary significantly throughout the year.
On the other hand, the magnitude of demanded
heat varies greatly between the summer season
and winter season.
Error of the NWI

Various error characteristics of the NWI can be
considered. However, the investigated case various NWI errors were not found. Accordingly, the
Table 2. GA parameters
328
Figure 5. Energy demand model. Load patterns for three individual houses on representative days in
Sapporo, Japan.
error pattern of the following two types is installed

in this study. Figure 6 shows the error pattern of
the two types used to analyze the proposed algorithm (Figure 4). As shown in Figure 6 (a), a
linear- -error and a quadratic error are installed
as error patterns. Target day operation plans are
determined by the NWI at 0:00 on the target day.
In this Section, the error that is proportional
(errort = const1 t ) to time is defined as the
linear error. On the other hand, the error that follows (errort = const 2 t 2 ) a secondary curve
relative to time is defined as the quadratic error.
Here, the integrated values of the two error types
are set equal to each other. Therefore, const1 and
const2 were decided so that Area A and Area B
(shown in Figure 6 (a)) might become equal (in
Figure 6 (a), Area A=Area B=1.0). The common
characteristic of these error patterns is the increase
of the NWI error as time increases. Moreover, as
shown in Figures 6 (b) and (c), fluctuation errors
of 20% and 40% at random are added to the

two error types. These fluctuation errors simulate
the instability of the solar insolation data.

Operation Planning
Figure 7 shows the results of the system operation
plan optimization analysis on representative February days (winter). Figures 7 (a) and (b) are the
optimal operation plans for a power system and a
heat system, respectively. Moreover, Figure 7 (c)
shows the fuel consumption plan in this case. The
fuel consumption is the sum total of each value
of a PEFC with a reformer and a boiler. Figure 8
shows the operation results of the system at the
time of the linear error and quadratic error on
the NWI. Similarly, Figure 9 shows the results of
the system operating plan optimization analysis
329
Figure 6. Error function and error function with random error in the numerical weather information
330
Figure 7. Analysis results of the proposed microgrid operation plan in February
Figure 8. Operation planning in the case of numerical weather information with two types of error; a
power system in February.)
331
Figure 9. Analysis results of the operation planning of the proposed microgrid in August
on representative August days (summer season).

Figure 10 shows the operation result with the two
error types on the NWI.
The battery operation plans shown in Figure
7 (a) and Figure 9 (a) differ greatly for each month.
Accordingly, the amount of the maximum electricity storage in August is clearly large compared
with that in February. This is because of the difference in the photovoltaic power generation in
February and August. Moreover, when Figure 7
(b) is compared with Figure 9 (b), the ratios of
the PEFC exhaust heat to heat demand vary
greatly for each month. The PEFC exhaust heat
to the heat demand ratio is very low in February.
As a result, heat supply on February representative
days is mainly boiler heat. The summer season
has little system fuel consumption based on the
difference in the heat power of a boiler (Figure 7
(c) and 9 (c)). Therefore, if the proposed compound
332
microgrid is optimized based on the objective

function in Equation (8), power should be generally optimized in the summer and heat should be
generally optimized in the winter.
Influence of the Numerical

Weather Information Error
The relation between the NWI error and the power
system operation results is investigated (Figure
4- and Figure 10). If an error is included in the
NWI, the amount of storage of electricity will
increase sharply for any month. From this result,
the time shift of power is conjectured to perform
an important role for optimizing system operation
with NWI error. Accordingly, because the operation plan is strongly influenced by battery capacity
setup, it is thought that the fuel consumption of
the system changes greatly.
Figure 10. Operation planning in the case of numerical weather information with two error types; power
system in August
When installing the linear error into the NWI,

large battery capacity is required compared with
the quadratic error. Therefore, the system operation
method changes with the error characteristics of
the NWI. To minimize the battery capacity, the
NWI quadratic error is desirable.
the fuel cell microgrid with unstable photovoltaics

achieves good operation. Even if it includes error
in the NWI, the system maintains good operation
and hardly suffers from the error.
Fuel Consumption
In this Section, the photovoltaic electricity production was predicted using numerical weather
information (NWI), and a system operation
optimization algorithm based on NWI was
proposed. The proposed algorithm uses a GA
(genetic algorithm), and optimizes the system
operation plan. However, since error exists
between NWI and meteorological data in real
time, the operation of an actual system differs
from the optimal operation plan defined beforehand. Accordingly, in this Section, the relation
between the error characteristic of the NWI and
fuel consumption of the system were clarified.
Moreover, the following was concluded. First,
when the proposal compound microgrid is installed in a cold region and optimized, power
is mainly optimized in the summer and heat is
primarily optimized in the winter. Second, for
system operation with NWI error, the power
time shift has an important role. Accordingly,
Figure 11 (a) shows the fuel consumption plan

in the case of the optimal system operation
plan. Moreover, Figures 11 (b) and (c) show the
operation results of the fuel consumption with
NWI error. The winter season (February and
December) with the large heat power of a boiler
requires significant fuel consumption. Moreover,
when Figures 11 (b) and (c) are compared with
the fuel consumption pattern for every month,
shown in Figure 11 (a), there is a clear difference.
Accordingly, the total fuel consumption on the
representative day was calculated about every
month (Figure 12). As shown in Figure 12, the
results of the fuel consumption plan in the optimal
system operation plan and the fuel consumption
when operating the system with NWI error were
small in value. From this result, it is surmised that
installing operation optimization using the NWI in
Conclusions
333
Figure 11. Analysis results of the fuel consumption
the operation plan changes greatly with the

magnitude of the battery capacity. As a result
system fuel consumption varies greatly from
month to month. Lastly, high-performance
operation can be achieved by installing the
334
operation optimization method based on the

NWI into the fuel cell microgrid with unstable
photovoltaics. Even if the error shown in this
Section is included in the NWI, the influence
on the system fuel consumption is small.
Figure 12. Analysis results of the fuel consumption on a representative day
ENERGY SUPPLY
CHARACTERISTICS OF A
COMBINED SOLAR CELL AND
DIESEL ENGINE SYSTEM WITH
A PREDICTION ALGORITHM FOR
SOLAR POWER GENERATION
Introduction
Recently, the neural network (NN) has been
proposed as a suitable statistical approach for
classification and prediction problems (Anstett &
Kreider, 1993; Kreider & Wang, (1992). A NN can
be easily used in prediction problems due to its
simplicity and adaptive pattern recognition ability.
The prediction of local weather using NN is low in
cost compared to large-scale prediction of weather
using a mainframe computer. The objectives of
this study are to develop an algorithm based on
a NN to predict the electricity production from a
solar cell and to optimize the operation plan of a
combined solar cell and diesel engine generator. A
layered NN is developed to learn and teach based
on weather data, which includes the amount of
solar radiation and the outside air temperature.
This system allows for the construction of a power
supply system with low environmental impact
that uses renewable energy. Several studies have

focused on combining a solar cell and a diesel
power plant (Muselli, Notton, & Louche, 1999;
Ashari & Nayar, 1999; Yamamoto et al., 2004;
Ismail et al., 2002). However, in these studies,
reductions in energy cost by predicting electricity production from the solar cell and power load
prediction were not investigated with respect to
the heat characteristics of the proposed systems.
The estimated power and generation capacities
of solar cells have been studied (Mondol et al.,
2007; Engin et al., 2009). The overall efficiency
performance of solar cell systems about a variety of
irradiation conditions was proposed by Mondol et
al. (2007), and the accuracy of an artificial NN for
estimating real-time maximum power generation
from solar cells is studied by Syafaruddin et al.(8).
A NN prediction algorithm (PAS) was developed
by Obara (2008b). The relationship between prediction errors of the PAS and the energy cost was
clarified, and the fuel consumption of an engine
generator was proposed. However, although the
output power from a solar cell was predicted using a NN in Ref. (9), the results were calculated
and predicted on the order of weeks and days for
the first weeks in February and August. Power
was only supplied to the demand side, and heat
335
from the proposed system was not considered. In

the present chapter, when the engine generator
operates, exhaust heat from the engine is used
to supply the demand. When the exhaust heat is
less than the demand, a back-up boiler is used
to supply the demand. Power and heat from the
proposed system are used to supply the demand.
In addition, calculations and predicted results
for solar cell electricity production as average
values for each month and the energy supply
characteristics of a solar cell and diesel engine generator are illustrated in this chapter. Two systems
are proposed in this work. The operation of the
diesel engine generator is based on the fluctuation
of the load in System 1, and a battery is not used.
Therefore, because the engine is operated over a
large area from a low to a high load, the average
engine operation efficiency is low. On the other
hand, in System 2 a battery is used to supply the
demand when solar power is less than the demand,
and the diesel engine generator operates at 25%
or less of the battery residual quantity to work in
safety Mode in the proposed system. Furthermore,
Figure 13. System scheme
336
operation of the engine generator is based on the

charge or discharge of the battery, with maximum
engine efficiency at maximum output power. Two
operating methods are used in the two systems.
In Method 1, we use actual electricity production
calculations from the solar cell, while in Method
2, NN electricity production prediction results are
used. This chapter illustrates that engine generator
operation time is shortened by introducing a NN
prediction algorithm. The analysis error of the
operation prediction is also considered. The heat
output characteristics of the diesel engine and the
back-up boiler are investigated.
Procedure
System Model
A block diagram of the combined solar cell and
diesel engine system is shown in Figure 13. As
shown in the figure, the proposed system consists
of a solar cell, a diesel engine generator, a battery,
a heat storage tank, a back-up boiler and a system
controller. The power output from the solar cell

can be supplied to the power demand through a
DC-AC converter and inverter, which also charges
the battery. The power output from the inverter
is supplied to a power grid or sold off to utilities
through an interconnection device. Table 3 shows
the specifications of the solar cell, battery, engine
and generator (SANYO Nickel-Metal Hydride
Production Information, 2009). Two operating
systems are used in this study. In System 1, a battery is not introduced into the system. When the
solar cell power is less than the power demand,
the diesel engine generator operates according
to the load fluctuation, and the surplus power
from the solar cell can be sold off. In System 2,
the surplus power from the solar cell is used to
charge a battery. The engine generator operates at
a fixed load (3 kW output power) for maximum
efficiency; the battery supplies the load when
the power output from the solar cell is less than
the demand. The battery capacity is measured
for every sampling period. If the battery capacity drops to 25% or less in the safety operation
mode, the engine generator operates and charges

the battery. The heat from the engine and boiler
is used to supply the heat demand.
Figure 14 shows the output characteristics of
a test diesel engine generator; the maximum
power output generated from the diesel engine
generator is 3 kW (Obara, 2007c). The heat exhaust
output from the engine generator includes enginecooling water and exhaust gas. The heat exhaust
is stored in a heat storage tank that supplies a
back-up boiler. The engine generator efficiency
is calculated using the approximate expression
shown in Figure 14 (b).
Examination Methods
Two methods were used to examine the two operating systems. In Method 1, the system operation
plan depends on actual calculations of the power
output from the solar cell. On the other hand,
in Method 2, the production of electricity from
the solar cell using a NN prediction algorithm
is introduced, and the system operation plan is
Table 3. System apparatus specifications
337
Figure 14. Output characteristics of the diesel engine generator at 1600 rpm
based on the NN predicted output results for the

solar cell. In this chapter, the operation plan of the
engine generator is investigated using the results
of Method 1 and Method 2.
supply and baths. The operating systems were

introduced into three apartments with the load
and heat patterns shown in Figure 15, multiplied
by three.
Energy Demand Patterns
Analysis Method
The power and heat demand patterns of a typical

household in Sapporo, Japan, are shown in Figure
15 (Obara, 2007c, 2007d). In this figure, the power
demand pattern does not change significantly each
month; this is because there is no cooling load in
the summer in Sapporo. The electricity demand
includes household appliances and electric lighting. Heat demand comes from heating, hot water
Amount of SlopeFace Solar Radiation and

Electricity Production of the Solar Cell
Direct solar radiation intensity (HD) and sky solar radiation intensity (HS) are used to calculate
the amount of slope-face solar radiation and the
electricity production from the solar cell (Obara,
2008b, 2007c, 2007d; SANYO Nickel-Metal
Hydride Production Information, 2009; Japan
Figure 15. Power load and heat load of one household in Sapporo in Japan
338
Solar Energy Society, 1985; NEDO Technical

Information Database, 2013). The formulas for
direct solar radiation and the sky solar radiation
are calculated using the following equations:
H D = I D cos
(9)
(10)
H s = 0.5 I S (1 + cos ) + 0.5 I H (1 cos )
(11)
cos = cos cot + sin cos ec tan

(12)
where ID is the direct solar radiation intensity, IH
is the global solar radiation intensity, IS is the
horizontal sky solar radiation, is the reflection
factor, is an incident angle to the acceptance
surface of the sunlight, is the latitude of the
setting point, is the solar celestial declination,
is the hour angle and is the angle of the gradient of the acceptance surface. HD and HS are
obtained from the standard weather, the solar
radiation data base in the meteorology government
office in Japan and AMEDAS (1990 to 2003,

NEDO Technical information database, (2013)).
Equations (9) to (12) are used to calculate the
slope-face solar radiation; the calculation results
are shown in Figure 16. In this figure, the average
values of slope-face solar radiation are calculated
for December, March, June and April.
The following equation is used to calculate the
average electricity production PS from the solar
module (Obara, 2008b).
Ps = SS S (H D + H S )
1 (TC TO ) (RT / 100)
(13)
where PS is the output power from the solar cell,

Tc is the temperature of the solar cell, Ss is the
area of the solar cell (72 m2), s is the generation
efficiency (14%), RT is the temperature coefficient
(0.4%/K) and To is the reference temperature (298
K). Equation (13) is used to calculate the solar
cell electricity production, the calculation results
are shown in Figure 17. As shown in this figure,
the average electricity production values from the
solar cell are calculated for December, March,
June and September.
Figure 16. Slope-face solar radiation
339
Figure 17. Solar cell electricity production
Proposed Neural Network Algorithm

The NN is suitable to predict the power output
from the solar cell because of its speed, simplicity and high prediction performance (Kil, R. M.
and Song, Y., (1997)). The prediction algorithm
of the electricity production of the solar cell uses
a layered NN, as shown in the block diagram in
Figure 18. The structure of the layered NN is
shown in Figure 19; it consists of three layers: the
input layer, the hidden layer and the output layer.
The successful implementation of a NN depends
on the training (learning) process. In the learning

process, the connection weights between layers
are determined following the total minimum error.
First, all weights are chosen randomly, and the past
weather pattern of a slope-face solar radiation and
outside air temperature are used as input signals
to the NN. The electricity production teaching
data from the solar cell are input into the output
layer. During the learning process, the learning
rate is specified as 0.1, and the sigmoid function
is utilized for the input-output characteristics of
the neurons. For each neuron, the teaching data
Figure 18. Prediction algorithm of the solar cell electricity production
340
Figure 19. The layered neural network of the prediction algorithm
is the calculation of output power from the solar

cell using Equation (13); the output of neuron j
in the n layer is given as:
O jn =
1
1 +e
I jn
(14)
patterns, amount of slope face solar radiation and

outside temperature are given to the NN, and the
learning data is the actual output power from the
solar cell. The mean squared errors (MSE) equation is described as:
Ln
The term I in Equation (14) is the input of

neuron j in layer n. It is calculated using the outn
j
n , n 1
put Okn1 and weight w j , k
of neuron k layer n-1,
as follows:
MSE N = 0.5 (t j O jN )2
(16)
j =1
where tj is the output target actual power, and O jN

is the estimated power value. The NN modifies
the weights so that the MSE approaches 0.0055%.
Weight Modification Equation (17) is used to
n , n 1
Ln 1
I jn = w j , k
k =1
n , n 1
Okn 1
(15)
calculate the corrected weighted w j , k
n , n 1
the weight before modification w j , k

n , n 1
where j =1,...., Ln, and k =1,......., Ln-1
Learning Process
First, all weights in the NN is determined randomly.
When the random initial values are input into the
proposed NN, the outputs agree with the correct
answer with high precision. The past weather
amount of modification w j , k
by using
and the
. The amount of
modification in Equation (18) is expressed in

Equation (19). The partial differential of Equation
(18) is calculated using Equations (19) and (20)
(Haykin, S., (1998))
n , n 1
w j , k
n , n 1
= w new w old = w j , k
n , n 1
w j,k

(17)
341
n , n 1
w j , k
MSE N
n , n 1
w j,k
MSE N
I jn
when n = N
MSE N
I jn
= (t j I jN ) O jN (1 O jN )
O jn 1
(18)
(19)
when n < N
Analysis Flow of the Learning Process

Figure 20 shows the proposed NN algorithm
n , n 1
learning process analysis flow. All weights w j , k
are determined randomly, and the learning rate

in Equation (18) is given to the program. The
input data I jn and teaching data tj are input into
the program. The input and output data of each
neuron are calculated. Equation (19) is calculated
using O jn and tj, as well as Equation (20). These
n , n 1
MSE N
I jn
Ln MSE
n +1, n
N
O n (1 O n )
=
w
L, j
j
j
L =1 I n +1
j
(20)
results are used in Equation (18), and w j , k
calculated. This value is given to Equation (17)

and the weight of each neuron is updated. The
Figure 20. The NN algorithm learning process calculation flow
342
is
analysis error is calculated using Equation (16).

When the analysis error is smaller than the previously defined value, the training process stops.
On the other hand, if it is larger than the threshold
value, the process is returns and calculates repeatedly.
Relation between the Input Data and

Analysis Error
In this section, the relation between input data
and analysis error is illustrated. Equation (16) is
used to calculate the analysis error, as shown in
Figure 21. In this figure, the relation between the
input data and analysis error is graphed for three
cases: in case 1, the input includes all outside
temperature data and solar radiation data that
are the average daily values; in case 2, 10% of
random fluctuation is given to solar radiation; and
in case 3, 20% random fluctuation is given to
solar radiation. As shown in Figure 21, the variation of solar radiation influences the analysis. The
biggest influence occurs in case3.

The predicted values for the electricity production
from the solar sell using NN prediction algorithm
are shown in Figure 22. Figure 23 presents the
average power generation prediction error values

from the solar cell; this figure was obtained by
subtracting Figure 22 from Figure 17. In Figure
23, the average error percentages are 25%, 29%,
19% and 26% for December, March, June and
September, respectively. These values are different because of the differences in slope-face solar
radiation and outside air temperature for each
month. Figure 25 (a) is obtained by subtracting
the power demand of three houses from the solar
cell electricity production (multiplying Figure 17
by 72 m2 to transfer values from kW/m2 to kW).
Furthermore, Figure 25 (b) is obtained by subtracting the power demand from the NN predictive
values of electricity output from the solar cell.
System 1 Results
The surplus power of the solar cell is sold as
shown in Figure 25. When the electricity output
from the solar cell is insufficient compared with
the power demand, the engine generator operates
and supplies the demand, as shown in Figures 26
and 27. The peak values of engine heat exhaust
are shown in Table 4. As shown in this table, the
peak values of heat exhaust are reduced by 3.3%,
7.5%, 1.1% and 2.7% by introducing the NN algorithm for each month. In addition, the peak value
of engine generator efficiency is 28.1 for the two
Figure 21. Relation between input data and analysis error in March
343
Figure 22. Predictive values of the solar cell

electricity
methods. The engine generator operates with low

efficiency because it operates according to load
fluctuation. When the engine heat exhaust is less
than the demand, the back-up boiler supplies the
demand side, as shown in Figure 28. In this figure,
the peak heat values from the back-up boiler in
case of Method 1 are 156 and 26 MJ at 6:00 a.m.
for September and June, respectively, and also 145
and 54 MJ at 7:00 a.m. for March and September,
respectively. Moreover, in Method 2, the peak heat
values from the back-up boiler are 15953, 26 and
Figure 23. Error in power generation prediction
Figure 24. Power results generated by subtracting the power demand from the solar cell output power
344
Figure 25. Electricity sales to utilities in System 1
Figure 26. Engine generator heat exhaust
Figure 27. Engine generator efficiency
345
Table 4. Peak values of exhaust heat from the engine generator in MJ
Figure 28. Heat from the back-up boiler in System 1
57 MJ for each month at 6:00 in the morning. In

System 1, the electricity is sold to the utilities, so
the overall system efficiency is good.
System 2 Results
A battery is used to supply the demand side when
the output power from the solar cell is lower than
the demand, as shown in Figure 29. The surplus
power from the solar cell is used to charge the
battery. The diesel engine generator operates according to the charge or discharge of the battery,
as shown in Figure 30. The engine generator
operates an average of 8, 6, 4, and 6 hours in
Method 1 and 7, 5, 4, and 5 hours in Method 2
for each month, respectively. During these engine
generator operating hours, the heat is supplied to
346
the demand side. When comparing Figure 30 (a)

with Figure 30 (b), the operating period of the
engine generator is shortened by introducing the
NN prediction algorithm. The engine operating
time is reduced by 12.5% in December and 16.7%
for March and September. The operation plan of
the back-up boiler is shown in Figure 31. In this
figure, the peak values of heat from the back-up
boiler are 8.6, 8.5, 8.3, 8.6 MJ for each month at
18:00 in Method 1. In addition, for Method 2, the
peak values are 8.6, 8.3 and 8.6 MJ for December,
June and September at 18:00, respectively, and
8.2 MJ for March at 17:00. The average total engine heat is9, 39, 26 and 38 MJ; furthermore, the
back-up boiler provides 77, 89, 97 and 90 MJ in
Method 1 for each month. Moreover, for Method
2, the average values of total engine heat is 39,
Figure 29. Battery operation plan
Figure 30. Analysis results of the engine generator operation plan
Figure 31. Heat from the back-up boiler in System 2
347
32, 26 and 32 MJ, respectively, and the average

heat from the back-up boiler is 87, 87, 96 and 95
MJ for each month.
Conclusions
A prediction algorithm using NN for electricity
production from a solar cell is developed. The
proposed system consists of two systems: System
1 and System 2. A battery is not introduced into
System 1. An engine generator is operated in
System 1 such that the solar cell power efficiency
may be covered. In this case, because the operating
range of the engine generator is wide, partial load
operation with low efficiency occurs frequently.
In addition, because the electricity is sold to utilities, the performance of System 1 is good. The
engine generator operates at a fixed load (3-kW
output power) of maximum efficiency in System
2; furthermore, the battery supplies the load with
electricity when the power from the solar cell is
insufficient compared to demand. The energy
supply characteristics of the combined solar cell
and diesel engine generator system are proposed
in the two methods for the two systems. In Method
1, the operation plan of the system depends on the
calculated solar cell electricity production results.
On the other hand, the system operation plan in
Method 2 introduces the output power results
from the NN prediction algorithm. The average
values for the prediction error of electricity production from the solar cell are 25%, 29%, 19% and
26% for December, March, June and September,
respectively. The operating period of the engine
generator is shortened by introducing the NN
prediction algorithm for the power and heat supplied to the demand side. The engine operating
time is reduced by 12.5% in December and 16.7%
348
for March and September. The diesel engine heat

characteristics are described, and a back-up boiler
operation plan is developed.
REFERENCES
Aki, H., Yamamoto, S., Kondoh, J., Maeda, T.,
Yamaguchi, H., Murata, A., & Ishii, I. (2006).
Fuel cells and energy networks of electricity, heat,
and hydrogen in residential areas. International
Journal of Hydrogen Energy, 31(8), 967980.
doi:10.1016/j.ijhydene.2005.12.016
Anstett, M., & Kreider, J. F. (1993). Application
of neural networking models to predict energy use.
ASHRAE Transaction, 99(1), 505517.
Ashari, M., & Nayar, C. V. (1999). An optimum
dispatch strategy using set points for a photovolatic (PV)-diesel-battery hybrid power system.
Solar Energy, 66(1), 19. doi:10.1016/S0038092X(99)00016-X
Data of Japan Meteorological Agency. (2009).
Retrieved from http://database.rish.kyoto-u.ac.jp/
arch/jmadata/gpv-original.html, Kyoto University.
Development of home fuel cell cogeneration
system in Tokyo Gas. (2009). ALIA News, 89.
Engin, K. S., & Hiyama, T. (2009). ANN based
real-time estimation of power generation of different PV module types. Transaction of Institute
of Electrical Engineers of Japan, B. Power and
Energy, 129(6), 783790.
Haykin, S. (1998). Neural networks: A comprehensive foundation (Prentice Hall International
Ed.). Upper Saddle River, NJ: Prentice-Hall.
Ismail, Y., Kemmoku, Y., Takikawa, H., &

Sakakibara, T. (2002). An operating method for
fuel saving in a stand-alone wing/diesel/battery
system. Journal of Japan Solar Energy Society,
28(2), 3138.
Japan meteorological business support center.
(2009). GPV/GSM (Grid point value / global
spectral model) Retrieved from http://www.jmbsc.
or.jp/hp/online/f-online0a.html
Japan Solar Energy Society. (1985). Solar Energy
Utilization Handbook. Japan: Ohmsha, Lt.
Jiayi, H., Chuanwen, J., & Rong, X. (2008).
A review on distributed energy resources and
microgrid. Renewable & Sustainable Energy
Reviews, 12(9), 24722483. doi:10.1016/j.
rser.2007.06.004
Kil, R. M., & Song, Y. (1997). Random Search
based on genetic operators. Lecture Notes in Artificial Intelligence., 1285, 196205.
Kreider, J. F., & Wang, X. A. (1992). Improved
Artificial neural networks for commercial building energy use prediction. Solar Engineering, 1,
361366.
Mondol, J. D., Yohanis, Y. G., & Norton, B.
(2007). Comparison of measured and predicted
long term performance of a gird connected
photovoltaic system. Energy Conversion and
Management, 48(4), 10651080. doi:10.1016/j.
enconman.2006.10.021
Muselli, M., Notton, G., & Louche, A. (1999).
Design of hybrid-photovoltaic power generator
with optimization of energy management. Solar
Energy, 65(3), 143157. doi:10.1016/S0038092X(98)00139-X
Narita, K. (1996). Research on unused energy of
cold region cities and utilization for district heat
and cooling (Ph.D. thesis). Hokkaido University,
Sapporo, Japan.
NEDO Technical Information Database. (2013).

Standard meteorology and Solar radiation data
(METPV-3). Retrieved from http://www.nedo.
co.jp
Obara, S. (2006a). Load response characteristics
Energy, 31(13), 18191830. doi:10.1016/j.
Obara, S. (2006b). Operating schedule of a
combined energy network system with fuel cell.
International Journal of Energy Research, 30(13),
10551073. doi:10.1002/er.1203
Obara, S. (2007a). Equipment arrangement planning of a fuel cell energy network optimized for
cost minimization. Renewable Energy, 32(3),
382406. doi:10.1016/j.renene.2006.02.012
Obara, S. (2007b). The exhaust heat use plan when
connecting solar modules to a fuel cell energy
network. Transactions of the ASME. Journal of
Energy Resources Technology, 129(1), 1828.
doi:10.1115/1.2424962
Obara, S. (2007c). Improvement of power generation efficiency of an independent microgrid
composed of distributed engine generators. ASME
Transl. Journal of Energy Resources Technology,
129(3), 190199. doi:10.1115/1.2748812
Obara, S. (2007d). Energy cost of an independent
microgrid with control of power output sharing
of a distributed engine generator. Journal of
Thermal Science and Technology, 2(1), 4253.
doi:10.1299/jtst.2.67
Obara, S. (2008a). Equipment plan of compound
interconnection microgrid composed from diesel
power plants and solid polymer membranetype fuel cell. International Journal of Hydrogen Energy, 33(1), 179188. doi:10.1016/j.
349
Obara, S. (2008b). Fuel reduction effect of the

solar cell and diesel engine hybrid system with a
prediction algorithm of solar power generation.
Journal of Power and Energy Systems, 2(4),
11661177. doi:10.1299/jpes.2.1166
SANYO nickel-metal hydride production information. (2009). Retrieved from http://www.sanyo.
co.jp
350
Yamamoto, S., Sumi, K., Nishikawa, E., &

Hashimoto, T. (2004). An operating method using prediction of photovoltaic power for a photovoltaic-diesel hybrid power generation system.
Transaction of Institute of Electrical Engineers of
Japan, B. Power and Energy, 124(4), 521530.
APPENDIX
Nomenclature

H : Heat [W]
I D : Direct solar radiation intensity [W/m2]
I H : Global-solar-radiation intensity [W/m2 ]
I M : Horizontal sky solar radiation intensity [W/m2]

H D : Intensity of direct solar [W/m2]
H M : Intensity of sky solar radiation [W/m2]

P : Power [W]
Q : Fuel consumption [W]
RT : Temperature coefficient [%/K]
Ss : Area of the solar cell [m2]

T : Temperature [K]
Tc : Temperature of the solar cell [K]
To : Reference temperature [K]
t : Sample time
Greek Symbols

: Angle of the acceptance surface gradient

: The solar celestial declination
: The latitude of a setting point
s : Photovoltaics efficiency at To [%]
: Incident angle to the sunlight acceptance surface
: Ground reflection factor
: Hour angle
Subscripts

bl : Boiler
bt : Battery
btc : Battery discharge
fc : PEFC with reformer
loss : Energy loss
need : Energy demand
pv : Photovoltaics
s : Solar module
st : Heat storage
sts : Heat storage output
351
352
Chapter 11
SOFC-PEFC Combined Microgrid
ABSTRACT
This chapter consists of two sections, Power Generation Efficiency of Photovoltaics and a SOFC-PEFC
Combined Microgrid with Time Shift Utilization of the SOFC Exhaust Heat and Power Generation
Efficiency of an SOFC-PEFC Combined System with Time Shift Utilization of SOFC Exhaust Heat.
The 1st section proposes the combined system of a solid-oxide fuel cell (SOFC) and a proton-exchange
membrane fuel cell (PEFC). The proposed system consists of a SOFC-PEFC combined system and a
photovoltaic system (PV) as the energy supplied to a microgrid. The 2nd section studies a microgrid by
introducing a combined SOFC and PEFC. This study investigates the operation of a SOFC-PEFC combined
system, with time shift operation of reformed gas, into a microgrid with 30 houses in Sapporo, Japan.
The summary of the 1st section is as follows. In this
section, the combined system of a solid-oxide fuel
cell (SOFC) and a proton-exchange membrane fuel
cell (PEFC) is developed. The proposed system
consists of a SOFC-PEFC combined system and
a photovoltaic system (PV) as the energy supply
to a microgrid. The exhaust heat of the SOFC is
used for the steam reforming of the bio-ethanol
gas with time shift utilization of the exhaust heat
of the SOFC in optional time. The SOFC-PEFC
combined system with the PV was introduced in
a microgrid of 30 residences in Sapporo, Japan.
The operation plan of the system has three cases:

without solar power, with 50% and with 100%
of solar output power. Moreover, three types of
system operation of using the SOFC independent
operation, PEFC independent operation and
SOFC-PEFC combined system are used to supply
the demand side. A comparative study between the
types of system operation is presented. The power
generation efficiency is investigated for different
load patterns: average load pattern, compressed
load pattern and extended load pattern. This study
reported that the power generation efficiencies of
the proposedsystem in consideration of these load
patterns are 27% to 48%.
DOI: 10.4018/978-1-4666-5796-0.ch011

A microgrid, with little environmental impact,
is developed by introducing a combined SOFC
(solid oxide fuel cell) and PEFC (proton exchange
membrane fuel cell) system. Although the SOFC
requires a higher operation temperature compared
to the PEFC, the power generation efficiency of
the SOFC is higher. However, if high temperature
exhaust heat may be used effectively, a system with
higher total power generation efficiency can be
built. Therefore, this study investigates the operation of a SOFC-PEFC combined system, with time
shift operation of reformed gas, into a microgrid
with 30 houses in Sapporo, Japan. The SOFC is
designed to correspond to base load operation, and
the exhaust heat of the SOFC is used for production of reformed gas. This reformed gas is used
for the production of electricity for the PEFC,
corresponding to fluctuation load of the next day.
Accordingly, the reformed gas is used with a time
shift operation. In this study, the relation between
operation method, power generation efficiency,
and amount of heat storage of the SOFC-PEFC
combined system to the difference in power load
pattern was investigated. The average power generation efficiency of the system can be maintained
at nearly 48% on a representative day in February
(winter season) and August (summer season).
POWER GENERATION EFFICIENCY

OF PHOTOVOLTAICS AND A SOFCPEFC COMBINED MICROGRID
WITH TIME SHIFT UTILIZATION
OF THE SOFC EXHAUST HEAT
Introduction
The solid-oxide fuel cell (SOFC) is recognized
as the most efficient fuel cell among various fuel
cells. The SOFC is considered an energy conversion device due to its several potential benefits,
including low pollutant emissions, high-energy efficiency, the possibility of using different kinds of
fuels and the possibility of building hybrid systems

(Naim & Maria, 2004; Muhsin & Zehra, 2009;
Erdmann, 2003). As a fuel cell for residences,
proton-exchange membrane fuel cell (PEFC) and
SOFC were developed. Though higher temperatures are required with SOFC in comparison with
PEFC, the power generation efficiency is high
and the uses for exhaust heat are wide. In order to
utilize the hot exhaust heat of SOFC effectively,
the SOFC and gas turbine (GT) compound system
was developed (Lim, Song, Shin, Yang, Jung,
Vinke, & Yang, 2008; Chan, Ho, & Tian, 2003;
Akkaya, Sahin, & Erdem, 2008; Sadegh & Ali,
2009; Haseli, Dincer, & Naterer, 2008). Although
high thermal efficiency is achieved in the SOFCGT compound system, maintenance of the high
thermal efficiency is difficult. Moreover, because
the partial load quality of GT is bad, it is difficult
to maintain the power generation efficiency of the
whole system. In this system, it is necessary to
perform synchronous operation the exhaust heat
output from the SOFC and the driving of GT.
Consideration of the load characteristic of the
demand side is indispensable to investigate the
power generation efficiency. Moreover, an operation plan with high power generation efficiency
by a high load factor is required.
In this Section, the exhaust heat of SOFC
is used for the steam reforming of bio-ethanol
fuel. The reformed gas is stored in a cylinder;
this reformed gas is supplied to the PEFC when
needed. In other words, time shift utilization of
the exhaust heat of the SOFC is installed. When
using this time shift utilization of the exhaust heat
of the SOFC, the operation of the system can be
planned so that a load peak is cut. In addition, in
this study, the SOFC-PEFC and PV combined
system are proposed to supply power and heat to
30 residences in Sapporo, Japan. The SOFC of
the proposal system is made to correspond to a
base load, and the PEFC is made to correspond
to fluctuating load. The exhaust heat of the SOFC
supplies the steam reformer and makes reformed
353
gas. This reformed gas is stored and used for the

operation of the PEFC the following day.
In this study, the optimized operation of a microgrid with the SOFC-PEFC and PV combined
system at different load patterns is clarified. A
comparative study is performed between SOFCindependent operation, PEFC-independent operation and SOFC-PEFC combined system operation.
The optimization analysis of the operation plan
considers the operation of a power system and a
heat system. Moreover, the power generation efficiency and the operation planning of the boiler
are investigated.
System Scheme
The proposed microgrid with a SOFC-PEFC and
PV combined system is shown in Figure 1. A
schematic figure of the proposed system is shown
in Figure 1(a). It consists of the SOFC-PEFC
combined system, PV system, power conditioning
system, heat supply system and system controller.
The details of each system are shown in Figures
1(b) to (e). Each piece of equipment of the proposal system is operated by the system controller
(Yoshiba, Izaki, & Watanabe, 2004). The electric
power and heat of the SOFC-PEFC combined
system and PV system are used to supply power
and heat to a grid of 30 residences in Sapporo in
Japan. The electricity production for every sample
time for each day is predicted by using NWI (the
numerical weather information) (Japan meteorological business support center, 2009; Data of
Japan Meteorological Agency, 2009) obtained by
the system controller, as shown in Figure 1(a).
The exhaust heat outputted from the SOFC is hot,
at 750 C - 900 C . This high temperature exhaust
heat is used for the steam reforming of bioethanol
and the heating of LNG (liquid natural gas) and
air supplied to the SOFC. The reformed gas is
stored in a cylinder. The power load peak of the
following day is cut by supplying the stored re-
354
formed gas to the PEFC. Such system management

is planned by the system controller. The SOFCPEFC combined system is examined by using the
different electrical load patterns of the microgrid.
SOFC and PEFC Combined System

A. Power Grid
The block diagram of the SOFC-PEFC combined
system used in the proposal system is shown in
Figure 1(b). The SOFC that is introduced into the
system has an internal reformer. As shown in this
figure, the exhaust heat of SOFC heats the LNG,
bio-ethanol gas and the air, which are supplied
to the SOFC cathode using the heat exchanger
(HEX). Supplying the exhaust heat of the SOFC
to the steam reformer (R/M) with a shift unit (S/U)
produces a reformed gas with a high hydrogen
density (Yasuda, 2005; Nakamura & Sei, 2002;
Oda, 1999). Moisture is present in the reformed
gas, so a condenser unit (C/U) for air cooling is
used. A CO oxidation device (C/O) is used to
remove CO from the reformed gas. After decreasing the CO content, the reforming gas is stored in
the cylinder by a compressor (C/P). The electric
power of the SOFC and PEFC is supplied to the
DC-DC and the DC-AC converter and inverter,
then to the power grid (Obara, 2010; Obara &
Abeer, 2009; Japan Solar Energy Society, 1985).
B. Heat Grid
The exhaust heat of the SOFC with an internal
reformer is used for heating the LNG, bio-ethanol
gas and the air supplied to the SOFC cathode. The
remaining exhaust heat is supplied to the steam
reformer (R/M). The exhaust heat of the reforming
unit, condenser unit (C/U) and PEFC is stored in
the heat storage tank. The heat exchanger is installed in the heat storage tank. Heat is exchanged
between the tap water and the heat medium in the
heat storage tank (Obara, 2010; Obara & Abeer,
2009; Japan Solar Energy Society, 1985). A boiler
is operated when there is little heat storage compared with the heat demand.
Figure 1. Proposal microgrid with photovoltaics and SOFC-PEFC combined power system
355
Photovoltaic (PV) System

The block diagram of the PV system is shown in
Figure 1(e). As shown in the figure, the proposed
PV system consists of a solar cell, DC-DC converter and DC-AC converter. The power output
from the solar cell can be supplied to the power
demand through an inverter. The type of solar cell
used is a polycrystalline silicon solar module. The
following equations are used to calculate the
electricity production Ps ,t from the solar module
(NEDO Technical information database, 2010;
Okawa, Shikazono, & Kasagi, 2007; Result report
symposium, 2007).
Ps ,t = SS S (H D ,t + H S ,t )
1 (TC ,t TO ) (RT / 100)
(1)
where Ps ,t is the output power from the solar cell,

H D ,t is the direct solar radiation intensity, H S ,t
is the sky solar radiation intensity, TC ,t is the
temperature of the solar cell, SS is the area of the
solar cell (72 m2), S is the generation efficiency
(14%), RT is the temperature coefficient (0.4%/K),
and TO is the reference temperature (298K).
Direct solar radiation and sky solar radiation are
used for power generation in a flat solar cell. The
global-solar-radiation intensity ( I H ,t ), horizontal
sky solar radiation intensity ( I S ,t ) and direct
solar radiation intensity ( I D ,t ) at time t (t =0, 1,
2, 63) can be determined from the NWI (NEDO
Technical information database, 2010; Okawa,
Shikazono & Kasagi, 2007).
H D ,t = I D ,t cos
(2)
H S ,t = 0.5 I S ,t (1 + cos )

+0.5 I H ,t (1 cos )
(4)
cos = cos cot + sin cos ec tan

(5)
where is the reflection factor, is the incident
angle to the acceptance surface of the sunlight,
is the latitude of the setting point, is the
solar celestial declination, is the hour angle,
and is the angle of the gradient of the acceptance
surface.
Operation of the SOFCPEFC Combined System

Figure 2 (a) shows the power division rate of the
SOFC-PEFC combined system that is introduced
to the demand side. As shown in this figure, the
operation plan of the SOFC corresponds to the
base load. On the other hand, the operation plan
of the PEFC is in accordance to the load fluctuation. The base load is set as a larger value than
the minimum value of the load fluctuation, so
the SOFC is operated with maximum generation
efficiency. The amount of exhaust heat from the
SOFC-PEFC combined system depends on the
operation plan of the SOFC, as shown in Figure
2 (b). In addition, the quantity of reformed gas
produced changes with the setting of the load of
the SOFC, as shown in Figure 2 (c). In Figure
2 (d), the stored reformed gas is supplied to the
PEFC, and it is used the next day.
Equations from (6) to (12) show the steam
reformation of bio-ethanol gas and the hydrogen
production process, as described before. Here, (l)
and (g) show the state of the liquid and gas, respectively.
H2O(l) H2O(l) 0.1924kJ/(mol K) (6)
(3)
C2 H5OH(l) C2 H5OH(l) 0.113kJ/(mol K)
(7)
356
Figure 2. Power division rate of the SOFC-PEFC combined system
H2O(l) H2O(g) 40.6kJ/mol
(8)
C2 H5OH(l) C2 H5OH(g) 38.6kJ/mol (9)

C2 H5OH + 3H2O 2CO2 + 6H2 173kJ/mol
(10)
C2 H5OH + 3O2 2CO2 + 3H2O + 1369kJ/mol
(11)
H2 + 1 / 2O2 3H2O + 287 kJ/mol
(12)
Partial Load Quality

Figure 3 (a) shows the power generation efficiency of the SOFC at different load factors
(Result report symposium, 2007; Takeda, et al.,
2004) for the SOFC accompanied by internal
reforming. Figure 3 (b) is the efficiency of the
PEFC at different load factors, where the PEFC
is accompanied by a steam reformer related to the
power generation efficiency (Mikkola, 2001;

Obara, 2007). When comparing Figs 3 (a) and 3
(b), the power generation efficiency of the SOFC
at a load factor of 25% or more is high in comparison with the PEFC. In addition, at a load
factor less than 25%, the power generation efficiency of the SOFC decreases greatly. Therefore,
a SOFC load factor of 25% or less is usually not
advisable. Moreover, the reformer efficiency R
in Figure 3 (b) is defined by the following equation:
R =
Heating value of reformed gas

Heating value of supply bioethanol
(13)
Operation Plan of the Proposal System

Figure 4 shows the power operation method of
the proposal system for three days ( Dayn , Dayn +1
and Dayn +2 ). The predicted output power from
357
Figure 3. Power generation efficiency of SOFC and PEFC, and reformer efficiency
the solar cell and the power demand are shown

in Figure 4 (a). In this figure, the operation plan
of the SOFC and PEFC are shown. Moreover, the
base load of the SOFC is shown. Figure 4 (b)
shows the amount of production of reformed gas
each day. After storing this reformed gas, it is

supplied to the PEFC to use the following day.
Figure 4 (c) shows the operation of the PEFC
using the reformed gas produced on the previous
day.
Figure 4. Power operation method of the proposal microgrid
358
Demand Pattern
Figures 5 (a) and (b) are the compressed model of
the width of daily power load fluctuations (compressed load pattern) and the extended model of
the width of daily load fluctuations (extended load
pattern), respectively. These patterns compress and
extend fluctuations of the average load of power
demand on a representative day to 50% and 150%,
respectively. Here the load integration value (the
amount of power demand on the representative
day) of both patterns is the same as the total power
demand under an average load. As a result, this
Section investigates the electricity demand model
of the microgrid of the three patterns, as shown
in Figure 6. This figure is an example of a representative day in February of 30 residences in
Sapporo (Narita, 1996). The electricity demand
includes appliances and electric lighting. The
thermal demand comes from heating, hot water
supply and baths. In the SOFC-PEFC combined
system, the production quantity of reformed gas
depends on the operation method of the SOFC, so
Figure 5. Power demand model of the microgrid
Figure 6. Power demand pattern of the microgrid

(Sapporo in Japan, 30 houses, February representative day)
the electric power demand pattern of the microgrid

affects the operation plan of the proposal system.
In the analysis, three load patterns, the average
load pattern, the extended load pattern and the
compressed load pattern, are investigated.
Analysis Method
The analysis procedure of the SOFC-PEFC combined system is described as follows.
1. First, the predictive values of the power load
pattern (Figure 6) and the production of
electricity of the photovoltaic power (Figure
4 (a)) are obtained for every sampling time.
However, such an analysis method of power
load estimation and output prediction of
photovoltaic power should be considered
separately. On the other hand, the production
of the electricity of the photovoltaic power is
predicted using the NWI described by previous Section. The amount of solar radiation
and the outside air temperature for every
sampling time can be obtained by the NWI.
Moreover, the output of photovoltaic power
359
2.
3.
4.
5.
6.
360
can be obtained by introducing the installation angles of the solar cell to Equations (1)
to (5).
The boundary value (i.e., capacity of the
SOFC) between the base load and the load
following is decided using the power demand
pattern obtained by (1) (Figure 2 (a)). Here,
this boundary value is decided using the
power demand pattern of a representative day
with the highest peak of the power load. In
this study, the extended power load pattern
of a representative day in February is used.
The amount of power demand and the production of electricity from photovoltaic power
generation are compared for each sampling
time. When there is a greater amount of
photovoltaic power than power demand, the
surplus power is charged to a battery. On the
other hand, the storage power is discharged
during the load peak in the evening.
The SOFC operates as shown in Figures 2
(a) to (c). On the other hand, the PEFC is
operated to respond to the fluctuating power
load. The reformed gas produced by the
exhaust heat of the SOFC on the previous
day is used for the PEFC operation.
The amount of exhaust heat of the SOFC is
obtained from Figure 3 (a) for every sampling
time. The amount of reformed gas (namely,
the amount of hydrogen) outputted from
the C/O in Figure 1 (b) is obtained from
the amount of SOFC exhaust heat using
Equations (6) to (12). The reformed gas is
stored in a cylinder and is used in the operation of the PEFC the next day. As described
in (2), the boundary value between the base
load zone and the factual load zone is decided
using the power load pattern of a representative day with the highest power load peak.
Therefore, a shortage of the storage reformed
gas used for the operation of the PEFC the
following day is not assumed.
The amount of exhaust heat from the R/M,
C/U and PEFC shown in Figure 1 (b) is ob-
tained from the heat balance. This exhaust

heat is stored in a heat storage tank. Heat
storage losses are 0.5%/hour supposing a real
system. The balance of heat is calculated for
every sampling time. When the amounts of
heat storage run short of the heat demanded,
a boiler with 90% efficiency is operated.
7. The fuel consumption (which is expressed
with the heating value) of the SOFC is
calculated in the case of the operation of
the proposal system according to (1) to (6),
R/M and a boiler. By the operation of the
proposed system based on (1) to (6), the
fuel consumption of the SOFC, R/M and the
boiler is calculated. The rate of the power
supply of the SOFC and PEFC to the heating value of the fuel supplied to the SOFC
and R/M is defined as the power generation
efficiency.

The maximum electrical load of the microgrid
appears in February, as shown in Figure 7 (a).
Therefore, the capacity of the SOFC or PEFC
will be decided with the extended load pattern in
a representative day in February. Thus, the capacity (30 kW) is decided for the operation analysis
with the independent operation of the SOFC and
PEFC. In the SOFC-PEFC combined system, the
total capacity of the SOFC and PEFC is 30 kW.
In addition, the capacities of the SOFC and PEFC
are 25.5 kW and 4.5 kW respectively.
Figure 8 shows the load factor of the SOFC
and PEFC at different load patterns (average,
compressed and extended load patterns) and at
0% (without solar power), 50% and 100% output
solar power. The amount of photovoltaic power
generated is calculated using the NWI and Equation (1) to (5). In this analysis, the average value
in Sapporo in 1990 to 2003 is used as the NWI.
The production of electricity at this time is set to
100%. In consideration of a cloudy sky etc., the
case of solar power at 50% and 0% (without solar
Figure 7. Energy demand pattern of the microgrid.

Sapporo-city in Japan, 30 houses.
Figure 8. Load factor of the SOFC and PEFF

without storage of electricity
361
power) is also analyzed. In the analysis of this

Section, 30 kW photovoltaic power generation
(145 W/m2, efficiency of the power generation is
14%) is introduced. The load factor of each fuel
cell in daytime falls by this photovoltaic power
generation.
The operation plan of only the SOFC to supply
the demand side is shown in Figures 9 to 11.
Moreover, the operation plan of only the PEFC
to supply the demand side is shown in Figures 12
to 14. As shown in Figures 9 and 12, the power
generation efficiency of the SOFC independent
operation is large compared with the PEFC independent operation. When the amount of photovoltaic power generated increases, the load of the
fuel cell falls. As a result, the power generation
efficiency of the fuel cell falls. The analysis results
for the fuel consumption of the SOFC and PEFC
are shown in Figures 10 and 13. When the amount
of photovoltaic power generated is large, the fuel
consumption of the fuel cell increases because
the partial-load operation with low fuel cell efficiency occurs. On the other hand, because the
amount of solar radiation fluctuates greatly, the
amount of photovoltaic power generated changes.
Therefore, as shown in Figures 10 and 9, the fuel
consumption of the fuel cell in the daytime has
sharp changes.
Figures 11 and 14 show the analysis results of
the boiler operation of the SOFC and PEFC independent systems, respectively. Because the heat
load is very large compared with the power load,
the difference between the SOFC independent
system and the PEFC independent system is small.
The operation plan of the SOFC-PEFC combined system is shown in Figures 15 to 18. Figure
15 shows the analysis results of the power generation efficiency of the proposed system. When
photovoltaic power generation is not introduced
(Figure 15 (a)), the change in the power generation
efficiency is small. However, the load factor of
the system falls when the amount of photovoltaic
power generated increases. Therefore, the proposed system differs in performance according
to the season and the weather.
362
Figure 9. Power generation efficiency of the SOFC
Figures 16 and 17 show the analysis results

for the consumption of the bioethanol supplied to
the reformer (R/M) and the natural gas supplied
to the SOFC, respectively. From the consumption
of natural gas shown in Figure 17, the power load
peak at 16:00 to3:00 shown in Figure 7 (a) is lev-
Figure 10. Natural gas supply of the SOFC
Figure 11. Boiler power supply of the SOFC
eled by the operation of the SOFC. On the other

hand, if the amount of photovoltaic power generated increases, the amount of daily bioethanol
supplied to the reformer (R/M) will increase, as
shown in Figure 16. Because the load factor of
the SOFC will fall if the power supply from pho-
tovoltaic power generation increases, the power

generation efficiency falls. Therefore, because
the amount of exhaust heat from the SOFC increases, the amount of bioethanol supplied to the
reformer increases. Figure 18 shows the analysis
results of the boiler operation. These results are
the same as the results of Figures 11 and 14.
Because the heat load (Figure 7 (b)) used in this
363
Figure 12. Power generation efficiency of the

PEFC
analysis is large compared with the system exhaust

heat (reformer (R/M), the condenser unit (C/U)
and the PEFC in Figure 1 (b)), the difference in
the boiler operation between each system is small.
Figure 19 shows the heat storage quantities in
case of average load patterns of a representative
day in each month. As shown in this figure, the
364
Figure 13. Bioethanol fuel supply of the PEFC
highest value of heat storage occurs in August

due to the smallest heat demand in this month
compared to the other two months.
Figure 20 shows the analysis results for the
fuel consumption in each system of a representative day every month. In the SOFC, the PEFC and
the SOFC-PEFC combined system, natural gas,
bioethanol, natural gas and bioethanol are consumed. Figure 20 (a) shows the results, not including boiler fuel. On the other hand, Figure 20 (b)
Figure 14. Boiler power supply of the PEFC
shows the results including boiler fuel. In Figure

20 (a), the fuel consumption of the SOFC-PEFC
combined system is reduced 10 to 35% compared
with the SOFC or PEFC independent system. If
boiler fuel is added to the fuel consumption of
the fuel cell, as shown in Figure 20 (b), the fuel
consumption in May and August will decrease
greatly. Because the heat load of the winter season

SOFCPEFC combined system
in a cold, snowy area is very large compared with

the power load, there are few reduction effects of
the fuel consumption of the proposed system with
boiler fuel. On the other hand, because there is
little heat load in mid-term and the summer season, the effect of the fuel reduction of the proposed
system is large.
Figure 21 shows the daily mean power generation efficiency of the SOFC-PEFC combined
system. The power generation efficiency consid-
365
Figure 16. Bio-ethanol supply of the SOFC-PEFC

combined system
crogrid, the improvement of the efficiency decrease due to the partial-load operation of the fuel
cells is important.
Conclusions
The SOFC-PEFC combined system with a time
shitting of reformed gas is proposed to supply
energy to a microgrid consisting of 30 residences
in Sapporo, Japan. Three cases are proposed for
the operation plan of the SOFC-PEFC combined
system without a PV system and with 50% and
100% solar power. The following Conclusions
are obtained:
ering all load patterns of the proposed system is

27% to 48%. Although the power generation efficiency changes according to the season and load
pattern, the photovoltaic power influences the
system efficiency most strongly. When introducing photovoltaic power into an independent mi-
366
1. When photovoltaic generation is not introduced into the SOFC-PEFC combined

system, the change in the power generation
efficiency is small. However, the load factor
of the proposal system falls when the amount
of photovoltaic power increases.
2. On the other hand, if the amount of photovoltaics power generated increases, the amount
of daily bioethanol supplied to the reformer
will increase. Because the load factor of the
SOFC decreases if the power supply due to
photovoltaic power increases, power generation efficiency decreases. Therefore, because
the amount of exhaust heat from the SOFC
increases, the amount of bioethanol supplied
to the reformer increases. In other words,
the proposed system differs in performance
according to the season and the weather.
3. The fuel consumption of the SOFC-PEFC
combined system is reduced 10 to 35%
compared with the SOFC or PEFC systems
independently. If boiler fuel is added to
the fuel consumption of the fuel cell, the
fuel consumption in May and August will
decrease greatly. Because the heat load of
the winter season in a cold, snowy area is
very large compared with the power load,
there are few reduction effects on the fuel
consumption of the proposed system.
Figure 17. Natural gas supply of the SOFC-PEFC

combined system
POWER GENERATION
EFFICIENCY OF AN SOFCPEFC COMBINED SYSTEM
WITH TIME SHIFT UTILIZATION
OF SOFC EXHAUST HEAT
Introduction
4. The power generation efficiency considering

three load patterns (average load pattern,
compressed load pattern and extended load
pattern) of the proposed system is 27% to
48%. However, because the heat load is very
large compared with the power load, the
difference between the SOFC independent
system and the PEFC independent system
is small.
PEFC (proton exchange membrane fuel cell) and

SOFC (solid oxide type fuel cell) have been developed as fuel cells for houses (Naim, Afgan, &
Maria, 2004; Gencoglu & Ural, 2009; Erdmann,
2003). The SOFC requires a higher operation
temperature compared to the PEFC. However, the
power generation efficiency is higher in SOFCs,
and high temperature exhaust heat can be widely
used. The SOFC-GT (gas turbine) combined
system was developed as an effective method to
use the high temperature exhaust heat of SOFC
(Lim, Song, Shin, Yang, Jung, Vinke, & Yang,
2008; Chan, Ho, & Tian, 2003; Akkaya, Sahin,
&Erdem, 2008; Motahar & Alemrajabi, 2009;
Haseli, Dincer, & Naterer, 2008). In this combined
system, the exhaust heat of the SOFC is used as
a heat source in the heat transfer medium of the
GT. Generally, calculations for the efficiency of
an energy system require knowledge about the
load pattern. Operation of the GT is dependent
on the amount of exhaust heat of the SOFC.
Furthermore, the exhaust heat of the SOFC and
operation of the GT synchronize. Therefore, the
load following operation of a microgrid using the
SOFC-GT combined system cannot maintain high
thermal efficiency. In the combined system with
the load following operation of the GT, and base
load operation of the SOFC, since the partialload characteristic of the GT is poor, the power
generation efficiency of the whole system is not
high. In the combined system with the base load
operation of the GT and load following operation
of the SOFC, since the exhaust heat of the SOFC
is not stabilized, the production of electricity
of the GT is restricted. Moreover, the power
367
Figure 18. Boiler power supply of the SOFC-PEFC combined system
Figure 19. Heat storage quantities in case of average load patternof the SOFC-PEFC combined system
368
Figure 20. Analysis result of the fuel consumption of a representative day
Figure 21. Total power generation efficiency on a representative day of the SOFC-PEFC combined system
generation efficiency during partial-load of the

SOFC is greatly reduced compared to full-load
operation. Consequently, in order to obtain high
power generation efficiency with the SOFC-GT
combined system, either storage-of-electricity
equipment needs to be installed or uniform system
load is required. Therefore, this chapter examines
a system which produces reformed gas with high
hydrogen concentration using the exhaust heat
of the SOFC. The system is characterized by the
ability to shift the utilization time of the produced

reformed gas. Flexible operation can be planned
because it is not necessary to synchronize the
exhaust heat output of the SOFC and utilization
of the reformed gas. In this chapter, the installation of a SOFC-PEFC combined system into a
microgrid used for supplying energy to 30 houses
in Sapporo, Japan is assumed. The SOFC is made
to correspond to the base load operation of the
microgrid, and the PEFC is made to correspond
369
to the fluctuation load in the proposal system. In

this case, reformed gas is produced by supplying
the exhaust heat of the SOFC to a steam reformer
using natural gas. This reformed gas is stored, and
the system operation method for the next day is
planned with reference to the amount of stored
reformed gas. In this study, fuel consumption of
all equipment, power generation efficiency, and
operation method of heat storage and boiler, when
operating the SOFC-PEFC combined system under different load patterns, is investigated. From
this result, the load characteristics of the system
and average power generation efficiency when
installing the proposed system into the microgrid
are elucidated.
System Scheme
Microgrid Model
Figure 22 shows a power system model of a microgrid with 30 houses. This microgrid is installed
into a residential area; power and heat are supplied
by introducing a SOFC-PEFC combined system.
The SOFC outputs high temperature exhaust heat
at 750 to 900 degrees Celsius. In this study, a
steam reformer of natural gas is operated using
Figure 22. Microgrid model of a power system
this high temperature exhaust heat. The reformed

gas is stored in a cylinder, and the stored gas can
be supplied to the PEFC at an arbitrary time. Flexible operation can be planned because the exhaust
heat of the SOFC and power generation of the
PEFC is not synchronized. On the other hand, load
with various fluctuations added to the microgrid
is expected. Accordingly, this study investigates
operation of the SOFC-PEFC combined system
using three different power load patterns.
SOFC-PEFC Combined System

1. Power System
Figure 23 is the block diagram of the SOFC-PEFC
combined system examined in this chapter. The
SOFC installed into the system assumes an internal
reforming type. Natural gas and air for supplying
the SOFC are heated by the SOFC exhaust through
a heat exchanger (HEX). The exhaust heat of the
SOFC is supplied to a steam reformer (R/M) of
natural gas, and reformed gas with high hydrogen
concentration is produced. Water contained in the
reformed gas is removed with an air-cooling condenser (C/S). After reducing the carbon monoxide
in the reformed gas to several ppm by using CO
oxidation equipment (C/O), the reformed gas is
stored in a cylinder. The reformed gas stored in
the cylinder can be supplied to the PEFC at a later
time. The power of the SOFC and PEFC is supplied to DC-DC and DC-AC converters. Finally,
after changing this power to the target voltage
and frequency through an inverter, the power is
supplied to the microgrid.
2. Heat System
First, the exhaust heat of the SOFC is used for
heating natural gas and air. The remaining exhaust
heat is supplied to the steam reformer (R/M) of
natural gas. The exhaust heat of the natural gas
steam reformer is stored in a heat storage tank.
A heat exchanger is installed in the heat storage
tank, and heat is exchanged between tap water
370
Figure 23. SOFC-PEFC combined power system
and heat medium. A boiler is operated when the

temperature of the heated tap water does not meet
demand.
Operation Method of the System

Figure 24 shows the power operation model of
the SOFC-PEFC combined system introduced
into the microgrid.
1. Model A
In Model A, shown in Figure 24 (a), Fuel cell A corresponds to base load operation of the microgrid,
and Fuel cell B corresponds to fluctuation load.
Here, the base load of Model A is set up smaller
than the minimum of the load fluctuation. In this
case, Fuel cell A is operated at maximum power
generation efficiency at all times.
2. Model B
In Model B shown in Figure 24 (b), in order to
make the power of Fuel cell A increase, the base
load by Fuel cell A is set as the minimum of load
fluctuation. Describe in the next Section, the
maximum power generation efficiency of SOFC

is high compared with PEFC. So, the SOFC is
introduced into Fuel cell A in each model in Figure 24. Because the amount of exhaust heat by
the SOFC-PEFC combined system changes with
the operating method of the SOFC, the amount
of production of reformed gas changes by load
setup of Fuel cell A.
3. Model C
In this section, the reforming gas consumed on the
next day is manufactured using the exhaust heat
of the SOFC by Fuel cell B (that is, the PEFC). In
this case, the operation method of Model C, shown
in Figure 24 (c), can be considered. In Model C,
the load of Fuel cell A increases compared with
Model B. When a part of the base load area is
included to the fluctuating load area, Fuel cell A
is accompanied by partial-load operation.
The characteristics of each model of Figure 24
are highly efficient SOFC under the base load and
PEFC with good efficiency under the partial load
corresponds to the fluctuation load. As a result,
the power generation efficiency of the system
371
Figure 24. Operation model of fuel cell systems

A and B
becomes flexible. Because the load of the SOFC of

Model C can be set up most highly, it is expected
that system overall efficiency improves greatly.
Partial-Load Characteristics
of Fuel Cells
Figure 25 (a) shows the relationship between the
load factor of the SOFC and power generation
efficiency with internal reforming (Development of a several 10kW class circular-flat-type
low-temperature operation SOFC system. 2007).
Figure 25 (b) is the relationship between the output
of the reforming gas and reformer efficiency of
steam reforming using natural gas (Yasuda, 2005).
Furthermore, Figure 25 (c) shows the relationship
between the load factor of the PEFC with the reformer, and power generation efficiency (Mathiak,
Heinzel, Roes, Kalk, Kraus, & Brandt, 2004; Oda,
1999; Dicks, Fellows, Mescal, & Seymour, 2000;
Mikkola, 2001; Alan & Donald, 2005). When
Figure 25 (a) is compared with Figure 25 (c), the
SOFC shows greater power generation efficiency
compared with the PEFC for 25% or more of the
load factor. When operating the SOFC with 25%
or less of the load factor, power generation efficiency decreases drastically. Therefore, operation
of the SOFC at 25% or less of the load factor is
not generally assumed. The power generation efficiency of the SOFC differs by nearly 21% for load
factors of 25% and 100%. On the other hand, the
power generation efficiency of the PEFC differs
by nearly 11% for load factors of 15% and 100%.
The performance of R/M, S/U, and C/O is
dependent on the reformer efficiency, shown in
Figure 25 (b). Equation (14) is the definitional
equation of reformer efficiency.
R =
increases. Furthermore, since the reforming gas

produced by the exhaust heat of the SOFC can
be used at a later time, operation of the system
372
Lower heating value of the reformed gas

100
Natural gas supplly (heating value)
(14)
According to the difference of the output of
reformed gas, reformer efficiency differs by less
than 10% (Figure 25 (b)). As for Fuel cell A,

SOFC and PEFC, and reformer efficiency
Time Shift Utilization of

the Reformed Gas
This section describes the analysis method of the
SOFC-PEFC combined system with time shift
utilization of the SOFC exhaust heat. In order to
obtain the reformed gas supplied to the PEFC,
it is necessary to increase the capacity of the
SOFC. On the other hand, to reduce equipment
costs, the installed capacity of the SOFC needs to
be decreased. Figure 26 (a) shows the pattern of
each operating method of the SOFC and PEFC.
If the SOFC is made to correspond to the base
load operation, shown in Figure 26 (a), exhaust
heat shown in Figure 26 (b) will be outputted. The
steam reformer of natural gas is operated using
this exhaust heat, and the reformed gas shown in
Figure 26 (c) is produced. This stored reformed
gas is used for the operation of the PEFC for the
next day (Figure 26 (d)).
Here, the capacity of the SOFC and PEFC is
set using the following procedure.
shown in Model C in Figure 24 (c), partial-load

operation is included. In this case, because the
amount of exhaust heat supplied to the reformer
is reduced, the output of reformed gas decreases.
However, production of reformed gas is stable
relative to reformer efficiency and reformed gas
output, shown in Figure 25 (b).
1. Randomly set the capacity of the SOFC and

PEFC at first. The value of the base load
shown in Figure 26 (a) is the same as the
capacity of the SOFC.
2. Obtain the amount of exhaust heat of the
SOFC (Figure 26 (b)) from Figure 25 (a).
3. The amount of reformed gas produced
(Figure 26 (c)) is calculated from the relationship between the reformer efficiency;
output of reformed gas shown in Figure 25
(b) and amount of exhaust heat expelled by
the SOFC (Figure 26 (b)).
4. Operation of the PEFC for the next day is
planned by using the reformed gas described
in c. The PEFC is made to correspond to the
fluctuation load operation shown in Figure
26 (a). Therefore, if the load pattern of the
next day is the same as the load pattern of a
representative day, the capacity of the SOFC
and PEFC will satisfy the power demand
(base and fluctuating loads), shown in Figure
26 (a).
373
Figure 26. Power division rate of fuel cell systems

A and B
6. Repeat calculation of b. to e. until the power

balance error of the system and reformed gas
is less than 0.1%. When the error is less than
0.1%, the capacity of the SOFC and PEFC
is considered optimized.
In the analysis of this chapter, operation is
planned so that the balance of power and heat
may be satisfied, and a detailed physical model
does not introduce.
Analysis Conditions
Load Model
5. Calculate the power balance of the system

and reformed gas from the operation plan
of the next day. When an error is found in
each balance, b. is recalculated for different
capacities of the SOFC and PEFC.
374
In this analysis example, the microgrid is installed into 30 houses in Sapporo, Japan. Figure
27 shows the power and heat demand model in
30 houses in Sapporo (Alan, & Donald, 2005).
Air conditioning is not used during the summer
season (from July to September) in Sapporo.
Moreover, the heating load during the winter
season (from November to March) is contained
in the heat load shown in Figure 27 (b). Therefore, the electricity demand model, shown in
Figure 27 (a), is mainly the load from lighting
and household appliances. In the SOFC-PEFC
combined system, the amount of reformed gas
produced changes with the operating methods of
the SOFC. For this reason, the power demand
pattern of the microgrid has a large influence on
operation plan of the system. Accordingly, this
chapter investigates two patterns based on the
power demand pattern shown in Figure 27 (a).
Figures 28 (a) and (b) are compressions of the
width of daily load fluctuations (compressed load
pattern) and extensions of the width of daily load
fluctuations (extended load pattern), respectively.
These patterns compress and extend fluctuations
of the average load of power demand on a representative day to 50%, and 150%, respectively.
Here the load integration value (the amount of
Figure 27. Energy demand pattern of the microgrid. 30 houses in Sapporo-city, Japan.
Figure 28. Power demand model of the microgrid
power demand on the representative day) of both

patterns is the same as the total power demand
under average load. As a result, this chapter investigates the electricity demand model of the
microgrid of three patterns, shown in Figure 29
(this figure is an example of a representative
day in February).
Figure 29. Power demand pattern of the microgrid
(30 houses in Sapporo, Japan, a representative
day in February)
375
Equipment Characteristic
The following equipment characteristics are used
in the analysis. The heat storage tank assumes
0.5% per hour of loss; and boiler efficiency is set
to 90%. In the analysis, power consumption of the
auxiliary machinery used to maintain operation
of the system, such as the blower, pump, and
controller, is not taken into consideration. Moreover, power consumption of the compressor used
for storage of the reformed gas is not taken into
consideration because it is small compared with
the power demand. Each setting value described
in the top is a general actual value.
Analysis Results
Operation Plan in a Representative Day
Figure 30 shows the result of the operation analysis
of the independent SOFC, independent PEFC, and
SOFC-PEFC combined systems in a representative day in February (winter season) and August
(summer season). The analysis results of the
relationship between the load factor and power
generation efficiency, fuel consumption in the
boiler and fuel cell, and heat storage plan of each
system are shown in Figure 30. The maximum
power load of the microgrid shown in Figure 27
(a) appears in February. Therefore, the capacity of
the SOFC and PEFC should be optimized under
electricity demand mode in a representative day in
February. As a result, considering the maximum
power load in a representative day in February,
the capacity of the SOFC and PEFC was set to
23 kW. The operation analysis in the individual
operation of the SOFC and PEFC also uses this
capacity (23 kW). On the other hand, the load of
the SOFC-PEFC combined system must combine
the power load and load from the production of
reformed gas. For this reason, the total capacity of
the SOFC and PEFC exceeds 23 kW. The capacity of the SOFC and PEFC with an average load
376
pattern (Figure 29), calculated from the operation

analysis described in previous section, is 17.6 kW
and 6.8 kW, respectively.
When the SOFC or PEFC is made to correspond
to load fluctuation independently (Figure 30a and
Figure 30b), there is little difference between the
fuel consumption of SOFC to PEFC. This reason
for the difference is in the load characteristic of
each fuel cell, shown in Figures 25 (a) and (c).
On the other hand, the SOFC independent system
consumes more boiler fuel in August. This is
because the power generation efficiency of the
SOFC is high, so there is little exhaust heat expelled. For the same reason, the capacity of the
heat storage tank of the SOFC independent system
is small compared with that of the PEFC independent system. Figure 30 (c) shows the analysis
result of the SOFC-PEFC combined system. The
result of the load factor and power generation
efficiency of the SOFC and PEFC is shown on
the left-hand side for every month. The load factor of the SOFC was almost constant except for
0:00 to 4:00. In other words, partial-load operation
was seen from 0:00 to 4:00. Therefore, the operation method of both months in Figure 30 (c)
is that of Model C, shown in Figure 24 (c). Since
reformed gas is produced in the SOFC-PEFC
combined system using the exhaust heat of the
SOFC, there is little remaining exhaust heat. As
a result, the capacity of the heat storage tank of
the SOFC-PEFC combined system is the smallest
of all the systems.
Influence of the Load Pattern

Figures 31 and 32 show the operation analysis
result of the SOFC-PEFC combined system
analyzed using the compressed (Figure 28) and
extended (Figure 29) load patterns. When load
patterns differ, the combination of the capacity
of SOFC and PEFC differs. The main reason for
this is that the maximum load changes with load
patterns. As a result of the operation analysis
Figure 30. Analysis results of daily operation planning
377
Figure 31. Analysis results of daily operation planning with the compressed load pattern (14.2 kW SOFC
and 4.9 kW PEFC combined system)
Figure 32. Analysis results of daily operation planning with the extended load pattern (21.8 kW SOFC
and 7.4 kW PEFC combined system)
described in previous section, the capacity of

the SOFC and PEFC is calculated to be 14.2 kW
and 4.9 kW under the compressed load pattern,
respectively. On the other hand, each capacity is
21.8 kW and 7.4 kW under the extended load pattern. As shown on the left side of Figure 31 (a),
the load factor of the SOFC in February under the
compressed load pattern is almost constant. The
378
load factor of the SOFC in a representative day in

August, shown on the left side of Figure 31 (b),
changes slightly between 0:00 and 4:00. On the
other hand, as shown on the left of Figure 32 (a)
and (b), there is a large change between 0:00 and
6:00 and a small change between 14:00 and 16:00
in the load factor of the SOFC in a representative
day in February and August under the extended
load pattern. The difference in the load pattern of

the microgrid gives the change in the load factor
of the SOFC-PEFC combined system, and influences power generation efficiency. As a result,
the fuel consumption of and amount of exhaust
heat generated by the system also change. During
extended load pattern operation, the capacity of
the heat storage tank is small, which results from
the difference in the amount of exhaust heat.
Figure 33. Fuel consumption of each system under

the average load pattern
Fuel Consumption and Power

Figure 33 shows the analysis result of fuel consumption of the independent SOFC, independent
PEFC, and SOFC-PEFC combined systems for
a representative day every month. The average
load pattern was used in this analysis. Total fuel
consumption is strongly influenced by heat load
every month, with the exception of the summer
season, since there is more heat compared to
electricity demand. On the other hand, since there
is no large monthly difference in power load, the
fuel consumption of the fuel cell shows few differences over the course of a year. The independent
PEFC system clearly consumes more total fuel
as compared with the other systems (Figure 33
(b)). This is because the average power generation
efficiency of the PEFC in the power load pattern
is low compared with the independent SOFC and
SOFC-PEFC combined systems.
Figure 34 shows the average value of the load
factor and power generation efficiency of each
system in a representative day for every month.
The average load pattern was used in this analysis.
In the SOFC-PEFC combined system, the load is
added to the SOFC by production of reformed
379
Figure 34. Load factor and efficiency of the fuel

cell under the average load pattern
Evaluation of the SOFC-PEFC

Combined System
Figure 35 shows analysis results of the average
power generation efficiency of each system in
representative day every month using three load
patterns (average, compressed and extended load
pattern). If the compressed load pattern is installed
into the SOFC-PEFC combined system, the
highest average power generation efficiency can
be obtained. On the other hand, the independent
PEFC load following operation shows the lowest
average power generation efficiency. Figure 36
shows the results of each system when using the
analysis results of the annual average of the SOFC
independent system under 100% average load. The
load factor of the SOFC-PEFC combined system
rises nearly 20% compared with other systems.
As a result, the power generation efficiency of
the system increases nearly 4% compared with
the SOFC independent system. Although the fuel
consumption in the fuel cell of the SOFC-PEFC
Figure 35. Analysis results of the average power
generation efficiency of each system
gas. By increasing the load factor in the SOFC

and using time shift utilization of the exhaust heat,
the power generation efficiency in the proposed
system is high compared with the other systems.
The average power generation efficiency of the
proposed system can be maintained at nearly 48%
every month.
380
Figure 36. Comparison of each system. The annual average result of the SOFC combined system when
using 100% average load
combined system is reduced by 11% compared

with the SOFC independent system, the boiler fuel
consumption of the proposal system increases as
a result of the amount of exhaust heat. Therefore,
the total fuel consumption of the SOFC-PEFC
combined system is only reduced by 1% compared
with the independent SOFC system. Figure 37
shows the operation result when installing each
load pattern into the SOFC-PEFC combined system. The average load pattern was installed in the
analysis of this figure. Moreover, the operation
result of the annual average was set to 100%. The
load factor of the fuel cell shows nearly 43% of
difference by the difference in the load pattern,
and power generation efficiency occurs 13% of

difference. However, when the fuel consumption
of the boiler is taken into consideration, the difference in total fuel consumption is 3%. In this
case, since there is a large space heating demand
in the microgrid, the fuel consumption reduction
of the fuel cell becomes small.
Conclusions
In this chapter, the operation plan of the SOFCPEFC combined system with time shift utilization
of the SOFC exhaust heat is investigated. The
analysis assumed installation of the proposed
Figure 37. Influence of the SOFC-PEFC combined system on the load characteristics. The annual average result of the SOFC-PEFC combined system when using 100% average load.
381
system into a microgrid which consists of 30

houses in Sapporo, Japan. The relationship between the difference in load pattern, operation
method of the SOFC-PEFC combined system,
power generation efficiency, and amount of heat
storage was examined. As a result, the following
conclusions were obtained.
The load factor of the SOFC-PEFC combined
system increases compared with other systems.
This is a result of reformed gas production using the exhaust heat of the SOFC. The reformed
gas is used for the operation of the PEFC on the
following day. The average power generation efficiency of the system can be maintained at nearly
48% (efficiency) on a representative day every
month. This value is a 6% (percentage changes)
improvement over independent SOFC operation.
Moreover, compared with the independent PEFC
operation, it is a 32% (percentage changes) improvement. However, because the load factor of
the fuel cell falls as the width of the daily power
load fluctuation increases, the average power
generation efficiency falls. When the width of the
power load fluctuation in a representative day is
50% of the average value, the difference in average power generation efficiency of the system is
nearly 13% (efficiency). As a result of reformed
gas production in the SOFC-PEFC combined
system using the SOFC exhaust heat, there is little
remaining exhaust heat. As a result, the capacity
of the heat storage tank of the proposed combined
system is small, and the fuel consumption of the
boiler increases. The total fuel consumption of
the SOFC-PEFC combined system serves as a
slight reduction compared with individual SOFC
operation.
382
REFERENCES
Akkaya, V. A., Sahin, B., & Erdem, H. H. (2008).
An analysis of SOFC/GT CHP system based on
exergetic performance criteria. International
doi:10.1016/j.ijhydene.2008.03.013
Alan, S. F., & Donald, F. R. Jr. (2005). Operating line analysis of fuel processors for PEM fuel
cell systems. International Journal of Hydrogen
Energy, 30(11), 12511257. doi:10.1016/j.
Chan, S. H., Ho, H. K., & Tian, Y. (2003).
Multi-level modeling of SOFCgas turbine hybrid system. International Journal of Hydrogen
Energy, 28(8), 889900. doi:10.1016/S03603199(02)00160-X
Data of Japan Meteorological Agency. (2009).
Recent Meteorological Data. Retrieved May 29,
2013, from http://database.rish.kyoto-u.ac.jp/arch/
jmadata/gpv-original.html
Development of a Several 10kW Class CircularFlat-Type Low-Temperature Operation SOFC System. (2007). Result report symposium 2007, New
Energy and Industrial Technology Development
Organization in Japan. Retrieved from http://www.
nedo.go.jp/informations/events/200623/26_7.pdf
In Japanese.
Dicks, L. A., Fellows, R. G., Mescal, M. C., &
Seymour, C. (2000). A study of SOFCPEM hybrid systems. Journal of Power Sources, 86(1-2),
501506. doi:10.1016/S0378-7753(99)00492-9
Erdmann, G. (2003). Future economics of the

fuel cell housing market. International Journal of
Hydrogen Energy, 28(7), 685694. doi:10.1016/
S0360-3199(02)00281-1
Gencoglu, T. M., & Ural, Z. (2009). Design of a
PEM fuel cell system for residential application.
52425248. doi:10.1016/j.ijhydene.2008.09.038
Haseli, Y., Dincer, I., & Naterer, G. F. (2008).
Thermodynamic modeling of a gas turbine cycle
combined with a solid oxide fuel cell. International
doi:10.1016/j.ijhydene.2008.05.036
Japan meteorological business support center.
(2009). Online data service, GPV/GSM (Grid
Point Value / GSM (Global Spectral Model)).
Retrieved October 22, 2012, from http://www.
jmbsc.or.jp/hp/online/f-online0a.html
Japan Solar Energy Society. (1985). Solar Energy
Utilization Handbook. Tokyo, Japan: Ohmsha,
Ltd.
Lim, T., Song, R., Shin, D., Yang, J., Jung, H.,
Vinke, I., & Yang, S. (2008). Operating characteristics of a 5 kW class anode-supported planar
SOFC stack for a fuel cell/gas turbine hybrid system. International Journal of Hydrogen Energy,
33(3), 10761083.
Mathiak, J., Heinzel, A., Roes, J., Kalk, T., Kraus,
H., & Brandt, H. (2004). Coupling of a 2.5 kW
steam reformer with a 1 kW PEM fuel cell.
Journal of Power Sources, 131(1-2), 112119.
doi:10.1016/j.jpowsour.2004.01.024
(Masters thesis) Helsinki University of Technology, Helsinki, Finland.
Motahar, S., & Alemrajabi, A. A. (2009). Exergy based performance analysis of a solid oxide
fuel cell and steam injected gas turbine hybrid
power system. International Journal of Hydrogen Energy, 34(5), 23962407. doi:10.1016/j.
Muhsin, T. G., & Zehra, U. (2009). Design of a
PEM fuel cell system for residential application.
52425248. doi:10.1016/j.ijhydene.2008.09.038
Naim, H. Afgan, & Maria G. C. (2004). Sustainability assessment of hydrogen energy systems.
13271342. doi:10.1016/j.ijhydene.2004.01.005
Nakamura, T., & Sei, M. (2002). Energy related
technology. High-efficiency fuel processor for fuel
cell system (Technical Report Vol. 77). Kadoma,
Japan: Matsushita Electric Works, Ltd.
University, Sapporo, Japan.
NEDO Technical information data base. (2010).
Standard Meteorology and Solar Radiation Data
(METPV-3). Retrieved October 13, 2001, from
http://www.nedo.go.jp/database/index.html
Obara, S. (2007). Improvement of power generation efficiency of an independent microgrid
composed of distributed engine generators.
Transactions of the ASME. Journal of Energy Resources Technology, 129(3), 190199.
doi:10.1115/1.2748812
Obara, S. (2010). Power generation efficiency of
an SOFC-PEFC combined system with time shift
utilization of SOFC exhaust heat. International
doi:10.1016/j.ijhydene.2009.11.032
383
Obara, S., & Abeer, G. E. (2009). compound microgrid installation operation planning of PEFC
and photovoltaics with prediction of electricity
production using GA and numerical weather
information. International Journal of Hydrogen Energy, 34(19), 82138222. doi:10.1016/j.
Oda, K., Sakamoto, S., Ueda, M., Fuji, A., &
Ouki, T. (1999). A small-scale reformer for fuel
Review, 31(2), 99106.
Okawa, D., Shikazono, N., & Kasagi, N. (2007).
Basic characteristics of gas turbine-solid oxide fuel
cell hybrid system with air reduction. Transactions
of the JSME, Series B, 73(730), 113-120.
Sadegh, M., & Ali, A. A. (2009). Exergy based
performance analysis of a solid oxide fuel cell and
steam injected gas turbine hybrid power system.
23962407. doi:10.1016/j.ijhydene.2008.12.065
384

start-up. In Proceedings of the 20th Energy System Economic and Environment Conference (pp.
343-344). Tokyo, Japan: Japan Society of Energy
and Resources.
Yasuda I. (2005). Development of hydrogen production technology for fuel cell. Energy Synthesis
Engineering, 28(2).
Yoshiba, F., Izaki, Y., & Watanabe, T. (2004).
Wide Range Load Controllable MCFC Cycle
with Pressure Swing Operation. Journal of
Power Sources, 137, 196205. doi:10.1016/j.
385
Chapter 12
Bioethanol Solar Reforming

System for Distributed Fuel Cell
ABSTRACT
This chapter consists of two sections, Hydrogen Production Characteristics of a Bioethanol Solar
Reforming System with Solar and Efficiency Analysis of a Combined PEFC and Bioethanol-SolarReforming System for Individual Houses. Heat transfer analysis applied in reforming the catalyst layer
of the reactor of FBSR (bioethanol steam reforming system) and the temperature distribution and transient response characteristics of the gas composition of the process are investigated in the 1st section
The overall efficiency of the production of electricity and heat power of the FBSR system is determined
by examining its thermal output characteristic in the 2nd section. It dependes for the overall efficiency
of the system on the amount of solar radiation fluctuation rather than the amount of solar radiation.
The summary of the 1st section is as follows.
The development of a bioethanol steam reforming system (FBSR) is considered as a means of
distributing energy using PEFCs. Small-scale
solar collectors (collection areas on the order of
several m2) are installed in a house as a method for
applying the FBSR. However, the temperature distribution of a reforming catalyst fluctuates under
conditions of unstable solar insulation. Therefore,

heat transfer analysis applied in reforming the
catalyst layer of the reactor and the temperature
distribution and transient response characteristics of the gas composition of the process were
investigated. As a case study, meteorological data
for representative days in March and August in
Sapporo, Japan were recorded, and the hydrogen
production speed, power generation output and
amount of electricity purchased were analyzed.
DOI: 10.4018/978-1-4666-5796-0.ch012
Bioethanol Solar Reforming System for Distributed Fuel Cell
The results showed that although fluctuations

in solar insolation affected the efficiency of the
FBSR, the average efficiency of each representative day exceeded 40%. By installing two solar
collectors, each with a collection area of 1 m2,
21% to 25% of the average power demand of an
individual house can be supplied.
In this section, the development of a FBSR using
sunlight as a heat source was investigated. The system was investigated using the experimental result
of catalyst performance, and numerical analysis.
If ethanol purity is high, the production method
of the bioethanol used for the proposal system
will not be limited. The overall efficiency of the
production of electricity and heat power of this
system was determined by examining its thermal
output characteristic. The FBSR was introduced
into standard individual houses in Sapporo, Japan
for analysis. The amount of hydrogen production,
the production-of-electricity characteristic, and
the thermal output characteristic were examined
using meteorological data on representative
days in March and August. Compared with the
representative day in March (28.0 MJ/Day), the
solar radiation of the representative day in August
(37.0 MJ/Day) is large. However, the amount of
solar radiation fluctuation of the representative
day in August in this analysis is large compared
with the representative day in March. It depends
for the overall efficiency of the system on the
amount of solar radiation fluctuation rather than
the amount of solar radiation. As a result, the
overall efficiency of the system, defined as the
rate of power and heat output compared to the
amount of solar heat collected, was calculated to
be 47.4% and 41.9% on the representative days
in March and August, respectively.
386
HYDROGEN PRODUCTION
CHARACTERISTICS OF A
BIOETHANOL SOLAR REFORMING
SYSTEM WITH SOLAR
INSOLATION FLUCTUATIONS
Introduction
The thermal efficiency in fuel cells generally depends on the supply method of hydrogen. A large
quantity of CO2 is discharged by reforming systems
that use fossil fuels as a heat source and reforming
fuel. So, there are researches on much hydrogen
production technology using solar energy currently
(for example, Obara & Tanno [2007a, 2007b,
2008]). Steam reforming of bioethanol, however,
is driven by heat from solar collectors (i.e., a fuel
cell system with a bioethanol solar reforming
system, or FBSR) (Obara & Tanno, 2007a). We
have previously proposed the production characteristics of a domestically installed reformed gas
system (Akpan, Akande, Aboudheir, Ibrahim, &
Idem, 2007), and a method for weather prediction
using a layered neural network (Obara & Tanno,
2007b, 2008). FBSR operation results in hydrogen
production, which results from supplying ethanol/
water vapour to a reforming catalyst layer. The fuel
conversion rate depends on the temperature of the
catalyst layer and the space velocity of the ethanol/
water vapour. Furthermore, the temperature of
the catalyst layer is strongly influenced by solar
radiation. If the amount of solar radiation input
to the acceptance surface of the FBSR fluctuates
sharply over a short-time period, the reforming
reaction in the catalyst layer will not fully advance
due to a response delay. Therefore, in this Section,
the unsteady heat transfer of the catalyst layer
installed in the reforming unit is analyzed, and the
temperature distribution and transient response

characteristics of the composition of process gas
are investigated. Next, the hydrogen production
characteristics of the FBSR are deduced from the
results of this analysis. Furthermore, the output
characteristics of the PEFC are examined. The
objective of this study is to clarify the characteristics of hydrogen production, efficiency of the
reforming component, and operation method for
the FBSR under conditions of short-time solar
insolation fluctuation. With this information, we
can compare the economic performances and
environmental impacts of photovoltaic and other
reforming systems. The efficiency of the reforming component is defined as the higher heating
value of hydrogen / the amount of collection of a
solar insolation per day.
Fuel Cell System with a Bioethanol

Solar Reforming System (FBSR)
System Block Diagram
Figure 1 (a) shows the power system block diagram of a fuel cell system with a bioethanol solar
reformer (FBSR) (Obara & Tanno, 2007a, 2007b,
2008). Two parabolic mirrors (solar collector) with
a solar tracking system are introduced into the
system. The solar energy obtained by solar collector A is used for vaporization of the bioethanol
solution (heat source of a vaporizer unit). The solar
energy obtained by solar collector B is used as
the heat source for the reforming reaction. In this
Section, the collecting areas of solar collectors A
and B are set to 1 m2 (unit area).
shown in Figure 1 (b). The reforming component

consists of solar collector B and the reactor, with
the catalyst layer shown in Figure 1 (c). The
reactor of the reforming component generates
the reformed gas with high concentrations of
hydrogen. A large amount of CO is contained in
this reformed gas. Consequently, reformed gas is
supplied to the shift and CO oxidation units, and
the CO concentration is subsequently reduced.
Moreover, because much of the water vapour is
contained in the reformed gas, water is removed
using the gas cooler. After completing these processes, the reformed gas is supplied to the PEFC,
and direct-current power is obtained. This power
is adjusted using a DC-DC converter, DC/AC
converter and inverter. However, when customer
demand exceeds power output from the system,
power is supplied commercially.
Reforming Component
The reactor shown in Figure 1 (c) is installed at
the focal spot of solar collector B, with diameter
Dcs and width Lcs , in the reforming component,
shown in Figure 1 (b). One of the end faces of the
reactor is the acceptance surface of the solar insolation, with area Ahs . Ethanol/water vapor is
supplied from the heat exchange surface of the
reactor. The solar insolation input to the heat
exchange surface is supplied directly to the catalyst layer in the reactor. In this Section, the fuel
vapor accompanying the reaction supplied to the
catalyst layer is described as process gas.
Heat Transfer Analysis
The Reformed Gas and Power System
Reactor Model
The molar ratio of steam to ethanol (S/C) of the

ethanol fuel supplied to the vaporizer unit from
the ethanol tank is set as 3.0. The ethanol solution is evaporated by supplying heat from solar
collector A to the vaporizer. The ethanol/water
vapor is supplied to the reforming component,
The reactor shown in Figure 1 (c) is filled with

spherical reforming catalyst several millimeters
in diameter. As shown in Figure 2 (a), all of the
walls except the solar insolation acceptance surface in the reactor are heat insulated. Moreover,
387
Figure 1. PEFC with bioethanol solar reforming (FBSR)
part of the thermal energy of the solar insolation

arrive to the heat exchange surface of the reactor
is emitted to the ambient air by convective
q con (Equation (1)) and radiative q rad (Equation
(2)) heat transfer. Solar insolation, except for the
388
aforementioned heat dissipation, is supplied to

the catalyst layer from the heat exchange surface,
and is used for reforming via heat conduction
between catalyst particles and convection of process gas.
Figure 2. Heat transfer analysis
qcon = Ahs h (Ths T )
(1)
Nu = 9.49 (Re Pr)
(2)
c
Dcl
0.516
1.43
qrad = hs Ahs Ths4 T4
Reactor
Equation (3) describes the heat convection of the
catalyst layer, and it contains the Damkohler correction number Da (Usami, Fukusako, & Yamada, 2000). The right-hand side of Equation (3)
consists of terms describing the convection and
chemical reaction of process gas. Here, Nu , Re
and Da are calculated using Equations (4) to (6),
respectively.
+ 27.2 Da 0.325
Nu =
hg Dc
Re =
ug Dc
(3)
(4)
(5)
Da = (H r ar ) Dc
ug C g Tg )
(6)
389
Reforming Reaction
Equation (7) is the reaction formula of steam
reforming of ethanol. However, the conversion rate from ethanol to hydrogen depends on
the temperature of the catalyst layer and space
velocity of the ethanol/water vapour. Therefore
in the analysis in this Section, the experimental
results of E. Akpan et. al., who investigated steam
reforming of ethanol (Figure 2 (b)) using a commercial catalyst, is applied. Figure 2 (b) shows the
relationship between the amount of the catalyst,
flow rate of ethanol, temperature of the catalyst
layer, and fuel conversion rate. The fuel conversion increases such that the catalyst is highly filled
when the temperature is high.
C2 H5OH + 3H2O

2CO2 + 6H2 173kJ/mol
(7)
Analysis Model of the Catalyst Layer

The temperature, coordinates of the radial direction and coordinates of the axial direction of the
catalyst layer are given by T , r and x , respectively. Equation (8) is the heat diffusion equation
in the catalyst layer. Here, qr is the amount of
endothermals in the catalyst layer, and c , C c
and c are the density, specific heat and heat
transfer coefficient of the catalyst, respectively;
t is the sampling time. The catalyst layer of the
cylinder type, shown in Figure 2 (a), is divided
into directions of r and x . These areas are used
as elements for the analysis. A two-dimensional
model is used in this analysis, and the following
assumptions are introduced.
Assumptions Used for the Analysis

1. The catalyst and process gas are at a local
thermal equilibrium.
390
2. The temperature dependence of the physicalproperty value is taken into account.

3. The flow velocity of the process gas is uniform in all cross-sections.
4. The pressure loss of process gas is not taken
into consideration.
Assumption c. and d. differ from an actual state
strictly. However, it is thought that these assumptions have the small influence on heat transmission.
Figure 2 (c) shows the division elements elx ,r of
the catalyst layer. Here, x = 1, 2,..., N x ,
r = 1, 2,..., N r , N r is the element number of the
catalyst layer.
Heat Diffusion Equation

Based on assumptions a. to d., described in the
previous Section, and the boundary conditions of
Equations (9) to (11), the temperature distribution
of the catalyst layer is analyzed using the diffusion
equation, Equation (8). A central finite difference
method is introduced into Equation (8) in this
analysis. Equation (12) is the mass flow rate of
process gas, and Equations (13) and (14) are the
boundary conditions. The volume flow rate of the
process gas in Equation (12) is ug , and g is the
mean density of the process gas. The value of u 0
in Equation (14) represents the space velocity of
the fuel vapor at the entrance of the catalyst layer.
This value is calculated by dividing the volume
flow rate of the fuel vapor by the cross Section
of the catalyst layer. Equation (15) is the value of
the endothermals from the reforming reaction.
Variables g g , and H r in Equation (15) are the
molar flow rate of process gas, conversion and
reaction heat, respectively. If the temperature T
of the catalyst layer is known, the conversion
can be obtained from the characteristics of the
catalyst. Because H r is determined by the reaction described by Equation (7), if g g is given, we
can calculate the amount of endothermals q r using the reforming reaction (Equation (7)).
2T 1 T 2T
C T
+ q r = c c
+
+
(8)
r 2
t
c
r r
x 2
Boundary Conditions
T
= 0 at r = Rcl , 0 x Lcl
r
T
= qs q rad q con =

x
qs Ts4 T4 h (Ts T )
c
(9)
(10)
at x = 0, 0 r Rcl
T
T
= 0 at x = Lcl ,
= 0
x
r
at r = 0 T = T for t = 0 (11)
( u ) = 0
x g g
(12)
u
u
= 0 at r = Rcl ,
= 0 at x = Lcl
r
x
(13)
u = u 0 at x = 0 ,
q s , convective heat transfer q con , and heat released

by radiative heat transfer q rad (Equations (1) and
(2)). Since the temperature of the catalyst is low,
radiation-heat transfer is considered to be small
compared with heat conduction and heat transmission. Influence of the radiation heat transmission
is not taken into consideration in this analysis.
The temperature distribution (Tx ,r ) of the catalyst
layer, shown in Figure 2 (c), is obtained by introducing the calculus of finite differences into
Equation (8). The temperature distribution is
analyzed for under boundary conditions (Equations (9) to (11)). The Gauss-Seidel method is
used for calculating the convergence of the calculus of finite differences. If the temperature
distribution Tx ,r is known, the conversion ( x ,r )
of each element will be determined from the relationship between the conversion and temperature
of the catalyst layer (Figure 2 (b)). According to
the value of x ,r , the amount of endothermals (
qr ,x ,r ) involved in the reforming reaction of the
process gas is obtained from Equation (15). Moreover, the process gas composition at the outlet of
element elx 1,r may be calculated from its composition at the outlet of element elx ,r . The tem-
u
= 0 at r = 0
r
q r = gg H r
(14)
(15)
perature distribution (Tx ,r ) of the catalyst layer is

then calculated from these results. This calculation
is repeated until the solution converges with that
from the diffusion equation (Equation (8)). For
each sample, the convergence solution of Tx ,r is
Analysis Method
obtained from the same calculation. If Tx ,r converges, the distribution of the conversion and gas
composition can be determined.
Analysis Procedure
Efficiency of the Reforming Component
The surface temperature Ths of the heat exchange

surface of the reactor is first calculated. This is
calculated by using the heat capacity of the heat
exchange surface, input heat of solar insolation
The rate of the amount of solar radiation obtained

by solar collectors A and B, and the higher calorific value of the produced hydrogen, is defined
as the efficiency of the reforming component.
391
Table 1. Analysis condition
Equation (16) is the formula for the efficiency of

the reforming component.
The higher calorific
value of hydrogen
Qh
s =

=
Amount of heat
QA + QB
colleected per day
(16)
age particle diameter of 3 mm and the packing

factor is 0.85. The transmissivity of the heat exchange wall of the reactor is set to 0.9, and the
collector efficiency (concentration factor of the
solar mirror) of the solar collectors is set to 90%.
The heat transfer coefficient h in Equation (1)
assumes natural convection. Here, h is set to 10
FBSR Operation
W/m2K. Moreover, hs in Equation (2) is 0.95.

The sampling interval time was 0.01 s and the
analysis is calculated for a maximum 600 s.
System Specifications
Analysis Conditions
Here, installation of the FBSR to the individual

house in Sapporo, Japan is described. Table 1
gives the analysis condition of the FBSR used in
this case study. The area Ahs of the solar insolation acceptance surface (heat exchange surface)
of the reactor is 0.005 m2. The diameter Dcl of
The length of the element is about 2 mm along

the r and x directions of the catalyst layer. The
number of elements is N x = 30 and N r = 40
for each direction. The sampling interval is 0.01
s, and analysis takes no longer than 600 s. For the
convergence calculation of the diffusion equation
shown in Figure (8), the analysis accuracy is less
than 10-5. S/C (molar ratio of steam to ethanol) of
the ethanol fuel supplied to the vaporizer was 3.0.
For the fuel (ethanol solution) supply referenced
the catalyst layer is 80 mm, and its width Lcl is

60 mm. The supporting structure of the reforming
catalyst is made of spherical alumina, the reactor
is filled with the reforming catalyst with an aver-
392
in Figure 2 (b), it was decided that the value

(amount of catalysts / ethanol flow rate) of the
horizontal axis in the figure should be set to 35000
kg/(kmol/s).
Analysis Results
Temperature Distribution
of the Catalyst Layer
Figure 3 (a) shows the results of the transient response characteristics of the catalyst layer temperature. In this analysis, after the solar insolation was
input into the heat exchange surface of the reactor,
0 s is the time when the maximum temperature
of the surface is stabilized. When the outside air
temperature is 293 K, the heat exchange surface
temperature, for a solar irradiance at 250 W/m2,
rises to about 500 K. On the other hand, at 1000
W/m2, the heat exchange surface temperature
rises to about 890 K. In Figure 3 (a), since the
solar irradiance is large, it is very sensitive to the
temperature distribution of the catalyst layer. This
is because the conversion of the ethanol/water
vapor increases, and the catalyst temperature in
the reactor is high. In each result of Figure 3 (a),
the temperature gradient along the x-axis has not
reached zero after 100 s. Since the supply direction
of fuel is the same as the supply direction of solar
radiation, the temperature gradient of the x-axis
direction becomes small.
Composition of the Process Gas

Figure 3 (b) shows the process gas composition
along the direction of the x axis of the catalyst
layer. The molar flow rate of hydrogen is larger
than for other gas compositions. The distribution
of the molar flow rate of hydrogen, and stable
time change as a function of the magnitude of the
solar insolation, were input into the reactor. At
the time of low solar irradiance, the hydrogen
production rate for short-time solar insolation
with a lot of fluctuation may result in an unstable
production rate. For example, in conditions with

less than 10 s of solar insolation fluctuation, the
hydrogen production rate does not yield the rated
speed (0.008 mol/s) in all areas of the catalyst
layer, for a solar irradiance of 250 W/m2.
Transient Response
Characteristics of the Reactor
Figure 4 shows the transient response characteristics of the hydrogen production rate of the reactor.
The period of stability of hydrogen production rate
is so short that the solar irradiance input into the
reactor is large, and the hydrogen rate affects the
outside temperature. This is because the result
of Equations (1) and (2) is affected by outside
temperature. However, under actual weather conditions, a solar insolation fluctuation interval on
the order of tens of seconds appears most often.
Therefore, in the following Section, the hydrogen
production rate of the FBSR is investigated using
observed solar irradiance data. This Section uses
observed solar irradiance data and outside air temperature from the Surface-weather-observation
1-minute data, 2007. Sapporo district meteorological observatory, Japan. Meteorological Business
Support Center.
Hydrogen Production Characteristics

Based on Observed Weather Data
1. Observed Weather Data
Figures 5 (a) and (b) are observed weather data
of solar irradiance and outside air temperature for
August 23 and March 1, 2007 in Sapporo. Moreover, Figure 6 shows observed data during daylight
hours for each day. Because the characteristics
of solar irradiance and outside air temperature in
March and August differ greatly, representative
days were chosen from these months. The solar
irradiance in Figures 5 (a) and (b) correspond with
daylight hours, shown in Figure 6. The daylight
hours is defined in period with direct solar radiation 0.12 kW/m2 or more. The amount of hydrogen
393
Figure 3. Temperature distribution and flow rate of process gas in the catalyst layer. Outside air temperature 293 K.
Figure 4. Flow rate of hydrogen production
394
Figure 5. Weather observation at 1-min intervals in Sapporo
Figure 6. Weather observation during daylight hours at 1-min intervals
production, reforming component efficiency, and

operation method of the FBSR are investigated
using these data.
2. The Amount of Hydrogen Production
Figure 7 shows the amount of hydrogen produced
during every minute of each representative day. As
shown in Figures 5 (a) and (b), the solar insolation
fluctuation from 6:00 to 11:00 differs greatly on
each representative day. As a result, the fluctuation
frequency of the amount of hydrogen produced on
August 23 appears greater than that of March 1.
The solar irradiance on the morning of March 1

is stable (Figure 5 (a)), while the solar insolation
fluctuation as a result of clouds was recorded on
August 23 (Figure 5 (b)).
Solar Insolation Fluctuation and

Efficiency of the Reforming Component
Results for solar irradiance, amount of hydrogen
production, and efficiency of the reforming component (Equation (16)) on each representative day
are shown in Table 2. The solar irradiance on the
395
Figure 7. Characteristics of the hydrogen flow rate of the FBSR
Figure 8. Analysis results of the operation plan on March 1, 2007
representative day in March is 1.32 times that of

the representative day in August. However, the
difference in the amount of hydrogen produced
is 1.17 times. On the other hand, the efficiency
of the reforming component in March was larger
than in August (47%, vs. 42%, respectively). As
396
shown in Figures 5 (a) and (b), solar insolation

fluctuated more in August as compared to March.
These figures show that fluctuation of the solar
irradiance influences the amount of hydrogen
produced by the FBSR. Accordingly, the amount
of hydrogen produced and efficiency of the re-
Table 2. Analysis results of the FBSR performance
forming component change with the magnitude

of solar insolation fluctuation. Moreover, frequent
occurrence of large solar insolation fluctuations is
disadvantageous for the efficiency of the reforming component.
Operational Plan for the FBSR

The power usage, when introducing the FBSR
into the average individual house in Sapporo, is
planned using the analysis results of Figures 5 to 7.
1. Power Load and Purchased Power
Sapporo is located in a cold district, and the cooling load of the summer season is not taken into
consideration. The heating load of the winter,
however, is supplied from the exhaust heat and
backup boiler of the system. Therefore, the power
load patterns for March and August (Figures 8 (a)
and 9 (a)) show that there is no large difference in
power usage. However, as Figures 5 and 6 show,
the solar irradiance and outside air temperature
differ greatly for each representative month.
Figures 8 (b) and 9 (b) show the amount of electricity produced every minute in the FBSR. Since
the power storage equipment is not connected to

the FBSR, the production of electricity strongly
depends on the characteristics of solar irradiance
and outside air temperature for that day. In this
analysis, the collection area of solar collectors A
and B were set to 1 m2. The characteristics of the
purchased power at this time are given in Figures
8 (c) and 9 (c). From these results, the load peak
at 8:00 in the morning can be cut by using power
generated by the FBSR. However, in order to cut
the peak around 19:00, it is necessary to use a
time-shifted power supply by introducing a battery and hydrogen storage system.
2. Exhaust Gas
Figure 10 shows the minute-by-minute discharge
pattern of CO2 by the FBSR. The CO2 emissions
on March 1 and August 23 are 0.732 kg/Day and
0.854 kg/Day, respectively. The greenhousegas-emission-factor for power generation in the
Greenhouse Gas Discharge Calculation Method
Investigative Commission Report (Ministry of
Environment in Japan, August 2000) is 0.378
kgCO2/kWh. On the other hand, the CO2 emissions with the FBSR in March and August were
397
Figure 9. Analysis results of operation on August 23, 2007
Figure 10. Analysis results for CO2 emissions
the same: 0.306 kgCO2/kWh. Using the FBSR,

the amount of greenhouse gas discharge can be
reduced by 19%, compared to commercial power.
3. The Utilization Rate of Renewable Energy
As shown in Table 2, the percentage utilization
of renewable energy by the FBSR in this Section (where the collecting area of the two solar
collectors is 2.0 m2) is 25.3% in August and
398
21.4% in March. These efficiencies are advantageous compared with the power generation of a
photovoltaic cell. They can be further increased
by increasing the utilization rate of renewable
energy, which can be done by increasing the
collecting area of solar collectors A and B.
When increasing the collecting area, it becomes
necessary to examine the method of storage of
produced hydrogen. Some potential methods for
the storage of produced hydrogen include using

a hydrogen cylinder and battery (e.g., blocks of
broken line in Figure 1 (a)).
Conclusions
The hydrogen production rate, power output, and
amount of power purchased for a PEFC using a
bioethanol reforming system (FBSR) were investigated by numerical analysis. In this Section, heat
transfer analysis was introduced into the catalyst
layer of the reforming component. As a result, the
relationship between the supply of solar insolation
and hydrogen production rate became clear, and
the following Conclusions were obtained.
1. The hydrogen molar flow rate and stability
time period of the steam reforming reaction
change with the magnitude of solar insolation. Moreover, fluctuations in the amount
of solar insolation have large effects on
the hydrogen production rate of the FBSR.
Accordingly, the efficiency of the reforming
component falls due to fluctuations in solar
insolation. If the solar insolation fluctuation
is mostly on the order of tens of seconds,
the reforming reaction may not be able to
obtain a stable production rate (rated speed)
of hydrogen, and the amount of hydrogen
produced and efficiency of the reforming
component will decrease.
2. When the amount of hydrogen produced by
the FBSR was analyzed using meteorological data from Sapporo, Japan on March 1
and August 23, 2007, the efficiency of the
reforming component was calculated to
be 40% or more on both days. Moreover,
the CO2 emissions on March 1 and August
23, 2007 from the FBSR were 0.732 kg/
Day and 0.854 kg/Day, respectively.
When the greenhouse gas emission factor for power generation specified in the
Greenhouse Gas Discharge Calculation
Method Investigative Commission report
(Ministry of Environment in Japan, August,

2000) is used, the amount of greenhouse
gas discharge in the FBSR is 19% lower
compared to commercial power.
3. The percent utilization of renewable energy
by the 2-m2 collecting area of the FBSR was
25.3% in August and 21.4% in March. These
efficiencies are superior to the efficiency
of power generation in a photovoltaic cell.
The efficiency performance of the FBSR
can sufficiently compete with a commercial
photovoltaic cell.
As uncertainty in this analysis, the difference
in assumptions described in previous Section, the
calculation error, the setting performance of each
equipment, etc. can be considered. Magnitude
of these influences is explained by future study.
EFFICIENCY ANALYSIS OF
A COMBINED PEFC AND
BIOETHANOL-SOLAR-REFORMING
SYSTEM FOR INDIVIDUAL HOUSES
Introduction
The development of a distributed power supply
with limited greenhouse gas emission is a global
issue of current interest and importance. PEFCs
(polymer electrolyte fuel cells) are one candidate
for a clean, distributed power supply. However,
the environmental impact of fuel cells changes
greatly depending on the method of hydrogen
production. For example, a large quantity of CO2
is discharged when using reforming methods that
employ fossil fuels. Alternatively, a fuel cell system that uses the heat of a small solar collector for
the steam reforming of bioethanol, a bioethanol
reforming system for fuel cells (FBSR system), has
been examined (Obara & Tanno, 2007a, 2007b,
2008). Furthermore, there are researches on much
hydrogen production technology using solar energy currently (for example, Azad, Kesavan, &
399
Al-Batty, 2009; Noglik & Roeb, 2009; Turner,

Sverdrup, Mann, Maness, Kroposki, Ghirardi,
Evans, & Blake, 2008; Petrasch & Steinfeld, 2007).
In this research, we have investigated the characteristics of the reformed gas production (Obara
& Tanno, 2007a), the production-of-electricity
characteristic of the FBSR using the weather
forecast (Obara &Tanno, 2007b, 2008), and the
relationship between solar insolation with fluctuation and a production-of-electricity characteristic
(Fresno, Fernndez-Saavedra, Gmez-Mancebo,
Vidal, Snchez, Rucandio, Quejido, & Romero,
2009). This chapter continues these reports. The
main objectives of this chapter are to compare
the efficiency of the overall efficiency of the
FBSR, and competition technology. If the overall
efficiency of the proposal system is higher than
competition technology, the trial production of
the proposal system is effective. Moreover, in this
study, fluctuation of solar radiation was taken into
consideration to the reaction rate of steam reforming. There is no example of investigation of the
solar reforming with solar radiation fluctuation
in the past. The conversion rate of the reforming
reaction of ethanol/water vapor is influenced by
the temperature of a catalyst layer and the space
velocity of a process gas (Fresno et al., 2009). A
pellet type reforming catalyst is installed in the
reactor of the FBSR. Reformed gas with a high
hydrogen composition is outputted by supplying
ethanol/water vapor to the reactor under the control
of reactor temperature. However, because the heat
source of a reactor is sunlight, the conversion rate
of the fuel is affected by weather. Unsteady heat
transfer analysis was introduced into the catalyst
layer in a previous examination (Fresno et al.,
2009). From this analysis, the transient response
characteristic concerning the temperature distribution in the catalyst layer and composition of
the process gas were investigated. As a result,
under the weather conditions with high levels of
solar radiation fluctuation over a short time, it
became clear that a reforming reaction was not
sufficient under the effect of a response delay.
400
Moreover, it clarified the generation efficiency

of the FBSR when taking into consideration the
transient response characteristic of the reforming
reaction. This chapter investigates the thermal
output characteristic of the FBSR. The overall efficiency of the system is clarified in consideration
of these results and the result of the last report.
It is the objective of this chapter to highlight the
differences between the FBSR and competing
technologies, such as photovoltaic cells.
Material and Method

System Block Diagram
Figure 11 (a) is a block diagram of the fuel cell
system with a bioethanol solar reforming system
(FBSR). The solar tracking system is introduced
into two solar collectors. Moreover, each collectors collecting area is nearly 1.0 m2. Bioethanol for
the boiler and the bioethanol solution for reforming
are contained in a fuel tank. The S/C (molar ratio
of steam to ethanol (Xuan, Leung, Leung, & Ni,
2009) of the bioethanol solution for reforming is
3.0. Two parabolic mirrors (solar collector) with
a solar tracking system are introduced into the
FBSR. The heat for fuel evaporation is condensed
in solar collector A while the heat for reforming
of the fuel vapor is condensed in solar collector B.
Fuel and Reformed Gas System

1. Flow of Fuel and Reformed Gas
The fuel for reforming (bioethanol solution) is
supplied to the vaporizer installed in solar collector
A from the fuel tank using a pump. The fuel vapor
is supplied to a reactor installed in solar collector
B. The reactor is filled with a reforming catalyst.
If the fuel vapor contacts the reforming catalyst,
it will be reformed into gas with a high hydrogen
content. Surplus water and CO are contained in the
reformed gas. Therefore, water is removed using
a gas cooler. Moreover, reformed gas is supplied
to a shift unit and a CO oxidation unit, and the
Figure 11. System design
401
amount of CO is oxidized. CO included in the

compressed hydrogen is also oxidized using the
CO oxidation unit.
2. Reaction Temperature and Chemical Reaction Heat
Figure 11 (b) shows the relationship between
the chemical reaction heat and the goal response
temperature in each component of the system. The
shifter, the gas cooler, the CO oxidation, and the
PEFC are accompanied by an exothermic reaction.
On the other hand, the vaporizer and the reformer
are accompanied by an endothermic reaction. The
amount of endothermals of the vaporizer and the
reformer of the FBSR is supplied by the solar
collectors. After supplying the exhaust heat of
the gas cooler, the CO oxidation, and the PEFC
to a heat storage tank, the heat is supplied to the
demand side. The boiler is operated when there
is less thermal storage than demand.
Electric Power System

If reformed gas is supplied to the PEFC, directcurrent power will be obtained. This power is
supplied to the DC-DC converter, the DC/AC
converter, and the inverter, and the power and
the frequency of regulation are supplied to the
demand side. When there are less power outputs
of the fuel cell than the quantity required, the
power is supplied to the demand side from the
commercial power.
Loss and Auxiliary-Machinery Power

Figure 11 (c) is a chart showing the efficiency
and conversion rate of all of the components of
the FBSR. The conversion rates for the reforming
reaction ( r ), CO oxidation reaction ( co ), and
hydrogen burning ( H ) are provided. The gen2
eration efficiency ( fc ) of the PEFC is described

in following section. On the other hand, in the
402
analysis in this chapter, auxiliary-machinery

power of blowers and the pump is not included.
The setting value of st , bl , dc , and it are
described in a later section.
Operation Method of the System

When fuel is supplied to vaporizer, the fuel gas is
fed to the reforming component. Reformed gas is
supplied to the shift unit and the gas cooler, and is
stored using cylinder. In order to store hydrogen,
the storage tank of the reformed gas is installed
in the system. While compressing and storing the
reformed gas, it removes the water vapor in the
gas using a cooler. Reformed gas is fed to the CO
oxidation unit from the cylinder or the gas cooler,
and is supplied to the fuel cell. To output power
to the commercial power grid (with a regular
frequency and voltage), the output of the fuel
cell is converted with a DC-DC converter and an
inverter. The exhaust heat from the gas cooler, the
fuel cell, and the CO oxidization unit is stored.
This heat is supplied to the demand side of the
FBSR. When supplied solar heat is insufficient,
a boiler is operated.
Heat Transfer Analysis

Efficiency of Reforming Component
The reforming component shown in Figure 11 (a)
consists of solar collector B and the reactor (Figure 12 (a)). A reactor end face is a solar insolation
acceptance surface with area Ahs . Other surfaces
are insulated. The solar insolation collected by
solar collector B is inputted into the solar insolation acceptance surface. Moreover, the heat input
to this solar insolation acceptance surface heats
the catalyst layer of the reactor. The efficiency of
the reforming component is determined by the
ratio of the rate of the higher heating value of the
hydrogen produced to that of the amount of con-
Figure 12. Reactor
densed solar radiation in solar collectors A and

B. Equation (17) is the formula for the efficiency
s of the reforming component.
s =
The higher heating

value of hydrogen (Qh )
Amount of heat
coollections per day ( QA + QB )
(17)
ceptance surface on the reactor is discharged to

the ambient air by convection heat transfer q con
and radiation heat transfer q rad . Equation (18) is
the heat convection of the catalyst layer, and it
contains the Damkohler correction number
(Usami et al., 2000). The value of Da in Equation
(18) is calculated using Equation (19).
Nu = 9.49 (Re Pr)
0.516
1.43
Heat Transfer in the Catalyst Layer
c
Dcl
The reactor is filled up with several millimeters

of the spherical reforming catalyst. In addition,
some of solar insolation inputted from the ac-
Da = (H r r ) Dc
+ 27.2 Da 0.325
ug C g Tg )
(18)
(19)
403
Reforming Reaction and Analytical

Model for the Catalyst Layer
The experimental result of ethanol steam reforming using the commercial catalyst by Akpan et
al., which is shown in Figure 12 (b), is introduced
into the analysis (Akpan et al., 2007). In this
figure, the relationship between the amount of
catalyst, the flow rate of ethanol, and the temperature of the catalyst layer and the fuel conversion rate of the catalyst is depicted. The fuel
conversion rate increases with an increasing
amount of catalyst and with increasing temperature, as shown in Figure 12 (b). Equation
(20) is the two-dimensional heat diffusion equation of the catalyst layer. The temperature of the
catalyst layer is represented byT , the coordinates
of the direction of the radial direction is represented by r , and the coordinates of the direction
of the catalyst layer is represented by x . A surface element is created in the direction of r , and
the direction of x , about the catalyst layer of
the cylinder type shown in Figure 12 (a). Figure
12 (c) portrays the element analysis model of
the catalyst layer. The element number of the
catalyst layer is expressed with elx ,r (however,
x = 1, 2,..., N x , r = 1, 2,..., N r ).
2T 1 T 2T
r 2 + r r + x 2

C T
+q r = c c
c
t
(20)
Boundary Conditions
T
= 0 at r = Rcl , 0 x Lcl
r
404
T
= qs q rad q con =

x
4
4
qs Ts T h (Ts T )
c
(21)
(22)
at x = 0, 0 r Rcl
T
T
= 0 at x = Lcl ,
= 0 at r = 0
x
r
T = T for t = 0
(23)
( u ) = 0
x g g
(24)
u
u
= 0 at r = Rcl ,
= 0 at x = Lcl
r
x
(25)
u = u 0 at x = 0 ,
u
= 0 at r = 0
r
q r = g g r H r
(26)
(27)
Heat Diffusion Equation

Equation (20) is discretized, and the temperature
distribution of the catalyst layer is analyzed under
the boundary conditions of Equations (21) to (23).
The central finite difference method (Mitchell &
Griffiths, 2001) is used to calculate the discretize
Equation (20) in this chapter. Here, qs is the heat
concerning the heat exchange wall of the reactor.
Equations (21) to (27) are given as the boundary
condition when calculating Eq (20). Equation (23)
is the boundary condition of temperature and
Equation (24) is the boundary condition of mass
flow of process gas. Equations (25) and (26) are
the boundary conditions of space velocity. Furthermore, Equation (27) is the boundary condition
of molar flow rate of the process gas. Equation

(24) is the mass flow rate of the process gas, and
Equations (25) and (26) are the boundary conditions in this analysis. The volume flow rate of the
process gas is ug in Equation (24), and g is the
mean density. The value u 0 in Equation (26) is
the space velocity of the fuel vapor at the entrance
of the catalyst layer. This value is the result of
dividing the volume flow rate of the fuel vapor
by the cross section of the catalyst layer. Equation
(27) expresses endothermals from the reforming
reaction. Variables g g , , and H r in Equation
(27) express the molar flow rate of process gas,
the conversion ratio, and the reaction heat, respectively. If the temperature T of the catalyst layer
is known, the conversion rate can be obtained
from the characteristic of the catalyst. Because
H r is determined by the reforming reaction, if
g g is given, the amount of endothermals q r can
be calculated using the reforming reaction.
Analysis Method
Temperature Distribution of the Catalyst
Layer, and the Composition Distribution
The temperature distribution (Tx ,r ) of the catalyst
layer shown in Figure 12 (c) is obtained by introducing calculus of finite differences into Equation
(20). The boundary conditions at this time are
shown in Equation (21) to Equation (26). The
Gauss Seidel method is used for convergence
calculation of the calculus of finite differences.
If the temperature distribution Tx ,r is decided, the
conversion rate ( x ,r ) in the each element is obtained based on the relationship between catalyst
temperature and conversion rate (Figure 12 (b)).
The amount of endothermals of process gas (qr ,x ,r )
is calculated based on the value of x ,r . The
temperature distribution (Tx ,r ) of the catalyst
layer is calculated from these results, and this

calculation is repeated to convergence of the
discretization equation of Equation (20).
Amount of Exhaust Heat

1. Gas Cooler
The exhaust heat of the gas cooler is calculated
using Equation (28) where the number of components in the process gas is N g and t is the
difference in temperature between the process gas
in the gas cooler entrance and exit. C n and Gn
are the specific heat and mass flow rates of the
gas composition n .
Ng
H gc = t (C n Gn )
(28)
n =1
2. CO Oxidation Unit
In the CO oxidation unit, the CO concentration
is reduced by CO burning. However, the hydrogen
contained in the process gas in the case of CO
burning may also burn. This selectivity is decided by control of the catalyst temperature. In
reality, some of the hydrogen also burns under
strict temperature control. In this chapter, the rate
of hydrogen burning in the process gas is set at
3%. The exhaust heat of the CO oxidation unit is
calculated using Equation (29). Here, V is the
volume flow rate of each gas and J is the calorific value.
H ox = VCO JCO + 0.03 VH J H
2
(29)
3. PEFC
When a PEFC of small capacity compared to the
power demand amount is introduced, the operation
time of a full load is long. Generally, the exhaust
heat power of a PEFC differs with load factor.
The PEFC in this chapter has many full-load
operation hours because the capacity is small
405
compared to the power demand. The PEFC often

has to operate with a high load factor. Therefore,
the generation efficiency ( fc ) of the PEFC is set
as 65%. On the other hand, the exhaust heat of
the PEFC is calculated to be 30% of the calorific
power of supply hydrogen.
Operation Case
Specification of the
Reforming Component
1. Catalyst Layer
The FBSR was introduced into individual houses
in Sapporo, Japan. The specification of the reforming component introduced into this case analysis
is described in Table 1. The diameter of the catalyst layer Dcl is 80 mm and its width Lcl is 60
mm. The reactor is filled up with the reforming
catalyst of 3 mm average particle diameter. The
catalyst filling factor is 0.85. The detailed experimental results of the ethanol steam reforming
by this reforming catalyst are described by reference 6. In addition, the catalyst layer is divided
into elements of 2 mm length in the directions of
r and x . The numbers of elements are N x = 30
and N r = 40 .
2. Heat System
The heat transfer coefficient h of the convective
heat transfer q con described by previous section
was set at 10 W/m2K (natural convection). Moreover, hs of the radiation heat transfer q rad gives
0.95 assuming a black body. The area Ahs on the
solar insolation acceptance surface of the reactor
is 0.005 m2. The transmissivity of the heat exchange
wall of the reactor was set at 0.9, and the condensing efficiency of the solar collector was set at
90%.
406
3. Solar Collector and Fuel Supply

In the analysis in this chapter, the collecting area
of solar collector A and solar collector B were
both set to be 1 m2 (unit area). In the convergence
calculation of the discretization equation, an
analytical accuracy of 10-5 was used. The use of
an ethanol solution fuel supply determined that
the value of the horizontal axis in Figure 12 (b)
(Amount of catalysts /Ethanol flow rate) is 35000
kg/(kmol/s).
Storage of the Reformed Gas

The reformed gas can be compressed and stored
in the FBSR (broken-line block in Figure 11 (a)).
Accordingly, the reformed gas with the pressure
P0 and the flow rate U 0,t outputted from the gas
cooler is pressurized to Pcp with a compressor.
Here, the subscript t is sampling time. The work
Wp,t of the compressor is assumed to be the work
of compression by an ideal gas and is calculated
using Equation (30). The compressor efficiency
cp in the equation includes the electricity consumption in an electric motor, the transmission
loss of power, loss due to insufficient air leak and
cooling, and other mechanical loss.
Wp,t = P0 U 0,t ln Pcp P0
cp
(30)
Installation Requirements of the

System and Demand Characteristic
The operation of the power and heat in the case
of introducing the FBSR into standard houses in
Sapporo is planned. However, since a detailed
report about the operation method of this power
has already been published (Fresno, 2009), this
chapter focuses on examining the overall efficiency
of the system.
The power load and the heat consumption of a

house in Sapporo on a representative day in both
March (March 1) and August (August 23) are
shown in Figures 13 (a) and (b) (Narita, 1996).
Because Sapporo is located in a cold district, there
is no cooling load in August. On the other hand,
the space heating load in March is supplied from
the exhaust heat of the system and the backup
boiler. As shown in Figure 13, the characteristics
concerning the amount of solar radiation and outside air temperature (Surface-weather-observation
1-minute data, 2008) will differ greatly in March
and August. In this chapter, as observational data
of the solar irradiance and the outside air temperature, Surface-Weather-Observation 1-Minute
Data and 2007 Sapporo District Meteorological
Observatory and Japan Meteorological Business
Support Center (Surface-weather-observation
1-minute data, 2008) are used. The figure at the
bottom of Figure 13 depicts daylight hours on each
representative day. The amount of solar radiation,
outside air temperature, and daylight hours are
data collected at 1-minute intervals.

Temperature Distribution
of the Catalyst Layer
Figure 14 shows the result of the transient response
characteristic of the catalyst layer temperature
when inputting constant solar insolation (250 W/
m2, 500 W/m2, 1000 W/m2) into the reactor (Fresno
2009). In Figure 14, 0 s is the stable time of the
acceptance surface temperature after inputting
solar insolation into the reactor. When the outside
air temperature is 293 K, the acceptance surface
with solar irradiance of 250 W/m2 rises to about
500 K. In the input of 1000 W/m2, the acceptance
surface rises to about 890 K. The conversion rate
of ethanol steam reforming increases, which leads
to a high catalyst temperature in the reactor. Therefore, the temperature distribution of the catalyst
layer shown in Figure 14 differs so greatly that
solar irradiance is large.
Composition of the Process Gas

Figure 15 shows the analysis result of the process
gas composition of the catalyst layer when inputting constant solar insolation (250 W/m2 and 500
W/m2) into the reactor (Fresno et al., 2009). This
figure shows the process gas composition along
the x axis of the catalyst layer as predicted by
the analysis. This figure shows the composition
of each gas with its respective molar flow rate.
The molar flow rate of hydrogen is larger than
other gases in the composition. Distribution of
the molar flow rate of hydrogen, and the time at
which the hydrogen flow rate becomes stable are
influenced by the magnitude of the solar irradiance input into the reactor. If there is little solar
irradiance and there a short period of solar radiation fluctuation, the hydrogen generation rate may
not reach the maximum possible. From this result,
when there is short-time fluctuation solar radiation
with little solar irradiance, the hydrogen generating rate may not reach a stable generation rate
(rated speed) by a response delay.
Amount of Hydrogen Generated

Figure 16 shows the analysis result of the amount
of hydrogen generated by the FBSR using the
amount of solar radiation, outside air temperature, and daylight hours shown in Figure 13. The
hydrogen generating rate fluctuates a lot when
comparing August 23 to March 1. As shown in
Figure 13, from 6:00 a.m. to 11:00 a.m., the solar
insolation fluctuation on the representative day
in March is stable compared to that in August.
The cause of this solar insolation fluctuation is a
407
Figure 13. Weather observation 1-minuite data in Sapporo
408
Figure 14. Temperature distribution in the catalyst layer of the reactor
Figure 15. Flow rate of process gas in the catalyst layer. Outside air temperature is 293 K.
shadow by clouds. As a result, in Figure 16, the

amount of hydrogen production on a representative
morning in March will be stabilized compared
with that in August. As shown in Figure 14, the
rate of hydrogen generation may not be less than
rated output in a weather condition with solar
insolation fluctuation.
Production of Electricity and Amount of

Purchased Powers
Figure 17 shows the analysis result of the production of electricity of the FBSR and the amount of
purchased powers. In this analysis, the power is
purchased when the production of electricity of
409
Figure 16. Characteristics of the hydrogen flow rate of the FBSR
the FBSR is less than the power demand amount

shown in Figure 13. As for the result of Figure 17,
the solar irradiance on the representative day of
each month and the characteristic of the outside air
temperature affect the purchase power. Moreover,
when introducing solar collector A and solar collector B with areas of 1.0 m2, the power load peak
at 8:00 of both representative days (Figure 13) can

be decreased. However, in order to decrease the
peak around 19:00, the increase in the compressed
hydrogen by extension of solar collecting area is
required. Moreover, a time shift of the electric
power supply is required using hydrogen storage equipment. When the amount of production
Figure 17. Analysis results of the power operation plan
410
of the reformed gas by solar collectors A and B

is introduced into the power load pattern shown
in Figure 13, storage of the reformed gas is not
required. The electric power supply rate of the
FBSR to a power demand amount is predicted to
be 21.4% and 25.3% on a representative day in
March and August, respectively.
Operation of the Exhaust Heat

Figures 18 and 19 show the analysis result of the
exhaust heat of the fuel cell, the CO oxidation
unit, the gas cooler, and the heat of the backup
boiler on representative days in March and August.
In terms of the solar insolation fluctuation from
Figure 18. Analysis results of the heat operation plan on March 1, 2007
411
Figure 19. Analysis results of the Heat Operation plan on August 23, 2007
6:00 a.m. to 11:00 a.m., the exhaust heat power of

the morning in March is stable compared to that
in August. This is due to the difference in solar
insolation fluctuation on a representative day in
each month shown in Figure 13. The heat supply
rates of the FBSR to heat demand is predicted
to be 1.2% and 13.7% on the representative day
in March and August (except for boiler power).
412
Overall Efficiency
The conversion rates to the electric power are
30.7% (March representative day) and 27.1%
(August representative day) every month among
the solar irradiance obtained by the 1 m2 solar
collectors A and B on the representative days. On
the other hand, the conversion rates to heat supply
of solar irradiance are 16.7% (March representative day) and 14.8% (August representative day).
Therefore, the overall efficiency of the FBSR by
this operation case is 47.4% (March representative
day) and 41.9% (August representative day). The
difference in solar irradiance will be 1.32 times in
August compared with that on the March representative day. However, the overall efficiency on
the March representative day is larger than that in
August. Therefore, the magnitude and the number
of occurrences of the solar insolation fluctuation
strongly influence the overall efficiency. Table 2
is the analysis results of the FBSR performance.
Conclusions
The overall efficiency of a PEFC with the bioethanol reforming system using a sunlight heat
source (FBSR) was investigated by numerical
analysis. In this chapter, the heat transmission
characteristics of the catalyst layer installed in
the reactor were investigated. The transient characteristic of hydrogen generation was examined
based on these results. Furthermore, the supply
and amount of purchase of electric power and
heat were investigated using the energy-demand
characteristic in a standard house in addition to
meteorological data on representative days in
March and August in Sapporo, Japan. The total
collecting area of the solar collectors was 2 m2.
The following conclusions were drawn from these
analysis results. As uncertainty in this analysis,
the calculation error, the setting performance of
each equipment, etc. can be considered. Magnitude
of these influences and error concerning the grid
system is explained by future study.
1. The magnitude of solar irradiance greatly
influences the temperature distribution and
composition distribution of process gas in
the catalyst layer. When there is short-time
fluctuation solar radiation with little solar
irradiance, the hydrogen generation rate
may not reach a stable generation rate (rated
speed) by a response delay.
2. The rate of converting sunlight into electrical power in the proposed system is 30.7%
and 27.1% on representative days in March
and August, respectively. On the other hand,
the rate converted into heat is 16.7% and
14.8%, respectively. As a result, the overall
efficiency of the FBSR by the analysis
case in this chapter is 47.4% and 41.9%,
respectively. These results indicate that the
proposed system is competitive with other
energy systems, such as a photovoltaic cell.
However, operation of the FBSR takes the
cost of the bioethanol.
REFERENCES
Akpan, E., Akande, A., Aboudheir, A., Ibrahim,
H., & Idem, R. (2007). Experimental, kinetic
and 2-D reactor modeling for simulation of the
production of hydrogen by the catalytic reforming
of concentrated crude ethanol (CRCCE) over a nibased commercial catalyst in a packed-bed tubular
reactor. Chemical Engineering Science, 62(12),
31123126. doi:10.1016/j.ces.2007.03.006
Azad, A. M., Kesavan, S., & Al-Batty, S. (2009).
A closed-loop proposal for hydrogen generation
using steel waste and a prototype solar concentrator. International Journal of Energy Research,
33(5), 481498. doi:10.1002/er.1491
Charvin, P., Stephane, A., Florent, L., & Gilles,
F. (2008). Analysis of solar chemical processes
for hydrogen production from water splitting
thermochemical cycles. Energy Conversion and
Management, 49, 15471556. doi:10.1016/j.
Clarke, R. E., Giddey, S., Ciacchi, F. T., Badwal,
S. P. S., Paul, B., & Andrews, J. (2009). Direct
coupling of an electrolyser to a solar PV system for
generating hydrogen. International Journal of Hydrogen Energy, 34(6), 25312542. doi:10.1016/j.
413
Fresno, F., Fernndez-Saavedra, R., GmezMancebo, B. M., Vidal, A., Snchez, M., &
Rucandio, M. I. et al. (2009). Solar hydrogen
production by two-step thermochemical cycles:
Evaluation of the activity of commercial ferrites.
29182924. doi:10.1016/j.ijhydene.2009.02.020
Japan Meteorological Business Support Center.
(2007). Surface-weather-observation 1-minute
data, sapporo district meteorological observatory.
Tokyo, Japan: Japan Meteorological Business
Support Center.
Mitchell, A. R., & Griffiths, D. F. (2001). Finite
difference and related methods for differential
equations. New York, NY: Wiley.
University, Sapporo, Japan.
Noglik, A., Roeb, M., Christian, S., & Robert, P.
P. (2009). experimental study on sulfur trioxide
decomposition in a volumetric solar receiverreactor. International Journal of Energy Research,
33(9), 799812. doi:10.1002/er.1514
Obara, S., & Tanno, I. (2007). development of
distributed energy system due to bio-ethanol pem
fuel cell with solar reforming, Part 1Evaluation
of basic performance. Transactions of the Society
of Heating. Air-Conditioning and Sanitary Engineers of Japan, 123, 2332.
414
Obara, S., & Tanno, I. (2007b). operation prediction of a bioethanol solar reforming system using
a neural network. Journal of Thermal Science and
Obara, S., & Tanno, I. (2008). Development of
distributed energy system due to bio-ethanol pem
fuel cell with solar reforming, Part 2High-speed
analysis of the operation plan using a neural
network. Transactions of the Society of Heating
[In Japanese]. Air-Conditioning and Sanitary
Engineers of Japan, 130, 3342.
Petrasch, J., & Steinfeld, A. (2007). dynamics of a
solar thermochemical reactor for steam-reforming
of methane. Chemical Engineering Science, 62,
42144228. doi:10.1016/j.ces.2007.04.036
Turner, J., Sverdrup, G., Mann, K. M., Maness,
P. C., Kroposki, B., & Ghirardi, M. etal. (2008).
Renewable hydrogen production. International
Journal of Energy Research, 32(5), 379407.
doi:10.1002/er.1372
Usami, Y., Fukusako, S., & Yamada, M. (2000).
Heat and mass transfer in a reforming catalyst bed
(Quantitative evaluation of the controlling factor
by experiment). Transactions of the JSME, Series
B, 67(659), 1801-1808.
Xuan, J., Leung, K. H. N., Leung, Y. C. D., &
Ni, M. (2009). A review of biomass-derived fuel
processors for fuel cell systems. Renewable &
Sustainable Energy Reviews, 13(6-7), 13011313.
doi:10.1016/j.rser.2008.09.027
APPENDIX
Nomenclature

A : Area [m2]
C : Specific heat [J/(gK)]
D : Diameter [m]
Da : Modified Damkohler number
el : Catalyst layer element number
G : Mass flow rate [g/s]
g g : Molar flow rate [mol/s]
H : Reaction heat [J/mol]
h : Heat transfer coefficient [W/(m2hK)]
J : Calorific value [J/m3]
L : Length, width [m]
Me : Mass flow rate of ethanol [g/s]
M h : Mass flow rate of hydrogen [g/s]
N : Number of elements
N g : Number of gas composition
Nu : Nusselt number
P : Power [W]
P0 : Inlet pressure of the process gas [Pa]
Pcp : Outlet pressure of the process gas [Pa]

Pr : Prandtl number
Q : Quantity of heat [J]
q : Heat [W]
R : Radius [m]
r : Radial direction of the catalyst layer
Re : Reynolds number
T : Temperature [K]
t : Temperature difference [K]
U 0 : The volume flow rate of the process gas [m3/s]
u : Flow rate [m/s]
u 0 : Initial flow rate [m/s]
Wp : Work of the compressor [W]

x : Axial direction of the catalyst layer
415
Greek Symbols

r : Reaction rate [mol/(m3s)]

: Layer of the element
: Emissivity
: Efficiency
cp : Efficiency of the compressor
s : Efficiency of the reforming component

: Heat conductivity [W/(mK)]
: Density [g/m3]
: Kinetic viscosity [m2/s]
: Stefan-Boltzmann constant
: Conversion ratio
Subscripts

416
A , B : Solar collectors A and B

bl : Boiler
c : Catalyst
cf : CO oxidation unit to the cell stack
cl : Catalyst layer
cm : Customer
con : Convective heat transfer
cp : Commercial power
cs : Solar collector
dc : DC-DC converter
fc : Cell stack
g : Process gas
gc : The gas cooler to the heat storage tank
h : The higher heating value of hydrogen
hs : Heat supply surface of the reactor
it : DC-AC converter and inverter
ox : The CO oxidation unit
pv : Vaporizer pump
r : Reforming
rad : Radiation
rs : The reactor to the shift unit
s : Sunlight
sc : Gas cooler to the CO oxidation unit
st : Storage tank
sg : The gas cooler shift unit
tb : The fuel tank to the boiler
: Ambient air
417
Compilation of References
Abdelaziz, L. (2004). Development of a radiant heating

and cooling model for building energy simulation software.
Building and Environment, 39(4), 421431. doi:10.1016/j.
buildenv.2003.09.016
Abu-Sharkh, S., Arnold, R. J., Kohler, J., Li, R., Markvart,
T., & Ross, J. N. etal. (2006). Canmicrogrids make a
major contribution to UK energy supply? Renewable &
Sustainable Energy Reviews, 10(2), 78127. doi:10.1016/j.
rser.2004.09.013
Al-Alawi, A., Al-Alawi, S. M., & Islam, S. M. (2007).

Predictive control of an integrated PV-diesel water and
power supply system using an artificial neural network.
Renewable Energy, 32(8), 14261439. doi:10.1016/j.
renene.2006.05.003
Alan, S. F., & Donald, F. R. Jr. (2005). Operating line
analysis of fuel processors for PEM fuel cell systems.
12511257. doi:10.1016/j.ijhydene.2005.02.011
Adam, H., & Matthew, L. (2005). Impacts of temporal

precision in optimization modeling of micro-combined
heat and power. Energy, 30(10), 17591779. doi:10.1016/j.
energy.2004.11.012
Alibhai, Z., et al. (2004). Coordination of distributed energy resources. In H. G. Stassen (Ed.), Systems, Man and
Cybernetics, IEEE International Conference on Volume
2, (pp. 1990-1995).
Aki, H., Yamamoto, S., Kondoh, J., Maeda, T., Yamaguchi,

H., Murata, A., & Ishii, I. (2006). Fuel cells and energy
networks of electricity, heat, and hydrogen in residential
areas. International Journal of Hydrogen Energy, 31(8),
967980. doi:10.1016/j.ijhydene.2005.12.016
Alibhai, Z., Lum, R., Huster, A., Gruver, W. A., & Kotak, D. B. (2004). Coordination of distributed energy
resources, Systems, man and cybernetics. In H. G. Stassen
(Ed.), IEEE International Conference on Vo. 2, Hague,
Netherlands, (pp. 1990-1995).
Akkaya, V. A., Sahin, B., & Erdem, H. H. (2008). An analysis of SOFC/GT CHP system based on exergetic performance criteria. International Journal of Hydrogen Energy,
33(10), 25662577. doi:10.1016/j.ijhydene.2008.03.013
Andrew, W., & Steven, F. (2007). The nature of domestic

electricity-loads and effects of time averaging on statistics
and on-site generation calculations. Applied Energy, 84(4),
389403. doi:10.1016/j.apenergy.2006.09.008
Akpan, E., Akande, A., Aboudheir, A., Ibrahim, H., &

Idem, R. (2007). Experimental, kinetic and 2-D reactor
modeling for simulation of the production of hydrogen
by the catalytic reforming of concentrated crude ethanol (CRCCE) over a ni-based commercial catalyst in a
packed-bed tubular reactor. Chemical Engineering Science, 62(12), 31123126. doi:10.1016/j.ces.2007.03.006
Anstett, M., & Kreider, J. F. (1993). Application of neural networking models to predict energy use. ASHRAE
Transaction, 99(1), 505517.
Architectural Institute of Japan, (2001). The nationwide
research study concerning the energy consumption in the
house in the 2001 fiscal year, 3, 3-6. In Japanese.
Architectural Institute of Japan. (2002). The nationwide

research study concerning the energy consumption in the
house in the 2001 fiscal year, 3, 3-6. In Japanese
Ashari, M., & Nayar, C. V. (1999). An optimum dispatch
strategy using set points for a photovolatic (PV)-dieselbattery hybrid power system. Solar Energy, 66(1), 19.
doi:10.1016/S0038-092X(99)00016-X
Ay, M., Midilli, A., & Dincer, I. (2006). Exergetic performance analysis of a PEM fuel cell. International Journal
of Energy Research, 30, 307321. doi:10.1002/er.1150
Azad, A. M., Kesavan, S., & Al-Batty, S. (2009). A closedloop proposal for hydrogen generation using steel waste
and a prototype solar concentrator. International Journal
of Energy Research, 33(5), 481498. doi:10.1002/er.1491
Badami, M., &Caldera, C. (2002). Dynamic model of a
load-following fuel cell vehicle: Impact of the air system.
SAE Technical Paper (SAE-2002-01-100), 1-10.
Baker, J. E. (1985). Adaptive selection methods for genetic algorithms. In Proc. 1st Int. Joint Conf. on Genetic
Algorithms, ICGA85, 101-111.
Bauen, A., & Hart, D. (2000). Opportunities for fuel
cell-based renewable energy supply in decentralised applications. Renewable Energy, 4, 25512554.
Bonnet, S., Alaphilippe, M., & Stouffs, P. (2005). Energy,
exergy and cost analysis of a micro-cogeneration system
based on an ericsson engine. International Journal of
Thermal Sciences, 44, 11611168. doi:10.1016/j.ijthermalsci.2005.09.005
Carlos, A., & Hernandez, A. (2005). Fuel consumption
minimization of a microgrid. IEEE Transactions on
Industry Applications, 41(3), 673681. doi:10.1109/
TIA.2005.847277
Chan, S. H., Ho, H. K., & Tian, Y. (2003). Multi-level
modeling of SOFCgas turbine hybrid system. International Journal of Hydrogen Energy, 28(8), 889900.
doi:10.1016/S0360-3199(02)00160-X
Charvin, P., Stephane, A., Florent, L., & Gilles, F. (2008).
Analysis of solar chemical processes for hydrogen production from water splitting thermochemical cycles.
Energy Conversion and Management, 49, 15471556.
doi:10.1016/j.enconman.2007.12.011
418
Clarke, R. E., Giddey, S., Ciacchi, F. T., Badwal, S. P.

S., Paul, B., & Andrews, J. (2009). Direct coupling of an
electrolyser to a solar PV system for generating hydrogen. International Journal of Hydrogen Energy, 34(6),
25312542. doi:10.1016/j.ijhydene.2009.01.053
Data of Japan Meteorological Agency. (2009). Recent
Meteorological Data. Retrieved May 29, 2013, from
http://database.rish.kyoto-u.ac.jp/arch/jmadata/gpvoriginal.html
De Ruyck, J. etal. (2007). Co-utilization of biomass and
natural gas in combined cycles through primary steam
reforming of the natural gas. Energy, 32(4), 371377.
doi:10.1016/j.energy.2006.07.010
Development of a Several 10kW Class Circular-Flat-Type
Low-Temperature Operation SOFC System. (2007).
Result report symposium 2007, New Energy and Industrial Technology Development Organization in Japan.
Retrieved from http://www.nedo.go.jp/informations/
events/200623/26_7.pdf In Japanese.
Development of home fuel cell cogeneration system in
Tokyo Gas. (2009). ALIA News, 89.
Dicks, L. A., Fellows, R. G., Mescal, M. C., & Seymour,
C. (2000). A study of SOFCPEM hybrid systems. Journal of Power Sources, 86(1-2), 501506. doi:10.1016/
S0378-7753(99)00492-9
Dormand, J. R., & Prince, P. J. (1980). A family of embedded Runge-Kutta formulae. Journal of Computational
and Applied Mathematics, 6, 1926. doi:10.1016/0771050X(80)90013-3
Eguchi, K., & Kurosawa, Y. (1988). Study on high load
swirl burners for Stirling engines. Reduction in NOx and
heat adsorption characteristics in exhaust gas recirculation. In Proc. of the 26th Japanese Symp. on Combustion,
(pp. 131-133).
Energy and Industrial Technology Development Organization in Japan. (1999). The researches-and-developments
trend of the low carbon fuel for vehicle. In Japanese.
Retrieved from http://www.nedo.go.jp/kankobutsu/foreigninfo/html9912/12119.html
Engin, K. S., & Hiyama, T. (2009). ANN based real-time
estimation of power generation of different PV module
types. Transaction of Institute of Electrical Engineers of
Japan, B. Power and Energy, 129(6), 783790.
Erdmann, G. (2003). Future economics of the fuel cell

housing market. International Journal of Hydrogen Energy, 28(7), 685694. doi:10.1016/S0360-3199(02)00281-1
Forsythe, G., Malcolm, M., & Moler, C. (1977). Computer
methods for mathematical computations. Upper Saddle
River, NJ: Prentice-Hall.
Fresno, F., Fernndez-Saavedra, R., Gmez-Mancebo,
B. M., Vidal, A., Snchez, M., & Rucandio, M. I. etal.
(2009). Solar hydrogen production by two-step thermochemical cycles: Evaluation of the activity of commercial
ferrites. International Journal of Hydrogen Energy, 34(7),
29182924. doi:10.1016/j.ijhydene.2009.02.020
Fuji Futures Co. Ltd. (2005). Commodity futures market
total information. Retrieved from http://www.fuji-ft.co.jp/
selection/toyu/index.htm
Fujiki K., Akagi S., Hirokawa T., & Yoshida K. (1997).
Optimal planning method of energy plant configurations
based on a genetic algorithm. Trans. Jpn. Soc. Mech. Eng.,
Series C, 64(617), 354-361. In Japanese
Gan, J., & Smith, C. T. (2006). A comparative analysis
of woody biomass and coal for electricity generation
under various CO2 emission reductions and taxes. Biomass and Bioenergy, 30(4), 296308. doi:10.1016/j.
biombioe.2005.07.006
Gencoglu, T. M., & Ural, Z. (2009). Design of a PEM
fuel cell system for residential application. International Journal of Hydrogen Energy, 34(12), 52425248.
doi:10.1016/j.ijhydene.2008.09.038
Goldberg, D. E. (1989). Genetic algorithms in search,
optimization and machine learning. Addison Wesley.
Gouda, T. etal. (2004). Microgrid. Japan: Denki Shimbun. In Japanese
Hamelin, J., Agbossou, K., Laperriere, A., Laurencelle,
F., & Bose, T. K. (2001). Dynamic behavior of a PEM
fuel cell stack for stationary applications. International
Journal of Hydrogen Energy, 26, 625629. doi:10.1016/
S0360-3199(00)00121-X
Haseli, Y., Dincer, I., & Naterer, G. F. (2008). Thermodynamic modeling of a gas turbine cycle combined
with a solid oxide fuel cell. International Journal of
Hydrogen Energy, 33(20), 58115822. doi:10.1016/j.
Hastie, T., Tibshireni, R., & Friedman, J. (2001). The
elements of statistical learning. Germany: Springer.
doi:10.1007/978-0-387-21606-5
Hatano, Y., et al. (2003). Investigation on the energy use
characteristics in the apartment house of hot water supply
and heating of an area. In Proceedings of the Meeting of
the Society of Heating, Air-Conditioning and Sanitary
Engineering, (pp. 1745-1748).
Hatziadoniu, C. J. (2002). A simplified dynamic model of
grid-connected fuel-cell generators. IEEE Transactions on
Power Delivery, 17(2), 467473. doi:10.1109/61.997919
Hawkes, A. D., & Leach, M. A. (2007). Cost-effective
operating strategy for residential micro-combined heat
and power. Energy, 32(5), 711723. doi:10.1016/j.energy.2006.06.001
Haykin, S. (1998). Neural networks: A comprehensive
foundation (Prentice Hall International Ed.). Upper Saddle
River, NJ: Prentice-Hall.
HC-TECH Inc. (1997). HC12a and HC22a Properties
and Performance Tests Data sheets.
He, W. (1995). Modeling a reformer in fuel cell powergeneration systems. In Proc. ASME Heat Transfer and
Fluids Eng., 321, 317-325.
Hongmei, Y., Haipeng, F., Pingjing, Y., & Yi, Y. (2000).
A combined genetic algorithm/simulated annealing algorithm for large scale system energy integration. Computers
& Chemical Engineering, 24, 20232035. doi:10.1016/
S0098-1354(00)00601-3
Hsin-Sen, C., Fanghei, T., Yi-Yie, Y., Kan-Lin, H., & FaLin, C. (2008). The development of a small PEMFC combined heat and power system. Journal of Power Sources,
176(2), 499514. doi:10.1016/j.jpowsour.2007.08.072
419
Hussain, M. M., Baschuk, J. J., Dincer, X., & Li, I. (2005).

Thermodynamic analysis of a PEM fuel cell power system.
International Journal of Thermal Sciences, 44, 903911.
doi:10.1016/j.ijthermalsci.2005.02.009
Japan Meteorological Business Support Center. (2007).

Surface-weather-observation 1-minute data, sapporo
district meteorological observatory. Tokyo, Japan: Japan
Meteorological Business Support Center.
Ibaraki Prefecture Government Office of Education.

(2002). Modeling of hydrogen energy system (High school
active science project research report). Ibaraki, Japan:
Author. In Japanese
Japan meteorological business support center. (2009).

GPV/GSM (Grid point value / global spectral model)
Retrieved from http://www.jmbsc.or.jp/hp/online/fonline0a.html
Ibe, S., et al. (2002). Development of fuel processor

for residential fuel cell cogeneration system. In K. Abe
(Ed.), Proc. 21th Annual Meeting of Japan Society of
Energy and Resources, Osaka, (pp. 493-496). In Japanese
National Astronomical Observatory. (2003). Rika Nenpyo. Chronological Scientific Tables CD-ROM. Japan:
Maruzen Co., Ltd.
Japan meteorological business support center. (2009).

Online data service, GPV/GSM (Grid Point Value / GSM
(Global Spectral Model)). Retrieved October 22, 2012,
from http://www.jmbsc.or.jp/hp/online/f-online0a.html
Ibe, S., Shinke, N., Takami, S., Yasuda, Y., Asatsu, H.,
& Echigo, M. (2002). Development of fuel processor for
residential fuel cell cogeneration system. In K. Abe (Ed.),
Proc. 21th Annual Meeting of Japan Society of Energy and
Resources, Osaka, (pp. 493-496).In Japanese.
Inoue, U. (1996). Air-conditioning handbook, 14. Japan:
Maruzen. In Japanese
Institute of Electrical Engineers of Japan. (2002). Technology of a fuel cell (pp. 247292). Japan: Ohmsha. In
Japanese
Ishihara, A., Mitsushima, S., Kamiya, N., & Ota, K.
(2004). Exergy analysis of polymer electrolyte fuel cell
systems using methanol. Journal of Power Sources, 126,
3440. doi:10.1016/j.jpowsour.2003.08.029
Ismail, Y., Kemmoku, Y., Takikawa, H., & Sakakibara, T.
(2002). An operating method for fuel saving in a standalone wing/diesel/battery system. Journal of Japan Solar
Energy Society, 28(2), 3138.
Ito, K., Shibata, T., & Yokoyama, R. (2002). Optimal
operation of a cogeneration plant in combination with
electric heat pumps. Trans. ASME J. Energy Resource
Technol., 116, 5664. doi:10.1115/1.2906010
Japan Meteorological Agency. (2007). Homepage. Retrieved from http://www.data.jma.go.jp/obd/stats/etrn/
index.php
420
Japan Solar Energy Society. (1985). Solar Energy Utilization Handbook. Japan: Ohmsha, Lt.
Japanese Ibaraki Prefecture Government Office of
Education. (2002). Modeling of hydrogen energy system
(High school active science project research report).
Ibaraki, Japan: Author. In Japanese.
Jiayi, H., Chuanwen, J., & Rong, X. (2008). A review
on distributed energy resources and microgrid. Renewable & Sustainable Energy Reviews, 12(9), 24722483.
doi:10.1016/j.rser.2007.06.004
Jorgensen, K. (2008). Technologies for electric, hybrid
and hydrogen vehicles: Electricity from renewable energy sources in transport. Utilities Policy, 16(2), 7279.
doi:10.1016/j.jup.2007.11.005
Kahaner, D., Moler, C., & Nash, S. (1989). Numerical
methods and software. Upper Saddle River, NJ: PrenticeHall.
Katrasnik, T. etal. (2005). Analysis of the dynamic response improvement of a turbocharged diesel engine driven
alternating current generating set. Energy Conversion
and Management, 46(18-19), 28382855. doi:10.1016/j.
Kazim, A. M. (2005). Exergoeconomic analysis of a
PEM electrolyser at various operating temperatures and
pressures. International Journal of Energy Research, 29,
539548. doi:10.1002/er.1073
Kil, R. M., & Song, Y. (1997). Random Search based on
genetic operators. Lecture Notes in Artificial Intelligence.,
1285, 196205.
Koike, S., et al. (1999). Demonstration program of 1kWclass PEMFC system for residential use by the Japan gas
association. In Proc. 3rd Int. Fuel Cell Conference, (pp.
497-498).
Larsen, J. F., & Wallace, J. S. (1995). Comparison of emissions and efficiency of a turbocharged lean-burn natural
gas and hythane-fuelled engine. ASME ICE, 24, 3140.
Kopac, M., & Kokturk, L. (2005). Determination of

optimum speed of an internal combustion engine by
exergy analysis. Int. J. Exergy, 2(1), 4054. doi:10.1504/
IJEX.2005.006432
Lim, T., Song, R., Shin, D., Yang, J., Jung, H., Vinke,
I., & Yang, S. (2008). Operating characteristics of a 5
kW class anode-supported planar SOFC stack for a fuel
cell/gas turbine hybrid system. International Journal of
Hydrogen Energy, 33(3), 10761083.
Kosaka, K., Tani, T., & Yoshida, S., (2000). Thermal

analysis of solid polymer water electrolysis system, Trans.
Jpn. Soc. Mech. Eng., 66 (642), B, 547-554. In Japanese
Lindstrom, B., & Petterson, L. (2003). Development of

a methanol fuelled reformer for fuel cell applications.
Journal of Power Sources, 118, 7178.
Kourosh, S. (2004). Dynamic and transient analysis of

power distribution system with fuel cells. Part 1: fuel
cell dynamic model. IEEE Trans. on Conversion, 19(2),
423428. doi:10.1109/TEC.2004.827039
Mathiak, J., Heinzel, A., Roes, J., Kalk, T., Kraus, H., &
Brandt, H. (2004). Coupling of a 2.5 kW steam reformer
with a 1 kW PEM fuel cell. Journal of Power Sources,
131(1-2), 112119. doi:10.1016/j.jpowsour.2004.01.024
Kreider, J. F., & Wang, X. A. (1992). Improved Artificial

neural networks for commercial building energy use
prediction. Solar Engineering, 1, 361366.
Meteorological Agency. (2007). Meteorological observation in Japan. Retrieved from http://www.data.kishou.

go.jp/index.htm
Krist, K., Gleason, K., & Wright, J. (1999). SOFC-based

residential cogeneration systems. In U. Stimming et al.
(Eds.), Proc. 6th Int. Symp. On Solid Oxide Fuel Cells
VI, Pennington, NJ, (pp.107-115).
Mikkola, M. (2001). Experimental studies on polymer

electrolyte membrane fuel cell stacks (Masters thesis).
Helsinki University of Technology, Finland. 58-79.
Kurata, S. (1999). Measurement of a domestic energy

consumption pattern. National Institute of Advanced
Industrial Science and Technology, News, (12), 1-14.
In Japanese.
Kuzgunkaya, E. H., & Hepbasli, Arif. (2006). Exergetic
performance assessment of a ground-source heat pump
drying system. International Journal of Energy Research,
31, 760777. doi:10.1002/er.1268
Kyakuno, T., Hattori, K., Ito, K., & Onda, K. (2003). Prediction of production power for high-pressure hydrogen by
high-pressure water electrolysis. J. Institute of Electrical
Engineers of Japan, 124-B(4), 605.
Kyoto Denkiki Co., Ltd. (2001). A system connection
inverter catalog and an examination data sheet.
Lariviere, E., Blanchot-Courtois, V., & Givois. (1998).
Simulation of a fuel cell in cogeneration. Int. Gas Res.
Conf., 4, 892-899.
Mitchell, A. R., & Griffiths, D. F. (2001). Finite difference and related methods for differential equations. New
York, NY: Wiley.
Mohamad, J. E., & Mofid, G. B. (2005). Exergetic
analysis of gas turbine plants. Int. J. Exergy, 2(1), 3139.
doi:10.1504/IJEX.2005.006431
Mohammadi A., et al. (2005). Development of highly
efficient and clean engine system using natural-gas and
hydrogen mixture fuel obtained from onboard reforming
(NEDO report ID:03B71006c). In Japanese
Mondol, J. D., Yohanis, Y. G., & Norton, B. (2007). Comparison of measured and predicted long term performance
of a gird connected photovoltaic system. Energy Conversion and Management, 48(4), 10651080. doi:10.1016/j.
Morner, S. O., & Klein, S. A. (2001). Experimental evaluation of the dynamic behavior of an air-breathing fuel
cell stack. ASME. Journal of Solar Energy Engineering,
123(3), 225231. doi:10.1115/1.1385202
421
Motahar, S., & Alemrajabi, A. A. (2009). Exergy based

performance analysis of a solid oxide fuel cell and steam
injected gas turbine hybrid power system. International Journal of Hydrogen Energy, 34(5), 23962407.
doi:10.1016/j.ijhydene.2008.12.065
Murakami and Izumida Lab. (2007). Homepage. Retrieved
from http://ipr20.cs.ehime-u.ac.jp/column/neural/index.
html
Muselli, M., Notton, G., & Louche, A. (1999). Design of
hybrid-photovoltaic power generator, with optimization
of energy management. Solar Energy, 65(3), 143157.
doi:10.1016/S0038-092X(98)00139-X
Nagano, S. (2002). Plate-type methanol steam reformer
using new catalytic combustion for a fuel cell. In Proceedings of SAE Technical Paper Series (Automotive Eng. 10).
Nagano, K., Mochida, T., Shimakura, K., Murashita,
K., & Takeda, S. (2002). Development of thermalphotovoltaic hybrid exterior wallboards incorporating
PV cells in and their winter performances. Solar Energy
Materials and Solar Cells, 77, 265282. doi:10.1016/
S0927-0248(02)00348-3
Nagase, O., et al. (2003). Quantification of the energy consumed in a house. In Proc. 19th Energy System Economic
and Environment Conference, (pp 461-466). In Japanese.
Nagase, O., Ikaga, T., & Chikamoto, T. (2003). Study on
the effect of energy saving methods and global warming
prevention in Tokyo. In Proceedings of the 19th Energy
System- Economics - Environment Conference, (pp. 461466). i Mikkola, M. In Japanese.
Nagata, Y., Hori, M., Sonai, A., & Shimotori, S. (1999).
Development of PEFC systems and components at
Toshiba. In I. Uchida et al. (Eds.), Proc. 3rd Int. Fuel
Cell Conference, Nagoya, Japan, (pp. 69-72).
Naim, H. Afgan, & Maria G. C. (2004). Sustainability
assessment of hydrogen energy systems. International
doi:10.1016/j.ijhydene.2004.01.005
422
Nakamura, T., & Sei, M. (2002). Energy related technology. High-efficiency fuel processor for fuel cell system
(Technical Report 77 4-9). Japan: Matsushita Electric
Works, Ltd. In Japanese.
Narita, K. (1996). The research on unused energy of the
cold region city and utilization for the district heat and
cooling (Ph.D. thesis), Hokkaido University, Japan.
National Astronomical Observatory. (2003). Rika Nenpyo, chronological scientific tables CD-ROM. Japan:
Maruzen Co., Ltd.
NEDO Technical Information Data Base. (2008). Standard
meteorology and solar radiation data (METPV-3). Retrieved from http://www.nedo.go.jp/database/index.html
NEDO Technical information data base. (2010). Standard
Meteorology and Solar Radiation Data (METPV-3). Retrieved October 13, 2001, from http://www.nedo.go.jp/
database/index.html
NEDO Technical Information Database. (2013). Standard
meteorology and Solar radiation data (METPV-3). Retrieved from http://www.nedo.co.jp
Nogare, D. D., Baggio, P., Tomasi, C., Mutri, L., & Canu,
P. (2007). A thermodynamic analysis of natural gas reforming processes for fuel cell application. Chemical Engineering Science, 62(18-20), 54185424. doi:10.1016/j.
ces.2006.12.065
Noglik, A., Roeb, M., Christian, S., & Robert, P. P. (2009).
experimental study on sulfur trioxide decomposition in a
volumetric solar receiver-reactor. International Journal
of Energy Research, 33(9), 799812. doi:10.1002/er.1514
Obara, S., & Kudo, K. (2000). Study of small-scale cogeneration system for cold region houses. In Proceedings
of the 3-rd Int. Conf. on Cold Climate Heating, Ventilating
and Air-Conditioning, (pp. 353-358).
Obara, S., & Kudo, K. (2003). Operational optimization and scheduling of multi-kind small capacity energy
devices for cold region houses. In Proceedings of 9th
ITES, 1, 297-302.
Obara, S., & Kudo, K. (2005). Relation between control

variables of PEM fuel cell system and power generation
efficiency. Transaction of JSME, Part B, 72(714), 447454. In Japanese.
Obara, S., & Kudo, K. (2005). Study on improvement in
efficiency of partial load driving of installing fuel cell
network with water electrolysis operation. Transactions
of the JSME, Series B, 71(701), 237-244. In Japanese
Obara, S., & Kudo, K. (2005). Installation planning of
small-scale fuel cell cogeneration in consideration of load
response characteristics (Load response characteristics of
electric power output). Transactions of the JSME, Series
B, 71(706), 1678-1685. In Japanese.
Obara, S., Tanno, I., Kito, S., Hoshi, A., & Sasaki, S.
(2007). Design of Environmental Energy System Using
Stirling Engine. In Proceedings of the 5th International
Symposium on Environmentally Conscious Design &
Inverse Manufacturing (EcoDesign), B2-1-4S, 1-8,
Tokyo, Japan.
Obara, S. etal. (2005). Study on small-scale co-generation
system for domestic house considering partial load and
load fluctuation.[In Japanese]. Journal of the Japan
Society of Mechanical Engineers, 71(704), 11691176.
Obara, S. (2006). Operating schedule of a combined energy
network system with fuel cell. International Journal of Energy Research, 30(13), 10551073. doi:10.1002/er.1203
Obara, S. (2006). Effective improvement in generation
efficiency due to partition cooperation management of
a fuel cell microgrid. Journal of Thermal Science and
Obara, S. (2006). The exhaust heat use plan when connecting solar modules to a fuel cell energy network.
Transactions of the ASME. Journal of Energy Resources
Technology, 129(1), 1828. doi:10.1115/1.2424962
Obara, S. (2006). Load response characteristics of a fuel
cell microgrid with control of number of units. International Journal of Hydrogen Energy, 31(13), 18191830.
doi:10.1016/j.ijhydene.2006.05.004
Obara, S. (2006). Load response characteristics of the
PEM fuel cell for individual cold-region houses. Transactions of the ASME. Journal of Fuel Cell Science and
Technology, 5(1), 011005-111.
Obara, S. (2007). Dynamic operation plan of a combined

fuel cell cogeneration, solar module, and geo-thermal
heat pump system using genetic algorithm. International Journal of Energy Research, 31(13), 12751291.
doi:10.1002/er.1300
Obara, S. (2007). Improvement of power generation
efficiency of an independent microgrid composed
of distributed engine generators. Transactions of the
ASME. Journal of Energy Resources Technology, 129(3),
190199. doi:10.1115/1.2748812
Obara, S. (2007). Equipment arrangement planning of
a fuel cell energy network optimized for cost minimization. Renewable Energy, 32(3), 382406. doi:10.1016/j.
renene.2006.02.012
Obara, S. (2007). Energy cost of an independent microgrid
with control of power output sharing of a distributed engine
generator. Journal of Thermal Science and Technology,
2(1), 4253. doi:10.1299/jtst.2.67
Obara, S. (2008). Equipment plan of compound interconnection microgrid composed from diesel power plants
and solid polymer membrane-type fuel cell. International Journal of Hydrogen Energy, 33(1), 179188.
doi:10.1016/j.ijhydene.2007.09.004
Obara, S. (2010). Power generation efficiency of an SOFCPEFC combined system with time shift utilization of SOFC
exhaust heat. International Journal of Hydrogen Energy,
35(2), 757767. doi:10.1016/j.ijhydene.2009.11.032
Obara, S., & Abeer, G. E. (2009). compound microgrid
installation operation planning of PEFC and photovoltaics
with prediction of electricity production using GA and
numerical weather information. International Journal
of Hydrogen Energy, 34(19), 82138222. doi:10.1016/j.
Obara, S., Kito, S., Hoshi, A., & Sasaki, S. (2007). Dynamic
characteristics of PEFC / woody biomass engine hybrid
microgrid. Journal of Thermal Science and Technology,
2(1), 111122. doi:10.1299/jtst.2.111
Obara, S., Kito, S., Hoshi, A., & Sasaki, S. (2007). Energy
balance of Stirling engine cogeneration and installation
potentiality to cold region houses. Journal of Environmental Engineering, 2(4), 696707. doi:10.1299/jee.2.696
423

small-scale fuel cell cogeneration in consideration of load
response characteristics (load response characteristics of
electric power output).[In Japanese]. Transactions of the
Japan Society of Mechanical Engineers, Series B, 71(706),
16781685. doi:10.1299/kikaib.71.1678
Obara, S., & Kudo, K. (2005). Study of efficiency improvements in a fuel-cell-powered vehicle using water
electrolysis by recovering regeneration energy and avoiding partial load operation. Transactions of the ASME.
Journal of Fuel Cell Science and Technology, 2, 202207.
doi:10.1115/1.1928930
Obara, S., & Kudo, K. (2005). Study on improvement in
efficiency of partial load driving of installing fuel cell
network with water electrolysis operation. Transactions
of the Japan Society of Mechanical Engineers, Series B,
71(701), 237244. doi:10.1299/kikaib.71.237
Obara, S., & Kudo, K. (2005). Study on small-scale fuel
cell cogeneration system with methanol steam reforming
considering partial load and load fluctuation. Trans. ASME.
Journal of Energy Resources Technology, 2, 202207.
Obara, S., & Kudo, K. (2006). Output response characteristics of a small-scale fuel cell for a house.[In Japanese]. Transactions of the Japan Society of Mechanical
Engineers, Series B, 72(713), 162169. doi:10.1299/
kikaib.72.162
Obara, S., & Kudo, K. (2006). Route planning of heat
supply piping in a fuel cell energy network. Transactions
of the JSME. Journal of Environmental Engineering, 1(1),
1728. doi:10.1299/jee.1.17
small-scale fuel cell cogeneration considering transient
response characteristics (load response characteristics of
electric power output). Journal of Environmental Engineering, 2(3), 621631. doi:10.1299/jee.2.621
Obara, S., Kudo, K., & Kuroda, A. (2005). Study on smallscale fuel cell cogeneration system with methanol steam
reforming considering partial load and load fluctuation.
Transactions of the ASME. Journal of Energy Resources
Technology, 127, 265271. doi:10.1115/1.1926310
424
Obara, S., & Tanno, I. (2007). Development of distributed

energy system due to bio-ethanol PEM fuel cell with solar
reforming, Part 1Evaluation of basic performance.
Trans. of the Society of Heating. Air-Conditioning and
Sanitary Engineers of Japan, 123, 2332.
Obara, S., & Tanno, I. (2007). development of distributed
energy system due to bio-ethanol pem fuel cell with solar
reforming, Part 1Evaluation of basic performance.
Transactions of the Society of Heating. Air-Conditioning
and Sanitary Engineers of Japan, 123, 2332.
Obara, S., & Tanno, I. (2007). Operation prediction of
a bioethanol solar reforming system using a neural network. Journal of Thermal Science and Technology, 2(2),
256267. doi:10.1299/jtst.2.256
Obara, S., & Tanno, I. (2008). Development of distributed
energy system due to bio-ethanol pem fuel cell with solar
reforming, Part 2High-speed analysis of the operation
plan using a neural network. Transactions of the Society
of Heating[In Japanese]. Air-Conditioning and Sanitary
Engineers of Japan, 130, 3342.
Obara, S., & Tanno, I. (2008). Fuel reduction effect of
the solar cell and diesel engine hybrid system with a prediction algorithm of solar power generation. Journal of
Power and Energy Systems, 2(4), 11661177. doi:10.1299/
jpes.2.1166
Oda, K. etal. (1999). Small-scale reformer for fuel cell
application.[In Japanese]. Sanyo Technical Review, 31(2),
99106.
Oda, K., Sakamoto, S., Ueda, M., Fuji, A., & Ouki, T.
(1999). A small-scale reformer for fuel cell application.
[In Japanese]. Sanyo Technical Review, 31(2), 99106.
Okawa, D., Shikazono, N., & Kasagi, N. (2007). Basic
characteristics of gas turbine-solid oxide fuel cell hybrid
system with air reduction. Transactions of the JSME,
Series B, 73(730), 113-120.
Osaka Gas Co. Ltd. (2006). Osaka Gas home use gas
appliance synthesis information site. Retrieved from
http://www.g-life.jp/html/scene/cogeneration/ecowill/
point/point09.html
Ozaki, S., & Tuziki, I. (1990).Trial calculation of the

quantity of public electric power and city gas to be replaced
by a distributed energy system. In Proceedings of the
9th Energy-resources seminar, 9, 174-179. In Japanese.
Ozeki, H. (1991). Research on a combustor for Stirling
engines. 4. Effects of a combustor engine operation parameter on exhaust characteristics. In Proc. of the 29th
Japanese Symp. on Combustion, (pp. 208-210).
Ozturk, A., Senel, A., & Onbastoglu, S. U. (2005).
Thermodynamic optimization of combined cycles. International Journal of Energy Research, 29, 657670.
doi:10.1002/er.1098
Padull, J., Ault, G. W., & McDonald, J. R. (2000). An
approach to the dynamic modeling of fuel cell characteristics for distributed generation operation. IEEE Power
Eng. Soc. Meeting, 1, 134-138.
Peacock, A. D., & Newborough, M. (2007). Controlling micro-CHP systems to modulate electrical load
profiles. Energy, 32(7), 10931103. doi:10.1016/j.
energy.2006.07.018
Pengmei, L., Zhenhong, Y., Longlong, M., Wu, C., Chen,
Y., & Zhu, J. (2007). Hydrogen-rich gas production from
biomass air and oxygen/steam gasification in a downdraft gasifier. Renewable Energy, 32(13), 21732185.
doi:10.1016/j.renene.2006.11.010
Petrasch, J., & Steinfeld, A. (2007). dynamics of a solar
thermochemical reactor for steam-reforming of methane. Chemical Engineering Science, 62, 42144228.
doi:10.1016/j.ces.2007.04.036
Qian, Y., Zuo, C., Tan, J., & He, J. (2007). Structural
analysis of bio-oils from sub-and supercritical water
liquefaction of woody biomass. Energy, 32(3), 96202.
doi:10.1016/j.energy.2006.03.027
Reduction in NOx and heat adsorption characteristics in
exhaust gas recirculation. In Proc. of the 26th Japanese
Symp. on Combustion, (pp. 131-133).
Rhn, C., hman, M., Gref, R., & Wsterlund, I. (2007).
Influence of fuel ash composition on high temperature
aerosol formation in fixed bed combustion of woody
biomass pellets. Fuel, 86(1-2), 181193. doi:10.1016/j.
fuel.2006.07.001
Robert, H. (2004). Microgrid: A conceptual solution. In

Proceedings of the 35th Annual IEEE Power Electronics
Specialists Conference, 6, 4285-4290.
Robert, H., et al. (2004). Microgrid: A conceptual solution.
In Proceedings of the 35th Annual IEEE Power Electronics Specialists Conference, 6, 4285-4290.
Sanino, L. A. M., & Reischel, R. A. R. (2007). Modeling
and identification of solar energy water heating system incorporating nonlinearities. Solar Energy, 81(5), 570580.
doi:10.1016/j.solener.2007.01.012
SANYO nickel-metal hydride production information.
(2009). Retrieved from http://www.sanyo.co.jp
Schenk, N. J., Moll, H. C., Potting, J., & Benders, R. M.
J. (2007). Wind energy, electricity, and hydrogen in the
Netherlands. Energy, 32(10), 19601971. doi:10.1016/j.
energy.2007.02.002
Shukuya, M. (2006). The way of looking and way of
thinking of exergy. In order to develop familiar green
engineering.[In Japanese]. IEEJ Journal, 126(4), 196197.
doi:10.1541/ieejjournal.126.196
Silva, G. J., Datta, A., & Bhattacharyya, S. P. (2003). PID
controllers for time-delay systems. Birkhauser.
Society of Living Amenity. (2006). Family-size fuel cell
cogeneration systems development of Tokyo Gas. ALIA
News, 89. In Japanese.
Srinivas, M., & Patnaik, L. M. (1994). Genetic algorithms: A survey. IEEE Computer, 27(6), 1726.
doi:10.1109/2.294849
Sun and Ice BmbH, Germany. (n. d.). Retrieved from
http://www.sun-and-ice.de
Swain M. R., & Usuf M. J. (1993). The effects of hydrogen
addition on natural gas engine operation. SAE Tech Pap
Ser, SAE-932775.
Takeda, Y., et al. (2004). Development of fuel processor
for rapid start-up. In K. Kimura (Ed.), Proc. 20th Energy
System Economic and Environment Conference, Tokyo,
(pp. 343-344). In Japanese
425
Takeda, Y., et al. (2004). Development of fuel processor

for rapid start-up. In K. Kimura (Ed.), Proc. 20th Energy
System Economic and Environment Conference, Tokyo,
(pp. 343-344). In Japanese.
Xuan, J., Leung, K. H. N., Leung, Y. C. D., & Ni, M. (2009).

A review of biomass-derived fuel processors for fuel cell
systems. Renewable & Sustainable Energy Reviews, 13(67), 13011313. doi:10.1016/j.rser.2008.09.027
Takeda, Y., Iwasaki, Y., Imada, N., & Miyata, T. (2004).

Development of fuel processor for rapid start-up. In K.
Kimura (Ed.), Proc. 20th Energy System Economic and Environment Conference, Tokyo, (pp. 343-344). In Japanese
Yamamoto, S., Sumi, K., Nishikawa, E., & Hashimoto,

T. (2004). An operating method using prediction of
photovoltaic power for a photovoltaic-diesel hybrid
power generation system. Transaction of Institute of
Electrical Engineers of Japan, B. Power and Energy,
124(4), 521530.
Takeda, Y., Iwasaki, Y., Imada, N., & Miyata, T. (2004).

Development of fuel processor for rapid start-up. In
Proceedings of the 20th Energy System Economic and
Environment Conference (pp. 343-344). Tokyo, Japan:
Japan Society of Energy and Resources.
Takuma, Y., & Goda, T. (2005). Microgrid for urban
energy.[in Japanese]. Transaction of the Society of Heating, Air-Conditioning and Sanitary Eng. of Japan, 79(7),
573579.
Taniguchi, Y., & Ishida, M. (2005). Passive load following method for PEFC system with reformer and its
efficiency improvement evaluated by using the concept of
exergy. Electrical Engineering in Japan, 152(4), 1726.
doi:10.1002/eej.20148
Tanrioven, M. (2005). Reliability and cost-benefits of adding alternate power sources to an independent microgrid
community. Journal of Power Sources, 150, 136149.
doi:10.1016/j.jpowsour.2005.02.071
Turner, J., Sverdrup, G., Mann, K. M., Maness, P. C.,
Kroposki, B., & Ghirardi, M. etal. (2008). Renewable
hydrogen production. International Journal of Energy
Research, 32(5), 379407. doi:10.1002/er.1372
Ueno, T., Sano, F., Saeki, O., & Tsuji, K. (2006). Effectiveness of an energy-consumption information system on
energy savings in residential houses based on monitored
data. Applied Energy, 83(2), 166183. doi:10.1016/j.
apenergy.2005.02.002
Urasaki, K. (2005). Application of lattice oxygen in metal
oxides to catalysts for hydrogen production (Ph.D. thesis).
Waseda University, Tokyo, Japan.
Usami, Y., Fukusako, S., & Yamada, M. (2000). Heat
and mass transfer in a reforming catalyst bed (Quantitative evaluation of the controlling factor by experiment).
Transactions of the JSME, Series B, 67(659), 1801-1808.
426
Yamano, Y. (2002). Development of a load-levelling

technique.[In Japanese]. Denki, 629, 5661.
YANMAR Co. Ltd. (2006). Cogeneration package. Retrieved from http://www.yanmar.co.jp/ products/mgasc/
cogene04.htm
Yap D., et al. (2004). Effect of hydrogen addition on
natural gas HCCI combustion. SAE Tech Pap Ser, SAE2004-01-1972.
Yasuda, I. (2005). Development of hydrogen production
technology for fuel cell. Energy Synthesis Engineering,
28(2). In Japanese.
Yerramalla, S., & Feliachi, A. (2003). Modeling and simulation of the dynamic behavior of a polymer electrolyte
membrane fuel cell. Journal of Power Sources, 124(1),
104113. doi:10.1016/S0378-7753(03)00733-X
Yoshiba, F., Izaki, Y., & Watanabe, T. (2004). Wide Range
Load Controllable MCFC Cycle with Pressure Swing
Operation. Journal of Power Sources, 137, 196205.
doi:10.1016/j.jpowsour.2004.06.001
Yoshida, K. (1999). Exergy engineering, as actual as
the theory. Japan: Kyoritsu Shuppann Ltd.In Japanese
Yoshinaga, M., et al. (2003). Investigation on the energy
consumption of a house, and a residents consciousness.
In Proceedings of the Meeting of the Society of Heating,
Air-Conditioning and Sanitary Engineering, (pp. 17291732). In Japanese
Zhang, Y., Ouyang, M., Lu, Q., Luo, J., & Li, X. (2004).
A model predicting performance of proton exchange
membrane fuel cell stack thermal systems. Applied
Thermal Engineering, 24, 501513. doi:10.1016/j.applthermaleng.2003.10.013
427
About the Author
Shinya Obara is a Professor at the Department of Electrical and Electronic Engineering at Kitami
Institute of Hokaido University, Japan. He received B.S. in mechanical engineering from Nagaoka University of Technology, Japan in 1987, M.S. in mechanical system from Nagaoka University of Technology, in 1989 and a Ph.D. in mechanical science from Hokkaido University in 2000, while he was working in companies and academic organizations. He worked as a researcher in the Department of
Mechanical Science of Hokkaido University from 2000-2001. Dr. Shinya Obara joined Tomakomai
National College of Technology in 2001 after an eight-year period in industry (as engineer and assistant
manager for research, in two different companies, namely Takasago Thermal Engineering Co., Ltd. and
Aisin AW Co., Ltd. in Japan). He was associate professor of the Department of Mechanical Engineering
of Tomakomai National College between 2001-2007. Moreover, he was a professor of the Department
of Mechanical Engineering of Tomakomai National College in 2008 and Professor of the Department
of Electrical and Electronic Engineering at Kitami Institute of Technology from 2008 to the present. His
research has been involved with power and heat energy and operation optimization analyses of energy
compound systems and energy efficiency, microgrid technology, and energy network system with renewable energy sources.
428
Index
auxiliary machinery 18, 24, 56, 238, 241, 264-265,

303, 376
Diesel Engine 87-88, 90, 198-200, 204-206,

217-222, 283, 321-322, 335-338, 346, 348
diesel-engine power 198, 204-205, 215-217, 219,
233
dioxide emissions 88, 167-169, 171-175, 178, 180,
182-183, 185, 188, 191, 193
distributed power 3, 39, 72, 189, 199, 216, 238, 399
Distributed Power Supply 3, 72, 189, 238, 399
Dynamic Characteristics 38-40, 43, 46-47, 52-53,
88, 94, 110, 118, 136-138, 140, 143, 146,
150-151, 153, 155, 157, 159-160, 164, 167168, 199, 216, 237-239, 242, 245, 247-250,
283
Dynamic Operation Plan 283, 287, 295, 303, 306
B
Base Load 38-41, 43, 52, 73, 87-88, 90, 93, 97-99,
137, 140, 151, 155, 161-162, 164, 170, 175,
177, 179-182, 187, 193, 238, 241-242, 245246, 248, 353, 356, 358, 367, 369, 371, 373
Bioethanol 282-283, 300-303, 309, 316, 354,
362-364, 385-388, 399-400, 413
biomass engine 237-239, 248
C
carbon dioxide emissions 88, 167-169, 171-175,
178, 180, 182-183, 185, 188, 191, 193
city gas 3, 18, 20, 40-41, 45, 104, 118-123, 126,
128-130, 143, 156-157, 167-174, 180, 182-185,
187-188, 237-239, 251, 254, 259, 262
Combined System 2, 237-238, 250-251, 254-255,
257-258, 261-262, 283, 322, 352-354, 356-357,
359-360, 362, 364-371, 373-374, 376, 378-379,
381-382
combustion exhaust 237, 248
comparative study 352, 354
Compound Microgrid 167-168, 321-324, 326, 328,
332-333
compound system 1-2, 13, 282, 322, 353
control variables 38-39, 45, 53-54, 62, 64-65, 68,
143, 146
cooperation management 72-73, 76
E
electrolysis 1-3, 5, 10-11, 15, 18, 21, 24-25, 29, 33,
103-104, 118-128, 130-131, 300
Energy Conversion 353
energy cost 137, 152, 167-168, 198-200, 215, 238,
321, 335
energy-demand model 199-200
Energy Network 1-4, 15, 18, 26-27, 33, 40, 103-104
Energy Storage 3, 7, 10-11, 16-17, 104
Engine Generator 38-41, 43, 47, 49-50, 52, 198-202,
204-207, 209-211, 213-215, 217-219, 221-227,
229, 231-233, 237-238, 251, 254, 321-322,
335-338, 343-348
exchange membrane 2, 18, 21, 137, 155, 167-169,
238, 322, 353, 367
Index
Fuel Cell Cogeneration 3, 5, 10, 15, 17, 53-54,

103-104, 167-168, 263-264, 324
Operational Planning 2-3, 9-13, 15-17, 118-119,

123, 126, 128, 130
Operation Planning 1-2, 12, 284-285, 289, 293,
296-299, 301, 321-322, 324, 326, 329,
331-333, 354, 377-378
Optimal Control 53, 64
Optimal Design 167, 181, 216
Optimization 1-2, 13, 33, 118, 123, 181, 239, 283,
287, 289, 303, 306, 321-324, 327, 329,
333-334, 354
G
gas discharge 136-137, 140, 155, 180, 193, 397-398
gas engine 38, 40, 167-169, 171-172, 180, 183-184,
193, 283-284
generation capacity 38, 40-41, 43, 52, 73, 81-82, 84,
93, 96-99, 113-114, 140, 151, 154, 199-201,
205, 217, 219, 221, 223, 225, 233, 265
Genetic Algorithm 1-2, 6, 13, 16, 26-27, 31, 104,
118, 123, 126-127, 129, 199-200, 202-203,
282-283, 287, 289, 301, 316, 321-322, 333
greenhouse gas 103, 136-137, 140, 155, 180, 193,
237-239, 250, 321, 397-399
H
Heat Pump 1, 5-6, 10, 12-13, 15-18, 20, 22, 24-25,
27, 175
Heat Transfer 18, 20, 27, 56, 201, 217, 245, 367,
385-392, 399-400, 402-403, 406
Hydrogen Production 24, 87, 300-301, 356, 385387, 393-395, 399, 409
K
kerosene diesel-engine 198, 200
M
meteorological data 272, 301, 322-323, 333,
385-386, 413
N
neural network (NN) 282-283, 287-288, 300-301,
304-305, 314-316, 321-322, 335, 340-341, 386
Numerical Weather Information 321-322, 330-333,
354
P
partial load 3, 38, 40, 90, 103-104, 118-121, 131,
140, 151-152, 154-155, 167-169, 174, 181,
193, 198-199, 216, 219, 225, 233, 348, 353,
357, 371
Photovoltaics 321-323, 325, 328, 333-334, 352-353,
355
power generation efficiency 10, 39, 53-55, 61,
63-64, 68, 72-73, 76, 82, 87-88, 90, 93-94,
96-100, 103-104, 113, 120-121, 143, 157, 162,
167-168, 175, 178, 181-185, 187, 189,
198-201, 205, 207, 210-213, 216-217, 219,
221, 223-224, 229, 233, 241, 324-326,
352-354, 357-358, 362-367, 369-373, 376,
379-382
power generators 155, 198-199, 215, 217, 233
power supply 3, 39, 41, 49-51, 53, 72, 75, 86, 97-98,
105, 109, 136-138, 140, 153-154, 156, 161162, 167-168, 177, 180, 189, 199, 216, 238,
243, 247, 250, 298, 302, 313, 335, 363, 365,
368, 397, 399, 410-411
proportional action 38, 53, 56
proton-exchange 38-39, 53, 73, 87, 90, 99, 104, 119,
138, 152, 181, 193, 352-353
proton exchange 2, 18, 137, 155, 167-169, 238, 322,
353, 367
429
Index
reforming system 251, 265, 282-283, 300-302, 316,

385-387, 399-400, 413
Renewable Energy 2-3, 5, 13, 15, 17, 136-138, 141,
152, 154-155, 168, 216, 233, 282-284, 301,
335, 398
representative days 10-11, 15-16, 27, 30-31, 43, 50,
81-83, 96, 109, 113-114, 150, 187, 223-224,
272, 329, 332, 385-386, 393, 410-413
thermal storage 2-3, 5-7, 11-12, 15-17, 56, 60, 65,

75, 108, 120, 122, 128, 130-131, 254, 402
transient response 44, 53-54, 56-57, 59, 62, 64-65,
138, 141, 242, 385, 387, 393, 400, 407
S
shift utilization 352-353, 367, 373, 380-381
Simulation 65, 109, 143, 167, 238, 291, 296
Solar Radiation 282-283, 286-288, 292-294,
298-299, 302-303, 307, 310, 312, 315-316,
323, 325-326, 335, 338-341, 343, 356, 362,
385-386, 391, 393, 400, 403, 407
Solar Reforming 282, 300-303, 316, 385-388, 400
solid-oxide fuel 352-353
solid-oxide fuel cell 352-353
Stirling cycle 237, 250-251
storage tank 5, 11-12, 15, 17-18, 20, 22, 24-25, 56,
60, 75, 88, 106, 108, 120, 122, 128, 168, 170,
182, 188-189, 191, 193, 198, 200-201, 215,
217, 239, 251, 254, 258, 265, 302, 304, 309,
312-313, 316, 322, 325-326, 328, 336-337,
354, 370, 376, 379, 382, 402
sustainable energy 282-283
430
U
urban area 40, 72-73, 78, 80-82, 84, 86-87, 95-100,
103-105, 110, 112-114, 117, 140, 168-169,
175, 177, 180-181, 238, 300
W
weather data 282-284, 289, 291, 293, 296-300, 335,
393
Weather Prediction 386
Wind Power Generation 136-138, 140-141, 146,
148-155, 157, 160-162, 167
woody biomass 237-239, 241, 248, 250-251, 254,
257-259, 262

Id Methods

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Id Methods

Încărcat de

Drepturi de autor:

Formate disponibile

Optimum Design of

A volume in the Advances in

Published in the United States of America by

Library of Congress Cataloging-in-Publication Data

IGI Global is currently accepting manuscripts

Titles in this Series

For a list of additional titles in this series, please visit: www.igi-global.com

701 E. Chocolate Ave., Hershey, PA 17033

the individual residence of Sapporo, Japan. From

Operating Schedule of a Combined Energy Network System

optimal capacity of accumulation-of-electricity

cooperative operation of complex energy devices,

Operating Schedule of a Combined Energy Network System

are changing greatly. Since estimating facility

The Combined Energy System

The Combined Energy

Operating Schedule of a Combined Energy Network System

Figure 1. Combined energy network system

Operating Schedule of a Combined Energy Network System

Operation Method of Combined

AC converter (8), and demand is fulfilled; (b)

Operating Schedule of a Combined Energy Network System

Figure 2. Outline of the proposal system

device Di , heat output H D ,t , amount of electric

Operating Schedule of a Combined Energy Network System

Table 1. Energy device initial specifications

energy storage S E ,D ,t , amount of thermal storage

Sst ,D ,t , and device selection switch SW ,D ,t

using the gene model with 0 and 1 (Obara and

Although a number of chromosome models N dv'

Operating Schedule of a Combined Energy Network System

Figure 3. Chromosome model

it. In the calculation of crossover, two parent

Operating Schedule of a Combined Energy Network System

chromosome model group of the last generation

chromosome models of N dv higher ranks are

Operating Schedule of a Combined Energy Network System

of the last generation, decided beforehand, is

Characteristics of Energy Devices

by the least-squares method with an equation of

electric energy supplied to the water electrolysis

ated from sampling time tk to t is calculated

Operating Schedule of a Combined Energy Network System

report on the water electrolysis bath for hydrogen

4. Thermal Storage Tank

0 SSt ,t SSt ,max

T,t TSt ,t TSt ,max

The following equation is an expression of the

H St ,in ,t and H St ,out ,t show the input and output

(4) when it depends on open-air temperature T,t .

Objective Function and

Operating Schedule of a Combined Energy Network System

flow FD ,t of the device multiplied by unit fuel

and H System ,t are decided on the basis of energy

price C fuel . The operational costs of the whole

demand patterns. The left-hand side of Equation

C System ,day = C System ,t

side of Equation (7) shows the electric power

electric power consumption of the heat pump (

by the electric heater ( E H ,t ), respectively. The

left-hand side of Equation (8) shows thermal

storage tank ( H St ,t ), respectively. FS , FFS ,t ,

and FS shows the calorific power of methanol

= ESystem ,t + E EL,t + E HP ,t +E H ,t (7)